Applied Evidence

In 1982, more than 20% of women surveyed in a nationally representative sample had discontinued oral contraceptives (OCs) on their own or at the recommendation of their physician due to bleeding or spotting.¹ Sadly, the percentage today has not decreased much.

Understandable concern, embarrassment, and annoyance lead these women to abandon OCs.^1,2 What they often don’t know, though, is that breakthrough bleeding generally is greatest in the first 3 to 4 months after starting OCs,³ and it steadily declines and stabilizes by the end of the fourth cycle.⁴ Timely counsel could enable many of these women to cope with the bleeding and stick with an effective contraceptive method. Additional incentives are noncontraceptive benefits of OCs: improved menstrual regularity and decreased menstrual blood loss, dysmenorrhea, and risk of ovarian and endometrial cancer.

Women who discontinue OCs on their own switch to less effective methods of birth control or use no method.^1,2 Consequences may be unexpected pregnancies and increased abortion rates.⁵ With patients who are using OCs, it would be appropriate to ask periodically whether they are satisfied with OC use.

In this review we discuss the mechanisms and management of breakthrough bleeding in women taking OCs, and provide tips for counseling that may help decrease the risk of discontinuation due to menstrual abnormalities in the initial months of use.

Breakthrough bleeding in this review refers to either unplanned spotting or bleeding, regardless of requirement for protection—unless defined otherwise by a specific study under discussion.

4 factors contribute to breakthrough bleeding

Breakthrough bleeding may be due to any the following factors: 1) physiologic effects of OCs on the endometrium, 2) OC-related parameters, including dose, formulation, and regimen, 3) patient behavior, including compliance, using concomitant medications, and smoking, and 4) benign or malignant pathology.

OCs and the endometrium: Estrogen-progestin balance significant

Progestin and estrogen in combination OCs have profound effects on the endometrium, which, though not contributing to contraception, do lead to a predictable pattern of bleeding or such problems as breakthrough bleeding or lack of withdrawal bleed.

Normally, estrogen causes the endometrium to proliferate. Progesterone stabilizes the growing uterine lining. Since the introduction of OCs in 1960, the trend in formulation has been to use the least amount of hormone necessary to inhibit ovulation. Given that the progestin is primarily responsible for the contraceptive efficacy of OCs, the risk of pregnancy is not altered with decreases in the estrogen component. However, significantly lowering the estrogen in OCs may account for breakthrough bleeding. Unplanned bleeding, though, is not dependent solely on the estrogen component as variations in the progestin can contribute to breakthrough bleeding.⁷

Most OC users in the US take low-dose formulations, so designated because the estrogen component is <50 μg. This level of estrogen in combination with a progestin provides excellent contraceptive efficacy, but may be insufficient to sustain endometrial integrity in some women.⁸ Studies that have compared OCs containing 20 μg ethinyl estradiol (EE) with those containing 30 μg or 35 μg EE have not been very useful for judging breakthrough bleeding rates because the products often also vary in the phasing and type of progestin. Some studies show more breakthrough bleeding with 20 μg EE pills,^9-11 but others show equal or improved cycle control with the lower EE dose.

Estrogen-progestin balance is more important than absolute level of estrogen. Endrikat et al¹² conducted a study to compare two 20 μg EE pills containing different progestins, and to compare 2 levonorgestrel-based formulations with differing EE amounts. An OC of 20 μg EE/100 μg levonorgestrel was compared with a preparation of 20 μg EE/500 mg norethisterone. A 30 μg EE/150 mg levonorgestrel pill was used as a standard reference preparation.

Overall, the 30 μg EE preparation showed a lower cumulative incidence of breakthrough bleeding compared with the 20 μg EE/100 μg levonorgestrel and 20 μg EE/500 μg norethisterone pills over 13 cycles (1.0% vs 4.1% vs 11.7%, respectively). However, the 20 μg EE/500 μg norethisterone pills consistently had higher breakthrough bleeding rates than the 20 μg EE/100 μg levonorgestrel pill. This suggests that, although the higher EE component in the 30 μg pill was important when comparing 2 formulations with the same progestin, the difference in progestins of the two 20 μg EE pills was most likely responsible for the differing rates of breakthrough bleeding.

This study highlights the ability to achieve greater cycle control by titrating the EE component of an OC in a balanced ratio with the same progestin, but suggests that the absolute quantity of EE in a given pill may be less important than maintaining a balance between the 2 hormones or less important than the impact of different progestins on breakthrough bleeding rates.

The delicate balance between estrogen and progesterone supplementation required for contraception may also lead to progestin-induced decidualization and endometrial atrophy, which can result in asynchronous, erratic bleeding.^7,13 This has been primarily studied in long-acting progestin-only contraceptives such as implants. Alterations in angiogenic factors¹⁴ may play a role. Hysteroscopic studies have shown abnormalities in superficial endometrial blood vessels in terms of size, proliferation, and fragility in women using Norplant.^13,15,16 Abnormalities in endothelial cells and extracellular matrix proteins,¹⁷ tissue factor,¹⁸ and endometrial lymphoid cells¹⁹ may contribute to breakthrough bleeding in progestin-dominant environments.

Available OC formulations, doses, and regimens

More than 30 formulations of combination OCs are available in the US, with different doses and types of estrogen and progestin ( TABLE 1 ).²⁰ Approved OCs have been studied in clinical trials to assess contraceptive efficacy and cycle control; however, comparisons between studies regarding bleeding phenomena are impaired by inconsistent terminology.³

While some studies describe breakthrough bleeding and spotting according to their recognized definitions, others simply refer to intermenstrual bleeding or use spotting to refer to any unexpected bleeding. In addition, cycle control studies of OC users frequently do not account for the effects of missed pills, use of concomitant medications, or smoking. The percentage of women who experience breakthrough bleeding in a given cycle varies widely even in different trials of the same formulation.

Pay attention to progestin level. Conventional wisdom holds that OCs with the lowest doses of EE (≤20 μg) are associated with more breakthrough bleeding.¹¹ However, even moderately low doses of either EE or progestin can increase the incidence of breakthrough bleeding. For example, when 3 pills with the same estrogen and progestin (50 μg EE/100 μg norethin-drone; 35 μg EE/100 μg norethindrone; and 35 μg EE/50 μg norethindrone) were compared in 192 women over 8 cycles, the pill containing the lowest amount of norethindrone (35 μg EE/50 μg norethisterone) caused the highest rates of breakthrough bleeding (decreasing to approximately 50% by cycle 8 as compared with 35% in the 35 μg EE/100 μg norethindrone pill and 25% in the 50 μg EE/100 μg norethindrone pill).²¹

In addition, the number of intermenstrual bleeding days plateaued more slowly as the amount of both hormones in the OC formulations decreased. This underscores the importance of the relative proportions of estrogen and progestin contained in combination OCs and its impact on breakthrough bleeding.

Similarly, a large comprehensive study in 1991 women compared 7 different formulations of combination OCs containing different dose combinations of EE and norgestimate—20/250, 50/250, 35/125, 20/60, 50/60, 30/90, 25/125.²⁵ Total inter-menstrual bleeding was more frequent at lower doses of either estrogen or progestin. However, as long as a similar estrogenprogestin ratio was maintained, bleeding rates were considered acceptable (approximately 10% of days per cycle with bleeding). The authors also noted that in the low-dose range of OCs, small changes in the absolute amount of either EE or norgestimate might result in noticeable changes in bleeding.

Type of progestin may affect breakthrough bleeding. All combination OCs contain either EE or mestranol. However, a variety of progestins have come into use. The 2 most common contraceptive progestins are derived from 19-nortestosterone, and are classified as gonanes or estranes ( TABLE 2 ).²²

Estranes include norethindrone and its derivatives, norethindrone acetate and ethinyodiol diacetate. Gonanes include levonorgestrel, norgestrel, desogestrel, gestodene, and norgestimate.

Each progestin differs in half-life, estrogenic, progestogenic, and androgenic properties, and these variations may explain differing rates of breakthrough bleeding among formulations.⁴ As shown by Endrikat et al,¹² pills with the same quantity of EE but different progestins can have marked differences in breakthrough bleeding rates.

Although gonanes have greater progestational activity, no trial has determined which progestin has the best bleeding profile. A recent Cochrane review comparing different progestins did find that, compared with pills containing levonorgestrel, those containing gestodene may be associated with less intermenstrual bleeding.²³

Regardless of the progestin used or the quantity of EE, breakthrough bleeding generally decreases with each successive cycle. One study that compared 2 combination OCs composed of EE/norgestimate and EE/norgestrel demonstrated bleeding rates of 11.3% and 10.6% during the first 6 cycles, that decreased to 5.1% and 6.3% in Cycles 13 to 24, respectively.²⁴ Additionally, all women using OCs can experience some cycles without a withdrawal bleed—a menstrual abnormality that may be concerning to those who desire a menstrual period as confirmation that they are not pregnant.

Comparing regimens. OC regimens are available as biphasic, triphasic, extended-cycle, and continuous use. Women using extended-cycle contraceptives may experience more breakthrough bleeding than those using a standard 28-day pill. However, in a 3-month cycle, there are only 7 days of planned bleeding. This is in contrast to 28-day cycles during 3 months in which there are 21 days of planned bleeding.

Though women on extended-cycle regimens may initially experience more breakthrough bleeding than women using 28-day regimens, the total number of planned and unplanned bleeding days may still decrease. Women using a 3-month cycle OC (30 μg EE/150 μg levonorgestrel) experienced more unscheduled bleeding than women using a standard 28-day cycle OC of the same formulation and dose.²⁶ The number of bleeding days decreased with each cycle. Another study examined continuous OC use (20 μg EE/100 μg levonorgestrel) over a period of 1 year, and reported a decreasing number of bleeding days over time.²⁷ In the case of continuous use, all bleeding is unscheduled, and any bleeding is considered breakthrough bleeding.

Multiphasic OC regimens were developed with the intention of decreasing breakthrough bleeding by mimicking the rising and falling pattern of estrogen and progesterone in the normal menstrual cycle.²⁸ After the introduction of the biphasic pill, an increase in breakthrough bleeding was noted, which led to the development of the triphasic pill.²⁹ Though the multiphasic hypothesis is physiologically plausible, recent reviews of the literature have found the evidence for its efficacy too limited and methodologically flawed to draw any definitive conclusions about a decrease in breakthrough bleeding.^30,31

TABLE 1
Available OCs by formulation and regimen

* Corrections were made to this chart following its publication in print.

TABLE 2
Progestins used in oral contraceptive formulations

Patient behavior contributing to breakthrough bleeding

Skipping a pill is a common cause of breakthrough bleeding.⁵ Compliance with any OC regimen is crucial to achieving a regular and predictable bleeding pattern. Of 6676 women surveyed retrospectively, 19% reported missing 1 or more pills per cycle, and 10% reported missing 2 or more pills per cycle.³² Prospective studies have found even higher rates of inconsistent use.

When an electronic device was used to monitor pill ingestion, up to 81% of women were found to miss at least 1 pill per cycle and up to 51% missed 3 or more pills per cycle.³³

Other side effects also adversely affect adherence. For example, women experiencing nausea may skip pills, which leads to breakthrough bleeding and, ultimately, discontinuation.³⁴ Patients need to understand the impact of skipping pills. Women who report irregular bleeding are 1.6 to 1.7 times more likely as those not reporting this side effect to miss 2 or more pills per cycle.⁵ Even 1 missed pill can increase the risk of bleeding irregularities.³⁵

Failure to take the pill at the same time every day, as well as poor comprehension of pill-taking instructions, are other strong predictors of inconsistent use and breakthrough bleeding.³²

Taking some prescription and over-the-counter medications, as well as herbal supplements, may interfere with the activity of OCs to alter bleeding patterns and contraceptive efficacy.³⁶ Medications that induce the cytochrome P-450 system (CYP450) in the liver increase the metabolism of OCs. Anticonvulsants, the antituberculosis agent rifampin, and antifungals such as griseofulvin, can increase the clearance of steroid hormones and thus lead to breakthrough bleeding. The herbal supplement St. John’s Wort, commonly used for mild to moderate depression, is associated with CYP450 induction. It has been shown to increase the incidence of breakthrough bleeding and probably ovulation in women taking OCs.³⁷

Smoking is associated with such anti-estrogenic effects as early menopause, osteoporosis, and menstrual abnormalities.³⁸ These effects may be related to the induction of hepatic estrogen and progesterone metabolism by cigarette smoking.^39,40

Before receiving OCs, women are made aware of the relationship between smoking, OCs, and an increased risk of myocardial infarction, stroke, and venous thromboembolism.⁴¹ They should also understand that the anti-estrogenic effect of smoking may lower estrogen levels and lead to breakthrough bleeding, even in women who are reliable pill-takers.^42,43

Smoking appears to have a dose-response relationship with breakthrough bleeding. Increasing levels of smoking have been associated with an increased risk of spotting or bleeding in each cycle.⁴⁴ The difference in cycle control between smokers and nonsmokers appears to be more pronounced with each cycle. Smokers demonstrate a 30% elevation in the risk of bleeding irregularities compared with nonsmokers in the first cycle of use, which rises to an 86% increased risk by the sixth cycle.

Reports conflict regarding the relationship between smoking and contraceptive efficacy, suggesting that confounding factors like compliance may be more important than the antihormonal effect of cigarettes.⁴⁵ Nevertheless, women who smoke should be informed of this potential complicating factor to OC use and as yet another reason to encourage smoking cessation.

Pathologic causes of breakthrough bleeding

When a woman experiences difficult cycle control after the first 3 to 4 months of OC use, consider the possibility of benign and malignant growths, including endometrial polyps, submucous myomas, and cervical or endometrial cancer.⁴⁶ Additionally, contraceptive failure must always be a consideration and that what appears to be breakthrough bleeding may actually represent bleeding in early pregnancy.

Cervicitis is an important but largely unrecognized source of unplanned bleeding in women using OCs. Causative organisms include chlamydia, gonorrhea, and trichomonas.²² Intermenstrual bleeding in women previously well controlled on OCs is particularly suggestive of asymptomatic chlamydia cervicitis.

Krettek et al³⁷ found that 29.2% of women who had been taking OCs for more than 3 months and presented with intermenstrual spotting had a positive test for Chlamydia trachomatis. By comparison, chlamydia cervicitis was found in 10.7% of matched controls taking OCs without spotting who were screened for symptoms of vaginitis or high-risk sexual behavior, and in just 6.1% of women undergoing routine screening before the initiation of contraception.

3 aspects of managing breakthrough bleeding

Managing breakthrough bleeding involves effective pretreatment and ongoing counseling, reassurance, and timely and appropriate diagnostic testing (TABLE 3). In some instances, pill-switching or other forms of medical management may be helpful, but these options are largely unproven.

TABLE 3
What to review with patients who are starting a combination OC

Counseling reduces anxiety and improves satisfaction, adherence

In a recent survey, 649 Canadian women who were picking up prescriptions for OCs were asked to complete a questionnaire at the pharmacy while they waited.⁴⁸ Over one third (34.5%) reported they had not received counseling from their health-care provider about breakthrough bleeding. Furthermore, only 28.3% of women who were counseled, and 26.1% of women who were not counseled, gave the optimal response to breakthrough bleeding as defined in this study (“continue taking pill and not call my doctor”).

Lack of counseling can lead to poor method satisfaction and significant cost expenditures due to visits and phone calls by women experiencing unexpected bothersome side-effects.⁵ Compared with women who reported the highest satisfaction with the care they received from their provider, those reporting the lowest scores were 1.6 to 2.2 times as likely to be dissatisfied with the pill.

Inform women that breakthrough bleeding is common in the first 3 to 4 cycles of OC use, that bleeding irregularities tend to decline with each successive cycle, and that they should not discontinue pill use without discussing their concerns with you. Remind women to keep sanitary protection with them during the first few months. The impact of poor counseling was underscored in a study of women enrolled in clinical trials of OCs, contraceptive vaginal rings, and Depo-Provera. Women taking OCs were the least likely to have been warned of menstrual irregularities and thus tended to stop using that method more often than those using a ring or Depo-Provera.⁴⁹ Of women who discontinue OCs, 47% use a less effective method and 19% use no method at all.¹

Give specific instructions for specific regimens. Given the array of OC regimens available, make sure women know how to take them properly. This will help ensure contraceptive efficacy and cycle control. Women who do not understand pill-package instructions are up to 2.8 times more likely to miss pills, which increases the risk of breakthrough bleeding and impacts contraceptive efficacy.⁵ Among women who were counseled about the consequences of missed pills, 76% reported knowing what to do in response (“use another form of birth control that month”). Of women who received no such counseling, only 48% gave the appropriate response (P<.001).⁴⁸ To improve adherence, advise women to establish a routine for pill-taking: taking the pill at the same time each day or linking pill ingestion with another daily activity, such as tooth brushing. Women without an established routine were 3.6 times more likely to miss 2 or more pills per cycle than women with a routine.⁵

Reassurance regarding contraceptive efficacy

Reassure users who take their pills routinely that breakthrough bleeding and contraceptive efficacy are not linked.⁵⁰ Breakthrough bleeding is not a sign that OCs are not working.⁴ On the other hand, approximately 1 million unintended pregnancies in the United States each year are associated with misuse or discontinuation of OCs.⁵¹

When to consider diagnostic testing

For OC users who continue to experience breakthrough bleeding beyond 3 to 4 cycles, other potential causes must be ruled out using appropriate diagnostic tests. A pregnancy test, appropriate testing for cervical infection, a pregnancy test, pelvic ultrasound, Pap smear, or endometrial biopsy may be warranted depending on clinical circumstances.

Fall-back options

If breakthrough bleeding continues beyond 3 months, and other reasons including poor adherence and pathologic processes are excluded, you could consider providing estrogen or switching to a different pill, though no clinical trials support definitive recommendations.

Aside from changing from a multiphasic to a monophasic formulation, altering the progestin component is often a first step in trying to control breakthrough bleeding.⁴⁶ An OC with a gonane rather than an estrane progestin may be beneficial as this class of progestins may provide more consistent hormonal effects on the endometrium.

Choosing an OC with a higher quantity of EE may also help, particularly for women using 20 μg pills. When possible, the same progestin should be used.

You may want to start a trial of conjugated estrogen, 1.25 mg, or estradiol, 2 mg, administered for 7 days when bleeding occurs. This can be repeated if necessary; however, if breakthrough bleeding continues despite this treatment, consideration of a different pill or method should be undertaken.

CORRESPONDENCE
Patricia A. Lohr, MD, University of Pittsburgh, Magee-Womens Hospital, 300 Halket Street, Pittsburgh, PA 15213-3180. E-mail: [email protected]

In 1982, more than 20% of women surveyed in a nationally representative sample had discontinued oral contraceptives (OCs) on their own or at the recommendation of their physician due to bleeding or spotting.¹ Sadly, the percentage today has not decreased much.

Understandable concern, embarrassment, and annoyance lead these women to abandon OCs.^1,2 What they often don’t know, though, is that breakthrough bleeding generally is greatest in the first 3 to 4 months after starting OCs,³ and it steadily declines and stabilizes by the end of the fourth cycle.⁴ Timely counsel could enable many of these women to cope with the bleeding and stick with an effective contraceptive method. Additional incentives are noncontraceptive benefits of OCs: improved menstrual regularity and decreased menstrual blood loss, dysmenorrhea, and risk of ovarian and endometrial cancer.

Women who discontinue OCs on their own switch to less effective methods of birth control or use no method.^1,2 Consequences may be unexpected pregnancies and increased abortion rates.⁵ With patients who are using OCs, it would be appropriate to ask periodically whether they are satisfied with OC use.

In this review we discuss the mechanisms and management of breakthrough bleeding in women taking OCs, and provide tips for counseling that may help decrease the risk of discontinuation due to menstrual abnormalities in the initial months of use.

Breakthrough bleeding in this review refers to either unplanned spotting or bleeding, regardless of requirement for protection—unless defined otherwise by a specific study under discussion.

4 factors contribute to breakthrough bleeding

Breakthrough bleeding may be due to any the following factors: 1) physiologic effects of OCs on the endometrium, 2) OC-related parameters, including dose, formulation, and regimen, 3) patient behavior, including compliance, using concomitant medications, and smoking, and 4) benign or malignant pathology.

OCs and the endometrium: Estrogen-progestin balance significant

Progestin and estrogen in combination OCs have profound effects on the endometrium, which, though not contributing to contraception, do lead to a predictable pattern of bleeding or such problems as breakthrough bleeding or lack of withdrawal bleed.

Normally, estrogen causes the endometrium to proliferate. Progesterone stabilizes the growing uterine lining. Since the introduction of OCs in 1960, the trend in formulation has been to use the least amount of hormone necessary to inhibit ovulation. Given that the progestin is primarily responsible for the contraceptive efficacy of OCs, the risk of pregnancy is not altered with decreases in the estrogen component. However, significantly lowering the estrogen in OCs may account for breakthrough bleeding. Unplanned bleeding, though, is not dependent solely on the estrogen component as variations in the progestin can contribute to breakthrough bleeding.⁷

Most OC users in the US take low-dose formulations, so designated because the estrogen component is <50 μg. This level of estrogen in combination with a progestin provides excellent contraceptive efficacy, but may be insufficient to sustain endometrial integrity in some women.⁸ Studies that have compared OCs containing 20 μg ethinyl estradiol (EE) with those containing 30 μg or 35 μg EE have not been very useful for judging breakthrough bleeding rates because the products often also vary in the phasing and type of progestin. Some studies show more breakthrough bleeding with 20 μg EE pills,^9-11 but others show equal or improved cycle control with the lower EE dose.

Estrogen-progestin balance is more important than absolute level of estrogen. Endrikat et al¹² conducted a study to compare two 20 μg EE pills containing different progestins, and to compare 2 levonorgestrel-based formulations with differing EE amounts. An OC of 20 μg EE/100 μg levonorgestrel was compared with a preparation of 20 μg EE/500 mg norethisterone. A 30 μg EE/150 mg levonorgestrel pill was used as a standard reference preparation.

Overall, the 30 μg EE preparation showed a lower cumulative incidence of breakthrough bleeding compared with the 20 μg EE/100 μg levonorgestrel and 20 μg EE/500 μg norethisterone pills over 13 cycles (1.0% vs 4.1% vs 11.7%, respectively). However, the 20 μg EE/500 μg norethisterone pills consistently had higher breakthrough bleeding rates than the 20 μg EE/100 μg levonorgestrel pill. This suggests that, although the higher EE component in the 30 μg pill was important when comparing 2 formulations with the same progestin, the difference in progestins of the two 20 μg EE pills was most likely responsible for the differing rates of breakthrough bleeding.

This study highlights the ability to achieve greater cycle control by titrating the EE component of an OC in a balanced ratio with the same progestin, but suggests that the absolute quantity of EE in a given pill may be less important than maintaining a balance between the 2 hormones or less important than the impact of different progestins on breakthrough bleeding rates.

The delicate balance between estrogen and progesterone supplementation required for contraception may also lead to progestin-induced decidualization and endometrial atrophy, which can result in asynchronous, erratic bleeding.^7,13 This has been primarily studied in long-acting progestin-only contraceptives such as implants. Alterations in angiogenic factors¹⁴ may play a role. Hysteroscopic studies have shown abnormalities in superficial endometrial blood vessels in terms of size, proliferation, and fragility in women using Norplant.^13,15,16 Abnormalities in endothelial cells and extracellular matrix proteins,¹⁷ tissue factor,¹⁸ and endometrial lymphoid cells¹⁹ may contribute to breakthrough bleeding in progestin-dominant environments.

Available OC formulations, doses, and regimens

More than 30 formulations of combination OCs are available in the US, with different doses and types of estrogen and progestin ( TABLE 1 ).²⁰ Approved OCs have been studied in clinical trials to assess contraceptive efficacy and cycle control; however, comparisons between studies regarding bleeding phenomena are impaired by inconsistent terminology.³

While some studies describe breakthrough bleeding and spotting according to their recognized definitions, others simply refer to intermenstrual bleeding or use spotting to refer to any unexpected bleeding. In addition, cycle control studies of OC users frequently do not account for the effects of missed pills, use of concomitant medications, or smoking. The percentage of women who experience breakthrough bleeding in a given cycle varies widely even in different trials of the same formulation.

Pay attention to progestin level. Conventional wisdom holds that OCs with the lowest doses of EE (≤20 μg) are associated with more breakthrough bleeding.¹¹ However, even moderately low doses of either EE or progestin can increase the incidence of breakthrough bleeding. For example, when 3 pills with the same estrogen and progestin (50 μg EE/100 μg norethin-drone; 35 μg EE/100 μg norethindrone; and 35 μg EE/50 μg norethindrone) were compared in 192 women over 8 cycles, the pill containing the lowest amount of norethindrone (35 μg EE/50 μg norethisterone) caused the highest rates of breakthrough bleeding (decreasing to approximately 50% by cycle 8 as compared with 35% in the 35 μg EE/100 μg norethindrone pill and 25% in the 50 μg EE/100 μg norethindrone pill).²¹

In addition, the number of intermenstrual bleeding days plateaued more slowly as the amount of both hormones in the OC formulations decreased. This underscores the importance of the relative proportions of estrogen and progestin contained in combination OCs and its impact on breakthrough bleeding.

Similarly, a large comprehensive study in 1991 women compared 7 different formulations of combination OCs containing different dose combinations of EE and norgestimate—20/250, 50/250, 35/125, 20/60, 50/60, 30/90, 25/125.²⁵ Total inter-menstrual bleeding was more frequent at lower doses of either estrogen or progestin. However, as long as a similar estrogenprogestin ratio was maintained, bleeding rates were considered acceptable (approximately 10% of days per cycle with bleeding). The authors also noted that in the low-dose range of OCs, small changes in the absolute amount of either EE or norgestimate might result in noticeable changes in bleeding.

Type of progestin may affect breakthrough bleeding. All combination OCs contain either EE or mestranol. However, a variety of progestins have come into use. The 2 most common contraceptive progestins are derived from 19-nortestosterone, and are classified as gonanes or estranes ( TABLE 2 ).²²

Estranes include norethindrone and its derivatives, norethindrone acetate and ethinyodiol diacetate. Gonanes include levonorgestrel, norgestrel, desogestrel, gestodene, and norgestimate.

Each progestin differs in half-life, estrogenic, progestogenic, and androgenic properties, and these variations may explain differing rates of breakthrough bleeding among formulations.⁴ As shown by Endrikat et al,¹² pills with the same quantity of EE but different progestins can have marked differences in breakthrough bleeding rates.

Although gonanes have greater progestational activity, no trial has determined which progestin has the best bleeding profile. A recent Cochrane review comparing different progestins did find that, compared with pills containing levonorgestrel, those containing gestodene may be associated with less intermenstrual bleeding.²³

Regardless of the progestin used or the quantity of EE, breakthrough bleeding generally decreases with each successive cycle. One study that compared 2 combination OCs composed of EE/norgestimate and EE/norgestrel demonstrated bleeding rates of 11.3% and 10.6% during the first 6 cycles, that decreased to 5.1% and 6.3% in Cycles 13 to 24, respectively.²⁴ Additionally, all women using OCs can experience some cycles without a withdrawal bleed—a menstrual abnormality that may be concerning to those who desire a menstrual period as confirmation that they are not pregnant.

Comparing regimens. OC regimens are available as biphasic, triphasic, extended-cycle, and continuous use. Women using extended-cycle contraceptives may experience more breakthrough bleeding than those using a standard 28-day pill. However, in a 3-month cycle, there are only 7 days of planned bleeding. This is in contrast to 28-day cycles during 3 months in which there are 21 days of planned bleeding.

Though women on extended-cycle regimens may initially experience more breakthrough bleeding than women using 28-day regimens, the total number of planned and unplanned bleeding days may still decrease. Women using a 3-month cycle OC (30 μg EE/150 μg levonorgestrel) experienced more unscheduled bleeding than women using a standard 28-day cycle OC of the same formulation and dose.²⁶ The number of bleeding days decreased with each cycle. Another study examined continuous OC use (20 μg EE/100 μg levonorgestrel) over a period of 1 year, and reported a decreasing number of bleeding days over time.²⁷ In the case of continuous use, all bleeding is unscheduled, and any bleeding is considered breakthrough bleeding.

Multiphasic OC regimens were developed with the intention of decreasing breakthrough bleeding by mimicking the rising and falling pattern of estrogen and progesterone in the normal menstrual cycle.²⁸ After the introduction of the biphasic pill, an increase in breakthrough bleeding was noted, which led to the development of the triphasic pill.²⁹ Though the multiphasic hypothesis is physiologically plausible, recent reviews of the literature have found the evidence for its efficacy too limited and methodologically flawed to draw any definitive conclusions about a decrease in breakthrough bleeding.^30,31

TABLE 1
Available OCs by formulation and regimen

* Corrections were made to this chart following its publication in print.

TABLE 2
Progestins used in oral contraceptive formulations

Patient behavior contributing to breakthrough bleeding

Skipping a pill is a common cause of breakthrough bleeding.⁵ Compliance with any OC regimen is crucial to achieving a regular and predictable bleeding pattern. Of 6676 women surveyed retrospectively, 19% reported missing 1 or more pills per cycle, and 10% reported missing 2 or more pills per cycle.³² Prospective studies have found even higher rates of inconsistent use.

When an electronic device was used to monitor pill ingestion, up to 81% of women were found to miss at least 1 pill per cycle and up to 51% missed 3 or more pills per cycle.³³

Other side effects also adversely affect adherence. For example, women experiencing nausea may skip pills, which leads to breakthrough bleeding and, ultimately, discontinuation.³⁴ Patients need to understand the impact of skipping pills. Women who report irregular bleeding are 1.6 to 1.7 times more likely as those not reporting this side effect to miss 2 or more pills per cycle.⁵ Even 1 missed pill can increase the risk of bleeding irregularities.³⁵

Failure to take the pill at the same time every day, as well as poor comprehension of pill-taking instructions, are other strong predictors of inconsistent use and breakthrough bleeding.³²

Taking some prescription and over-the-counter medications, as well as herbal supplements, may interfere with the activity of OCs to alter bleeding patterns and contraceptive efficacy.³⁶ Medications that induce the cytochrome P-450 system (CYP450) in the liver increase the metabolism of OCs. Anticonvulsants, the antituberculosis agent rifampin, and antifungals such as griseofulvin, can increase the clearance of steroid hormones and thus lead to breakthrough bleeding. The herbal supplement St. John’s Wort, commonly used for mild to moderate depression, is associated with CYP450 induction. It has been shown to increase the incidence of breakthrough bleeding and probably ovulation in women taking OCs.³⁷

Smoking is associated with such anti-estrogenic effects as early menopause, osteoporosis, and menstrual abnormalities.³⁸ These effects may be related to the induction of hepatic estrogen and progesterone metabolism by cigarette smoking.^39,40

Before receiving OCs, women are made aware of the relationship between smoking, OCs, and an increased risk of myocardial infarction, stroke, and venous thromboembolism.⁴¹ They should also understand that the anti-estrogenic effect of smoking may lower estrogen levels and lead to breakthrough bleeding, even in women who are reliable pill-takers.^42,43

Smoking appears to have a dose-response relationship with breakthrough bleeding. Increasing levels of smoking have been associated with an increased risk of spotting or bleeding in each cycle.⁴⁴ The difference in cycle control between smokers and nonsmokers appears to be more pronounced with each cycle. Smokers demonstrate a 30% elevation in the risk of bleeding irregularities compared with nonsmokers in the first cycle of use, which rises to an 86% increased risk by the sixth cycle.

Reports conflict regarding the relationship between smoking and contraceptive efficacy, suggesting that confounding factors like compliance may be more important than the antihormonal effect of cigarettes.⁴⁵ Nevertheless, women who smoke should be informed of this potential complicating factor to OC use and as yet another reason to encourage smoking cessation.

Pathologic causes of breakthrough bleeding

When a woman experiences difficult cycle control after the first 3 to 4 months of OC use, consider the possibility of benign and malignant growths, including endometrial polyps, submucous myomas, and cervical or endometrial cancer.⁴⁶ Additionally, contraceptive failure must always be a consideration and that what appears to be breakthrough bleeding may actually represent bleeding in early pregnancy.

Cervicitis is an important but largely unrecognized source of unplanned bleeding in women using OCs. Causative organisms include chlamydia, gonorrhea, and trichomonas.²² Intermenstrual bleeding in women previously well controlled on OCs is particularly suggestive of asymptomatic chlamydia cervicitis.

Krettek et al³⁷ found that 29.2% of women who had been taking OCs for more than 3 months and presented with intermenstrual spotting had a positive test for Chlamydia trachomatis. By comparison, chlamydia cervicitis was found in 10.7% of matched controls taking OCs without spotting who were screened for symptoms of vaginitis or high-risk sexual behavior, and in just 6.1% of women undergoing routine screening before the initiation of contraception.

3 aspects of managing breakthrough bleeding

Managing breakthrough bleeding involves effective pretreatment and ongoing counseling, reassurance, and timely and appropriate diagnostic testing (TABLE 3). In some instances, pill-switching or other forms of medical management may be helpful, but these options are largely unproven.

TABLE 3
What to review with patients who are starting a combination OC

Counseling reduces anxiety and improves satisfaction, adherence

In a recent survey, 649 Canadian women who were picking up prescriptions for OCs were asked to complete a questionnaire at the pharmacy while they waited.⁴⁸ Over one third (34.5%) reported they had not received counseling from their health-care provider about breakthrough bleeding. Furthermore, only 28.3% of women who were counseled, and 26.1% of women who were not counseled, gave the optimal response to breakthrough bleeding as defined in this study (“continue taking pill and not call my doctor”).

Lack of counseling can lead to poor method satisfaction and significant cost expenditures due to visits and phone calls by women experiencing unexpected bothersome side-effects.⁵ Compared with women who reported the highest satisfaction with the care they received from their provider, those reporting the lowest scores were 1.6 to 2.2 times as likely to be dissatisfied with the pill.

Inform women that breakthrough bleeding is common in the first 3 to 4 cycles of OC use, that bleeding irregularities tend to decline with each successive cycle, and that they should not discontinue pill use without discussing their concerns with you. Remind women to keep sanitary protection with them during the first few months. The impact of poor counseling was underscored in a study of women enrolled in clinical trials of OCs, contraceptive vaginal rings, and Depo-Provera. Women taking OCs were the least likely to have been warned of menstrual irregularities and thus tended to stop using that method more often than those using a ring or Depo-Provera.⁴⁹ Of women who discontinue OCs, 47% use a less effective method and 19% use no method at all.¹

Give specific instructions for specific regimens. Given the array of OC regimens available, make sure women know how to take them properly. This will help ensure contraceptive efficacy and cycle control. Women who do not understand pill-package instructions are up to 2.8 times more likely to miss pills, which increases the risk of breakthrough bleeding and impacts contraceptive efficacy.⁵ Among women who were counseled about the consequences of missed pills, 76% reported knowing what to do in response (“use another form of birth control that month”). Of women who received no such counseling, only 48% gave the appropriate response (P<.001).⁴⁸ To improve adherence, advise women to establish a routine for pill-taking: taking the pill at the same time each day or linking pill ingestion with another daily activity, such as tooth brushing. Women without an established routine were 3.6 times more likely to miss 2 or more pills per cycle than women with a routine.⁵

Reassurance regarding contraceptive efficacy

Reassure users who take their pills routinely that breakthrough bleeding and contraceptive efficacy are not linked.⁵⁰ Breakthrough bleeding is not a sign that OCs are not working.⁴ On the other hand, approximately 1 million unintended pregnancies in the United States each year are associated with misuse or discontinuation of OCs.⁵¹

When to consider diagnostic testing

For OC users who continue to experience breakthrough bleeding beyond 3 to 4 cycles, other potential causes must be ruled out using appropriate diagnostic tests. A pregnancy test, appropriate testing for cervical infection, a pregnancy test, pelvic ultrasound, Pap smear, or endometrial biopsy may be warranted depending on clinical circumstances.

Fall-back options

If breakthrough bleeding continues beyond 3 months, and other reasons including poor adherence and pathologic processes are excluded, you could consider providing estrogen or switching to a different pill, though no clinical trials support definitive recommendations.

Aside from changing from a multiphasic to a monophasic formulation, altering the progestin component is often a first step in trying to control breakthrough bleeding.⁴⁶ An OC with a gonane rather than an estrane progestin may be beneficial as this class of progestins may provide more consistent hormonal effects on the endometrium.

Choosing an OC with a higher quantity of EE may also help, particularly for women using 20 μg pills. When possible, the same progestin should be used.

You may want to start a trial of conjugated estrogen, 1.25 mg, or estradiol, 2 mg, administered for 7 days when bleeding occurs. This can be repeated if necessary; however, if breakthrough bleeding continues despite this treatment, consideration of a different pill or method should be undertaken.

CORRESPONDENCE
Patricia A. Lohr, MD, University of Pittsburgh, Magee-Womens Hospital, 300 Halket Street, Pittsburgh, PA 15213-3180. E-mail: [email protected]

In 1982, more than 20% of women surveyed in a nationally representative sample had discontinued oral contraceptives (OCs) on their own or at the recommendation of their physician due to bleeding or spotting.¹ Sadly, the percentage today has not decreased much.

Understandable concern, embarrassment, and annoyance lead these women to abandon OCs.^1,2 What they often don’t know, though, is that breakthrough bleeding generally is greatest in the first 3 to 4 months after starting OCs,³ and it steadily declines and stabilizes by the end of the fourth cycle.⁴ Timely counsel could enable many of these women to cope with the bleeding and stick with an effective contraceptive method. Additional incentives are noncontraceptive benefits of OCs: improved menstrual regularity and decreased menstrual blood loss, dysmenorrhea, and risk of ovarian and endometrial cancer.

Women who discontinue OCs on their own switch to less effective methods of birth control or use no method.^1,2 Consequences may be unexpected pregnancies and increased abortion rates.⁵ With patients who are using OCs, it would be appropriate to ask periodically whether they are satisfied with OC use.

In this review we discuss the mechanisms and management of breakthrough bleeding in women taking OCs, and provide tips for counseling that may help decrease the risk of discontinuation due to menstrual abnormalities in the initial months of use.

Breakthrough bleeding in this review refers to either unplanned spotting or bleeding, regardless of requirement for protection—unless defined otherwise by a specific study under discussion.

4 factors contribute to breakthrough bleeding

Breakthrough bleeding may be due to any the following factors: 1) physiologic effects of OCs on the endometrium, 2) OC-related parameters, including dose, formulation, and regimen, 3) patient behavior, including compliance, using concomitant medications, and smoking, and 4) benign or malignant pathology.

OCs and the endometrium: Estrogen-progestin balance significant

Progestin and estrogen in combination OCs have profound effects on the endometrium, which, though not contributing to contraception, do lead to a predictable pattern of bleeding or such problems as breakthrough bleeding or lack of withdrawal bleed.

Normally, estrogen causes the endometrium to proliferate. Progesterone stabilizes the growing uterine lining. Since the introduction of OCs in 1960, the trend in formulation has been to use the least amount of hormone necessary to inhibit ovulation. Given that the progestin is primarily responsible for the contraceptive efficacy of OCs, the risk of pregnancy is not altered with decreases in the estrogen component. However, significantly lowering the estrogen in OCs may account for breakthrough bleeding. Unplanned bleeding, though, is not dependent solely on the estrogen component as variations in the progestin can contribute to breakthrough bleeding.⁷

Most OC users in the US take low-dose formulations, so designated because the estrogen component is <50 μg. This level of estrogen in combination with a progestin provides excellent contraceptive efficacy, but may be insufficient to sustain endometrial integrity in some women.⁸ Studies that have compared OCs containing 20 μg ethinyl estradiol (EE) with those containing 30 μg or 35 μg EE have not been very useful for judging breakthrough bleeding rates because the products often also vary in the phasing and type of progestin. Some studies show more breakthrough bleeding with 20 μg EE pills,^9-11 but others show equal or improved cycle control with the lower EE dose.

Estrogen-progestin balance is more important than absolute level of estrogen. Endrikat et al¹² conducted a study to compare two 20 μg EE pills containing different progestins, and to compare 2 levonorgestrel-based formulations with differing EE amounts. An OC of 20 μg EE/100 μg levonorgestrel was compared with a preparation of 20 μg EE/500 mg norethisterone. A 30 μg EE/150 mg levonorgestrel pill was used as a standard reference preparation.

Overall, the 30 μg EE preparation showed a lower cumulative incidence of breakthrough bleeding compared with the 20 μg EE/100 μg levonorgestrel and 20 μg EE/500 μg norethisterone pills over 13 cycles (1.0% vs 4.1% vs 11.7%, respectively). However, the 20 μg EE/500 μg norethisterone pills consistently had higher breakthrough bleeding rates than the 20 μg EE/100 μg levonorgestrel pill. This suggests that, although the higher EE component in the 30 μg pill was important when comparing 2 formulations with the same progestin, the difference in progestins of the two 20 μg EE pills was most likely responsible for the differing rates of breakthrough bleeding.

This study highlights the ability to achieve greater cycle control by titrating the EE component of an OC in a balanced ratio with the same progestin, but suggests that the absolute quantity of EE in a given pill may be less important than maintaining a balance between the 2 hormones or less important than the impact of different progestins on breakthrough bleeding rates.

The delicate balance between estrogen and progesterone supplementation required for contraception may also lead to progestin-induced decidualization and endometrial atrophy, which can result in asynchronous, erratic bleeding.^7,13 This has been primarily studied in long-acting progestin-only contraceptives such as implants. Alterations in angiogenic factors¹⁴ may play a role. Hysteroscopic studies have shown abnormalities in superficial endometrial blood vessels in terms of size, proliferation, and fragility in women using Norplant.^13,15,16 Abnormalities in endothelial cells and extracellular matrix proteins,¹⁷ tissue factor,¹⁸ and endometrial lymphoid cells¹⁹ may contribute to breakthrough bleeding in progestin-dominant environments.

Available OC formulations, doses, and regimens

More than 30 formulations of combination OCs are available in the US, with different doses and types of estrogen and progestin ( TABLE 1 ).²⁰ Approved OCs have been studied in clinical trials to assess contraceptive efficacy and cycle control; however, comparisons between studies regarding bleeding phenomena are impaired by inconsistent terminology.³

While some studies describe breakthrough bleeding and spotting according to their recognized definitions, others simply refer to intermenstrual bleeding or use spotting to refer to any unexpected bleeding. In addition, cycle control studies of OC users frequently do not account for the effects of missed pills, use of concomitant medications, or smoking. The percentage of women who experience breakthrough bleeding in a given cycle varies widely even in different trials of the same formulation.

Pay attention to progestin level. Conventional wisdom holds that OCs with the lowest doses of EE (≤20 μg) are associated with more breakthrough bleeding.¹¹ However, even moderately low doses of either EE or progestin can increase the incidence of breakthrough bleeding. For example, when 3 pills with the same estrogen and progestin (50 μg EE/100 μg norethin-drone; 35 μg EE/100 μg norethindrone; and 35 μg EE/50 μg norethindrone) were compared in 192 women over 8 cycles, the pill containing the lowest amount of norethindrone (35 μg EE/50 μg norethisterone) caused the highest rates of breakthrough bleeding (decreasing to approximately 50% by cycle 8 as compared with 35% in the 35 μg EE/100 μg norethindrone pill and 25% in the 50 μg EE/100 μg norethindrone pill).²¹

In addition, the number of intermenstrual bleeding days plateaued more slowly as the amount of both hormones in the OC formulations decreased. This underscores the importance of the relative proportions of estrogen and progestin contained in combination OCs and its impact on breakthrough bleeding.

Similarly, a large comprehensive study in 1991 women compared 7 different formulations of combination OCs containing different dose combinations of EE and norgestimate—20/250, 50/250, 35/125, 20/60, 50/60, 30/90, 25/125.²⁵ Total inter-menstrual bleeding was more frequent at lower doses of either estrogen or progestin. However, as long as a similar estrogenprogestin ratio was maintained, bleeding rates were considered acceptable (approximately 10% of days per cycle with bleeding). The authors also noted that in the low-dose range of OCs, small changes in the absolute amount of either EE or norgestimate might result in noticeable changes in bleeding.

Type of progestin may affect breakthrough bleeding. All combination OCs contain either EE or mestranol. However, a variety of progestins have come into use. The 2 most common contraceptive progestins are derived from 19-nortestosterone, and are classified as gonanes or estranes ( TABLE 2 ).²²

Estranes include norethindrone and its derivatives, norethindrone acetate and ethinyodiol diacetate. Gonanes include levonorgestrel, norgestrel, desogestrel, gestodene, and norgestimate.

Each progestin differs in half-life, estrogenic, progestogenic, and androgenic properties, and these variations may explain differing rates of breakthrough bleeding among formulations.⁴ As shown by Endrikat et al,¹² pills with the same quantity of EE but different progestins can have marked differences in breakthrough bleeding rates.

Although gonanes have greater progestational activity, no trial has determined which progestin has the best bleeding profile. A recent Cochrane review comparing different progestins did find that, compared with pills containing levonorgestrel, those containing gestodene may be associated with less intermenstrual bleeding.²³

Regardless of the progestin used or the quantity of EE, breakthrough bleeding generally decreases with each successive cycle. One study that compared 2 combination OCs composed of EE/norgestimate and EE/norgestrel demonstrated bleeding rates of 11.3% and 10.6% during the first 6 cycles, that decreased to 5.1% and 6.3% in Cycles 13 to 24, respectively.²⁴ Additionally, all women using OCs can experience some cycles without a withdrawal bleed—a menstrual abnormality that may be concerning to those who desire a menstrual period as confirmation that they are not pregnant.

Comparing regimens. OC regimens are available as biphasic, triphasic, extended-cycle, and continuous use. Women using extended-cycle contraceptives may experience more breakthrough bleeding than those using a standard 28-day pill. However, in a 3-month cycle, there are only 7 days of planned bleeding. This is in contrast to 28-day cycles during 3 months in which there are 21 days of planned bleeding.

Though women on extended-cycle regimens may initially experience more breakthrough bleeding than women using 28-day regimens, the total number of planned and unplanned bleeding days may still decrease. Women using a 3-month cycle OC (30 μg EE/150 μg levonorgestrel) experienced more unscheduled bleeding than women using a standard 28-day cycle OC of the same formulation and dose.²⁶ The number of bleeding days decreased with each cycle. Another study examined continuous OC use (20 μg EE/100 μg levonorgestrel) over a period of 1 year, and reported a decreasing number of bleeding days over time.²⁷ In the case of continuous use, all bleeding is unscheduled, and any bleeding is considered breakthrough bleeding.

Multiphasic OC regimens were developed with the intention of decreasing breakthrough bleeding by mimicking the rising and falling pattern of estrogen and progesterone in the normal menstrual cycle.²⁸ After the introduction of the biphasic pill, an increase in breakthrough bleeding was noted, which led to the development of the triphasic pill.²⁹ Though the multiphasic hypothesis is physiologically plausible, recent reviews of the literature have found the evidence for its efficacy too limited and methodologically flawed to draw any definitive conclusions about a decrease in breakthrough bleeding.^30,31

TABLE 1
Available OCs by formulation and regimen

* Corrections were made to this chart following its publication in print.

TABLE 2
Progestins used in oral contraceptive formulations

Patient behavior contributing to breakthrough bleeding

Skipping a pill is a common cause of breakthrough bleeding.⁵ Compliance with any OC regimen is crucial to achieving a regular and predictable bleeding pattern. Of 6676 women surveyed retrospectively, 19% reported missing 1 or more pills per cycle, and 10% reported missing 2 or more pills per cycle.³² Prospective studies have found even higher rates of inconsistent use.

When an electronic device was used to monitor pill ingestion, up to 81% of women were found to miss at least 1 pill per cycle and up to 51% missed 3 or more pills per cycle.³³

Other side effects also adversely affect adherence. For example, women experiencing nausea may skip pills, which leads to breakthrough bleeding and, ultimately, discontinuation.³⁴ Patients need to understand the impact of skipping pills. Women who report irregular bleeding are 1.6 to 1.7 times more likely as those not reporting this side effect to miss 2 or more pills per cycle.⁵ Even 1 missed pill can increase the risk of bleeding irregularities.³⁵

Failure to take the pill at the same time every day, as well as poor comprehension of pill-taking instructions, are other strong predictors of inconsistent use and breakthrough bleeding.³²

Taking some prescription and over-the-counter medications, as well as herbal supplements, may interfere with the activity of OCs to alter bleeding patterns and contraceptive efficacy.³⁶ Medications that induce the cytochrome P-450 system (CYP450) in the liver increase the metabolism of OCs. Anticonvulsants, the antituberculosis agent rifampin, and antifungals such as griseofulvin, can increase the clearance of steroid hormones and thus lead to breakthrough bleeding. The herbal supplement St. John’s Wort, commonly used for mild to moderate depression, is associated with CYP450 induction. It has been shown to increase the incidence of breakthrough bleeding and probably ovulation in women taking OCs.³⁷

Smoking is associated with such anti-estrogenic effects as early menopause, osteoporosis, and menstrual abnormalities.³⁸ These effects may be related to the induction of hepatic estrogen and progesterone metabolism by cigarette smoking.^39,40

Before receiving OCs, women are made aware of the relationship between smoking, OCs, and an increased risk of myocardial infarction, stroke, and venous thromboembolism.⁴¹ They should also understand that the anti-estrogenic effect of smoking may lower estrogen levels and lead to breakthrough bleeding, even in women who are reliable pill-takers.^42,43

Smoking appears to have a dose-response relationship with breakthrough bleeding. Increasing levels of smoking have been associated with an increased risk of spotting or bleeding in each cycle.⁴⁴ The difference in cycle control between smokers and nonsmokers appears to be more pronounced with each cycle. Smokers demonstrate a 30% elevation in the risk of bleeding irregularities compared with nonsmokers in the first cycle of use, which rises to an 86% increased risk by the sixth cycle.

Reports conflict regarding the relationship between smoking and contraceptive efficacy, suggesting that confounding factors like compliance may be more important than the antihormonal effect of cigarettes.⁴⁵ Nevertheless, women who smoke should be informed of this potential complicating factor to OC use and as yet another reason to encourage smoking cessation.

Pathologic causes of breakthrough bleeding

When a woman experiences difficult cycle control after the first 3 to 4 months of OC use, consider the possibility of benign and malignant growths, including endometrial polyps, submucous myomas, and cervical or endometrial cancer.⁴⁶ Additionally, contraceptive failure must always be a consideration and that what appears to be breakthrough bleeding may actually represent bleeding in early pregnancy.

Cervicitis is an important but largely unrecognized source of unplanned bleeding in women using OCs. Causative organisms include chlamydia, gonorrhea, and trichomonas.²² Intermenstrual bleeding in women previously well controlled on OCs is particularly suggestive of asymptomatic chlamydia cervicitis.

Krettek et al³⁷ found that 29.2% of women who had been taking OCs for more than 3 months and presented with intermenstrual spotting had a positive test for Chlamydia trachomatis. By comparison, chlamydia cervicitis was found in 10.7% of matched controls taking OCs without spotting who were screened for symptoms of vaginitis or high-risk sexual behavior, and in just 6.1% of women undergoing routine screening before the initiation of contraception.

3 aspects of managing breakthrough bleeding

Managing breakthrough bleeding involves effective pretreatment and ongoing counseling, reassurance, and timely and appropriate diagnostic testing (TABLE 3). In some instances, pill-switching or other forms of medical management may be helpful, but these options are largely unproven.

TABLE 3
What to review with patients who are starting a combination OC

Counseling reduces anxiety and improves satisfaction, adherence

In a recent survey, 649 Canadian women who were picking up prescriptions for OCs were asked to complete a questionnaire at the pharmacy while they waited.⁴⁸ Over one third (34.5%) reported they had not received counseling from their health-care provider about breakthrough bleeding. Furthermore, only 28.3% of women who were counseled, and 26.1% of women who were not counseled, gave the optimal response to breakthrough bleeding as defined in this study (“continue taking pill and not call my doctor”).

Lack of counseling can lead to poor method satisfaction and significant cost expenditures due to visits and phone calls by women experiencing unexpected bothersome side-effects.⁵ Compared with women who reported the highest satisfaction with the care they received from their provider, those reporting the lowest scores were 1.6 to 2.2 times as likely to be dissatisfied with the pill.

Inform women that breakthrough bleeding is common in the first 3 to 4 cycles of OC use, that bleeding irregularities tend to decline with each successive cycle, and that they should not discontinue pill use without discussing their concerns with you. Remind women to keep sanitary protection with them during the first few months. The impact of poor counseling was underscored in a study of women enrolled in clinical trials of OCs, contraceptive vaginal rings, and Depo-Provera. Women taking OCs were the least likely to have been warned of menstrual irregularities and thus tended to stop using that method more often than those using a ring or Depo-Provera.⁴⁹ Of women who discontinue OCs, 47% use a less effective method and 19% use no method at all.¹

Give specific instructions for specific regimens. Given the array of OC regimens available, make sure women know how to take them properly. This will help ensure contraceptive efficacy and cycle control. Women who do not understand pill-package instructions are up to 2.8 times more likely to miss pills, which increases the risk of breakthrough bleeding and impacts contraceptive efficacy.⁵ Among women who were counseled about the consequences of missed pills, 76% reported knowing what to do in response (“use another form of birth control that month”). Of women who received no such counseling, only 48% gave the appropriate response (P<.001).⁴⁸ To improve adherence, advise women to establish a routine for pill-taking: taking the pill at the same time each day or linking pill ingestion with another daily activity, such as tooth brushing. Women without an established routine were 3.6 times more likely to miss 2 or more pills per cycle than women with a routine.⁵

Reassurance regarding contraceptive efficacy

Reassure users who take their pills routinely that breakthrough bleeding and contraceptive efficacy are not linked.⁵⁰ Breakthrough bleeding is not a sign that OCs are not working.⁴ On the other hand, approximately 1 million unintended pregnancies in the United States each year are associated with misuse or discontinuation of OCs.⁵¹

When to consider diagnostic testing

For OC users who continue to experience breakthrough bleeding beyond 3 to 4 cycles, other potential causes must be ruled out using appropriate diagnostic tests. A pregnancy test, appropriate testing for cervical infection, a pregnancy test, pelvic ultrasound, Pap smear, or endometrial biopsy may be warranted depending on clinical circumstances.

Fall-back options

If breakthrough bleeding continues beyond 3 months, and other reasons including poor adherence and pathologic processes are excluded, you could consider providing estrogen or switching to a different pill, though no clinical trials support definitive recommendations.

Aside from changing from a multiphasic to a monophasic formulation, altering the progestin component is often a first step in trying to control breakthrough bleeding.⁴⁶ An OC with a gonane rather than an estrane progestin may be beneficial as this class of progestins may provide more consistent hormonal effects on the endometrium.

Choosing an OC with a higher quantity of EE may also help, particularly for women using 20 μg pills. When possible, the same progestin should be used.

You may want to start a trial of conjugated estrogen, 1.25 mg, or estradiol, 2 mg, administered for 7 days when bleeding occurs. This can be repeated if necessary; however, if breakthrough bleeding continues despite this treatment, consideration of a different pill or method should be undertaken.

CORRESPONDENCE
Patricia A. Lohr, MD, University of Pittsburgh, Magee-Womens Hospital, 300 Halket Street, Pittsburgh, PA 15213-3180. E-mail: [email protected]

Polyneuropathy has an estimated prevalence of about 2% in the general population.^1,2 Despite being common, polyneuropathy remains a diagnostic challenge for most clinicians for many reasons, including the large number of potential causes and the fact that a specific cause often cannot be identified even after appropriate testing. These factors can contribute to uncertainty about the direction and level of aggressiveness of the evaluation. The result is often a “one size fits all” strategy—from an unnecessarily expensive “shotgun” approach to a “defeatist” attitude that too quickly deems a neuropathy as idiopathic.

A number of experts have developed algorithms for the evaluation of neuropathy.^1,3-8 We have incorporated and modified many of these suggestions to develop a simple, user-friendly approach; one that characterizes the neuropathy using “what?”, “where?”, “when?” and “what setting?” parameters (FIGURE).

The focus of this article is on the polyneuropathy that presents with widespread sensory or sensorimotor symptoms and signs. We will not discuss the evaluation of other peripheral neuropathic processes—eg, mononeuropathies, plexopathies, or motor neuronopathies such as amyotrophic lateral sclerosis. We emphasize the fundamental step of determining whether the polyneuropathy is a distal, symmetrical, sensory or sensorimotor polyneuropathy, in which case metabolic/toxic, inherited and idiopathic causes are more probable. In contrast, alternative presentations suggest immune-mediated and infectious causes.

What: Sensory, motor, or autonomic?

The first step is to assign the patient’s complaints and your examination findings to specific nerve fiber types of the peripheral nervous system: sensory, motor, or autonomic.⁵

FIGURE
Characterizing your patient’s neuropathy

Sensory findings help narrow your search

Once you have ruled out a nonneuropathic process (eg, arthritis) as the cause of sensory symptoms and signs, turn your attention to distinguishing between peripheral or central nervous system dysfunction. When the pathology resides in the central nervous system (eg, multiple sclerosis) accompanying symptoms and signs usually assist in localization. If a patient acknowledges clinical features suggesting past or present involvement of cerebellar, urinary, or visual (eg, optic neuritis) systems, for example, magnetic resonance imaging rather than electrodiagnostic testing may be warranted.

In a patient with peripheral nervous system dysfunction, sensory abnormalities help exclude neuromuscular diseases not associated with sensory dysfunction, such as myopathies, neuromuscular transmission disorders, or disease of the motor neuron.

Positive neuropathic sensory symptoms suggest acquired polyneuropathies. Patients may describe “prickling,” “tingling,” “swelling,” “asleep-like numbness,” or a sensation of “bunched-up socks.” Patients with acquired polyneuropathies usually complain of positive neuropathic sensory symptoms (PNSS), whereas patients with inherited polyneuropathies only rarely do (FIGURE).

Pain suggests toxic, metabolic, or ischemic causes. Patients may describe “electric shocks,” “burning,” “freezing,” “tightness,” or “throbbing.” They may complain of discomfort or pain to sensory stimuli that under normal circumstances would not be painful (allodynia)—eg, discomfort evoked by a bed sheet resting on the feet. They may also describe exaggerated pain to stimuli that would normally evoke low levels of discomfort or pain (hyperalgesia).

A painful neuropathy narrows the differential diagnosis to diseases that affect smaller nerve fibers, which generally convey pain and temperature input. Causes may be toxic, metabolic, ischemic, or idiopathic. For example, a pure “small-fiber” polyneuropathy commonly occurs in patients aged >60 years and typically causes painful feet. It is often idiopathic, but diabetes or impaired glucose tolerance and alcohol toxicity should be explored.⁹ Patients with painful neuropathy usually also exhibit reduced or absent sensation of pinprick and temperature in the distribution of their sensory complaints.

Negative neuropathic sensory symptoms. In addition to positive neuropathic symptoms, patients may complain of “negative” symptoms such as loss of sensation and imbalance. Examination usually reveals abnormalities of proprioception and sensation to vibration with reduced or absent deep tendon reflexes and ataxia. These features can occur in acquired or inherited causes of neuropathy.

Important sensory tests. Test sensation on the toes and fingertips, and more proximally (eg, ankle and shin) if any abnormality is found at these distal sites. Test vibration with a 128 Hz tuning fork, pinprick with disposable safety pin, and light touch with a cotton swab.

You may test temperature sensation by warming or cooling the handle or prong of the tuning fork and applying it to the patient’s skin.

Joint position testing is performed by asking the patient to avert his eyes, then moving the distal phalanx of a finger or toe up or down by small increments and asking the patient to tell you the direction of movement. Assess a patient’s casual and tandem gait for unsteadiness or ataxia.

Motor symptoms: Weigh them against sensory findings

Most patients with neuropathy have some degree of weakness, but it is usually overshadowed by sensory complaints. Distal lower extremity weakness may manifest as “foot drop,” which, if it affects ankle dorsiflexion, may cause a “slapping” or noisy step due to the forefoot hitting the ground with abnormal force.

Distal upper extremity weakness may cause trouble with fine motor skills of the hands.

Proximal weakness may present as difficulty in rising from a chair or lifting objects above the shoulders.

There may be muscle atrophy or fasciculations.

Motor symptoms are seldom the sole complaint. When motor and sensory symptoms are combined, it is helpful to rank them in order of symptom predominance—ie, motor greater than sensory, or vice versa. For instance, many immune-mediated disorders, such as Guillain-Barré syndrome and chronic inflammatory demyelinating polyradiculoneuropathy (CIDP), produce chiefly motor abnormalities and fewer sensory symptoms. Sensory complaints predominate in many other polyneuropathies, especially the “length-dependent” polyneuropathies (ie, those affecting the longer nerves initially) caused by metabolic or toxic disorders.

Autonomic symptom evaluation

The number of processes that affect both autonomic and somatic nerves are relatively few (TABLE 1).^5,7,10 It is particularly important to assess symptoms suggesting involvement of the autonomic nervous system.

Autonomic symptoms include lightheadedness, syncope, diarrhea, constipation, postprandial bloating, early satiety, urinary complaints, erectile dysfunction, abnormal or absent sweating, and dry mouth and eyes. Many of these complaints are common in the general population, so their relevance should be based on severity and temporal evolution, as well as comorbidities and medication use.

When complaints do not clearly implicate pathology in the autonomic nervous system, autonomic testing may be helpful, targeted for the domain that may be impaired. For example, bedside orthostatics or tilt-table testing are used for pre-syncopal symptoms, but gastric emptying testing can assist the evaluation of complaints of early satiety or postprandial bloating.

TABLE 1
Relatively common acquired polyneuropathies with autonomic nervous system involvement

Where: The distribution of nerve involvement

“Where” refers to distribution of nerve involvement 1) globally throughout the body and 2) locally along the nerve(s). During the history taking and examination, determine the nature of general distribution (eg, symmetric or asymmetric) and where along the length of the nerve(s) (proximal and/or distal) the dysfunction exists.

Polyneuropathy most commonly presents in a “length-dependent” distribution, with clinical features appearing initially most distally and symmetrically (ie, in the feet). Asymmetry and involvement of the proximal parts of a nerve are “red flags” for an uncommon cause that may require referral to a neurologist (FIGURE).

Comparative vs absolute measurements. At bedside, 2 approaches are used to assess the distribution of nerve involvement: comparative and absolute. The comparative approach searches for a relative difference in sensory thresholds or weakness between sites. It can assess side-to-side or one nerve (or root or region) territory to another. It is useful for establishing sensory or motor impairment in a radicular, plexus, or single nerve distribution.

Testing for an absolute reduction in sensation (eg, decreased vibration in the toes) can be more challenging because it requires experience in judging what is normal and abnormal according to expectations for a particular site and modality. Take into account that sensory thresholds are normally increased with the patient’s age and height.¹¹ For example, we commonly encounter elderly patients whose vibration sensation in the toes is said to be decreased, when in fact the reduced sensory threshold is only an age-related change.

Most assessments of sensory thresholds use the absolute approach because most generalized polyneuropathies are “length-dependent.”

Perform motor testing for appendicular (upper and lower extremities) and axial (neck and trunk) muscles, assessing particularly for weakness, atrophy, and fasciculations.

The typical polyneuropathy caused by metabolic, toxic, inherited, or unknown causes is distal and symmetric.¹² Neuropathies caused by other mechanisms, such as immune-mediated or infectious, are rarely length-dependent. Examples include motor neuronopathies (eg, amyotrophic lateral sclerosis), sensory neuronopathies (eg, paraneoplastic), polyradiculoneuropathies (Guillain-Barré syndrome, CIDP), and mononeuritis multiplex (caused by vasculitis).

When: The time course of signs and symptoms

Knowing whether the onset of neuropathy was definite and abrupt or was gradual is the most helpful temporal clue to possible underlying causes. The time course (ie, tempo) following onset is also important.

An acute/subacute onset with a definite date often suggests an immune-mediated or an infectious process (FIGURE). With respect to immune-mediated neuropathies, consider primarily autoimmune conditions (eg, Guillain-Barré syndrome, vasculitic neuropathy) and also paraneoplastic autoimmune syndromes (eg, subacute sensory neuropathy). In the latter case, a cancer presents to the immune system an epitope that is similarly found in the nervous system, prompting an autoimmune attack of the nervous system. Both autoimmune and infectious processes almost always have a rapid start on a definite date.

With an insidious onset, the patient won’t recollect a definite date on which the neuropathy began. The underlying mechanism usually is an inherited, metabolic, or toxic process—or idiopathic, if a cause cannot be identified.

The disease course sheds additional light on the causative mechanism. Onset and subsequent progression often correlate in a predictable manner, owing in part to the underlying mechanism. For example, an acute onset neuropathy often is followed by rapid disease progression, especially when caused by an autoimmune process (Guillain-Barré and vasculitic neuropathy). On the other hand, a neuropathy of insidious onset usually follows a slow or even static course. But there are exceptions that may make the diagnosis challenging and illustrate the need for clinical follow-up.

What setting

The patient’s medical history, medications, social and family history, and a review of systems can uncover known risk factors for neuropathic processes.

Common causes of acquired polyneuropathies are diabetes mellitus, chronic renal disease, and alcohol dependence. If a patient with distal, symmetric sensory complaints also has any of these conditions, a causal relationship should be considered. Another example is the patient with a history of bariatric surgery, which could lead to neuropathy as a result of malnutrition, particularly vitamin B₁ and B₁₂ deficiencies.¹³

The clinical setting may indicate a need for further evaluation by a neurologist. For example, a family history of inherited neuropathy in a patient with high arches and curled “hammer” toes would strongly suggest Charcot-Marie-Tooth (CMT) disease, also known as hereditary motor and sensory neuropathy. A patient with a monoclonal gammopathy may have a paraproteinemic neuropathy that warrants further evaluation by a neuromuscular specialist. Although more rare, a paraneoplastic cause would warrant consideration in a smoker, especially if the neuropathy was subacute. In many cases, neuropathic symptoms are the first clue of a new medical condition (eg, impaired glucose tolerance).

Electrodiagnostic testing: What it can and can’t tell you

Most polyneuropathies warrant additional electrodiagnostic evaluation in an electromyography (EMG) laboratory. Electro-diagnostic testing comprises 2 procedures: nerve conduction studies and needle electrode examination.

Preparing your patient. When ordering this study, be sure to discuss it thoroughly with the patient. The entire study typically takes an hour or longer. And it can be painful, though in our experience nearly every patient tolerates the procedure. Most important for the patient to understand is that information gleaned from electrodiagnostic testing may be essential to the diagnosis, as explained below.

Many benefits of the study. First, electrodiagnostic testing can confirm a peripheral neuropathic basis for a patient’s complaints.

Second, electrodiagnostic testing helps characterize a neuropathy as primarily demyelinating, primarily axonal, or mixed demyelinating and axonal. An example of this benefit is that a primary demyelinating characterization greatly narrows the list of possible causes (eg, Guillain-Barré syndrome, chronic inflammatory demyelinating polyradiculoneuropathy, Charcot-Marie-Tooth disease type 1).

Third, electrodiagnostic testing assists in characterizing the neuropathic process as sensory, motor or sensorimotor.

Fourth, electrodiagnostic testing helps localize the neuropathic process (the “where”).

Fifth, electrodiagnostic testing can gauge the severity of the neuropathic process.

Limitations of the study. When the study returns normal results, keep in mind it has limited sensitivity. For example, nerve conduction studies are only able to assess the function of larger myelinated nerve fibers; a neuropathic process solely in small fibers will not be evident with this test. Likewise, the needle electrode examination is unable to assess small nerve fiber status. We include this caveat in our electrodiagnostic testing reports of patients who have symptoms suggestive of pure small fiber polyneuropathy.

When might blood studies be useful?

Laboratory testing of blood is often of great value, but only after a particular polyneuropathy has been characterized and placed into one or more potential etiologic subgroups.

For example, laboratory testing for a distal, symmetric sensory polyneuropathy (TABLE 2) should be much different than testing for another presentation (eg, mononeuritis multiplex). TABLE 3 details some of the laboratory tests we recommend for the more common polyneuropathies that don’t typically present in a distal, symmetric sensory fashion or that are accompanied by other distinctive features. In our experience, clinicians too frequently order unnecessary and expensive tests for disorders only rarely associated with neuropathy; the rare causes of neuropathy are intentionally not the subject of this review.

TABLE 2
Common blood tests for the evaluation of a distal, symmetrical neuropathy^1,3,5,6,8

TABLE 3
Other diagnostic tests for acquired neuropathies in specific clinical situations^2,4,5,7

Common causes of distal, symmetric polyneuropathies

Distal, symmetric polyneuropathies are usually due to metabolic/toxic, inherited, or idiopathic causes (FIGURE, TABLE 2). As such, it is often unnecessary to obtain diagnostic tests searching for active infectious, autoimmune or paraneoplastic etiologies.³ In electrodiagnostic terms, these neuropathies are almost always primarily axonal rather than demyelinating, usually involving both large and small nerve fibers.

Diabetes is the most common cause of neuropathy in developed Western countries, occurring in more than 50% of patients who require insulin.¹⁴ Diabetic neuropathy typically presents as a distal, symmetric neuropathy syndrome, though diabetic lumbosacral radiculoplexus neuropathy (diabetic amyotrophy) and other presentations may be seen less commonly.

Recent evidence suggests that neuropathy—particularly a sensory and often painful, distal, symmetric “small-fiber” neuropathy—sometimes occurs in patients with impaired fasting glucose (IFG) or impaired glucose tolerance (IGT).^15-19 The 2-hour oral glucose tolerance test has been deemed more sensitive for the early diabetic state of IGT/IFG associated with neuropathy.¹⁹ It remains to be seen whether this early diabetic state neuropathy will turn out to represent a large percentage of cases of what were previously thought of as idiopathic small-fiber neuropathy.

Alcoholism is another common cause of a distal, symmetric neuropathy that is predominantly sensory with a painful, burning sensation.^20-22 Alcohol abuse and dependence occurs in 10% to 20% of the primary care population.^23,24 The prevalence of neuropathy in alcoholics is uncertain, though 1 study of hospitalized patients admitting to daily alcohol intake of over 100 g for men or 80 g for women (10 oz of beer, 1 oz liquor and 3–4 oz wine each have 10 g of alcohol) for 2 years or more (mean of 238±120 g for a period of 22.7±10.2 years) demonstrated that one third of patients had electrophysiologic evidence of peripheral neuropathy and one fourth of autonomic neuropathy.²⁰ Most subjects in this cohort did not show any clinical or laboratory evidence of malnutrition.

Frequently alcoholic neuropathy coexists with thiamine-deficient neuropathy,²² warranting assessment of thiamine status in all alcoholics with neuropathy. Pain is a prominent complaint in alcoholic neuropathy but a less common complaint in thiamine-deficient neuropathy.²² Cobalamin (vitamin B₁₂)-deficient neuropathy is more likely to occur suddenly and to involve the hands or hands and feet simultaneously, and is less likely to be painful.²⁵ Myelopathy is frequent in cobalamin deficiency, sometimes serving as a clue to diagnosis. It may be difficult to determine whether sensory symptoms are caused by myelopathy or neuropathy. Electro-diagnostic testing, somatosensory evoked potentials, and radiological investigation may be helpful.

Uremic neuropathy may occur in patients with chronic renal failure on dialysis.^26,27 Today it is less common by virtue of the widespread implementation of dialysis and renal transplantation. Generally reserve this diagnosis for patients with end-stage renal failure with a creatinine clearance of less than 10 mL/min. Other systemic disorders associated with renal failure must be considered; these include diabetes mellitus, amyloidosis, and vasculitis. Also consider drug-induced neuropathy in this population.

Pharmaceutical agents, industrial and environmental agents, and substances of abuse may cause neuropathy, most commonly in the form of a length-dependent axonal neuropathy. Offending chemotherapeutic agents include colchicine, pyridoxine, and amiodarone. However, toxic polyneuropathies probably represent a rather small proportion of cases.

The list of potentially offending substances is long and includes many medications and agents that the general population is commonly exposed to in low doses. Consider the inherent risk of neuropathy of the particular agent in question. To establish a causal link, verify exposure, determine that symptoms are temporally related to the toxin, and rule out other causes of neuropathy. Furthermore, some clinical improvement or at least stabilization should occur following removal of the offending agent, though this may take months to years.²⁸

Medication-induced neuropathy is more common than industrial and environmental neuropathies. The neuropathy of heavy metal intoxication is rare, and is usually accompanied by a combination of gastrointestinal, hematologic, and central nervous system problems.

HIV neuropathy is predominantly sensory and length-dependent. It occurs in about one third of infected patients. This is an exception to the tenet that infectious neuropathies typically present in a non– length-dependent pattern. However, in almost all circumstances, the diagnosis of HIV is well established and so it usually doesn’t present a diagnostic challenge. It is not our routine practice to check for HIV in those presenting with a new-onset distal, symmetric neuropathy, though we will in the proper clinical setting. Whether HIV neuropathy should be included in the differential diagnosis depends on the incidence of HIV in your patient population and the patient’s specifics.⁶ Rarely, Guillain-Barré syndrome and CIDP can be the presentation of a recent HIV infection.

For inherited neuropathies, there are an overwhelming number of commercially available panels for DNA testing.^3,29-31 A detailed discussion of inherited neuropathies is beyond the scope of this review, and many excellent reviews have been written on the subject. We will, however, make a few general comments.

First, Charcot-Marie-Tooth (CMT) disease is the most common inherited neuromuscular disorder, and it is encountered in primary care.

Second, foot deformities (eg, high arches, curled toes) commonly accompany an inherited polyneuropathy and are valuable clues.

Third, the inheritence pattern should be sought through a detailed family history and examination of other family members (including query and examination for foot deformities). Inherited neuropathies can be autosomal dominant, autosomal recessive, or X-linked, based largely on what gene harbors the mutation. Furthermore, spontaneous “de novo” mutations are not uncommon, being responsible for probably 25% of cases of CMT type 1.

Fourth, inherited neuropathies should be characterized just like acquired neuropathies, with particular attention paid to whether the neuropathy is demyelinating (eg, CMT type 1) or axonal (CMT type 2).

With respect to DNA testing of blood, focused genetic testing is almost always possible once you further characterize the inherited neuropathy and also take into account prevalence estimates of the various inherited neuropathies (eg, CMT type 1A caused by a duplication of the PMP-22 gene is responsible for the majority of cases on CMT), obviating the need for expensive, comprehensive genetic testing.

Other presentations

The evaluation of neuropathies that are not distal and symmetric can be complex and in many cases may warrant referral to a neurologist for further evaluation. With such presentations, diagnostic considerations include chest imaging and antineuronal (paraneoplastic) antibody serology for smokers with subacute neuropathies, cerebrospinal fluid analysis for an acquired demyelinating neuropathy, infectious neuropathy or a polyradiculopathy (eg, Lyme disease),^3,7 and sensory nerve biopsy (eg, sural nerve) when vasculitis, amyloidosis, or sarcoidosis is suspected.

CORRESPONDENCE
Ted M. Burns, MD, University of Virginia Health System Department of Neurology, Box 800394, Charlottesville, VA 22908. E-mail: [email protected]

Polyneuropathy has an estimated prevalence of about 2% in the general population.^1,2 Despite being common, polyneuropathy remains a diagnostic challenge for most clinicians for many reasons, including the large number of potential causes and the fact that a specific cause often cannot be identified even after appropriate testing. These factors can contribute to uncertainty about the direction and level of aggressiveness of the evaluation. The result is often a “one size fits all” strategy—from an unnecessarily expensive “shotgun” approach to a “defeatist” attitude that too quickly deems a neuropathy as idiopathic.

A number of experts have developed algorithms for the evaluation of neuropathy.^1,3-8 We have incorporated and modified many of these suggestions to develop a simple, user-friendly approach; one that characterizes the neuropathy using “what?”, “where?”, “when?” and “what setting?” parameters (FIGURE).

The focus of this article is on the polyneuropathy that presents with widespread sensory or sensorimotor symptoms and signs. We will not discuss the evaluation of other peripheral neuropathic processes—eg, mononeuropathies, plexopathies, or motor neuronopathies such as amyotrophic lateral sclerosis. We emphasize the fundamental step of determining whether the polyneuropathy is a distal, symmetrical, sensory or sensorimotor polyneuropathy, in which case metabolic/toxic, inherited and idiopathic causes are more probable. In contrast, alternative presentations suggest immune-mediated and infectious causes.

What: Sensory, motor, or autonomic?

The first step is to assign the patient’s complaints and your examination findings to specific nerve fiber types of the peripheral nervous system: sensory, motor, or autonomic.⁵

FIGURE
Characterizing your patient’s neuropathy

Sensory findings help narrow your search

Once you have ruled out a nonneuropathic process (eg, arthritis) as the cause of sensory symptoms and signs, turn your attention to distinguishing between peripheral or central nervous system dysfunction. When the pathology resides in the central nervous system (eg, multiple sclerosis) accompanying symptoms and signs usually assist in localization. If a patient acknowledges clinical features suggesting past or present involvement of cerebellar, urinary, or visual (eg, optic neuritis) systems, for example, magnetic resonance imaging rather than electrodiagnostic testing may be warranted.

In a patient with peripheral nervous system dysfunction, sensory abnormalities help exclude neuromuscular diseases not associated with sensory dysfunction, such as myopathies, neuromuscular transmission disorders, or disease of the motor neuron.

Positive neuropathic sensory symptoms suggest acquired polyneuropathies. Patients may describe “prickling,” “tingling,” “swelling,” “asleep-like numbness,” or a sensation of “bunched-up socks.” Patients with acquired polyneuropathies usually complain of positive neuropathic sensory symptoms (PNSS), whereas patients with inherited polyneuropathies only rarely do (FIGURE).

Pain suggests toxic, metabolic, or ischemic causes. Patients may describe “electric shocks,” “burning,” “freezing,” “tightness,” or “throbbing.” They may complain of discomfort or pain to sensory stimuli that under normal circumstances would not be painful (allodynia)—eg, discomfort evoked by a bed sheet resting on the feet. They may also describe exaggerated pain to stimuli that would normally evoke low levels of discomfort or pain (hyperalgesia).

A painful neuropathy narrows the differential diagnosis to diseases that affect smaller nerve fibers, which generally convey pain and temperature input. Causes may be toxic, metabolic, ischemic, or idiopathic. For example, a pure “small-fiber” polyneuropathy commonly occurs in patients aged >60 years and typically causes painful feet. It is often idiopathic, but diabetes or impaired glucose tolerance and alcohol toxicity should be explored.⁹ Patients with painful neuropathy usually also exhibit reduced or absent sensation of pinprick and temperature in the distribution of their sensory complaints.

Negative neuropathic sensory symptoms. In addition to positive neuropathic symptoms, patients may complain of “negative” symptoms such as loss of sensation and imbalance. Examination usually reveals abnormalities of proprioception and sensation to vibration with reduced or absent deep tendon reflexes and ataxia. These features can occur in acquired or inherited causes of neuropathy.

Important sensory tests. Test sensation on the toes and fingertips, and more proximally (eg, ankle and shin) if any abnormality is found at these distal sites. Test vibration with a 128 Hz tuning fork, pinprick with disposable safety pin, and light touch with a cotton swab.

You may test temperature sensation by warming or cooling the handle or prong of the tuning fork and applying it to the patient’s skin.

Joint position testing is performed by asking the patient to avert his eyes, then moving the distal phalanx of a finger or toe up or down by small increments and asking the patient to tell you the direction of movement. Assess a patient’s casual and tandem gait for unsteadiness or ataxia.

Motor symptoms: Weigh them against sensory findings

Most patients with neuropathy have some degree of weakness, but it is usually overshadowed by sensory complaints. Distal lower extremity weakness may manifest as “foot drop,” which, if it affects ankle dorsiflexion, may cause a “slapping” or noisy step due to the forefoot hitting the ground with abnormal force.

Distal upper extremity weakness may cause trouble with fine motor skills of the hands.

Proximal weakness may present as difficulty in rising from a chair or lifting objects above the shoulders.

There may be muscle atrophy or fasciculations.

Motor symptoms are seldom the sole complaint. When motor and sensory symptoms are combined, it is helpful to rank them in order of symptom predominance—ie, motor greater than sensory, or vice versa. For instance, many immune-mediated disorders, such as Guillain-Barré syndrome and chronic inflammatory demyelinating polyradiculoneuropathy (CIDP), produce chiefly motor abnormalities and fewer sensory symptoms. Sensory complaints predominate in many other polyneuropathies, especially the “length-dependent” polyneuropathies (ie, those affecting the longer nerves initially) caused by metabolic or toxic disorders.

Autonomic symptom evaluation

The number of processes that affect both autonomic and somatic nerves are relatively few (TABLE 1).^5,7,10 It is particularly important to assess symptoms suggesting involvement of the autonomic nervous system.

Autonomic symptoms include lightheadedness, syncope, diarrhea, constipation, postprandial bloating, early satiety, urinary complaints, erectile dysfunction, abnormal or absent sweating, and dry mouth and eyes. Many of these complaints are common in the general population, so their relevance should be based on severity and temporal evolution, as well as comorbidities and medication use.

When complaints do not clearly implicate pathology in the autonomic nervous system, autonomic testing may be helpful, targeted for the domain that may be impaired. For example, bedside orthostatics or tilt-table testing are used for pre-syncopal symptoms, but gastric emptying testing can assist the evaluation of complaints of early satiety or postprandial bloating.

TABLE 1
Relatively common acquired polyneuropathies with autonomic nervous system involvement

Where: The distribution of nerve involvement

“Where” refers to distribution of nerve involvement 1) globally throughout the body and 2) locally along the nerve(s). During the history taking and examination, determine the nature of general distribution (eg, symmetric or asymmetric) and where along the length of the nerve(s) (proximal and/or distal) the dysfunction exists.

Polyneuropathy most commonly presents in a “length-dependent” distribution, with clinical features appearing initially most distally and symmetrically (ie, in the feet). Asymmetry and involvement of the proximal parts of a nerve are “red flags” for an uncommon cause that may require referral to a neurologist (FIGURE).

Comparative vs absolute measurements. At bedside, 2 approaches are used to assess the distribution of nerve involvement: comparative and absolute. The comparative approach searches for a relative difference in sensory thresholds or weakness between sites. It can assess side-to-side or one nerve (or root or region) territory to another. It is useful for establishing sensory or motor impairment in a radicular, plexus, or single nerve distribution.

Testing for an absolute reduction in sensation (eg, decreased vibration in the toes) can be more challenging because it requires experience in judging what is normal and abnormal according to expectations for a particular site and modality. Take into account that sensory thresholds are normally increased with the patient’s age and height.¹¹ For example, we commonly encounter elderly patients whose vibration sensation in the toes is said to be decreased, when in fact the reduced sensory threshold is only an age-related change.

Most assessments of sensory thresholds use the absolute approach because most generalized polyneuropathies are “length-dependent.”

Perform motor testing for appendicular (upper and lower extremities) and axial (neck and trunk) muscles, assessing particularly for weakness, atrophy, and fasciculations.

The typical polyneuropathy caused by metabolic, toxic, inherited, or unknown causes is distal and symmetric.¹² Neuropathies caused by other mechanisms, such as immune-mediated or infectious, are rarely length-dependent. Examples include motor neuronopathies (eg, amyotrophic lateral sclerosis), sensory neuronopathies (eg, paraneoplastic), polyradiculoneuropathies (Guillain-Barré syndrome, CIDP), and mononeuritis multiplex (caused by vasculitis).

When: The time course of signs and symptoms

Knowing whether the onset of neuropathy was definite and abrupt or was gradual is the most helpful temporal clue to possible underlying causes. The time course (ie, tempo) following onset is also important.

An acute/subacute onset with a definite date often suggests an immune-mediated or an infectious process (FIGURE). With respect to immune-mediated neuropathies, consider primarily autoimmune conditions (eg, Guillain-Barré syndrome, vasculitic neuropathy) and also paraneoplastic autoimmune syndromes (eg, subacute sensory neuropathy). In the latter case, a cancer presents to the immune system an epitope that is similarly found in the nervous system, prompting an autoimmune attack of the nervous system. Both autoimmune and infectious processes almost always have a rapid start on a definite date.

With an insidious onset, the patient won’t recollect a definite date on which the neuropathy began. The underlying mechanism usually is an inherited, metabolic, or toxic process—or idiopathic, if a cause cannot be identified.

The disease course sheds additional light on the causative mechanism. Onset and subsequent progression often correlate in a predictable manner, owing in part to the underlying mechanism. For example, an acute onset neuropathy often is followed by rapid disease progression, especially when caused by an autoimmune process (Guillain-Barré and vasculitic neuropathy). On the other hand, a neuropathy of insidious onset usually follows a slow or even static course. But there are exceptions that may make the diagnosis challenging and illustrate the need for clinical follow-up.

What setting

The patient’s medical history, medications, social and family history, and a review of systems can uncover known risk factors for neuropathic processes.

Common causes of acquired polyneuropathies are diabetes mellitus, chronic renal disease, and alcohol dependence. If a patient with distal, symmetric sensory complaints also has any of these conditions, a causal relationship should be considered. Another example is the patient with a history of bariatric surgery, which could lead to neuropathy as a result of malnutrition, particularly vitamin B₁ and B₁₂ deficiencies.¹³

The clinical setting may indicate a need for further evaluation by a neurologist. For example, a family history of inherited neuropathy in a patient with high arches and curled “hammer” toes would strongly suggest Charcot-Marie-Tooth (CMT) disease, also known as hereditary motor and sensory neuropathy. A patient with a monoclonal gammopathy may have a paraproteinemic neuropathy that warrants further evaluation by a neuromuscular specialist. Although more rare, a paraneoplastic cause would warrant consideration in a smoker, especially if the neuropathy was subacute. In many cases, neuropathic symptoms are the first clue of a new medical condition (eg, impaired glucose tolerance).

Electrodiagnostic testing: What it can and can’t tell you

Most polyneuropathies warrant additional electrodiagnostic evaluation in an electromyography (EMG) laboratory. Electro-diagnostic testing comprises 2 procedures: nerve conduction studies and needle electrode examination.

Preparing your patient. When ordering this study, be sure to discuss it thoroughly with the patient. The entire study typically takes an hour or longer. And it can be painful, though in our experience nearly every patient tolerates the procedure. Most important for the patient to understand is that information gleaned from electrodiagnostic testing may be essential to the diagnosis, as explained below.

Many benefits of the study. First, electrodiagnostic testing can confirm a peripheral neuropathic basis for a patient’s complaints.

Second, electrodiagnostic testing helps characterize a neuropathy as primarily demyelinating, primarily axonal, or mixed demyelinating and axonal. An example of this benefit is that a primary demyelinating characterization greatly narrows the list of possible causes (eg, Guillain-Barré syndrome, chronic inflammatory demyelinating polyradiculoneuropathy, Charcot-Marie-Tooth disease type 1).

Third, electrodiagnostic testing assists in characterizing the neuropathic process as sensory, motor or sensorimotor.

Fourth, electrodiagnostic testing helps localize the neuropathic process (the “where”).

Fifth, electrodiagnostic testing can gauge the severity of the neuropathic process.

Limitations of the study. When the study returns normal results, keep in mind it has limited sensitivity. For example, nerve conduction studies are only able to assess the function of larger myelinated nerve fibers; a neuropathic process solely in small fibers will not be evident with this test. Likewise, the needle electrode examination is unable to assess small nerve fiber status. We include this caveat in our electrodiagnostic testing reports of patients who have symptoms suggestive of pure small fiber polyneuropathy.

When might blood studies be useful?

Laboratory testing of blood is often of great value, but only after a particular polyneuropathy has been characterized and placed into one or more potential etiologic subgroups.

For example, laboratory testing for a distal, symmetric sensory polyneuropathy (TABLE 2) should be much different than testing for another presentation (eg, mononeuritis multiplex). TABLE 3 details some of the laboratory tests we recommend for the more common polyneuropathies that don’t typically present in a distal, symmetric sensory fashion or that are accompanied by other distinctive features. In our experience, clinicians too frequently order unnecessary and expensive tests for disorders only rarely associated with neuropathy; the rare causes of neuropathy are intentionally not the subject of this review.

TABLE 2
Common blood tests for the evaluation of a distal, symmetrical neuropathy^1,3,5,6,8

TABLE 3
Other diagnostic tests for acquired neuropathies in specific clinical situations^2,4,5,7

Common causes of distal, symmetric polyneuropathies

Distal, symmetric polyneuropathies are usually due to metabolic/toxic, inherited, or idiopathic causes (FIGURE, TABLE 2). As such, it is often unnecessary to obtain diagnostic tests searching for active infectious, autoimmune or paraneoplastic etiologies.³ In electrodiagnostic terms, these neuropathies are almost always primarily axonal rather than demyelinating, usually involving both large and small nerve fibers.

Diabetes is the most common cause of neuropathy in developed Western countries, occurring in more than 50% of patients who require insulin.¹⁴ Diabetic neuropathy typically presents as a distal, symmetric neuropathy syndrome, though diabetic lumbosacral radiculoplexus neuropathy (diabetic amyotrophy) and other presentations may be seen less commonly.

Recent evidence suggests that neuropathy—particularly a sensory and often painful, distal, symmetric “small-fiber” neuropathy—sometimes occurs in patients with impaired fasting glucose (IFG) or impaired glucose tolerance (IGT).^15-19 The 2-hour oral glucose tolerance test has been deemed more sensitive for the early diabetic state of IGT/IFG associated with neuropathy.¹⁹ It remains to be seen whether this early diabetic state neuropathy will turn out to represent a large percentage of cases of what were previously thought of as idiopathic small-fiber neuropathy.

Alcoholism is another common cause of a distal, symmetric neuropathy that is predominantly sensory with a painful, burning sensation.^20-22 Alcohol abuse and dependence occurs in 10% to 20% of the primary care population.^23,24 The prevalence of neuropathy in alcoholics is uncertain, though 1 study of hospitalized patients admitting to daily alcohol intake of over 100 g for men or 80 g for women (10 oz of beer, 1 oz liquor and 3–4 oz wine each have 10 g of alcohol) for 2 years or more (mean of 238±120 g for a period of 22.7±10.2 years) demonstrated that one third of patients had electrophysiologic evidence of peripheral neuropathy and one fourth of autonomic neuropathy.²⁰ Most subjects in this cohort did not show any clinical or laboratory evidence of malnutrition.

Frequently alcoholic neuropathy coexists with thiamine-deficient neuropathy,²² warranting assessment of thiamine status in all alcoholics with neuropathy. Pain is a prominent complaint in alcoholic neuropathy but a less common complaint in thiamine-deficient neuropathy.²² Cobalamin (vitamin B₁₂)-deficient neuropathy is more likely to occur suddenly and to involve the hands or hands and feet simultaneously, and is less likely to be painful.²⁵ Myelopathy is frequent in cobalamin deficiency, sometimes serving as a clue to diagnosis. It may be difficult to determine whether sensory symptoms are caused by myelopathy or neuropathy. Electro-diagnostic testing, somatosensory evoked potentials, and radiological investigation may be helpful.

Uremic neuropathy may occur in patients with chronic renal failure on dialysis.^26,27 Today it is less common by virtue of the widespread implementation of dialysis and renal transplantation. Generally reserve this diagnosis for patients with end-stage renal failure with a creatinine clearance of less than 10 mL/min. Other systemic disorders associated with renal failure must be considered; these include diabetes mellitus, amyloidosis, and vasculitis. Also consider drug-induced neuropathy in this population.

Pharmaceutical agents, industrial and environmental agents, and substances of abuse may cause neuropathy, most commonly in the form of a length-dependent axonal neuropathy. Offending chemotherapeutic agents include colchicine, pyridoxine, and amiodarone. However, toxic polyneuropathies probably represent a rather small proportion of cases.

The list of potentially offending substances is long and includes many medications and agents that the general population is commonly exposed to in low doses. Consider the inherent risk of neuropathy of the particular agent in question. To establish a causal link, verify exposure, determine that symptoms are temporally related to the toxin, and rule out other causes of neuropathy. Furthermore, some clinical improvement or at least stabilization should occur following removal of the offending agent, though this may take months to years.²⁸

Medication-induced neuropathy is more common than industrial and environmental neuropathies. The neuropathy of heavy metal intoxication is rare, and is usually accompanied by a combination of gastrointestinal, hematologic, and central nervous system problems.

HIV neuropathy is predominantly sensory and length-dependent. It occurs in about one third of infected patients. This is an exception to the tenet that infectious neuropathies typically present in a non– length-dependent pattern. However, in almost all circumstances, the diagnosis of HIV is well established and so it usually doesn’t present a diagnostic challenge. It is not our routine practice to check for HIV in those presenting with a new-onset distal, symmetric neuropathy, though we will in the proper clinical setting. Whether HIV neuropathy should be included in the differential diagnosis depends on the incidence of HIV in your patient population and the patient’s specifics.⁶ Rarely, Guillain-Barré syndrome and CIDP can be the presentation of a recent HIV infection.

For inherited neuropathies, there are an overwhelming number of commercially available panels for DNA testing.^3,29-31 A detailed discussion of inherited neuropathies is beyond the scope of this review, and many excellent reviews have been written on the subject. We will, however, make a few general comments.

First, Charcot-Marie-Tooth (CMT) disease is the most common inherited neuromuscular disorder, and it is encountered in primary care.

Second, foot deformities (eg, high arches, curled toes) commonly accompany an inherited polyneuropathy and are valuable clues.

Third, the inheritence pattern should be sought through a detailed family history and examination of other family members (including query and examination for foot deformities). Inherited neuropathies can be autosomal dominant, autosomal recessive, or X-linked, based largely on what gene harbors the mutation. Furthermore, spontaneous “de novo” mutations are not uncommon, being responsible for probably 25% of cases of CMT type 1.

Fourth, inherited neuropathies should be characterized just like acquired neuropathies, with particular attention paid to whether the neuropathy is demyelinating (eg, CMT type 1) or axonal (CMT type 2).

With respect to DNA testing of blood, focused genetic testing is almost always possible once you further characterize the inherited neuropathy and also take into account prevalence estimates of the various inherited neuropathies (eg, CMT type 1A caused by a duplication of the PMP-22 gene is responsible for the majority of cases on CMT), obviating the need for expensive, comprehensive genetic testing.

Other presentations

The evaluation of neuropathies that are not distal and symmetric can be complex and in many cases may warrant referral to a neurologist for further evaluation. With such presentations, diagnostic considerations include chest imaging and antineuronal (paraneoplastic) antibody serology for smokers with subacute neuropathies, cerebrospinal fluid analysis for an acquired demyelinating neuropathy, infectious neuropathy or a polyradiculopathy (eg, Lyme disease),^3,7 and sensory nerve biopsy (eg, sural nerve) when vasculitis, amyloidosis, or sarcoidosis is suspected.

CORRESPONDENCE
Ted M. Burns, MD, University of Virginia Health System Department of Neurology, Box 800394, Charlottesville, VA 22908. E-mail: [email protected]

Polyneuropathy has an estimated prevalence of about 2% in the general population.^1,2 Despite being common, polyneuropathy remains a diagnostic challenge for most clinicians for many reasons, including the large number of potential causes and the fact that a specific cause often cannot be identified even after appropriate testing. These factors can contribute to uncertainty about the direction and level of aggressiveness of the evaluation. The result is often a “one size fits all” strategy—from an unnecessarily expensive “shotgun” approach to a “defeatist” attitude that too quickly deems a neuropathy as idiopathic.

A number of experts have developed algorithms for the evaluation of neuropathy.^1,3-8 We have incorporated and modified many of these suggestions to develop a simple, user-friendly approach; one that characterizes the neuropathy using “what?”, “where?”, “when?” and “what setting?” parameters (FIGURE).

The focus of this article is on the polyneuropathy that presents with widespread sensory or sensorimotor symptoms and signs. We will not discuss the evaluation of other peripheral neuropathic processes—eg, mononeuropathies, plexopathies, or motor neuronopathies such as amyotrophic lateral sclerosis. We emphasize the fundamental step of determining whether the polyneuropathy is a distal, symmetrical, sensory or sensorimotor polyneuropathy, in which case metabolic/toxic, inherited and idiopathic causes are more probable. In contrast, alternative presentations suggest immune-mediated and infectious causes.

What: Sensory, motor, or autonomic?

The first step is to assign the patient’s complaints and your examination findings to specific nerve fiber types of the peripheral nervous system: sensory, motor, or autonomic.⁵

FIGURE
Characterizing your patient’s neuropathy

Sensory findings help narrow your search

Once you have ruled out a nonneuropathic process (eg, arthritis) as the cause of sensory symptoms and signs, turn your attention to distinguishing between peripheral or central nervous system dysfunction. When the pathology resides in the central nervous system (eg, multiple sclerosis) accompanying symptoms and signs usually assist in localization. If a patient acknowledges clinical features suggesting past or present involvement of cerebellar, urinary, or visual (eg, optic neuritis) systems, for example, magnetic resonance imaging rather than electrodiagnostic testing may be warranted.

In a patient with peripheral nervous system dysfunction, sensory abnormalities help exclude neuromuscular diseases not associated with sensory dysfunction, such as myopathies, neuromuscular transmission disorders, or disease of the motor neuron.

Positive neuropathic sensory symptoms suggest acquired polyneuropathies. Patients may describe “prickling,” “tingling,” “swelling,” “asleep-like numbness,” or a sensation of “bunched-up socks.” Patients with acquired polyneuropathies usually complain of positive neuropathic sensory symptoms (PNSS), whereas patients with inherited polyneuropathies only rarely do (FIGURE).

Pain suggests toxic, metabolic, or ischemic causes. Patients may describe “electric shocks,” “burning,” “freezing,” “tightness,” or “throbbing.” They may complain of discomfort or pain to sensory stimuli that under normal circumstances would not be painful (allodynia)—eg, discomfort evoked by a bed sheet resting on the feet. They may also describe exaggerated pain to stimuli that would normally evoke low levels of discomfort or pain (hyperalgesia).

A painful neuropathy narrows the differential diagnosis to diseases that affect smaller nerve fibers, which generally convey pain and temperature input. Causes may be toxic, metabolic, ischemic, or idiopathic. For example, a pure “small-fiber” polyneuropathy commonly occurs in patients aged >60 years and typically causes painful feet. It is often idiopathic, but diabetes or impaired glucose tolerance and alcohol toxicity should be explored.⁹ Patients with painful neuropathy usually also exhibit reduced or absent sensation of pinprick and temperature in the distribution of their sensory complaints.

Negative neuropathic sensory symptoms. In addition to positive neuropathic symptoms, patients may complain of “negative” symptoms such as loss of sensation and imbalance. Examination usually reveals abnormalities of proprioception and sensation to vibration with reduced or absent deep tendon reflexes and ataxia. These features can occur in acquired or inherited causes of neuropathy.

Important sensory tests. Test sensation on the toes and fingertips, and more proximally (eg, ankle and shin) if any abnormality is found at these distal sites. Test vibration with a 128 Hz tuning fork, pinprick with disposable safety pin, and light touch with a cotton swab.

You may test temperature sensation by warming or cooling the handle or prong of the tuning fork and applying it to the patient’s skin.

Joint position testing is performed by asking the patient to avert his eyes, then moving the distal phalanx of a finger or toe up or down by small increments and asking the patient to tell you the direction of movement. Assess a patient’s casual and tandem gait for unsteadiness or ataxia.

Motor symptoms: Weigh them against sensory findings

Most patients with neuropathy have some degree of weakness, but it is usually overshadowed by sensory complaints. Distal lower extremity weakness may manifest as “foot drop,” which, if it affects ankle dorsiflexion, may cause a “slapping” or noisy step due to the forefoot hitting the ground with abnormal force.

Distal upper extremity weakness may cause trouble with fine motor skills of the hands.

Proximal weakness may present as difficulty in rising from a chair or lifting objects above the shoulders.

There may be muscle atrophy or fasciculations.

Motor symptoms are seldom the sole complaint. When motor and sensory symptoms are combined, it is helpful to rank them in order of symptom predominance—ie, motor greater than sensory, or vice versa. For instance, many immune-mediated disorders, such as Guillain-Barré syndrome and chronic inflammatory demyelinating polyradiculoneuropathy (CIDP), produce chiefly motor abnormalities and fewer sensory symptoms. Sensory complaints predominate in many other polyneuropathies, especially the “length-dependent” polyneuropathies (ie, those affecting the longer nerves initially) caused by metabolic or toxic disorders.

Autonomic symptom evaluation

The number of processes that affect both autonomic and somatic nerves are relatively few (TABLE 1).^5,7,10 It is particularly important to assess symptoms suggesting involvement of the autonomic nervous system.

Autonomic symptoms include lightheadedness, syncope, diarrhea, constipation, postprandial bloating, early satiety, urinary complaints, erectile dysfunction, abnormal or absent sweating, and dry mouth and eyes. Many of these complaints are common in the general population, so their relevance should be based on severity and temporal evolution, as well as comorbidities and medication use.

When complaints do not clearly implicate pathology in the autonomic nervous system, autonomic testing may be helpful, targeted for the domain that may be impaired. For example, bedside orthostatics or tilt-table testing are used for pre-syncopal symptoms, but gastric emptying testing can assist the evaluation of complaints of early satiety or postprandial bloating.

TABLE 1
Relatively common acquired polyneuropathies with autonomic nervous system involvement

Where: The distribution of nerve involvement

“Where” refers to distribution of nerve involvement 1) globally throughout the body and 2) locally along the nerve(s). During the history taking and examination, determine the nature of general distribution (eg, symmetric or asymmetric) and where along the length of the nerve(s) (proximal and/or distal) the dysfunction exists.

Polyneuropathy most commonly presents in a “length-dependent” distribution, with clinical features appearing initially most distally and symmetrically (ie, in the feet). Asymmetry and involvement of the proximal parts of a nerve are “red flags” for an uncommon cause that may require referral to a neurologist (FIGURE).

Comparative vs absolute measurements. At bedside, 2 approaches are used to assess the distribution of nerve involvement: comparative and absolute. The comparative approach searches for a relative difference in sensory thresholds or weakness between sites. It can assess side-to-side or one nerve (or root or region) territory to another. It is useful for establishing sensory or motor impairment in a radicular, plexus, or single nerve distribution.

Testing for an absolute reduction in sensation (eg, decreased vibration in the toes) can be more challenging because it requires experience in judging what is normal and abnormal according to expectations for a particular site and modality. Take into account that sensory thresholds are normally increased with the patient’s age and height.¹¹ For example, we commonly encounter elderly patients whose vibration sensation in the toes is said to be decreased, when in fact the reduced sensory threshold is only an age-related change.

Most assessments of sensory thresholds use the absolute approach because most generalized polyneuropathies are “length-dependent.”

Perform motor testing for appendicular (upper and lower extremities) and axial (neck and trunk) muscles, assessing particularly for weakness, atrophy, and fasciculations.

The typical polyneuropathy caused by metabolic, toxic, inherited, or unknown causes is distal and symmetric.¹² Neuropathies caused by other mechanisms, such as immune-mediated or infectious, are rarely length-dependent. Examples include motor neuronopathies (eg, amyotrophic lateral sclerosis), sensory neuronopathies (eg, paraneoplastic), polyradiculoneuropathies (Guillain-Barré syndrome, CIDP), and mononeuritis multiplex (caused by vasculitis).

When: The time course of signs and symptoms

Knowing whether the onset of neuropathy was definite and abrupt or was gradual is the most helpful temporal clue to possible underlying causes. The time course (ie, tempo) following onset is also important.

An acute/subacute onset with a definite date often suggests an immune-mediated or an infectious process (FIGURE). With respect to immune-mediated neuropathies, consider primarily autoimmune conditions (eg, Guillain-Barré syndrome, vasculitic neuropathy) and also paraneoplastic autoimmune syndromes (eg, subacute sensory neuropathy). In the latter case, a cancer presents to the immune system an epitope that is similarly found in the nervous system, prompting an autoimmune attack of the nervous system. Both autoimmune and infectious processes almost always have a rapid start on a definite date.

With an insidious onset, the patient won’t recollect a definite date on which the neuropathy began. The underlying mechanism usually is an inherited, metabolic, or toxic process—or idiopathic, if a cause cannot be identified.

The disease course sheds additional light on the causative mechanism. Onset and subsequent progression often correlate in a predictable manner, owing in part to the underlying mechanism. For example, an acute onset neuropathy often is followed by rapid disease progression, especially when caused by an autoimmune process (Guillain-Barré and vasculitic neuropathy). On the other hand, a neuropathy of insidious onset usually follows a slow or even static course. But there are exceptions that may make the diagnosis challenging and illustrate the need for clinical follow-up.

What setting

The patient’s medical history, medications, social and family history, and a review of systems can uncover known risk factors for neuropathic processes.

Common causes of acquired polyneuropathies are diabetes mellitus, chronic renal disease, and alcohol dependence. If a patient with distal, symmetric sensory complaints also has any of these conditions, a causal relationship should be considered. Another example is the patient with a history of bariatric surgery, which could lead to neuropathy as a result of malnutrition, particularly vitamin B₁ and B₁₂ deficiencies.¹³

The clinical setting may indicate a need for further evaluation by a neurologist. For example, a family history of inherited neuropathy in a patient with high arches and curled “hammer” toes would strongly suggest Charcot-Marie-Tooth (CMT) disease, also known as hereditary motor and sensory neuropathy. A patient with a monoclonal gammopathy may have a paraproteinemic neuropathy that warrants further evaluation by a neuromuscular specialist. Although more rare, a paraneoplastic cause would warrant consideration in a smoker, especially if the neuropathy was subacute. In many cases, neuropathic symptoms are the first clue of a new medical condition (eg, impaired glucose tolerance).

Electrodiagnostic testing: What it can and can’t tell you

Most polyneuropathies warrant additional electrodiagnostic evaluation in an electromyography (EMG) laboratory. Electro-diagnostic testing comprises 2 procedures: nerve conduction studies and needle electrode examination.

Preparing your patient. When ordering this study, be sure to discuss it thoroughly with the patient. The entire study typically takes an hour or longer. And it can be painful, though in our experience nearly every patient tolerates the procedure. Most important for the patient to understand is that information gleaned from electrodiagnostic testing may be essential to the diagnosis, as explained below.

Many benefits of the study. First, electrodiagnostic testing can confirm a peripheral neuropathic basis for a patient’s complaints.

Second, electrodiagnostic testing helps characterize a neuropathy as primarily demyelinating, primarily axonal, or mixed demyelinating and axonal. An example of this benefit is that a primary demyelinating characterization greatly narrows the list of possible causes (eg, Guillain-Barré syndrome, chronic inflammatory demyelinating polyradiculoneuropathy, Charcot-Marie-Tooth disease type 1).

Third, electrodiagnostic testing assists in characterizing the neuropathic process as sensory, motor or sensorimotor.

Fourth, electrodiagnostic testing helps localize the neuropathic process (the “where”).

Fifth, electrodiagnostic testing can gauge the severity of the neuropathic process.

Limitations of the study. When the study returns normal results, keep in mind it has limited sensitivity. For example, nerve conduction studies are only able to assess the function of larger myelinated nerve fibers; a neuropathic process solely in small fibers will not be evident with this test. Likewise, the needle electrode examination is unable to assess small nerve fiber status. We include this caveat in our electrodiagnostic testing reports of patients who have symptoms suggestive of pure small fiber polyneuropathy.

When might blood studies be useful?

Laboratory testing of blood is often of great value, but only after a particular polyneuropathy has been characterized and placed into one or more potential etiologic subgroups.

For example, laboratory testing for a distal, symmetric sensory polyneuropathy (TABLE 2) should be much different than testing for another presentation (eg, mononeuritis multiplex). TABLE 3 details some of the laboratory tests we recommend for the more common polyneuropathies that don’t typically present in a distal, symmetric sensory fashion or that are accompanied by other distinctive features. In our experience, clinicians too frequently order unnecessary and expensive tests for disorders only rarely associated with neuropathy; the rare causes of neuropathy are intentionally not the subject of this review.

TABLE 2
Common blood tests for the evaluation of a distal, symmetrical neuropathy^1,3,5,6,8

TABLE 3
Other diagnostic tests for acquired neuropathies in specific clinical situations^2,4,5,7

Common causes of distal, symmetric polyneuropathies

Distal, symmetric polyneuropathies are usually due to metabolic/toxic, inherited, or idiopathic causes (FIGURE, TABLE 2). As such, it is often unnecessary to obtain diagnostic tests searching for active infectious, autoimmune or paraneoplastic etiologies.³ In electrodiagnostic terms, these neuropathies are almost always primarily axonal rather than demyelinating, usually involving both large and small nerve fibers.

Diabetes is the most common cause of neuropathy in developed Western countries, occurring in more than 50% of patients who require insulin.¹⁴ Diabetic neuropathy typically presents as a distal, symmetric neuropathy syndrome, though diabetic lumbosacral radiculoplexus neuropathy (diabetic amyotrophy) and other presentations may be seen less commonly.

Recent evidence suggests that neuropathy—particularly a sensory and often painful, distal, symmetric “small-fiber” neuropathy—sometimes occurs in patients with impaired fasting glucose (IFG) or impaired glucose tolerance (IGT).^15-19 The 2-hour oral glucose tolerance test has been deemed more sensitive for the early diabetic state of IGT/IFG associated with neuropathy.¹⁹ It remains to be seen whether this early diabetic state neuropathy will turn out to represent a large percentage of cases of what were previously thought of as idiopathic small-fiber neuropathy.

Alcoholism is another common cause of a distal, symmetric neuropathy that is predominantly sensory with a painful, burning sensation.^20-22 Alcohol abuse and dependence occurs in 10% to 20% of the primary care population.^23,24 The prevalence of neuropathy in alcoholics is uncertain, though 1 study of hospitalized patients admitting to daily alcohol intake of over 100 g for men or 80 g for women (10 oz of beer, 1 oz liquor and 3–4 oz wine each have 10 g of alcohol) for 2 years or more (mean of 238±120 g for a period of 22.7±10.2 years) demonstrated that one third of patients had electrophysiologic evidence of peripheral neuropathy and one fourth of autonomic neuropathy.²⁰ Most subjects in this cohort did not show any clinical or laboratory evidence of malnutrition.

Frequently alcoholic neuropathy coexists with thiamine-deficient neuropathy,²² warranting assessment of thiamine status in all alcoholics with neuropathy. Pain is a prominent complaint in alcoholic neuropathy but a less common complaint in thiamine-deficient neuropathy.²² Cobalamin (vitamin B₁₂)-deficient neuropathy is more likely to occur suddenly and to involve the hands or hands and feet simultaneously, and is less likely to be painful.²⁵ Myelopathy is frequent in cobalamin deficiency, sometimes serving as a clue to diagnosis. It may be difficult to determine whether sensory symptoms are caused by myelopathy or neuropathy. Electro-diagnostic testing, somatosensory evoked potentials, and radiological investigation may be helpful.

Uremic neuropathy may occur in patients with chronic renal failure on dialysis.^26,27 Today it is less common by virtue of the widespread implementation of dialysis and renal transplantation. Generally reserve this diagnosis for patients with end-stage renal failure with a creatinine clearance of less than 10 mL/min. Other systemic disorders associated with renal failure must be considered; these include diabetes mellitus, amyloidosis, and vasculitis. Also consider drug-induced neuropathy in this population.

Pharmaceutical agents, industrial and environmental agents, and substances of abuse may cause neuropathy, most commonly in the form of a length-dependent axonal neuropathy. Offending chemotherapeutic agents include colchicine, pyridoxine, and amiodarone. However, toxic polyneuropathies probably represent a rather small proportion of cases.

The list of potentially offending substances is long and includes many medications and agents that the general population is commonly exposed to in low doses. Consider the inherent risk of neuropathy of the particular agent in question. To establish a causal link, verify exposure, determine that symptoms are temporally related to the toxin, and rule out other causes of neuropathy. Furthermore, some clinical improvement or at least stabilization should occur following removal of the offending agent, though this may take months to years.²⁸

Medication-induced neuropathy is more common than industrial and environmental neuropathies. The neuropathy of heavy metal intoxication is rare, and is usually accompanied by a combination of gastrointestinal, hematologic, and central nervous system problems.

HIV neuropathy is predominantly sensory and length-dependent. It occurs in about one third of infected patients. This is an exception to the tenet that infectious neuropathies typically present in a non– length-dependent pattern. However, in almost all circumstances, the diagnosis of HIV is well established and so it usually doesn’t present a diagnostic challenge. It is not our routine practice to check for HIV in those presenting with a new-onset distal, symmetric neuropathy, though we will in the proper clinical setting. Whether HIV neuropathy should be included in the differential diagnosis depends on the incidence of HIV in your patient population and the patient’s specifics.⁶ Rarely, Guillain-Barré syndrome and CIDP can be the presentation of a recent HIV infection.

For inherited neuropathies, there are an overwhelming number of commercially available panels for DNA testing.^3,29-31 A detailed discussion of inherited neuropathies is beyond the scope of this review, and many excellent reviews have been written on the subject. We will, however, make a few general comments.

First, Charcot-Marie-Tooth (CMT) disease is the most common inherited neuromuscular disorder, and it is encountered in primary care.

Second, foot deformities (eg, high arches, curled toes) commonly accompany an inherited polyneuropathy and are valuable clues.

Third, the inheritence pattern should be sought through a detailed family history and examination of other family members (including query and examination for foot deformities). Inherited neuropathies can be autosomal dominant, autosomal recessive, or X-linked, based largely on what gene harbors the mutation. Furthermore, spontaneous “de novo” mutations are not uncommon, being responsible for probably 25% of cases of CMT type 1.

Fourth, inherited neuropathies should be characterized just like acquired neuropathies, with particular attention paid to whether the neuropathy is demyelinating (eg, CMT type 1) or axonal (CMT type 2).

With respect to DNA testing of blood, focused genetic testing is almost always possible once you further characterize the inherited neuropathy and also take into account prevalence estimates of the various inherited neuropathies (eg, CMT type 1A caused by a duplication of the PMP-22 gene is responsible for the majority of cases on CMT), obviating the need for expensive, comprehensive genetic testing.

Other presentations

The evaluation of neuropathies that are not distal and symmetric can be complex and in many cases may warrant referral to a neurologist for further evaluation. With such presentations, diagnostic considerations include chest imaging and antineuronal (paraneoplastic) antibody serology for smokers with subacute neuropathies, cerebrospinal fluid analysis for an acquired demyelinating neuropathy, infectious neuropathy or a polyradiculopathy (eg, Lyme disease),^3,7 and sensory nerve biopsy (eg, sural nerve) when vasculitis, amyloidosis, or sarcoidosis is suspected.

CORRESPONDENCE
Ted M. Burns, MD, University of Virginia Health System Department of Neurology, Box 800394, Charlottesville, VA 22908. E-mail: [email protected]

A 39-year-old female nurse with a history of moderate to severe asthma was preparing medications for patients at an extended care facility. While pouring a bulk psyllium product from a large container into a medicine cup, a clumped mass of the powder fell from the container, striking the counter, and raising a cloud of psyllium dust. On inhaling this dust, the nurse became dyspneic. She did not respond to a beta-agonist and suffered respiratory arrest within 5 minutes. Emergency Medical Services and staff at the receiving hospital were unable to resuscitate her. Time of death was 1 hour and 25 minutes after exposure to psyllium dust. Autopsy showed severe bronchospasm with mucus plugging, but no signs of foreign material in the airways. Anaphylaxis secondary to inhaled psyllium was the final diagnosis.

Risk is significant and underestimated

Thousands of health care workers, particularly nurses, are at risk for hypersensitivity responses to psyllium, including rhinitis, conjunctivitis, cough, wheeze, asthma exacerbation, nausea, and anaphylaxis.^1-7 Exposure in the workplace usually results from inhalation of dust during the preparation of the product for patient use.

One review found that up to 73% of residents in extended care facilities receive psyllium products.⁸ Hospitalized and home care patients also receive psyllium. Nurses especially are constantly exposed, and many are sensitized.

Despite the significant number of persons at risk, awareness and precaution are inadequate. Personal communication reveals that no health care providers—nurses, physicians, medical directors—were aware of the degree of risk. There was no routine use of protective devices or precautions when working with psyllium products in the facilities reviewed.

Quality and safety of psyllium products vary

Psyllium is a well-known powdered fiber supplement derived from the seed husk of Plantago ovata.^9,10 The supplement contains varying ratios of endosperm, seed embryo, and psyllium (a translucent hydrophilic colloid material).

The endosperm and seed embryo proteins are allergenic; the colloid is not.¹¹ Concentrations of each component vary according to the means of processing and the country of origin.^9-11

Independent review indicates that the concentration of non-psyllium allergenic components is inversely proportional to the cost of the commercial product. The product dispensed for the patient in our case was particularly heterogenous.

Those at particular risk

Multiple instances of anaphylaxis have been linked to psyllium since the initial description in 1941.¹² Ingestion of cereals containing psyllium is the most common cause of psyllium-related anaphylaxis.¹⁰ In a number of cases of anaphylaxis secondary to ingestion, the initial sensitization was related to psyllium dust inhalation.¹ Twenty cases of anaphylaxis to a psyllium-rich cereal were reviewed shortly after its introduction. Fourteen of those affected were nurses. It was concluded that these nurses were sensitized from job-related inhalation exposure.¹³

IgE sensitization to psyllium has been shown in as many as 12% of health care personnel. History of asthma or atopy in a health care worker, as in the patient presented, seems to make IgE sensitization to psyllium more likely and raises the risk for fatal anaphylaxis.²

One review of workers with documented IgE sensitivity to psyllium showed that 42% complained of symptoms of conjunctivitis, rhinitis, or exacerbation of asthma.³ In another study, 29% of subjects were shown to have significant bronchial hyper-responsiveness to psyllium dust.²

One 1987 investigation of health care workers in an extended care facility concluded that the response to inhaled psyllium in these employees could range from mild to disabling.⁵ In a similar study, a nurse who had experienced only mild symptoms previously required intubation for severe bronchospasm after only 1 minute of exposure to psyllium dust.³ As happened with the patient in our case, she was symptomatic (moderate to severe asthma) and exposed, which led to fatal anaphylaxis.

Urge precautions to limit risk

Package inserts for the most popular—but not all—brands of psyllium describe precautions for dispensing psyllium, particularly for those with asthma.

1. Don’t pour the powder. Pouring the bulk form of the powder into medicine or beverage cups disperses significant amounts of dust. Yet this seems to be the most common manner of dispensing this medication. The standard set by the Occupational Safety and Health Administration (OSHA) limits the time-weighted average concentration for nuisance dust to 15 mg/m³. Dust concentration exhibited when psyllium powder is poured ranges from 4 to 28.5 mg/m³, with a mean concentration of 12.96 mg/m³. Given the frequency of inappropriate preparation, it is probable that OSHA recommended safe levels are frequently exceeded.³

Spoon the powder … To limit the amount of aerosolized dust, the powder should only be spooned into cups. While this recommendation is in the package insert, most health care workers are unaware of the guideline.

… or pour solvent over powder. Even better, our experiments showed there was less dust produced if the solvent (water or orange juice) was poured slowly over the powder than if the powder was put into solvent.

… or immerse powder in liquid. Alternatively, a device such as a scoop with a lid could be used that would submerge the powder into solution and then release it, creating virtually no dust.

2. Ventilate properly … Respiratory protection must be offered to the preparer and anyone else in the area. One maneuver to prevent exposure to psyllium dust would be to prepare the product in a chemical hood to ventilate dust away from the preparer.

… or use a mask. A simple mask, as is used to prevent the spread of communicable diseases, would offer some protection, though incomplete. The Material Safety Data Sheets for psyllium recommends that personnel wear powered purifying respirators with HEPA (high-efficiency particulate air) filters and Tyvek face seals while cleaning up a psyllium spill. In our query, however, these items were not readily available.

3. Substitute other preparations? A third alternative to prevent psyllium dust exposure would be to use other forms of psyllium, such as granules, capsules, or wafers. The drawback to this alternative, clearly, is greater cost. As already mentioned, the cost of a psyllium product is inversely proportional to the number of antigenic components.

4. Assess risk beforehand. A final means of preventing serious hypersensitivity reactions is to screen staff for their degree of risk: pre-employment screening as well during subsequent review of systems by a physician. Stratify risk based on such factors as history of allergies, atopy or asthma, and frequency of respiratory symptoms, particularly as may be related to work.

5. Consider cross-reactivity. For example, there is evidence that reactivity to psyllium may be seen in those with dust and grass allergies.¹⁵

Social history should focus on employment and exposure both at work and at home (eg, those who care for elderly relatives in the home). Smoking status is also important, as cigarette smoke increases the risk of bronchial hyper-responsiveness. Appropriate precautions and warnings could then be offered based on perceived risk.

CORRESPONDENCE
E-mail: [email protected]

A 39-year-old female nurse with a history of moderate to severe asthma was preparing medications for patients at an extended care facility. While pouring a bulk psyllium product from a large container into a medicine cup, a clumped mass of the powder fell from the container, striking the counter, and raising a cloud of psyllium dust. On inhaling this dust, the nurse became dyspneic. She did not respond to a beta-agonist and suffered respiratory arrest within 5 minutes. Emergency Medical Services and staff at the receiving hospital were unable to resuscitate her. Time of death was 1 hour and 25 minutes after exposure to psyllium dust. Autopsy showed severe bronchospasm with mucus plugging, but no signs of foreign material in the airways. Anaphylaxis secondary to inhaled psyllium was the final diagnosis.

Risk is significant and underestimated

Thousands of health care workers, particularly nurses, are at risk for hypersensitivity responses to psyllium, including rhinitis, conjunctivitis, cough, wheeze, asthma exacerbation, nausea, and anaphylaxis.^1-7 Exposure in the workplace usually results from inhalation of dust during the preparation of the product for patient use.

One review found that up to 73% of residents in extended care facilities receive psyllium products.⁸ Hospitalized and home care patients also receive psyllium. Nurses especially are constantly exposed, and many are sensitized.

Despite the significant number of persons at risk, awareness and precaution are inadequate. Personal communication reveals that no health care providers—nurses, physicians, medical directors—were aware of the degree of risk. There was no routine use of protective devices or precautions when working with psyllium products in the facilities reviewed.

Quality and safety of psyllium products vary

Psyllium is a well-known powdered fiber supplement derived from the seed husk of Plantago ovata.^9,10 The supplement contains varying ratios of endosperm, seed embryo, and psyllium (a translucent hydrophilic colloid material).

The endosperm and seed embryo proteins are allergenic; the colloid is not.¹¹ Concentrations of each component vary according to the means of processing and the country of origin.^9-11

Independent review indicates that the concentration of non-psyllium allergenic components is inversely proportional to the cost of the commercial product. The product dispensed for the patient in our case was particularly heterogenous.

Those at particular risk

Multiple instances of anaphylaxis have been linked to psyllium since the initial description in 1941.¹² Ingestion of cereals containing psyllium is the most common cause of psyllium-related anaphylaxis.¹⁰ In a number of cases of anaphylaxis secondary to ingestion, the initial sensitization was related to psyllium dust inhalation.¹ Twenty cases of anaphylaxis to a psyllium-rich cereal were reviewed shortly after its introduction. Fourteen of those affected were nurses. It was concluded that these nurses were sensitized from job-related inhalation exposure.¹³

IgE sensitization to psyllium has been shown in as many as 12% of health care personnel. History of asthma or atopy in a health care worker, as in the patient presented, seems to make IgE sensitization to psyllium more likely and raises the risk for fatal anaphylaxis.²

One review of workers with documented IgE sensitivity to psyllium showed that 42% complained of symptoms of conjunctivitis, rhinitis, or exacerbation of asthma.³ In another study, 29% of subjects were shown to have significant bronchial hyper-responsiveness to psyllium dust.²

One 1987 investigation of health care workers in an extended care facility concluded that the response to inhaled psyllium in these employees could range from mild to disabling.⁵ In a similar study, a nurse who had experienced only mild symptoms previously required intubation for severe bronchospasm after only 1 minute of exposure to psyllium dust.³ As happened with the patient in our case, she was symptomatic (moderate to severe asthma) and exposed, which led to fatal anaphylaxis.

Urge precautions to limit risk

Package inserts for the most popular—but not all—brands of psyllium describe precautions for dispensing psyllium, particularly for those with asthma.

1. Don’t pour the powder. Pouring the bulk form of the powder into medicine or beverage cups disperses significant amounts of dust. Yet this seems to be the most common manner of dispensing this medication. The standard set by the Occupational Safety and Health Administration (OSHA) limits the time-weighted average concentration for nuisance dust to 15 mg/m³. Dust concentration exhibited when psyllium powder is poured ranges from 4 to 28.5 mg/m³, with a mean concentration of 12.96 mg/m³. Given the frequency of inappropriate preparation, it is probable that OSHA recommended safe levels are frequently exceeded.³

Spoon the powder … To limit the amount of aerosolized dust, the powder should only be spooned into cups. While this recommendation is in the package insert, most health care workers are unaware of the guideline.

… or pour solvent over powder. Even better, our experiments showed there was less dust produced if the solvent (water or orange juice) was poured slowly over the powder than if the powder was put into solvent.

… or immerse powder in liquid. Alternatively, a device such as a scoop with a lid could be used that would submerge the powder into solution and then release it, creating virtually no dust.

2. Ventilate properly … Respiratory protection must be offered to the preparer and anyone else in the area. One maneuver to prevent exposure to psyllium dust would be to prepare the product in a chemical hood to ventilate dust away from the preparer.

… or use a mask. A simple mask, as is used to prevent the spread of communicable diseases, would offer some protection, though incomplete. The Material Safety Data Sheets for psyllium recommends that personnel wear powered purifying respirators with HEPA (high-efficiency particulate air) filters and Tyvek face seals while cleaning up a psyllium spill. In our query, however, these items were not readily available.

3. Substitute other preparations? A third alternative to prevent psyllium dust exposure would be to use other forms of psyllium, such as granules, capsules, or wafers. The drawback to this alternative, clearly, is greater cost. As already mentioned, the cost of a psyllium product is inversely proportional to the number of antigenic components.

4. Assess risk beforehand. A final means of preventing serious hypersensitivity reactions is to screen staff for their degree of risk: pre-employment screening as well during subsequent review of systems by a physician. Stratify risk based on such factors as history of allergies, atopy or asthma, and frequency of respiratory symptoms, particularly as may be related to work.

5. Consider cross-reactivity. For example, there is evidence that reactivity to psyllium may be seen in those with dust and grass allergies.¹⁵

Social history should focus on employment and exposure both at work and at home (eg, those who care for elderly relatives in the home). Smoking status is also important, as cigarette smoke increases the risk of bronchial hyper-responsiveness. Appropriate precautions and warnings could then be offered based on perceived risk.

CORRESPONDENCE
E-mail: [email protected]

A 39-year-old female nurse with a history of moderate to severe asthma was preparing medications for patients at an extended care facility. While pouring a bulk psyllium product from a large container into a medicine cup, a clumped mass of the powder fell from the container, striking the counter, and raising a cloud of psyllium dust. On inhaling this dust, the nurse became dyspneic. She did not respond to a beta-agonist and suffered respiratory arrest within 5 minutes. Emergency Medical Services and staff at the receiving hospital were unable to resuscitate her. Time of death was 1 hour and 25 minutes after exposure to psyllium dust. Autopsy showed severe bronchospasm with mucus plugging, but no signs of foreign material in the airways. Anaphylaxis secondary to inhaled psyllium was the final diagnosis.

Risk is significant and underestimated

Thousands of health care workers, particularly nurses, are at risk for hypersensitivity responses to psyllium, including rhinitis, conjunctivitis, cough, wheeze, asthma exacerbation, nausea, and anaphylaxis.^1-7 Exposure in the workplace usually results from inhalation of dust during the preparation of the product for patient use.

One review found that up to 73% of residents in extended care facilities receive psyllium products.⁸ Hospitalized and home care patients also receive psyllium. Nurses especially are constantly exposed, and many are sensitized.

Despite the significant number of persons at risk, awareness and precaution are inadequate. Personal communication reveals that no health care providers—nurses, physicians, medical directors—were aware of the degree of risk. There was no routine use of protective devices or precautions when working with psyllium products in the facilities reviewed.

Quality and safety of psyllium products vary

Psyllium is a well-known powdered fiber supplement derived from the seed husk of Plantago ovata.^9,10 The supplement contains varying ratios of endosperm, seed embryo, and psyllium (a translucent hydrophilic colloid material).

The endosperm and seed embryo proteins are allergenic; the colloid is not.¹¹ Concentrations of each component vary according to the means of processing and the country of origin.^9-11

Independent review indicates that the concentration of non-psyllium allergenic components is inversely proportional to the cost of the commercial product. The product dispensed for the patient in our case was particularly heterogenous.

Those at particular risk

Multiple instances of anaphylaxis have been linked to psyllium since the initial description in 1941.¹² Ingestion of cereals containing psyllium is the most common cause of psyllium-related anaphylaxis.¹⁰ In a number of cases of anaphylaxis secondary to ingestion, the initial sensitization was related to psyllium dust inhalation.¹ Twenty cases of anaphylaxis to a psyllium-rich cereal were reviewed shortly after its introduction. Fourteen of those affected were nurses. It was concluded that these nurses were sensitized from job-related inhalation exposure.¹³

IgE sensitization to psyllium has been shown in as many as 12% of health care personnel. History of asthma or atopy in a health care worker, as in the patient presented, seems to make IgE sensitization to psyllium more likely and raises the risk for fatal anaphylaxis.²

One review of workers with documented IgE sensitivity to psyllium showed that 42% complained of symptoms of conjunctivitis, rhinitis, or exacerbation of asthma.³ In another study, 29% of subjects were shown to have significant bronchial hyper-responsiveness to psyllium dust.²

One 1987 investigation of health care workers in an extended care facility concluded that the response to inhaled psyllium in these employees could range from mild to disabling.⁵ In a similar study, a nurse who had experienced only mild symptoms previously required intubation for severe bronchospasm after only 1 minute of exposure to psyllium dust.³ As happened with the patient in our case, she was symptomatic (moderate to severe asthma) and exposed, which led to fatal anaphylaxis.

Urge precautions to limit risk

Package inserts for the most popular—but not all—brands of psyllium describe precautions for dispensing psyllium, particularly for those with asthma.

1. Don’t pour the powder. Pouring the bulk form of the powder into medicine or beverage cups disperses significant amounts of dust. Yet this seems to be the most common manner of dispensing this medication. The standard set by the Occupational Safety and Health Administration (OSHA) limits the time-weighted average concentration for nuisance dust to 15 mg/m³. Dust concentration exhibited when psyllium powder is poured ranges from 4 to 28.5 mg/m³, with a mean concentration of 12.96 mg/m³. Given the frequency of inappropriate preparation, it is probable that OSHA recommended safe levels are frequently exceeded.³

Spoon the powder … To limit the amount of aerosolized dust, the powder should only be spooned into cups. While this recommendation is in the package insert, most health care workers are unaware of the guideline.

… or pour solvent over powder. Even better, our experiments showed there was less dust produced if the solvent (water or orange juice) was poured slowly over the powder than if the powder was put into solvent.

… or immerse powder in liquid. Alternatively, a device such as a scoop with a lid could be used that would submerge the powder into solution and then release it, creating virtually no dust.

2. Ventilate properly … Respiratory protection must be offered to the preparer and anyone else in the area. One maneuver to prevent exposure to psyllium dust would be to prepare the product in a chemical hood to ventilate dust away from the preparer.

… or use a mask. A simple mask, as is used to prevent the spread of communicable diseases, would offer some protection, though incomplete. The Material Safety Data Sheets for psyllium recommends that personnel wear powered purifying respirators with HEPA (high-efficiency particulate air) filters and Tyvek face seals while cleaning up a psyllium spill. In our query, however, these items were not readily available.

3. Substitute other preparations? A third alternative to prevent psyllium dust exposure would be to use other forms of psyllium, such as granules, capsules, or wafers. The drawback to this alternative, clearly, is greater cost. As already mentioned, the cost of a psyllium product is inversely proportional to the number of antigenic components.

4. Assess risk beforehand. A final means of preventing serious hypersensitivity reactions is to screen staff for their degree of risk: pre-employment screening as well during subsequent review of systems by a physician. Stratify risk based on such factors as history of allergies, atopy or asthma, and frequency of respiratory symptoms, particularly as may be related to work.

5. Consider cross-reactivity. For example, there is evidence that reactivity to psyllium may be seen in those with dust and grass allergies.¹⁵

Social history should focus on employment and exposure both at work and at home (eg, those who care for elderly relatives in the home). Smoking status is also important, as cigarette smoke increases the risk of bronchial hyper-responsiveness. Appropriate precautions and warnings could then be offered based on perceived risk.

CORRESPONDENCE
E-mail: [email protected]

The Adult Treatment Panel III and the Dietary Guidelines for Americans recommend that adults eat a healthy diet as defined in TABLE 1—including more than 20 to 30 g of dietary fiber per day, 5 to 10 g of which should be viscous soluble fiber.^1,2,3 The benefits of such a diet are supported by growing epidemiologic data:

• Risk of CVD drops by 20% to 30%
• Morbidity and mortality from cardiovascular diseases fall by more than 10%
• Foods rich in viscous soluble fiber, at a dose of 5 to 10 g/d, can reduce LDL-C levels by 10% to 15%, improve insulin sensitivity, and decrease blood pressure
• Pooled analysis of 10 prospective cohort studies showed that, for adults at low risk of heart attack (estimated with Framingham scores), every additional 10 g of fiber consumed daily may prevent 1 coronary death in 74 adults (95% confidence interval [CI], 51–154), and 1 in 19 adults at high risk (95% CI, 13–38).⁴

Despite these hopeful findings, the average American adult eats less than half the recommended amount of fiber: 10 to 15 g/d; just 2.4 g of viscous soluble fiber.^2,5Healthy People 2010 has set a national objective to reduce coronary heart disease (CHD) mortality from 208 deaths to 166 deaths per 100,000 people.⁶ The US Preventive Services Task Force (USPSTF) recommends that primary care clinicians intensively counsel adults with elevated risk of CVD to improve daily intake of essential components of a healthy diet, including dietary fiber.⁷

In this article, we review practical approaches you can use in motivating patients to eat more fiber, types of dietary fiber and their sources, and the evidence for fiber’s impact on CVD risk factors and prevention of CVD events.

TABLE 1
Recommended ATP-III dietary change reduces risk of CVD by 20% to 30%

Making sure patients get the right dietary fiber

Dietary fiber (nondigestible non-starch polysaccharides) is classified as insoluble or viscous soluble, based on its solubility in water and other fluids in the digestive tract.⁸

Insoluble fibers, present in most fruits, vegetables, and whole grain products, constitute over 80% of the total fiber in the typical US diet. Insoluble fibers shorten bowel transit time and produce a laxative effect but have no significant impact on blood lipids.⁹

Viscous soluble fiber is present in smaller quantities in foods, but is responsible for the fiber-related physiological effects such as decreased blood cholesterol.⁵ The US Food and Drug Administration (FDA) has also approved health claims for 2 such fibers—beta-glucan from whole oat sources and psyllium—to be used as supplements in reducing CVD risk.^10,11

Viscous soluble fiber absorbs and holds water in the gastrointestinal tract, forming a thick gel. Viscous soluble bulk reduces reabsorption of bile acids from the ileum. Consequently, the liver shifts sterol metabolism from cholesterol production to replacement of bile acids lost in the stool. Fecal excretion of total bile acids increases in conjunction with serum cholesterol reduction when diets high in viscous soluble fiber are eaten.^12-14

Moreover, when viscous soluble fiber is fermented by colonic microflora, it produces short-chain fatty acids that directly inhibit intestinal fat absorption, and cholesterol transport and synthesis.¹⁵ Viscous soluble fiber may delay the absorption of simple sugars from the gut, thus decreasing postprandial glucose level and the demand for insulin secretion, and increasing insulin sensitivity.¹⁶

Viscous soluble fibers reduce CVD risk factors: The evidence

The goal of the cholesterol-lowering diet is to prevent CVD events by reducing LDLC, blood pressure, and glucose intolerance. Evidence from observational studies and randomized controlled trials confirms that foods rich in viscous soluble fiber lower blood lipids and arterial blood pressure and enhance insulin sensitivity.

Soluble fiber reduces LDL-C and may allow lower drug doses. In randomized controlled trials, 2 viscous soluble fiber sources at different doses resulted in a modest but clinically significant reduction in LDLC.^17,18 In addition, the reduction was seen in healthy men and women and in those with hypercholesterolemia (TABLE 2).^19-24 A 1% decrease in LDL-C reduces the risk of CVD events by 1%.¹

For those on lipid-lowering drug therapy, supplementation with viscous fiber may help achieve the recommended reduction in LDL-C with lower doses of drugs.²⁵ Moreyra et al²⁵ treated hypercholesterolemic adults with 20 mg of simvastatin plus placebo, 10 mg of simvastatin plus placebo, or 10 mg of simvastatin plus 15 g of psyllium. A significant LDL-C reduction (36%) was observed in the group receiving 10 mg of simvastatin plus psyllium and 20 mg of simvastatin. The authors concluded that psyllium supplementation augmented the lipid-lowering effects of the statin.

Blood pressure decreases, as does the need for medications. In a prospective study of 51,529 male health professionals followed for 5 years, Ascherio et al²⁶ found an inverse association between incidence of hypertension and fruit fiber intake in young men, independent of age and other risk factors. Moderate but significant reductions in 24-hour monitored systolic blood pressure—by 5.9 mm Hg after psyllium²⁷ and by 5 mm Hg after oat administration²⁸—were reported in randomized controlled trials. A 5 mm Hg reduction of systolic blood pressure in the population would result in a 9% reduction in fatal CHD and a 7% decrease in all-cause mortality.²⁹

Eighty-eight patients with a history of essential moderate hypertension treated with medications were randomized to consume 5.4 g/d of beta-glucan in oats or placebo for 12 weeks. Oat consumption reduced the need for blood pressure medications in 73% of participants in the treatment group, compared with 42% in the control group.³⁰

The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure recommends a dietary plan rich in fiber (24 g/1600 kcal/d) from fruits and vegetables among other life style modifications to treat hypertension.²⁹

Insulin sensitivity improves and HbA_1c drops. Foods rich in viscous soluble fiber lower glycemic index compared with reference food in healthy people and in patients with type II diabetes.^31,32 In a randomized trial, barley meal enriched with 1.5 to 15 g of viscous soluble fiber reduced postprandial glucose concentration by 19% to 28% compared with wheat bread, and by 15% to 30% compared with glucose solution.³¹ Consumption of barley meal containing 3.5 g of soluble viscous fiber improved insulin sensitivity as assessed by the standard oral glucose tolerance test; it significantly reduced fasting insulin by 11.5% and postprandial increase in plasma glucose levels by 8.3%, and increased glucose disappearance by 7.5%.³³

In patients with diabetes, postprandial glucose was significantly less by 10.6% after a meal of barley bread compared with control wheat bread.³² Intake of barley bread with 5.2 g of viscous soluble fiber for 12 weeks resulted in 20% reduced postprandial glucose response and in decreased doses of oral hypoglycemic medications.³⁴

In a meta-analysis of 12 clinical trials, diets moderate in carbohydrate intake with increased consumption of foods rich in fiber provided the best control of fasting blood glucose and a significant decrease in glycosylated hemoglobin (HbA_1c) in patients with diabetes.³⁵

TABLE 2
Viscous soluble fiber reduces LDL-cholesterol (results of RCTs)

CVD events decline with soluble-fiber diets, and diverse fiber sources yield equivalent effects

TABLE 3 summarizes findings from prospective observational studies of adults followed for more than 5 years to evaluate the incidence of CVD in groups consuming different kinds of dietary fiber. These studies show that risk reduction is consistent in different populations and with different sources of dietary fiber,^4,36-38 regardless of patient age, gender, treatment status, smoking, and other risk factors for CVD.

In a pooled analysis of 10 large cohort studies, an additional intake of 10 g/d of fruit fiber reduced the risk of CHD death by 30%.⁴

The Women’s Health Study followed health professionals for 6 years and reported that an increase of 4.9 g/d of viscous soluble fiber was associated with a 44% reduction in myocardial infarction incidents.³⁸

Among participants in the National Health and Nutrition Examination Survey (NHANES) I Follow-up Study, each 5 g/d increase in viscous soluble fiber intake diminished the risk of CVD by 6% and coronary heart disease by 8%.³⁷

Estimated risk reduction is adjusted for all known risk factors of CVD; therefore it is independent of Framingham baseline CHD risk.

For another perspective on the evidence for fiber preventing CVD events, consider the number needed to treat (NNT) with increased fiber intake to prevent the particular CVD-related patient outcome for each study (TABLE 3). Based on the findings of Pereira and colleagues’ pooled cohort study,⁴ among persons at less than 5% risk for CHD according to the Framingham baseline,¹ 67 subjects (95% CI, 44–182) need to increase their fruit fiber intake by 10 g/d to prevent 1 CHD death. For persons at baseline CHD risk greater than 20%, the NNT is reduced to 17 (95% CI, 11–45). In the Women’s Health Study, an additional 4.9 g/d of viscous soluble fiber intake among 45 women (95% CI, 31–167) was estimated to prevent 1 CHD event.³⁶

TABLE 3
Increased daily fiber consumption, regardless of source, reduces risk for cardiovascular disease

What should patients eat to increase viscous soluble fiber?

The 2005 Dietary Guidelines for Americans recommend a daily intake of 14 g of total dietary fiber for every 1000 kcal consumed.² This would include 2.5 to 6.5 cups of fruits and vegetables and at least three 1-ounce servings of whole-grain foods daily.^39-42 Common foods and their dietary total and soluble fiber content per serving are listed in FIGURE 1, which is a modified nutrition prescription from the American Academy of Family Physicians (familydoctor.org/298.xml) incorporating the recommendations for viscous soluble fiber intake from guidelines discussed in this paper.^1-3,6

Four servings per day of the FDA-recommended whole oats (0.75 g per serving of viscous soluble fiber) and psyllium (1.78 g per serving) are required to reduce CVD risk.^10,11

An estimated 23% of Americans eat 2 to 3 servings a day of whole grains and 4 to 5 servings a day of fruits and vegetables (all sources of viscous soluble fiber). Healthy People 2010 has called for this proportion to increase to 50% to 75%.⁶

FIGURE 1
Common foods high in fiber for a nutrition prescription*

Ways to approach dietary counseling

The USPSTF found good evidence that, for adult patients at increased risk for diet-related chronic disease, medium- to high-intensity counseling improves the average daily intake of core components of a healthy diet (including saturated fat, fiber, fruit, and vegetables).⁷ It recommends that clinicians routinely provide intensive behavioral counseling to promote a healthy diet in adult patients with hyperlipidemia and other known risk factors for cardiovascular diseases. It found no evidence sufficient to recommend for or against routine counseling of unselected patients in primary care.

Both USPSTF⁷ and Adult Treatment Panel III¹ offer frameworks to engage patients in therapeutic lifestyle changes.

USPSTF recommends the “5-A” framework to behavioral counseling⁷:

Assess dietary practices and related risk factors

Advise to change dietary practices

Agree on individual diet change goals

Assist to change dietary practices

Arrange regular follow-up and support for refer to more intensive behavioral nutritional counseling (eg, medical nutrition therapy) if needed.

Physicians may use face-to-face and interactive health communications, staff-performed education, and nutritional prescriptions (FIGURE 1) to influence patients’ food choices that increase dietary fiber consumption and achieve the doses recommended for adults at risk of CVD.^7,43

Low-intensity interventions involve 5 minutes of professional counseling supplemented by patient self-help materials, telephone counseling, or other interactive health communications via Internet.^44,45

Medium-intensity interventions involve face-to-face dietary counseling (2 to 3 group or individual sessions) delivered by a dietitian or a specially trained primary care physician.

High-intensity interventions yield the largest effect. These involve multiple sessions lasting 30 minutes or longer⁷ and use office-level supports and algorithms to estimate the baseline risk of CVD, individual calorie and nutrients intake, and energy expenditure.⁴⁵

The Adult Treatment Panel III outlines a model for initiating and monitoring dietary therapy for adults with risk of CVD (FIGURE 2). You may start by assessing the risk of CVD with Framingham scores and current dietary patterns, then counsel patients directly or engage dietitians to help patients institute therapeutic lifestyle changes. After 6 weeks of the therapeutic diet, repeat the lipoprotein examination to determine recommendations to continue or intensify the lifestyle changes. After another 6 weeks, you may consider adding drug therapy and further reinforcing dietary recommendations.

Regardless of the approach you use to assess a patient’s diet, a comprehensive individualized nutritional prescription will effectively increase daily intake of dietary fiber, with success similar to that achieved in a major randomized clinical trial.⁴⁶

FIGURE 2
ATP III guidance on initiating dietary changes: Determine your patient’s 10-year risk for CVD

Acknowledgments

The authors thank Heather Haley, MS for editing the manuscript, and Elizabeth Greene, for her help in its preparation. Dr. Shamliyan thanks Professor E. B. Arushanian, MD, PhD, Department of Pharmacology, Stavropol Medical Academy, Russia for motivating her enthusiasm for scientific research.

CORRESPONDENCE
Joseph M. Keenan, MD, Department of Family Medicine and Community Health, University of Minnesota Medical School, MMC 381, 420 Delaware Street SE, Minneapolis, MN 55455. E-mail: [email protected]

The Adult Treatment Panel III and the Dietary Guidelines for Americans recommend that adults eat a healthy diet as defined in TABLE 1—including more than 20 to 30 g of dietary fiber per day, 5 to 10 g of which should be viscous soluble fiber.^1,2,3 The benefits of such a diet are supported by growing epidemiologic data:

• Risk of CVD drops by 20% to 30%
• Morbidity and mortality from cardiovascular diseases fall by more than 10%
• Foods rich in viscous soluble fiber, at a dose of 5 to 10 g/d, can reduce LDL-C levels by 10% to 15%, improve insulin sensitivity, and decrease blood pressure
• Pooled analysis of 10 prospective cohort studies showed that, for adults at low risk of heart attack (estimated with Framingham scores), every additional 10 g of fiber consumed daily may prevent 1 coronary death in 74 adults (95% confidence interval [CI], 51–154), and 1 in 19 adults at high risk (95% CI, 13–38).⁴

Despite these hopeful findings, the average American adult eats less than half the recommended amount of fiber: 10 to 15 g/d; just 2.4 g of viscous soluble fiber.^2,5Healthy People 2010 has set a national objective to reduce coronary heart disease (CHD) mortality from 208 deaths to 166 deaths per 100,000 people.⁶ The US Preventive Services Task Force (USPSTF) recommends that primary care clinicians intensively counsel adults with elevated risk of CVD to improve daily intake of essential components of a healthy diet, including dietary fiber.⁷

In this article, we review practical approaches you can use in motivating patients to eat more fiber, types of dietary fiber and their sources, and the evidence for fiber’s impact on CVD risk factors and prevention of CVD events.

TABLE 1
Recommended ATP-III dietary change reduces risk of CVD by 20% to 30%

Making sure patients get the right dietary fiber

Dietary fiber (nondigestible non-starch polysaccharides) is classified as insoluble or viscous soluble, based on its solubility in water and other fluids in the digestive tract.⁸

Insoluble fibers, present in most fruits, vegetables, and whole grain products, constitute over 80% of the total fiber in the typical US diet. Insoluble fibers shorten bowel transit time and produce a laxative effect but have no significant impact on blood lipids.⁹

Viscous soluble fiber is present in smaller quantities in foods, but is responsible for the fiber-related physiological effects such as decreased blood cholesterol.⁵ The US Food and Drug Administration (FDA) has also approved health claims for 2 such fibers—beta-glucan from whole oat sources and psyllium—to be used as supplements in reducing CVD risk.^10,11

Viscous soluble fiber absorbs and holds water in the gastrointestinal tract, forming a thick gel. Viscous soluble bulk reduces reabsorption of bile acids from the ileum. Consequently, the liver shifts sterol metabolism from cholesterol production to replacement of bile acids lost in the stool. Fecal excretion of total bile acids increases in conjunction with serum cholesterol reduction when diets high in viscous soluble fiber are eaten.^12-14

Moreover, when viscous soluble fiber is fermented by colonic microflora, it produces short-chain fatty acids that directly inhibit intestinal fat absorption, and cholesterol transport and synthesis.¹⁵ Viscous soluble fiber may delay the absorption of simple sugars from the gut, thus decreasing postprandial glucose level and the demand for insulin secretion, and increasing insulin sensitivity.¹⁶

Viscous soluble fibers reduce CVD risk factors: The evidence

The goal of the cholesterol-lowering diet is to prevent CVD events by reducing LDLC, blood pressure, and glucose intolerance. Evidence from observational studies and randomized controlled trials confirms that foods rich in viscous soluble fiber lower blood lipids and arterial blood pressure and enhance insulin sensitivity.

Soluble fiber reduces LDL-C and may allow lower drug doses. In randomized controlled trials, 2 viscous soluble fiber sources at different doses resulted in a modest but clinically significant reduction in LDLC.^17,18 In addition, the reduction was seen in healthy men and women and in those with hypercholesterolemia (TABLE 2).^19-24 A 1% decrease in LDL-C reduces the risk of CVD events by 1%.¹

For those on lipid-lowering drug therapy, supplementation with viscous fiber may help achieve the recommended reduction in LDL-C with lower doses of drugs.²⁵ Moreyra et al²⁵ treated hypercholesterolemic adults with 20 mg of simvastatin plus placebo, 10 mg of simvastatin plus placebo, or 10 mg of simvastatin plus 15 g of psyllium. A significant LDL-C reduction (36%) was observed in the group receiving 10 mg of simvastatin plus psyllium and 20 mg of simvastatin. The authors concluded that psyllium supplementation augmented the lipid-lowering effects of the statin.

Blood pressure decreases, as does the need for medications. In a prospective study of 51,529 male health professionals followed for 5 years, Ascherio et al²⁶ found an inverse association between incidence of hypertension and fruit fiber intake in young men, independent of age and other risk factors. Moderate but significant reductions in 24-hour monitored systolic blood pressure—by 5.9 mm Hg after psyllium²⁷ and by 5 mm Hg after oat administration²⁸—were reported in randomized controlled trials. A 5 mm Hg reduction of systolic blood pressure in the population would result in a 9% reduction in fatal CHD and a 7% decrease in all-cause mortality.²⁹

Eighty-eight patients with a history of essential moderate hypertension treated with medications were randomized to consume 5.4 g/d of beta-glucan in oats or placebo for 12 weeks. Oat consumption reduced the need for blood pressure medications in 73% of participants in the treatment group, compared with 42% in the control group.³⁰

The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure recommends a dietary plan rich in fiber (24 g/1600 kcal/d) from fruits and vegetables among other life style modifications to treat hypertension.²⁹

Insulin sensitivity improves and HbA_1c drops. Foods rich in viscous soluble fiber lower glycemic index compared with reference food in healthy people and in patients with type II diabetes.^31,32 In a randomized trial, barley meal enriched with 1.5 to 15 g of viscous soluble fiber reduced postprandial glucose concentration by 19% to 28% compared with wheat bread, and by 15% to 30% compared with glucose solution.³¹ Consumption of barley meal containing 3.5 g of soluble viscous fiber improved insulin sensitivity as assessed by the standard oral glucose tolerance test; it significantly reduced fasting insulin by 11.5% and postprandial increase in plasma glucose levels by 8.3%, and increased glucose disappearance by 7.5%.³³

In patients with diabetes, postprandial glucose was significantly less by 10.6% after a meal of barley bread compared with control wheat bread.³² Intake of barley bread with 5.2 g of viscous soluble fiber for 12 weeks resulted in 20% reduced postprandial glucose response and in decreased doses of oral hypoglycemic medications.³⁴

In a meta-analysis of 12 clinical trials, diets moderate in carbohydrate intake with increased consumption of foods rich in fiber provided the best control of fasting blood glucose and a significant decrease in glycosylated hemoglobin (HbA_1c) in patients with diabetes.³⁵

TABLE 2
Viscous soluble fiber reduces LDL-cholesterol (results of RCTs)

CVD events decline with soluble-fiber diets, and diverse fiber sources yield equivalent effects

TABLE 3 summarizes findings from prospective observational studies of adults followed for more than 5 years to evaluate the incidence of CVD in groups consuming different kinds of dietary fiber. These studies show that risk reduction is consistent in different populations and with different sources of dietary fiber,^4,36-38 regardless of patient age, gender, treatment status, smoking, and other risk factors for CVD.

In a pooled analysis of 10 large cohort studies, an additional intake of 10 g/d of fruit fiber reduced the risk of CHD death by 30%.⁴

The Women’s Health Study followed health professionals for 6 years and reported that an increase of 4.9 g/d of viscous soluble fiber was associated with a 44% reduction in myocardial infarction incidents.³⁸

Among participants in the National Health and Nutrition Examination Survey (NHANES) I Follow-up Study, each 5 g/d increase in viscous soluble fiber intake diminished the risk of CVD by 6% and coronary heart disease by 8%.³⁷

Estimated risk reduction is adjusted for all known risk factors of CVD; therefore it is independent of Framingham baseline CHD risk.

For another perspective on the evidence for fiber preventing CVD events, consider the number needed to treat (NNT) with increased fiber intake to prevent the particular CVD-related patient outcome for each study (TABLE 3). Based on the findings of Pereira and colleagues’ pooled cohort study,⁴ among persons at less than 5% risk for CHD according to the Framingham baseline,¹ 67 subjects (95% CI, 44–182) need to increase their fruit fiber intake by 10 g/d to prevent 1 CHD death. For persons at baseline CHD risk greater than 20%, the NNT is reduced to 17 (95% CI, 11–45). In the Women’s Health Study, an additional 4.9 g/d of viscous soluble fiber intake among 45 women (95% CI, 31–167) was estimated to prevent 1 CHD event.³⁶

TABLE 3
Increased daily fiber consumption, regardless of source, reduces risk for cardiovascular disease

What should patients eat to increase viscous soluble fiber?

The 2005 Dietary Guidelines for Americans recommend a daily intake of 14 g of total dietary fiber for every 1000 kcal consumed.² This would include 2.5 to 6.5 cups of fruits and vegetables and at least three 1-ounce servings of whole-grain foods daily.^39-42 Common foods and their dietary total and soluble fiber content per serving are listed in FIGURE 1, which is a modified nutrition prescription from the American Academy of Family Physicians (familydoctor.org/298.xml) incorporating the recommendations for viscous soluble fiber intake from guidelines discussed in this paper.^1-3,6

Four servings per day of the FDA-recommended whole oats (0.75 g per serving of viscous soluble fiber) and psyllium (1.78 g per serving) are required to reduce CVD risk.^10,11

An estimated 23% of Americans eat 2 to 3 servings a day of whole grains and 4 to 5 servings a day of fruits and vegetables (all sources of viscous soluble fiber). Healthy People 2010 has called for this proportion to increase to 50% to 75%.⁶

FIGURE 1
Common foods high in fiber for a nutrition prescription*

Ways to approach dietary counseling

The USPSTF found good evidence that, for adult patients at increased risk for diet-related chronic disease, medium- to high-intensity counseling improves the average daily intake of core components of a healthy diet (including saturated fat, fiber, fruit, and vegetables).⁷ It recommends that clinicians routinely provide intensive behavioral counseling to promote a healthy diet in adult patients with hyperlipidemia and other known risk factors for cardiovascular diseases. It found no evidence sufficient to recommend for or against routine counseling of unselected patients in primary care.

Both USPSTF⁷ and Adult Treatment Panel III¹ offer frameworks to engage patients in therapeutic lifestyle changes.

USPSTF recommends the “5-A” framework to behavioral counseling⁷:

Assess dietary practices and related risk factors

Advise to change dietary practices

Agree on individual diet change goals

Assist to change dietary practices

Arrange regular follow-up and support for refer to more intensive behavioral nutritional counseling (eg, medical nutrition therapy) if needed.

Physicians may use face-to-face and interactive health communications, staff-performed education, and nutritional prescriptions (FIGURE 1) to influence patients’ food choices that increase dietary fiber consumption and achieve the doses recommended for adults at risk of CVD.^7,43

Low-intensity interventions involve 5 minutes of professional counseling supplemented by patient self-help materials, telephone counseling, or other interactive health communications via Internet.^44,45

Medium-intensity interventions involve face-to-face dietary counseling (2 to 3 group or individual sessions) delivered by a dietitian or a specially trained primary care physician.

High-intensity interventions yield the largest effect. These involve multiple sessions lasting 30 minutes or longer⁷ and use office-level supports and algorithms to estimate the baseline risk of CVD, individual calorie and nutrients intake, and energy expenditure.⁴⁵

The Adult Treatment Panel III outlines a model for initiating and monitoring dietary therapy for adults with risk of CVD (FIGURE 2). You may start by assessing the risk of CVD with Framingham scores and current dietary patterns, then counsel patients directly or engage dietitians to help patients institute therapeutic lifestyle changes. After 6 weeks of the therapeutic diet, repeat the lipoprotein examination to determine recommendations to continue or intensify the lifestyle changes. After another 6 weeks, you may consider adding drug therapy and further reinforcing dietary recommendations.

Regardless of the approach you use to assess a patient’s diet, a comprehensive individualized nutritional prescription will effectively increase daily intake of dietary fiber, with success similar to that achieved in a major randomized clinical trial.⁴⁶

FIGURE 2
ATP III guidance on initiating dietary changes: Determine your patient’s 10-year risk for CVD

Acknowledgments

The authors thank Heather Haley, MS for editing the manuscript, and Elizabeth Greene, for her help in its preparation. Dr. Shamliyan thanks Professor E. B. Arushanian, MD, PhD, Department of Pharmacology, Stavropol Medical Academy, Russia for motivating her enthusiasm for scientific research.

CORRESPONDENCE
Joseph M. Keenan, MD, Department of Family Medicine and Community Health, University of Minnesota Medical School, MMC 381, 420 Delaware Street SE, Minneapolis, MN 55455. E-mail: [email protected]

The Adult Treatment Panel III and the Dietary Guidelines for Americans recommend that adults eat a healthy diet as defined in TABLE 1—including more than 20 to 30 g of dietary fiber per day, 5 to 10 g of which should be viscous soluble fiber.^1,2,3 The benefits of such a diet are supported by growing epidemiologic data:

• Risk of CVD drops by 20% to 30%
• Morbidity and mortality from cardiovascular diseases fall by more than 10%
• Foods rich in viscous soluble fiber, at a dose of 5 to 10 g/d, can reduce LDL-C levels by 10% to 15%, improve insulin sensitivity, and decrease blood pressure
• Pooled analysis of 10 prospective cohort studies showed that, for adults at low risk of heart attack (estimated with Framingham scores), every additional 10 g of fiber consumed daily may prevent 1 coronary death in 74 adults (95% confidence interval [CI], 51–154), and 1 in 19 adults at high risk (95% CI, 13–38).⁴

Despite these hopeful findings, the average American adult eats less than half the recommended amount of fiber: 10 to 15 g/d; just 2.4 g of viscous soluble fiber.^2,5Healthy People 2010 has set a national objective to reduce coronary heart disease (CHD) mortality from 208 deaths to 166 deaths per 100,000 people.⁶ The US Preventive Services Task Force (USPSTF) recommends that primary care clinicians intensively counsel adults with elevated risk of CVD to improve daily intake of essential components of a healthy diet, including dietary fiber.⁷

In this article, we review practical approaches you can use in motivating patients to eat more fiber, types of dietary fiber and their sources, and the evidence for fiber’s impact on CVD risk factors and prevention of CVD events.

TABLE 1
Recommended ATP-III dietary change reduces risk of CVD by 20% to 30%

Making sure patients get the right dietary fiber

Dietary fiber (nondigestible non-starch polysaccharides) is classified as insoluble or viscous soluble, based on its solubility in water and other fluids in the digestive tract.⁸

Insoluble fibers, present in most fruits, vegetables, and whole grain products, constitute over 80% of the total fiber in the typical US diet. Insoluble fibers shorten bowel transit time and produce a laxative effect but have no significant impact on blood lipids.⁹

Viscous soluble fiber is present in smaller quantities in foods, but is responsible for the fiber-related physiological effects such as decreased blood cholesterol.⁵ The US Food and Drug Administration (FDA) has also approved health claims for 2 such fibers—beta-glucan from whole oat sources and psyllium—to be used as supplements in reducing CVD risk.^10,11

Viscous soluble fiber absorbs and holds water in the gastrointestinal tract, forming a thick gel. Viscous soluble bulk reduces reabsorption of bile acids from the ileum. Consequently, the liver shifts sterol metabolism from cholesterol production to replacement of bile acids lost in the stool. Fecal excretion of total bile acids increases in conjunction with serum cholesterol reduction when diets high in viscous soluble fiber are eaten.^12-14

Moreover, when viscous soluble fiber is fermented by colonic microflora, it produces short-chain fatty acids that directly inhibit intestinal fat absorption, and cholesterol transport and synthesis.¹⁵ Viscous soluble fiber may delay the absorption of simple sugars from the gut, thus decreasing postprandial glucose level and the demand for insulin secretion, and increasing insulin sensitivity.¹⁶

Viscous soluble fibers reduce CVD risk factors: The evidence

The goal of the cholesterol-lowering diet is to prevent CVD events by reducing LDLC, blood pressure, and glucose intolerance. Evidence from observational studies and randomized controlled trials confirms that foods rich in viscous soluble fiber lower blood lipids and arterial blood pressure and enhance insulin sensitivity.

Soluble fiber reduces LDL-C and may allow lower drug doses. In randomized controlled trials, 2 viscous soluble fiber sources at different doses resulted in a modest but clinically significant reduction in LDLC.^17,18 In addition, the reduction was seen in healthy men and women and in those with hypercholesterolemia (TABLE 2).^19-24 A 1% decrease in LDL-C reduces the risk of CVD events by 1%.¹

For those on lipid-lowering drug therapy, supplementation with viscous fiber may help achieve the recommended reduction in LDL-C with lower doses of drugs.²⁵ Moreyra et al²⁵ treated hypercholesterolemic adults with 20 mg of simvastatin plus placebo, 10 mg of simvastatin plus placebo, or 10 mg of simvastatin plus 15 g of psyllium. A significant LDL-C reduction (36%) was observed in the group receiving 10 mg of simvastatin plus psyllium and 20 mg of simvastatin. The authors concluded that psyllium supplementation augmented the lipid-lowering effects of the statin.

Blood pressure decreases, as does the need for medications. In a prospective study of 51,529 male health professionals followed for 5 years, Ascherio et al²⁶ found an inverse association between incidence of hypertension and fruit fiber intake in young men, independent of age and other risk factors. Moderate but significant reductions in 24-hour monitored systolic blood pressure—by 5.9 mm Hg after psyllium²⁷ and by 5 mm Hg after oat administration²⁸—were reported in randomized controlled trials. A 5 mm Hg reduction of systolic blood pressure in the population would result in a 9% reduction in fatal CHD and a 7% decrease in all-cause mortality.²⁹

Eighty-eight patients with a history of essential moderate hypertension treated with medications were randomized to consume 5.4 g/d of beta-glucan in oats or placebo for 12 weeks. Oat consumption reduced the need for blood pressure medications in 73% of participants in the treatment group, compared with 42% in the control group.³⁰

The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure recommends a dietary plan rich in fiber (24 g/1600 kcal/d) from fruits and vegetables among other life style modifications to treat hypertension.²⁹

Insulin sensitivity improves and HbA_1c drops. Foods rich in viscous soluble fiber lower glycemic index compared with reference food in healthy people and in patients with type II diabetes.^31,32 In a randomized trial, barley meal enriched with 1.5 to 15 g of viscous soluble fiber reduced postprandial glucose concentration by 19% to 28% compared with wheat bread, and by 15% to 30% compared with glucose solution.³¹ Consumption of barley meal containing 3.5 g of soluble viscous fiber improved insulin sensitivity as assessed by the standard oral glucose tolerance test; it significantly reduced fasting insulin by 11.5% and postprandial increase in plasma glucose levels by 8.3%, and increased glucose disappearance by 7.5%.³³

In patients with diabetes, postprandial glucose was significantly less by 10.6% after a meal of barley bread compared with control wheat bread.³² Intake of barley bread with 5.2 g of viscous soluble fiber for 12 weeks resulted in 20% reduced postprandial glucose response and in decreased doses of oral hypoglycemic medications.³⁴

In a meta-analysis of 12 clinical trials, diets moderate in carbohydrate intake with increased consumption of foods rich in fiber provided the best control of fasting blood glucose and a significant decrease in glycosylated hemoglobin (HbA_1c) in patients with diabetes.³⁵

TABLE 2
Viscous soluble fiber reduces LDL-cholesterol (results of RCTs)

CVD events decline with soluble-fiber diets, and diverse fiber sources yield equivalent effects

TABLE 3 summarizes findings from prospective observational studies of adults followed for more than 5 years to evaluate the incidence of CVD in groups consuming different kinds of dietary fiber. These studies show that risk reduction is consistent in different populations and with different sources of dietary fiber,^4,36-38 regardless of patient age, gender, treatment status, smoking, and other risk factors for CVD.

In a pooled analysis of 10 large cohort studies, an additional intake of 10 g/d of fruit fiber reduced the risk of CHD death by 30%.⁴

The Women’s Health Study followed health professionals for 6 years and reported that an increase of 4.9 g/d of viscous soluble fiber was associated with a 44% reduction in myocardial infarction incidents.³⁸

Among participants in the National Health and Nutrition Examination Survey (NHANES) I Follow-up Study, each 5 g/d increase in viscous soluble fiber intake diminished the risk of CVD by 6% and coronary heart disease by 8%.³⁷

Estimated risk reduction is adjusted for all known risk factors of CVD; therefore it is independent of Framingham baseline CHD risk.

For another perspective on the evidence for fiber preventing CVD events, consider the number needed to treat (NNT) with increased fiber intake to prevent the particular CVD-related patient outcome for each study (TABLE 3). Based on the findings of Pereira and colleagues’ pooled cohort study,⁴ among persons at less than 5% risk for CHD according to the Framingham baseline,¹ 67 subjects (95% CI, 44–182) need to increase their fruit fiber intake by 10 g/d to prevent 1 CHD death. For persons at baseline CHD risk greater than 20%, the NNT is reduced to 17 (95% CI, 11–45). In the Women’s Health Study, an additional 4.9 g/d of viscous soluble fiber intake among 45 women (95% CI, 31–167) was estimated to prevent 1 CHD event.³⁶

TABLE 3
Increased daily fiber consumption, regardless of source, reduces risk for cardiovascular disease

What should patients eat to increase viscous soluble fiber?

The 2005 Dietary Guidelines for Americans recommend a daily intake of 14 g of total dietary fiber for every 1000 kcal consumed.² This would include 2.5 to 6.5 cups of fruits and vegetables and at least three 1-ounce servings of whole-grain foods daily.^39-42 Common foods and their dietary total and soluble fiber content per serving are listed in FIGURE 1, which is a modified nutrition prescription from the American Academy of Family Physicians (familydoctor.org/298.xml) incorporating the recommendations for viscous soluble fiber intake from guidelines discussed in this paper.^1-3,6

Four servings per day of the FDA-recommended whole oats (0.75 g per serving of viscous soluble fiber) and psyllium (1.78 g per serving) are required to reduce CVD risk.^10,11

An estimated 23% of Americans eat 2 to 3 servings a day of whole grains and 4 to 5 servings a day of fruits and vegetables (all sources of viscous soluble fiber). Healthy People 2010 has called for this proportion to increase to 50% to 75%.⁶

FIGURE 1
Common foods high in fiber for a nutrition prescription*

Ways to approach dietary counseling

The USPSTF found good evidence that, for adult patients at increased risk for diet-related chronic disease, medium- to high-intensity counseling improves the average daily intake of core components of a healthy diet (including saturated fat, fiber, fruit, and vegetables).⁷ It recommends that clinicians routinely provide intensive behavioral counseling to promote a healthy diet in adult patients with hyperlipidemia and other known risk factors for cardiovascular diseases. It found no evidence sufficient to recommend for or against routine counseling of unselected patients in primary care.

Both USPSTF⁷ and Adult Treatment Panel III¹ offer frameworks to engage patients in therapeutic lifestyle changes.

USPSTF recommends the “5-A” framework to behavioral counseling⁷:

Assess dietary practices and related risk factors

Advise to change dietary practices

Agree on individual diet change goals

Assist to change dietary practices

Arrange regular follow-up and support for refer to more intensive behavioral nutritional counseling (eg, medical nutrition therapy) if needed.

Physicians may use face-to-face and interactive health communications, staff-performed education, and nutritional prescriptions (FIGURE 1) to influence patients’ food choices that increase dietary fiber consumption and achieve the doses recommended for adults at risk of CVD.^7,43

Low-intensity interventions involve 5 minutes of professional counseling supplemented by patient self-help materials, telephone counseling, or other interactive health communications via Internet.^44,45

Medium-intensity interventions involve face-to-face dietary counseling (2 to 3 group or individual sessions) delivered by a dietitian or a specially trained primary care physician.

High-intensity interventions yield the largest effect. These involve multiple sessions lasting 30 minutes or longer⁷ and use office-level supports and algorithms to estimate the baseline risk of CVD, individual calorie and nutrients intake, and energy expenditure.⁴⁵

The Adult Treatment Panel III outlines a model for initiating and monitoring dietary therapy for adults with risk of CVD (FIGURE 2). You may start by assessing the risk of CVD with Framingham scores and current dietary patterns, then counsel patients directly or engage dietitians to help patients institute therapeutic lifestyle changes. After 6 weeks of the therapeutic diet, repeat the lipoprotein examination to determine recommendations to continue or intensify the lifestyle changes. After another 6 weeks, you may consider adding drug therapy and further reinforcing dietary recommendations.

Regardless of the approach you use to assess a patient’s diet, a comprehensive individualized nutritional prescription will effectively increase daily intake of dietary fiber, with success similar to that achieved in a major randomized clinical trial.⁴⁶

FIGURE 2
ATP III guidance on initiating dietary changes: Determine your patient’s 10-year risk for CVD

Acknowledgments

The authors thank Heather Haley, MS for editing the manuscript, and Elizabeth Greene, for her help in its preparation. Dr. Shamliyan thanks Professor E. B. Arushanian, MD, PhD, Department of Pharmacology, Stavropol Medical Academy, Russia for motivating her enthusiasm for scientific research.

CORRESPONDENCE
Joseph M. Keenan, MD, Department of Family Medicine and Community Health, University of Minnesota Medical School, MMC 381, 420 Delaware Street SE, Minneapolis, MN 55455. E-mail: [email protected]

Though hyaluronic acid may reduce symptoms related to osteoarthritis of the knee, the relatively small and transient response in the population studied in our analysis does not provide sufficient reason to recommend or not recommend this therapy.

Those who might want to opt for hyaluronic acid injections

With the relatively low risk of complications, some patients may still opt to try hyaluronic acid injections as opposed to other osteoarthritis management strategies. Potential candidates include those whose only other option is surgery, in the hope that HA injection might postpone having to make that decision. Also, those whose pain or stiffness have not been relieved with other therapies might want to consider HA.

Two instruments for assessing osteoarthritis

There are many claims to the efficacy of hyaluronic acid injections for decreasing the pain associated with osteoarthritis of the knee. This meta-analysis was an attempt to collapse the data for hyaluronic acid treatment, using a reduction in score on the Western Ontario McMaster Universities Index (WOMAC) or the Lequesne index as its outcome measurement.

The WOMAC is a disease-specific, self-administered instrument for patients with osteoarthritis of the knee or hip. It has 3 separate dimensions (with 24 individual scenarios), measuring pain (5 scenarios), stiffness (2 scenarios), and physical function (17 scenarios). It may be administered using a 100 mm/10 cm visual analog scale (VAS) (where 0=none, 100 or 10=extreme) or a Likert scale (0 to 4, where 0=none, 4=extreme). These results are then scored on a 0 to 20 scale for pain, a 0 to 8 scale for stiffness, and a 0 to 68 scale for physical function. Lower scores for both scales indicate a lesser degree of pain, stiffness, or physical dysfunction. In a double-blind, randomized controlled trial, WOMAC was found to be a valid and reliable tool for determining self-reported status for osteoarthritis of the knee or hip.¹²

The Lequesne index is a 10-question interview-style survey given to patients with osteoarthritis of the knee. It has 5 questions pertaining to pain or discomfort, 1 question dealing with maximum distance walked, and 4 questions about activities of daily living. The total questionnaire is scored on a 0 to 24 scale, with lower scores meaning less functional impairment. A study by Faucher et al¹³ found the Lequesne index to be a reliable questionnaire.

Methods

Selection of studies

Two researchers (JM and AT) performed a computerized literature search of PubMed (1950–2004), CINAHL (1982–2004), and Medline (1966–2004) to identify citations concerning the efficacy of hyaluronic acid injection for management of osteoarthritis of the knee. Four separate searches were conducted. The first used the terms knee, osteoarthritis, WOMAC, and hyaluronic acid. The second used the same terms as the first, replacing WOMAC with Lequesne. The third and fourth searched WOMAC and validity and Lequesne and validity, respectively. All 4 searches were limited to human randomized clinical trials, in English-language journal reports. A hand search of the reference lists of all retrieved studies was performed to ensure that no eligible studies were excluded.

Studies were selected independently by the same 2 researchers. The search was performed independently to ensure an exhaustive review of the literature. All studies were considered eligible until disqualified based on exclusion criteria.

Studies were eligible for inclusion if they addressed hyaluronic acid injection for osteoarthritis at the knee and used the WOMAC or Lequesne indexes as outcomes measurements. It was also necessary that they provided means and standard deviations in order to perform statistical analysis. We attempted to contact authors who did not provide necessary statistics for meta-analysis; however, we received no responses.

Assessment of methodological quality and data abstraction

The methodological quality of each study was assessed independently by the reviewers using the Physiological Evidence Database (PEDro) rating scale.¹⁴ It was determined that studies must include a control group that used placebo saline injections, provide means and standard deviations at baseline for the WOMAC or Lequesne, and also means and standard deviations after the intervention for both the treatment and control groups.

All abstracted data were converted into a percentage of the total possible score for each outcome measurement using a method described by a statistical consultant and an algorithm developed by one investigator (JM) using MATLAB 7.0 (MathWorks, Inc, Natick, Mass). This allowed for comparison between results of the WOMAC and Lequesne studies. This also allowed for comparison to the results of the Wang et al¹⁵ meta-analysis.

Statistical analysis

Homogeneity of the included studies, as determined by inclusion and exclusion criteria, allowed for meta-analysis. One meta-analysis was performed on the 3 studies that used the WOMAC scale,^16-18 2 on the 4 studies that used the Lequesne index.^7-9,19 Most studies that used the Lequesne index reported outcomes for a variety of follow-up points (4, 5, 8, 20, 24, 26, 49, 52 weeks); however, there was not a consistent reporting timeline among these studies. We grouped the follow-up reports into “up to 6 months post-treatment” and “6 months or greater post-treatment.” The first Lequesne analysis evaluated the difference between treatment and placebo groups at with data points less than 6 months post-treatment. The second Lequesne analysis examined the difference between groups at 6 months and greater post-treatment. In the study by Karlsson,⁷ 2 different but similar types of hyaluronic acid were used. These groups were separated and analyzed as independent studies; 95% confidence intervals were calculated for all analyses. Effect sizes were also calculated to determine the between-group effect of the treatment and placebo.

Results

Selection of studies

The literature search resulted in 35 potentially eligible studies, of which 7 were selected after application of the inclusion criteria.^7-9,16-19 Three of the selected studies used the WOMAC as an outcome assessment tool,^16-18 4 used the Lequesne index.^7-9,19 The remaining 28 studies were excluded. TABLE W1 details the rejection of studies.

Methodological quality and study characteristics

Each investigator then independently scored the methodological quality for the remaining studies by using the PEDro scale.¹⁴ There was total agreement for all studies in regards to the PEDro score. No study received a score under 5. The average PEDro rating among the studies was 8.2 out of 10.

Homogeneity

All groups across all included studies were similar at baseline. All patients had mild to severe osteoarthritis of the knee. Mean ages of the groups were similar (treatment groups=67.1 years, placebo=66.7 years). Not all patients received treatment as per manufacturer’s recommendations ( TABLES 1 AND 2 ). A saline placebo was used for control groups in all studies. The volume of saline placebo for each study was equal to volume of hyaluronic acid injection plus saline vehicle used for that study. All studies used either the WOMAC scale or the Lequesne Functional Index as an outcome measurement tool. All studies included a follow-up outcome measurement; however, the range of follow-up time periods was wide (4 to 52 weeks).

Meta-analysis

The analysis for the outcome measurements on the WOMAC scale revealed no significant difference between groups in regards to pain or disability (95% confidence interval [CI] for pain, –0.6043 to 5.4755; for disability, –0.8282 to 4.8619). The outcome measurement for stiffness demonstrated a significant difference between treatment and control (95% CI for stiffness, 2.1780 to 8.7955).

The analysis for the outcome measurement on the Lequesne index revealed a significant difference between groups for measurements taken up to 6 months post-treatment (95% CI, 1.2315 to 2.6268). However, these differences were not seen in the 6+ month outcomes analysis (95% CI, –.8489 to .04787]. Confidence intervals are presented in FIGURE 1 . Mean differences for each outcome at baseline and post-treatment between the active treatment and placebo are presented in TABLE 3 .

TABLE 1
Hyaluronic acid: Manufacturer’s recommendations

TABLE 2
Details of included studies

FIGURE 1
Hyaluronic acid vs control: 95% confidence intervals

TABLE 3
Mean improvements: treatment vs. control

FIGURE 2
Comparing effect sizes of hyaluronic acid therapy and saline placebo

Discussion: Benefits are limited

Based on this meta-analysis, we cannot conclude that hyaluronic acid performs better than saline placebo for a reduction of pain or disability on the WOMAC. Some indication may be warranted for reduction in stiffness. It is notable that hyaluronic acid injection and saline placebo groups both experienced an improvement in pain, stiffness, and disability scores on the WOMAC.

We also cannot conclude that hyaluronic acid enhances functionality as evaluated on the Lequesne Index past 6 months post-treatment. In 2 studies,^7,19 patients with placebo actually showed reduced disability compared with patients in the hyaluronic acid group. While a short-term improvement in functionality was observed in other studies, this effect was not seen in follow-up reports greater than 6 months.

Comparisons with other meta-analyses

A recently published meta-analysis by Modawal et al²⁰ also pooled data to compare hyaluronic acid treatment to saline placebo. That meta-analysis was similar to ours in that it had relatively small sample size (n=11) and examined self-reported patient outcomes as the item of interest. The difference between this meta-analysis and ours is that the length of the evaluation periods for the other analysis was fairly short; the longest follow-up was 26 weeks. Our study examined the pooled effects of hyaluronic acid carried out to at least 24 weeks and up to 52 weeks. Their results for the short-term outcomes demonstrated significant decrease in pain in the active treatment group; however, these results diminished considerably when controlled for study quality.

Our meta-analysis was not consistent with a previous meta-analysis of hyaluronic acid treatment with regard to pain relief and reduction of disability. This study¹⁵ revealed that hyaluronic acid has a therapeutic effect on osteoarthritis of the knee, specifically decreasing pain and disability. The analysis included 20 studies, but differed from ours in that it failed to control for between-group heterogeneity and used an original formula to derive mean differences in efficacy scores between hyaluronic acid and placebo. Each study in our meta-analysis used an identical method for reporting outcome scores. Our study also synthesized the original data values as reported in each original study.

In addition, our results have to be interpreted with attention to a large effect of saline injection, which were much greater than the expected effect of placebo. Calculated effect sizes demonstrated similar effects for both treatment and placebo groups, with no clustering that was different between groups ( FIGURE 2 ). Treatment and placebo groups showed equal response when evaluated for greatest effect.

Refocusing research

The effects of hyaluronic acid injection appear to be transient and offer only slight improvement for older patients with mild to advanced osteoarthritis. Attention should be given to other groups, including younger patients and patients with mild degenerative changes. These studies should attempt to better stratify patient groups to identify those who are most likely to benefit from intervention. In addition, there must be better reporting of treatment protocols and results will allow clinicians to make rational decisions regarding treatments.

Cost and side effects

Consider the degree of benefit and costs when making treatment decisions. The average treatment cost was approximately $900 in 2003.²¹ Obviously one issue is whether a patient’s insurance provider will pay for the treatment.

Side effects are uncommon, and complications with hyaluronic acid injections have been relatively mild.

A few cases of hyaluronic acid allergy have been reported,²² including sweating, paleness, feelings of pressure in the chest and stomach, skin turning blue, or a decrease in blood pressure.²²

Acknowledgments

The authors would like to thank Esra Akdeniz from the Statistical Consulting Center at the Pennsylvania State University for assistance with the methods for performing statistical meta-analysis.

CORRESPONDENCE
Jennifer M. Medina, 146 Recreation Building, University Park, PA 16802. E-mail: [email protected]