Clinical Review

Dr. Gariepy reports no financial relationships relevant to this article. Dr. Creinin reports that he is a consultant to, and a speaker for, Schering-Plough.

Progestin-only contraception—a diverse group of oral (progestin-only pills, or so-called minipills), injectable (depot medroxyprogesterone acetate), intrauterine (the levonorgestrel intrauterine system), and implantable (etonogestrel implant) methods—may offer advantages over estrogen-containing contraception:

• the flexibility of distinctive methods of delivery
• the ability to initiate the method in postpartum breastfeeding women
• enhanced safety in women who should not be exposed to exogenous estrogens.

Unpredictable bleeding is a major disadvantage of progestin-only contraception, however, and can cause women to discontinue these methods—and discontinuation without an effective backup method creates a high risk of unplanned pregnancy. The significant variability in bleeding patterns among progestin-only contraceptive methods hinders our ability to counsel patients about them.

Furthermore, the lack of uniform definitions of bleeding patterns with hormonal contraception, including progestin-only methods, makes it difficult to counsel women accurately and compare bleeding patterns among methods.

Accurate prediction of the bleeding patterns associated with progestin-only contraception could lower the discontinuation rate. For example, studies have shown that pretreatment counseling about expected side effects increases approximately fourfold the acceptability and continuation of depot medroxyprogesterone acetate.^1,2

In this Update, we review the data on bleeding patterns associated with progestin-only contraceptives, including the likelihood of 1) amenorrhea and 2) discontinuation due to changes in the bleeding pattern.

We also discuss what has been learned about the treatment of changes in bleeding patterns induced by progestin-only contraception.

Our goal? To summarize the findings in a comprehensive way that makes it easier for you to discuss expected bleeding patterns with your patients—so that women can choose the method of contraception that is the best fit for them.

Describing bleeding patterns is a challenging task

One of the difficulties of interpreting clinical data on bleeding patterns—with any type of contraception—is the lack of a universally accepted standard for collecting and reporting these data. The first suggestions for standardization were made in 1976, when Rodriguez and colleagues proposed using 90-day reference periods for analysis, as a way to minimize variability among individual menstrual cycles.³

Subsequently, the World Health Organization’s (WHO’s) Special Programme of Research, Development and Research Training in Human Reproduction developed recommendations for data collection, terminology, presentation, and data analysis when reporting vaginal bleeding during clinical trials of hormonal contraception. These recommendations became known as the WHO Belsey criteria (TABLE 1). They remain the standard.⁴

Under the WHO Belsey system:

• vaginal blood loss for which a woman uses sanitary protection is classified as bleeding
• vaginal blood loss that does not result in the use of sanitary protection is considered spotting.

This system also specifies indices for evaluating the bleeding pattern for each woman and reference period, including the number of bleeding-spotting days, number of bleeding-spotting episodes, lengths of bleeding-spotting episodes, and bleeding-spotting-free intervals. A bleeding-spotting episode is defined as one or more consecutive days during which blood loss (bleeding or spotting) has been recorded, each episode being bounded by bleeding-spotting-free days. The WHO Belsey criteria also identified subgroups that have “clinically important bleeding patterns” (TABLE 1).

But not all researchers use the WHO Belsey criteria. Many trials use, and report, their own system of analysis. Some researchers have chosen reference periods of other durations and study periods that range from 1 to 5 years. Some studies report bleeding patterns by number of days, and others report the percentage of women experiencing a given bleeding pattern during a reference period. The lack of uniformity results in data that are difficult to compare from one study to the next—and to explain to our patients.

It’s unclear whether any of our research definitions of clinically significant bleeding have ever been validated as clinically important to our patients. Multiple studies do show that changes in menstrual bleeding patterns are a significant cause of dissatisfaction with any given contraceptive method, but we don’t know if the number of days of bleeding-spotting or the predictability of this bleeding-spotting is the critical piece of information we should be relating to our patients.

In other words, do our beliefs about clinically important bleeding patterns reflect women’s beliefs?

TABLE 1

The WHO Belsey system of bleeding patterns

Implantable contraception

The etonogestrel (ENG) implant (Implanon) is the only implantable contraceptive available in the United States. This single-rod contraceptive can be used for as long as 3 years.

Contraceptive implants, including the levonorgestrel implants once sold in the United States and still available in some parts of the world, are highly effective. The Implanon prescribing information reports a first-year failure rate of 0.38 pregnancies for every 100 woman-years of use; Hatcher and co-workers reported a failure rate of 0.⁵ The difference is based on how the FDA defines pregnancy in contraceptive trials. In fact, the only pregnancies reported with the ENG implant happened after it was removed. Importantly, the studies evaluated by the FDA included only women not using any medications known to induce liver metabolism (the cytochrome P450 pathway) and who were between 80% and 130% of ideal body weight. The efficacy of the ENG implant for women who are taking medications that induce liver metabolism or who are greater than 130% of their ideal body weight is unknown.

The efficacy of the ENG implant is likely derived from suppression of ovulation and increased cervical mucus viscosity. Associated changes in the endometrium that occur with this low dosage of progestin are likely the primary cause of irregular and unpredictable bleeding.

Several studies have sought to describe the bleeding patterns experienced with the ENG implant.^6-8 During the first 3 months, approximately 50% of all women using the ENG implant reported bleeding-spotting (TABLE 2) for 30 days, on average (TABLE 3). The number of days decreases to approximately 20 bleeding-spotting days for each 90-day reference period at 6 to 24 months, with wide variability. For example: From 3 to 6 months, women reported 22 days of bleeding-spotting (standard deviation, 20 days); from months 21 to 24, 20 days of bleeding-spotting (standard deviation, 14 days).⁷

After using the ENG implant for 2 years, therefore, most women can expect the number of bleeding-spotting days for every 90-day reference period to range between 6 and 34 days. These days of bleeding-spotting are often noncontinuous, however. On average, women reported three separate bleeding-spotting episodes for every 90-day reference period.⁷

Although individual bleeding patterns are unpredictable, women who had no bleeding, or infrequent bleeding, at the beginning of use of the ENG implant had only a “small chance” of bleeding frequently.⁶ The most common bleeding pattern observed throughout the study was infrequent bleeding, defined as fewer than three episodes of bleeding in a 90-day reference period (excluding amenorrhea).⁷

Amenorrhea may not persist. The amenorrhea rate at 6 months of use and beyond ranges from 10% to 20% (TABLE 4). Importantly, women who are amenorrheic in one 90-day reference period are not necessarily the ones who are amenorrheic in another reference period. So, unlike what is more commonly seen with other progestin-only methods, such as injectables, amenorrhea is not sustained for most women.

This unpredictable pattern affects continuation of the ENG implant (TABLE 5). Irregular bleeding is the most common reason women cite for discontinuation, accounting for 30% to 60% of all women who discontinue early.^7,9

Overall, approximately 4% of ENG users discontinue the method at 1 year. Ten percent to 20% discontinue each year thereafter because of intolerance to bleeding changes.^6-9

There are, however, differences in discontinuation rates across cultures. In an integrated analysis of 13 different trials that evaluated patterns of vaginal bleeding with the ENG implant where the rate of menstrual changes was similar, women from Europe and Canada were much more likely (23%) to discontinue the implant because of those changes than women from Southeast Asia and Chile were (2%).⁶ This finding may reflect differences in cultural beliefs or disparate access to other contraceptive options.¹⁰

TABLE 2

What percentage of women taking progestin-only contraception report bleeding-spotting?

TABLE 3

How many days of bleeding-spotting do women have
when they use progestin-only contraception?

TABLE 4

What percentage of women taking progestin-only contraception report amenorrhea?

TABLE 5

What percentage discontinue progestin-only contraception
because of a change in bleeding pattern?

PERSPECTIVE AND GUIDANCE FOR YOUR PRACTICE

The pattern of bleeding seen with the ENG implant is like the activity of the heart in atrial fibrillation: irregularly irregular. Still, most (80%) women continue to use it beyond 1 year. In fact, the discontinuation rate for the ENG implant is less than that of depot medroxyprogesterone acetate (DMPA) and progestin-only pills.

Most ENG implant users report no difficulty tolerating the associated unpredictable bleeding; it’s possible that they had unpredictable bleeding at baseline, or were drawn to the improvement in their dysmenorrhea.⁶

Importantly, unpredictable bleeding does not affect efficacy; the ENG implant remains one of the most effective long-acting reversible contraceptives. For women who can tolerate unpredictable bleeding, the ENG implant is a highly effective contraceptive option.

Injectable contraception

Approved by the FDA in 1992, DMPA (Depo-Provera) has good efficacy and long-acting protection. Disadvantages include unpredictable bleeding, weight gain, acne, depression, hair loss, and the controversial issue of decreased bone loss with prolonged use.

What are the expected changes in bleeding patterns with DMPA? Women often have unpredictable patterns, with infrequent but prolonged bleeding-spotting episodes.¹¹ The overall incidence of irregular bleeding can be as high as 70% in the first year of use.¹² Irregular bleeding decreases with continued use, to as low as 10% after the first year (TABLE 2).

Although the number of bleeding-spotting days decreases over time, women have reported as many as 10 days of irregular bleeding-spotting between 9 and 12 months of use (TABLE 3). The rates of irregular bleeding and amenorrhea are similar for the subcutaneous formulation of DMPA.¹³

DMPA is often used because of the high likelihood of amenorrhea. However, amenorrhea is not accomplished in most women in a short time. At 3 months of use, 10% to 45% of women report amenorrhea; after 1 year, the rate increases to 40% to 70% (TABLE 4). At 5 years, 80% of women report amenorrhea.¹²

A source of frustration. DMPA’s high discontinuation rate, compared with what is seen with other contraceptives, can be frustrating for patients and clinicians. Irregular bleeding is the most common reason for discontinuation. Approximately 35% of women who start DMPA discontinue it during the first 3 months of use because of irregular bleeding (TABLE 5). The cumulative discontinuation rate rises over time: At 1 year, 40% to 60% of women who started DMPA will have discontinued it because of changes in bleeding patterns (TABLE 5). Furthermore, 70% of women reporting DMPA discontinuation due to bleeding changes stopped the method after the first injection.¹⁴

Paul and colleagues conducted a telephone survey to determine the patterns of use and reasons for discontinuation among DMPA users.¹⁵ Of 252 DMPA users surveyed, 20% cited menstrual disturbances as the reason for discontinuation. These changes were equally distributed: amenorrhea, irregular bleeding, and heavy bleeding, all 6.8%.

Of approximately 7,000 women who participated in the 2002 National Survey for Family Growth, 600 had used DMPA in the past. Thirty-four percent pointed to a dislike of changes in menstrual periods as the reason for discontinuation.¹⁶

Not surprisingly, helping your patient develop realistic expectations about bleeding patterns with DMPA can decrease the discontinuation rate. Women who received repeated, structured information about DMPA were less likely to discontinue it because of menstrual disturbances (amenorrhea and irregular and heavy bleeding) than were women in a routine counseling group (OR, 0.20; 95% CI: 0.11, 0.37).¹⁷

Other investigators have reported similar findings, with a fourfold to sixfold lower likelihood of discontinuation because of bleeding changes among women who received detailed counseling about DMPA.^1,2

PERSPECTIVE AND GUIDANCE FOR YOUR PRACTICE

DMPA is effective and convenient, but unpredictable bleeding in the first year of use is not uncommon. The irregularity is similar to that seen with the ENG implant in the first 6 months of use. Thereafter, DMPA users are more likely to achieve and maintain amenorrhea, compared to ENG implant users.

Intrauterine contraception

The main mechanism of contraceptive action in the levonorgestrel intrauterine system (LNG-IUS) (Mirena) is significant thickening of cervical mucus, resulting in a physical barrier to sperm penetration; ovulation inhibition may also contribute. In a study of women who had been using the LNG-IUS for 4 years, 88% (15/17 cycles) were still ovulatory according to progesterone levels, but only 47% (8/17 cycles) showed normal follicular growth and rupture by ultrasonography.¹⁸ The efficacy of the LNG-IUS is 99.8%.⁶

Advantages of the LNG-IUS include its high effectiveness; long-term action; increased rate of menstrual cycles that are shorter, lighter, and marked by less cramping as use continues; and a high likelihood of amenorrhea as duration of use lengthens.

As with other progestin-only contraceptives, the major disadvantage of the LNG-IUS is associated irregular bleeding that, as is the case with DMPA, appears to decrease with duration of use for most women.

What are the expected changes in bleeding patterns with LNG-IUS? Local effects of the LNG-IUS on the endometrial lining include stromal pseudodecidualization, glandular atrophy, and increased infiltration of leukocytes in the endometrium. These effects, combined with partial inhibition of ovulatory function, commonly result in irregular bleeding.

The number of days of bleeding-spotting is pronounced in the first 3 to 6 months after insertion. Approximately 18% of women reported bleeding-spotting in the first 3 months; 6% to 25%, at 6 months; and only 1% of women, approximately, at 12 months (TABLE 2).

In a survey of Finnish women who used the LNG-IUS, 45.2% reported irregular bleeding, and 18.1% reported spotting, at some point during use.¹⁹ Importantly, the prevalence of bleeding-spotting does decrease with duration of use. Nevertheless, as many as 10% of women still report irregular bleeding-spotting at 2 years (TABLE 2).

As with other progestin-only contraceptives, amenorrhea rates for the LNG-IUS vary (TABLE 4). In a Brazilian study of 256 women, 44% reported amenorrhea at 6 months; 50%, at 12 and 24 months.²⁰ In a larger study of 1,821 Finnish women, however, only 17% of women reported amenorrhea at 12 months.²¹ A survey study of approximately 16,000 Finnish women who used the LNG-IUS found that 75% reported that they “totally or occasionally missed menses” at any time during as long as 5 years of use.¹⁹

The discontinuation rate for the LNG-IUS is lower than for the ENG implant or DMPA. Still, changes in bleeding patterns are the most common reason for discontinuation. At 1 year of use, approximately 10% of women discontinue the LNG-IUS because of changes in the bleeding pattern (TABLE 5).

In the most comprehensive study of early removal of the LNG-IUS, the total discontinuation rate—for all reasons—increased to 13% at 2 years, 19% at 3 years, 25% at 4 years, and 35% at 5 years.¹⁹ Women who reported excessive bleeding were almost three times more likely to discontinue the LNG-IUS early than women who did not report such a problem (RR, 2.77; 95% CI: 2.5, 3.07). Women who experience spotting are almost twice as likely to discontinue early (RR, 1.89; 95% CI: 1.75, 2.05). Others have reported the cumulative discontinuation rate to be as low as 14.4% at 5 years (when measuring discontinuation because of changes in menstrual bleeding) and as high as 35% at 5 years (when considering the total discontinuation rate for all reasons).²¹

Amenorrhea lowers the discontinuation rate. In one analysis, women who reported that they “totally or occasionally missed periods” were half as likely to discontinue the LNG-IUS as those who didn’t make such a report (RR, 0.46; 95% CI: 0.43, 0.50).¹⁹

PERSPECTIVE AND GUIDANCE FOR YOUR PRACTICE

Irregular bleeding is common with the LNG-IUS in the first 3 to 6 months of use, but overall discontinuation is relatively low—probably because of the high likelihood that bleeding patterns improve over time. Still, irregular bleeding remains the most common reason for discontinuation. Realistic expectations about bleeding patterns and the lower likelihood for amenorrhea, in comparison with DMPA, are important variables to discuss with women who are considering the LNG-IUS.

Progestin-only pills

Progestin-only pills (POPs) have a failure rate that ranges from 1.1 to 9.6 for every 100 users in the first year.²² A POP is used most often by women in whom estrogen is contraindicated, including those who are breastfeeding.²³

Disadvantages. POPs require precise adherence and make irregular vaginal bleeding likely. Although 40% to 50% of women who take a POP have normal menstrual cycles, 40% have short, irregular cycles, and another 10% experience even more markedly irregular cycles—from spotting to amenorrhea.²²

Studies that precede the WHO Belsey system showed that 1) as many as 70% of women who use a POP reported “breakthrough bleeding-spotting” in one or more cycles and 2) 6% to 16% have “breakthrough bleeding or inter-menstrual spotting” in all cycles (TABLES 2 AND 3).^24,25

On average, 25% of women discontinue POPs because of changes in their menstrual cycle (TABLE 5).

PERSPECTIVE AND GUIDANCE FOR YOUR PRACTICE

The mechanism that results in irregular vaginal bleeding in women taking a POP is unclear; evidence suggests that incomplete suppression of ovulation and direct endometrial effects are possible. To the frustration of patients and clinicians, it isn’t possible to predict who will have irregular bleeding—i.e., there is no association between body weight, or age, and the risk of irregular bleeding. As with other progestin-only methods, irregular bleeding is the most common reason for discontinuing POPs.²⁶ A Cochrane review of POPs is under way.²⁷

Is unpredictable bleeding with progestin-only contraceptives treatable?

Bleeding and discontinuation rates associated with progestin-only contraceptives that are observed in clinical trials, especially rates used for FDA review and approval of a product, don’t always translate to real-life medicine. Typically, in such trials, no treatment for irregular or unacceptable bleeding patterns is permitted: If an effective treatment is available, overall acceptability and continuation of the contraceptive could, potentially, be boosted. This matter is most relevant with injectable, intrauterine, and implantable progestin-only methods.

Findings of one meta-analysis. A recent Cochrane review evaluated the literature until December 2006 on the treatment of vaginal bleeding irregularities induced by progestin-only contraceptives.²⁸ Twenty-three randomized controlled trials, encompassing 2,674 subjects, were included. Seventy percent of the trials that were included were determined to reflect a low or moderate risk of bias.

Treatment with estrogen alone reduced the number of days of an ongoing bleeding episode among DMPA and levonorgestrel implant (Norplant) users. Treatment often led to individuals’ discontinuation in a study, however, because of gastrointestinal upset. Combined oral contraceptives can treat amenorrhea with success among DMPA users.

Antiprogestins such as mifepristone cause a reduction in bleeding among women using the levonorgestrel implant, but are not of benefit for ENG implant users.

Last, use of NSAIDs to treat irregular bleeding has shown variable efficacy. Additional small studies cited in the Cochrane review suggest that the following treatments were more effective than placebo for terminating an episode of bleeding among women using progestin-only contraception: the antiprogestin mifepristone for DMPA and POP users; mifepristone plus an estrogen for ENG implant users; and doxycycline for ENG implant users.²⁸

Overall, some women benefit from attempts at treatment. The authors of the Cochrane review caution that their findings do not support the routine clinical use of any of the regimens included in the trials, particularly for obtaining a long-term effect.²⁸

Newer trials, different findings? A more recent double-blind, randomized trial, in which the subjects were 100 Thai women, showed that irregular bleeding with DMPA ceased completely in 88% of those treated with tranexamic acid, 250 mg QID for 5 days, compared with 8% of women in whom bleeding ceased after treatment with placebo.²⁹

Another recent randomized trial found that mifepristone, combined with ethinyl estradiol or doxycycline, was significantly more effective than placebo in ending an episode of bleeding in ENG implant users. No improvement was seen, however, in subsequent bleeding patterns, and improvement with treatment, compared with placebo, amounted to a decrease of only about 2 days.³⁰

Noticeably missing from the literature are large trials that evaluate the use of combined hormonal contraceptives for bleeding irregularities in women using long-acting progestin-only contraceptives. True, some women use these methods because of a contraindication to estrogen-containing methods, but, in reality, most women who use these methods do so because of their high efficacy and ease of use.

PERSPECTIVE AND GUIDANCE FOR YOUR PRACTICE

For women who use the ENG implant or LNG-IUS and have no contraindication to estrogen-containing contraceptives, we often provide a short (1 or 2 months) course of a combined hormonal contraceptive when they find bleeding irregularities bothersome.

Because the serum progestin level provided with these methods is extremely low, adding a low-dose combined oral contraceptive, contraceptive patch, or contraceptive vaginal ring is not that different than using any of the combined hormonal contraceptives. A woman will not become pregnant if she forgets to take the pill or the ring falls out because she still has the progestin-only method in place. And if the short course of a combined hormonal contraceptive helps her continue the more effective method, then the overall goal of avoiding unintended pregnancy is better accomplished.

Large trials to evaluate the use of combined hormonal methods in such circumstances would, of course, be of great benefit.

Good counseling → informed choice → adherence and continuation

With all forms of progestin-only contraception, unpredictable bleeding occurs often and is the most common reason for method discontinuation.

Counseling that explicitly discusses the high likelihood of unpredictable menstrual bleeding allows women to prioritize this issue in their choice of a contraceptive.

Informed choice leads to a better continuation rate for progestin-only methods.

Seeking understanding. We lack full understanding of exactly what it is about changes in bleeding patterns that matter to women. Have definitions of bleeding and spotting that researchers utilize missed quality of life concerns that are more relevant to women? Are women concerned about how many days are spent avoiding sexual activity? Do religious restrictions figure prominently for some? How dissatisfied are they with days of cramping or bloating without bleeding? What do women want to know when they consider the bleeding patterns for their contraceptive options?

The answers to these questions likely vary from patient to patient—and that observation leads us back to grasping the art of contraceptive counseling: Our counseling needs to be concise, relatable, and honest.

Dr. Gariepy reports no financial relationships relevant to this article. Dr. Creinin reports that he is a consultant to, and a speaker for, Schering-Plough.

Progestin-only contraception—a diverse group of oral (progestin-only pills, or so-called minipills), injectable (depot medroxyprogesterone acetate), intrauterine (the levonorgestrel intrauterine system), and implantable (etonogestrel implant) methods—may offer advantages over estrogen-containing contraception:

• the flexibility of distinctive methods of delivery
• the ability to initiate the method in postpartum breastfeeding women
• enhanced safety in women who should not be exposed to exogenous estrogens.

Unpredictable bleeding is a major disadvantage of progestin-only contraception, however, and can cause women to discontinue these methods—and discontinuation without an effective backup method creates a high risk of unplanned pregnancy. The significant variability in bleeding patterns among progestin-only contraceptive methods hinders our ability to counsel patients about them.

Furthermore, the lack of uniform definitions of bleeding patterns with hormonal contraception, including progestin-only methods, makes it difficult to counsel women accurately and compare bleeding patterns among methods.

Accurate prediction of the bleeding patterns associated with progestin-only contraception could lower the discontinuation rate. For example, studies have shown that pretreatment counseling about expected side effects increases approximately fourfold the acceptability and continuation of depot medroxyprogesterone acetate.^1,2

In this Update, we review the data on bleeding patterns associated with progestin-only contraceptives, including the likelihood of 1) amenorrhea and 2) discontinuation due to changes in the bleeding pattern.

We also discuss what has been learned about the treatment of changes in bleeding patterns induced by progestin-only contraception.

Our goal? To summarize the findings in a comprehensive way that makes it easier for you to discuss expected bleeding patterns with your patients—so that women can choose the method of contraception that is the best fit for them.

Describing bleeding patterns is a challenging task

One of the difficulties of interpreting clinical data on bleeding patterns—with any type of contraception—is the lack of a universally accepted standard for collecting and reporting these data. The first suggestions for standardization were made in 1976, when Rodriguez and colleagues proposed using 90-day reference periods for analysis, as a way to minimize variability among individual menstrual cycles.³

Subsequently, the World Health Organization’s (WHO’s) Special Programme of Research, Development and Research Training in Human Reproduction developed recommendations for data collection, terminology, presentation, and data analysis when reporting vaginal bleeding during clinical trials of hormonal contraception. These recommendations became known as the WHO Belsey criteria (TABLE 1). They remain the standard.⁴

Under the WHO Belsey system:

• vaginal blood loss for which a woman uses sanitary protection is classified as bleeding
• vaginal blood loss that does not result in the use of sanitary protection is considered spotting.

This system also specifies indices for evaluating the bleeding pattern for each woman and reference period, including the number of bleeding-spotting days, number of bleeding-spotting episodes, lengths of bleeding-spotting episodes, and bleeding-spotting-free intervals. A bleeding-spotting episode is defined as one or more consecutive days during which blood loss (bleeding or spotting) has been recorded, each episode being bounded by bleeding-spotting-free days. The WHO Belsey criteria also identified subgroups that have “clinically important bleeding patterns” (TABLE 1).

But not all researchers use the WHO Belsey criteria. Many trials use, and report, their own system of analysis. Some researchers have chosen reference periods of other durations and study periods that range from 1 to 5 years. Some studies report bleeding patterns by number of days, and others report the percentage of women experiencing a given bleeding pattern during a reference period. The lack of uniformity results in data that are difficult to compare from one study to the next—and to explain to our patients.

It’s unclear whether any of our research definitions of clinically significant bleeding have ever been validated as clinically important to our patients. Multiple studies do show that changes in menstrual bleeding patterns are a significant cause of dissatisfaction with any given contraceptive method, but we don’t know if the number of days of bleeding-spotting or the predictability of this bleeding-spotting is the critical piece of information we should be relating to our patients.

In other words, do our beliefs about clinically important bleeding patterns reflect women’s beliefs?

TABLE 1

The WHO Belsey system of bleeding patterns

Implantable contraception

The etonogestrel (ENG) implant (Implanon) is the only implantable contraceptive available in the United States. This single-rod contraceptive can be used for as long as 3 years.

Contraceptive implants, including the levonorgestrel implants once sold in the United States and still available in some parts of the world, are highly effective. The Implanon prescribing information reports a first-year failure rate of 0.38 pregnancies for every 100 woman-years of use; Hatcher and co-workers reported a failure rate of 0.⁵ The difference is based on how the FDA defines pregnancy in contraceptive trials. In fact, the only pregnancies reported with the ENG implant happened after it was removed. Importantly, the studies evaluated by the FDA included only women not using any medications known to induce liver metabolism (the cytochrome P450 pathway) and who were between 80% and 130% of ideal body weight. The efficacy of the ENG implant for women who are taking medications that induce liver metabolism or who are greater than 130% of their ideal body weight is unknown.

The efficacy of the ENG implant is likely derived from suppression of ovulation and increased cervical mucus viscosity. Associated changes in the endometrium that occur with this low dosage of progestin are likely the primary cause of irregular and unpredictable bleeding.

Several studies have sought to describe the bleeding patterns experienced with the ENG implant.^6-8 During the first 3 months, approximately 50% of all women using the ENG implant reported bleeding-spotting (TABLE 2) for 30 days, on average (TABLE 3). The number of days decreases to approximately 20 bleeding-spotting days for each 90-day reference period at 6 to 24 months, with wide variability. For example: From 3 to 6 months, women reported 22 days of bleeding-spotting (standard deviation, 20 days); from months 21 to 24, 20 days of bleeding-spotting (standard deviation, 14 days).⁷

After using the ENG implant for 2 years, therefore, most women can expect the number of bleeding-spotting days for every 90-day reference period to range between 6 and 34 days. These days of bleeding-spotting are often noncontinuous, however. On average, women reported three separate bleeding-spotting episodes for every 90-day reference period.⁷

Although individual bleeding patterns are unpredictable, women who had no bleeding, or infrequent bleeding, at the beginning of use of the ENG implant had only a “small chance” of bleeding frequently.⁶ The most common bleeding pattern observed throughout the study was infrequent bleeding, defined as fewer than three episodes of bleeding in a 90-day reference period (excluding amenorrhea).⁷

Amenorrhea may not persist. The amenorrhea rate at 6 months of use and beyond ranges from 10% to 20% (TABLE 4). Importantly, women who are amenorrheic in one 90-day reference period are not necessarily the ones who are amenorrheic in another reference period. So, unlike what is more commonly seen with other progestin-only methods, such as injectables, amenorrhea is not sustained for most women.

This unpredictable pattern affects continuation of the ENG implant (TABLE 5). Irregular bleeding is the most common reason women cite for discontinuation, accounting for 30% to 60% of all women who discontinue early.^7,9

Overall, approximately 4% of ENG users discontinue the method at 1 year. Ten percent to 20% discontinue each year thereafter because of intolerance to bleeding changes.^6-9

There are, however, differences in discontinuation rates across cultures. In an integrated analysis of 13 different trials that evaluated patterns of vaginal bleeding with the ENG implant where the rate of menstrual changes was similar, women from Europe and Canada were much more likely (23%) to discontinue the implant because of those changes than women from Southeast Asia and Chile were (2%).⁶ This finding may reflect differences in cultural beliefs or disparate access to other contraceptive options.¹⁰

TABLE 2

What percentage of women taking progestin-only contraception report bleeding-spotting?

TABLE 3

How many days of bleeding-spotting do women have
when they use progestin-only contraception?

TABLE 4

What percentage of women taking progestin-only contraception report amenorrhea?

TABLE 5

What percentage discontinue progestin-only contraception
because of a change in bleeding pattern?

PERSPECTIVE AND GUIDANCE FOR YOUR PRACTICE

The pattern of bleeding seen with the ENG implant is like the activity of the heart in atrial fibrillation: irregularly irregular. Still, most (80%) women continue to use it beyond 1 year. In fact, the discontinuation rate for the ENG implant is less than that of depot medroxyprogesterone acetate (DMPA) and progestin-only pills.

Most ENG implant users report no difficulty tolerating the associated unpredictable bleeding; it’s possible that they had unpredictable bleeding at baseline, or were drawn to the improvement in their dysmenorrhea.⁶

Importantly, unpredictable bleeding does not affect efficacy; the ENG implant remains one of the most effective long-acting reversible contraceptives. For women who can tolerate unpredictable bleeding, the ENG implant is a highly effective contraceptive option.

Injectable contraception

Approved by the FDA in 1992, DMPA (Depo-Provera) has good efficacy and long-acting protection. Disadvantages include unpredictable bleeding, weight gain, acne, depression, hair loss, and the controversial issue of decreased bone loss with prolonged use.

What are the expected changes in bleeding patterns with DMPA? Women often have unpredictable patterns, with infrequent but prolonged bleeding-spotting episodes.¹¹ The overall incidence of irregular bleeding can be as high as 70% in the first year of use.¹² Irregular bleeding decreases with continued use, to as low as 10% after the first year (TABLE 2).

Although the number of bleeding-spotting days decreases over time, women have reported as many as 10 days of irregular bleeding-spotting between 9 and 12 months of use (TABLE 3). The rates of irregular bleeding and amenorrhea are similar for the subcutaneous formulation of DMPA.¹³

DMPA is often used because of the high likelihood of amenorrhea. However, amenorrhea is not accomplished in most women in a short time. At 3 months of use, 10% to 45% of women report amenorrhea; after 1 year, the rate increases to 40% to 70% (TABLE 4). At 5 years, 80% of women report amenorrhea.¹²

A source of frustration. DMPA’s high discontinuation rate, compared with what is seen with other contraceptives, can be frustrating for patients and clinicians. Irregular bleeding is the most common reason for discontinuation. Approximately 35% of women who start DMPA discontinue it during the first 3 months of use because of irregular bleeding (TABLE 5). The cumulative discontinuation rate rises over time: At 1 year, 40% to 60% of women who started DMPA will have discontinued it because of changes in bleeding patterns (TABLE 5). Furthermore, 70% of women reporting DMPA discontinuation due to bleeding changes stopped the method after the first injection.¹⁴

Paul and colleagues conducted a telephone survey to determine the patterns of use and reasons for discontinuation among DMPA users.¹⁵ Of 252 DMPA users surveyed, 20% cited menstrual disturbances as the reason for discontinuation. These changes were equally distributed: amenorrhea, irregular bleeding, and heavy bleeding, all 6.8%.

Of approximately 7,000 women who participated in the 2002 National Survey for Family Growth, 600 had used DMPA in the past. Thirty-four percent pointed to a dislike of changes in menstrual periods as the reason for discontinuation.¹⁶

Not surprisingly, helping your patient develop realistic expectations about bleeding patterns with DMPA can decrease the discontinuation rate. Women who received repeated, structured information about DMPA were less likely to discontinue it because of menstrual disturbances (amenorrhea and irregular and heavy bleeding) than were women in a routine counseling group (OR, 0.20; 95% CI: 0.11, 0.37).¹⁷

Other investigators have reported similar findings, with a fourfold to sixfold lower likelihood of discontinuation because of bleeding changes among women who received detailed counseling about DMPA.^1,2

PERSPECTIVE AND GUIDANCE FOR YOUR PRACTICE

DMPA is effective and convenient, but unpredictable bleeding in the first year of use is not uncommon. The irregularity is similar to that seen with the ENG implant in the first 6 months of use. Thereafter, DMPA users are more likely to achieve and maintain amenorrhea, compared to ENG implant users.

Intrauterine contraception

The main mechanism of contraceptive action in the levonorgestrel intrauterine system (LNG-IUS) (Mirena) is significant thickening of cervical mucus, resulting in a physical barrier to sperm penetration; ovulation inhibition may also contribute. In a study of women who had been using the LNG-IUS for 4 years, 88% (15/17 cycles) were still ovulatory according to progesterone levels, but only 47% (8/17 cycles) showed normal follicular growth and rupture by ultrasonography.¹⁸ The efficacy of the LNG-IUS is 99.8%.⁶

Advantages of the LNG-IUS include its high effectiveness; long-term action; increased rate of menstrual cycles that are shorter, lighter, and marked by less cramping as use continues; and a high likelihood of amenorrhea as duration of use lengthens.

As with other progestin-only contraceptives, the major disadvantage of the LNG-IUS is associated irregular bleeding that, as is the case with DMPA, appears to decrease with duration of use for most women.

What are the expected changes in bleeding patterns with LNG-IUS? Local effects of the LNG-IUS on the endometrial lining include stromal pseudodecidualization, glandular atrophy, and increased infiltration of leukocytes in the endometrium. These effects, combined with partial inhibition of ovulatory function, commonly result in irregular bleeding.

The number of days of bleeding-spotting is pronounced in the first 3 to 6 months after insertion. Approximately 18% of women reported bleeding-spotting in the first 3 months; 6% to 25%, at 6 months; and only 1% of women, approximately, at 12 months (TABLE 2).

In a survey of Finnish women who used the LNG-IUS, 45.2% reported irregular bleeding, and 18.1% reported spotting, at some point during use.¹⁹ Importantly, the prevalence of bleeding-spotting does decrease with duration of use. Nevertheless, as many as 10% of women still report irregular bleeding-spotting at 2 years (TABLE 2).

As with other progestin-only contraceptives, amenorrhea rates for the LNG-IUS vary (TABLE 4). In a Brazilian study of 256 women, 44% reported amenorrhea at 6 months; 50%, at 12 and 24 months.²⁰ In a larger study of 1,821 Finnish women, however, only 17% of women reported amenorrhea at 12 months.²¹ A survey study of approximately 16,000 Finnish women who used the LNG-IUS found that 75% reported that they “totally or occasionally missed menses” at any time during as long as 5 years of use.¹⁹

The discontinuation rate for the LNG-IUS is lower than for the ENG implant or DMPA. Still, changes in bleeding patterns are the most common reason for discontinuation. At 1 year of use, approximately 10% of women discontinue the LNG-IUS because of changes in the bleeding pattern (TABLE 5).

In the most comprehensive study of early removal of the LNG-IUS, the total discontinuation rate—for all reasons—increased to 13% at 2 years, 19% at 3 years, 25% at 4 years, and 35% at 5 years.¹⁹ Women who reported excessive bleeding were almost three times more likely to discontinue the LNG-IUS early than women who did not report such a problem (RR, 2.77; 95% CI: 2.5, 3.07). Women who experience spotting are almost twice as likely to discontinue early (RR, 1.89; 95% CI: 1.75, 2.05). Others have reported the cumulative discontinuation rate to be as low as 14.4% at 5 years (when measuring discontinuation because of changes in menstrual bleeding) and as high as 35% at 5 years (when considering the total discontinuation rate for all reasons).²¹

Amenorrhea lowers the discontinuation rate. In one analysis, women who reported that they “totally or occasionally missed periods” were half as likely to discontinue the LNG-IUS as those who didn’t make such a report (RR, 0.46; 95% CI: 0.43, 0.50).¹⁹

PERSPECTIVE AND GUIDANCE FOR YOUR PRACTICE

Irregular bleeding is common with the LNG-IUS in the first 3 to 6 months of use, but overall discontinuation is relatively low—probably because of the high likelihood that bleeding patterns improve over time. Still, irregular bleeding remains the most common reason for discontinuation. Realistic expectations about bleeding patterns and the lower likelihood for amenorrhea, in comparison with DMPA, are important variables to discuss with women who are considering the LNG-IUS.

Progestin-only pills

Progestin-only pills (POPs) have a failure rate that ranges from 1.1 to 9.6 for every 100 users in the first year.²² A POP is used most often by women in whom estrogen is contraindicated, including those who are breastfeeding.²³

Disadvantages. POPs require precise adherence and make irregular vaginal bleeding likely. Although 40% to 50% of women who take a POP have normal menstrual cycles, 40% have short, irregular cycles, and another 10% experience even more markedly irregular cycles—from spotting to amenorrhea.²²

Studies that precede the WHO Belsey system showed that 1) as many as 70% of women who use a POP reported “breakthrough bleeding-spotting” in one or more cycles and 2) 6% to 16% have “breakthrough bleeding or inter-menstrual spotting” in all cycles (TABLES 2 AND 3).^24,25

On average, 25% of women discontinue POPs because of changes in their menstrual cycle (TABLE 5).

PERSPECTIVE AND GUIDANCE FOR YOUR PRACTICE

The mechanism that results in irregular vaginal bleeding in women taking a POP is unclear; evidence suggests that incomplete suppression of ovulation and direct endometrial effects are possible. To the frustration of patients and clinicians, it isn’t possible to predict who will have irregular bleeding—i.e., there is no association between body weight, or age, and the risk of irregular bleeding. As with other progestin-only methods, irregular bleeding is the most common reason for discontinuing POPs.²⁶ A Cochrane review of POPs is under way.²⁷

Is unpredictable bleeding with progestin-only contraceptives treatable?

Bleeding and discontinuation rates associated with progestin-only contraceptives that are observed in clinical trials, especially rates used for FDA review and approval of a product, don’t always translate to real-life medicine. Typically, in such trials, no treatment for irregular or unacceptable bleeding patterns is permitted: If an effective treatment is available, overall acceptability and continuation of the contraceptive could, potentially, be boosted. This matter is most relevant with injectable, intrauterine, and implantable progestin-only methods.

Findings of one meta-analysis. A recent Cochrane review evaluated the literature until December 2006 on the treatment of vaginal bleeding irregularities induced by progestin-only contraceptives.²⁸ Twenty-three randomized controlled trials, encompassing 2,674 subjects, were included. Seventy percent of the trials that were included were determined to reflect a low or moderate risk of bias.

Treatment with estrogen alone reduced the number of days of an ongoing bleeding episode among DMPA and levonorgestrel implant (Norplant) users. Treatment often led to individuals’ discontinuation in a study, however, because of gastrointestinal upset. Combined oral contraceptives can treat amenorrhea with success among DMPA users.

Antiprogestins such as mifepristone cause a reduction in bleeding among women using the levonorgestrel implant, but are not of benefit for ENG implant users.

Last, use of NSAIDs to treat irregular bleeding has shown variable efficacy. Additional small studies cited in the Cochrane review suggest that the following treatments were more effective than placebo for terminating an episode of bleeding among women using progestin-only contraception: the antiprogestin mifepristone for DMPA and POP users; mifepristone plus an estrogen for ENG implant users; and doxycycline for ENG implant users.²⁸

Overall, some women benefit from attempts at treatment. The authors of the Cochrane review caution that their findings do not support the routine clinical use of any of the regimens included in the trials, particularly for obtaining a long-term effect.²⁸

Newer trials, different findings? A more recent double-blind, randomized trial, in which the subjects were 100 Thai women, showed that irregular bleeding with DMPA ceased completely in 88% of those treated with tranexamic acid, 250 mg QID for 5 days, compared with 8% of women in whom bleeding ceased after treatment with placebo.²⁹

Another recent randomized trial found that mifepristone, combined with ethinyl estradiol or doxycycline, was significantly more effective than placebo in ending an episode of bleeding in ENG implant users. No improvement was seen, however, in subsequent bleeding patterns, and improvement with treatment, compared with placebo, amounted to a decrease of only about 2 days.³⁰

Noticeably missing from the literature are large trials that evaluate the use of combined hormonal contraceptives for bleeding irregularities in women using long-acting progestin-only contraceptives. True, some women use these methods because of a contraindication to estrogen-containing methods, but, in reality, most women who use these methods do so because of their high efficacy and ease of use.

PERSPECTIVE AND GUIDANCE FOR YOUR PRACTICE

For women who use the ENG implant or LNG-IUS and have no contraindication to estrogen-containing contraceptives, we often provide a short (1 or 2 months) course of a combined hormonal contraceptive when they find bleeding irregularities bothersome.

Because the serum progestin level provided with these methods is extremely low, adding a low-dose combined oral contraceptive, contraceptive patch, or contraceptive vaginal ring is not that different than using any of the combined hormonal contraceptives. A woman will not become pregnant if she forgets to take the pill or the ring falls out because she still has the progestin-only method in place. And if the short course of a combined hormonal contraceptive helps her continue the more effective method, then the overall goal of avoiding unintended pregnancy is better accomplished.

Large trials to evaluate the use of combined hormonal methods in such circumstances would, of course, be of great benefit.

Good counseling → informed choice → adherence and continuation

With all forms of progestin-only contraception, unpredictable bleeding occurs often and is the most common reason for method discontinuation.

Counseling that explicitly discusses the high likelihood of unpredictable menstrual bleeding allows women to prioritize this issue in their choice of a contraceptive.

Informed choice leads to a better continuation rate for progestin-only methods.

Seeking understanding. We lack full understanding of exactly what it is about changes in bleeding patterns that matter to women. Have definitions of bleeding and spotting that researchers utilize missed quality of life concerns that are more relevant to women? Are women concerned about how many days are spent avoiding sexual activity? Do religious restrictions figure prominently for some? How dissatisfied are they with days of cramping or bloating without bleeding? What do women want to know when they consider the bleeding patterns for their contraceptive options?

The answers to these questions likely vary from patient to patient—and that observation leads us back to grasping the art of contraceptive counseling: Our counseling needs to be concise, relatable, and honest.

Dr. Gariepy reports no financial relationships relevant to this article. Dr. Creinin reports that he is a consultant to, and a speaker for, Schering-Plough.

Progestin-only contraception—a diverse group of oral (progestin-only pills, or so-called minipills), injectable (depot medroxyprogesterone acetate), intrauterine (the levonorgestrel intrauterine system), and implantable (etonogestrel implant) methods—may offer advantages over estrogen-containing contraception:

• the flexibility of distinctive methods of delivery
• the ability to initiate the method in postpartum breastfeeding women
• enhanced safety in women who should not be exposed to exogenous estrogens.

Unpredictable bleeding is a major disadvantage of progestin-only contraception, however, and can cause women to discontinue these methods—and discontinuation without an effective backup method creates a high risk of unplanned pregnancy. The significant variability in bleeding patterns among progestin-only contraceptive methods hinders our ability to counsel patients about them.

Furthermore, the lack of uniform definitions of bleeding patterns with hormonal contraception, including progestin-only methods, makes it difficult to counsel women accurately and compare bleeding patterns among methods.

Accurate prediction of the bleeding patterns associated with progestin-only contraception could lower the discontinuation rate. For example, studies have shown that pretreatment counseling about expected side effects increases approximately fourfold the acceptability and continuation of depot medroxyprogesterone acetate.^1,2

In this Update, we review the data on bleeding patterns associated with progestin-only contraceptives, including the likelihood of 1) amenorrhea and 2) discontinuation due to changes in the bleeding pattern.

We also discuss what has been learned about the treatment of changes in bleeding patterns induced by progestin-only contraception.

Our goal? To summarize the findings in a comprehensive way that makes it easier for you to discuss expected bleeding patterns with your patients—so that women can choose the method of contraception that is the best fit for them.

Describing bleeding patterns is a challenging task

One of the difficulties of interpreting clinical data on bleeding patterns—with any type of contraception—is the lack of a universally accepted standard for collecting and reporting these data. The first suggestions for standardization were made in 1976, when Rodriguez and colleagues proposed using 90-day reference periods for analysis, as a way to minimize variability among individual menstrual cycles.³

Subsequently, the World Health Organization’s (WHO’s) Special Programme of Research, Development and Research Training in Human Reproduction developed recommendations for data collection, terminology, presentation, and data analysis when reporting vaginal bleeding during clinical trials of hormonal contraception. These recommendations became known as the WHO Belsey criteria (TABLE 1). They remain the standard.⁴

Under the WHO Belsey system:

• vaginal blood loss for which a woman uses sanitary protection is classified as bleeding
• vaginal blood loss that does not result in the use of sanitary protection is considered spotting.

This system also specifies indices for evaluating the bleeding pattern for each woman and reference period, including the number of bleeding-spotting days, number of bleeding-spotting episodes, lengths of bleeding-spotting episodes, and bleeding-spotting-free intervals. A bleeding-spotting episode is defined as one or more consecutive days during which blood loss (bleeding or spotting) has been recorded, each episode being bounded by bleeding-spotting-free days. The WHO Belsey criteria also identified subgroups that have “clinically important bleeding patterns” (TABLE 1).

But not all researchers use the WHO Belsey criteria. Many trials use, and report, their own system of analysis. Some researchers have chosen reference periods of other durations and study periods that range from 1 to 5 years. Some studies report bleeding patterns by number of days, and others report the percentage of women experiencing a given bleeding pattern during a reference period. The lack of uniformity results in data that are difficult to compare from one study to the next—and to explain to our patients.

It’s unclear whether any of our research definitions of clinically significant bleeding have ever been validated as clinically important to our patients. Multiple studies do show that changes in menstrual bleeding patterns are a significant cause of dissatisfaction with any given contraceptive method, but we don’t know if the number of days of bleeding-spotting or the predictability of this bleeding-spotting is the critical piece of information we should be relating to our patients.

In other words, do our beliefs about clinically important bleeding patterns reflect women’s beliefs?

TABLE 1

The WHO Belsey system of bleeding patterns

Implantable contraception

The etonogestrel (ENG) implant (Implanon) is the only implantable contraceptive available in the United States. This single-rod contraceptive can be used for as long as 3 years.

Contraceptive implants, including the levonorgestrel implants once sold in the United States and still available in some parts of the world, are highly effective. The Implanon prescribing information reports a first-year failure rate of 0.38 pregnancies for every 100 woman-years of use; Hatcher and co-workers reported a failure rate of 0.⁵ The difference is based on how the FDA defines pregnancy in contraceptive trials. In fact, the only pregnancies reported with the ENG implant happened after it was removed. Importantly, the studies evaluated by the FDA included only women not using any medications known to induce liver metabolism (the cytochrome P450 pathway) and who were between 80% and 130% of ideal body weight. The efficacy of the ENG implant for women who are taking medications that induce liver metabolism or who are greater than 130% of their ideal body weight is unknown.

The efficacy of the ENG implant is likely derived from suppression of ovulation and increased cervical mucus viscosity. Associated changes in the endometrium that occur with this low dosage of progestin are likely the primary cause of irregular and unpredictable bleeding.

Several studies have sought to describe the bleeding patterns experienced with the ENG implant.^6-8 During the first 3 months, approximately 50% of all women using the ENG implant reported bleeding-spotting (TABLE 2) for 30 days, on average (TABLE 3). The number of days decreases to approximately 20 bleeding-spotting days for each 90-day reference period at 6 to 24 months, with wide variability. For example: From 3 to 6 months, women reported 22 days of bleeding-spotting (standard deviation, 20 days); from months 21 to 24, 20 days of bleeding-spotting (standard deviation, 14 days).⁷

After using the ENG implant for 2 years, therefore, most women can expect the number of bleeding-spotting days for every 90-day reference period to range between 6 and 34 days. These days of bleeding-spotting are often noncontinuous, however. On average, women reported three separate bleeding-spotting episodes for every 90-day reference period.⁷

Although individual bleeding patterns are unpredictable, women who had no bleeding, or infrequent bleeding, at the beginning of use of the ENG implant had only a “small chance” of bleeding frequently.⁶ The most common bleeding pattern observed throughout the study was infrequent bleeding, defined as fewer than three episodes of bleeding in a 90-day reference period (excluding amenorrhea).⁷

Amenorrhea may not persist. The amenorrhea rate at 6 months of use and beyond ranges from 10% to 20% (TABLE 4). Importantly, women who are amenorrheic in one 90-day reference period are not necessarily the ones who are amenorrheic in another reference period. So, unlike what is more commonly seen with other progestin-only methods, such as injectables, amenorrhea is not sustained for most women.

This unpredictable pattern affects continuation of the ENG implant (TABLE 5). Irregular bleeding is the most common reason women cite for discontinuation, accounting for 30% to 60% of all women who discontinue early.^7,9

Overall, approximately 4% of ENG users discontinue the method at 1 year. Ten percent to 20% discontinue each year thereafter because of intolerance to bleeding changes.^6-9

There are, however, differences in discontinuation rates across cultures. In an integrated analysis of 13 different trials that evaluated patterns of vaginal bleeding with the ENG implant where the rate of menstrual changes was similar, women from Europe and Canada were much more likely (23%) to discontinue the implant because of those changes than women from Southeast Asia and Chile were (2%).⁶ This finding may reflect differences in cultural beliefs or disparate access to other contraceptive options.¹⁰

TABLE 2

What percentage of women taking progestin-only contraception report bleeding-spotting?

TABLE 3

How many days of bleeding-spotting do women have
when they use progestin-only contraception?

TABLE 4

What percentage of women taking progestin-only contraception report amenorrhea?

TABLE 5

What percentage discontinue progestin-only contraception
because of a change in bleeding pattern?

PERSPECTIVE AND GUIDANCE FOR YOUR PRACTICE

The pattern of bleeding seen with the ENG implant is like the activity of the heart in atrial fibrillation: irregularly irregular. Still, most (80%) women continue to use it beyond 1 year. In fact, the discontinuation rate for the ENG implant is less than that of depot medroxyprogesterone acetate (DMPA) and progestin-only pills.

Most ENG implant users report no difficulty tolerating the associated unpredictable bleeding; it’s possible that they had unpredictable bleeding at baseline, or were drawn to the improvement in their dysmenorrhea.⁶

Importantly, unpredictable bleeding does not affect efficacy; the ENG implant remains one of the most effective long-acting reversible contraceptives. For women who can tolerate unpredictable bleeding, the ENG implant is a highly effective contraceptive option.

Injectable contraception

Approved by the FDA in 1992, DMPA (Depo-Provera) has good efficacy and long-acting protection. Disadvantages include unpredictable bleeding, weight gain, acne, depression, hair loss, and the controversial issue of decreased bone loss with prolonged use.

What are the expected changes in bleeding patterns with DMPA? Women often have unpredictable patterns, with infrequent but prolonged bleeding-spotting episodes.¹¹ The overall incidence of irregular bleeding can be as high as 70% in the first year of use.¹² Irregular bleeding decreases with continued use, to as low as 10% after the first year (TABLE 2).

Although the number of bleeding-spotting days decreases over time, women have reported as many as 10 days of irregular bleeding-spotting between 9 and 12 months of use (TABLE 3). The rates of irregular bleeding and amenorrhea are similar for the subcutaneous formulation of DMPA.¹³

DMPA is often used because of the high likelihood of amenorrhea. However, amenorrhea is not accomplished in most women in a short time. At 3 months of use, 10% to 45% of women report amenorrhea; after 1 year, the rate increases to 40% to 70% (TABLE 4). At 5 years, 80% of women report amenorrhea.¹²

A source of frustration. DMPA’s high discontinuation rate, compared with what is seen with other contraceptives, can be frustrating for patients and clinicians. Irregular bleeding is the most common reason for discontinuation. Approximately 35% of women who start DMPA discontinue it during the first 3 months of use because of irregular bleeding (TABLE 5). The cumulative discontinuation rate rises over time: At 1 year, 40% to 60% of women who started DMPA will have discontinued it because of changes in bleeding patterns (TABLE 5). Furthermore, 70% of women reporting DMPA discontinuation due to bleeding changes stopped the method after the first injection.¹⁴

Paul and colleagues conducted a telephone survey to determine the patterns of use and reasons for discontinuation among DMPA users.¹⁵ Of 252 DMPA users surveyed, 20% cited menstrual disturbances as the reason for discontinuation. These changes were equally distributed: amenorrhea, irregular bleeding, and heavy bleeding, all 6.8%.

Of approximately 7,000 women who participated in the 2002 National Survey for Family Growth, 600 had used DMPA in the past. Thirty-four percent pointed to a dislike of changes in menstrual periods as the reason for discontinuation.¹⁶

Not surprisingly, helping your patient develop realistic expectations about bleeding patterns with DMPA can decrease the discontinuation rate. Women who received repeated, structured information about DMPA were less likely to discontinue it because of menstrual disturbances (amenorrhea and irregular and heavy bleeding) than were women in a routine counseling group (OR, 0.20; 95% CI: 0.11, 0.37).¹⁷

Other investigators have reported similar findings, with a fourfold to sixfold lower likelihood of discontinuation because of bleeding changes among women who received detailed counseling about DMPA.^1,2

PERSPECTIVE AND GUIDANCE FOR YOUR PRACTICE

DMPA is effective and convenient, but unpredictable bleeding in the first year of use is not uncommon. The irregularity is similar to that seen with the ENG implant in the first 6 months of use. Thereafter, DMPA users are more likely to achieve and maintain amenorrhea, compared to ENG implant users.

Intrauterine contraception

The main mechanism of contraceptive action in the levonorgestrel intrauterine system (LNG-IUS) (Mirena) is significant thickening of cervical mucus, resulting in a physical barrier to sperm penetration; ovulation inhibition may also contribute. In a study of women who had been using the LNG-IUS for 4 years, 88% (15/17 cycles) were still ovulatory according to progesterone levels, but only 47% (8/17 cycles) showed normal follicular growth and rupture by ultrasonography.¹⁸ The efficacy of the LNG-IUS is 99.8%.⁶

Advantages of the LNG-IUS include its high effectiveness; long-term action; increased rate of menstrual cycles that are shorter, lighter, and marked by less cramping as use continues; and a high likelihood of amenorrhea as duration of use lengthens.

As with other progestin-only contraceptives, the major disadvantage of the LNG-IUS is associated irregular bleeding that, as is the case with DMPA, appears to decrease with duration of use for most women.

What are the expected changes in bleeding patterns with LNG-IUS? Local effects of the LNG-IUS on the endometrial lining include stromal pseudodecidualization, glandular atrophy, and increased infiltration of leukocytes in the endometrium. These effects, combined with partial inhibition of ovulatory function, commonly result in irregular bleeding.

The number of days of bleeding-spotting is pronounced in the first 3 to 6 months after insertion. Approximately 18% of women reported bleeding-spotting in the first 3 months; 6% to 25%, at 6 months; and only 1% of women, approximately, at 12 months (TABLE 2).

In a survey of Finnish women who used the LNG-IUS, 45.2% reported irregular bleeding, and 18.1% reported spotting, at some point during use.¹⁹ Importantly, the prevalence of bleeding-spotting does decrease with duration of use. Nevertheless, as many as 10% of women still report irregular bleeding-spotting at 2 years (TABLE 2).

As with other progestin-only contraceptives, amenorrhea rates for the LNG-IUS vary (TABLE 4). In a Brazilian study of 256 women, 44% reported amenorrhea at 6 months; 50%, at 12 and 24 months.²⁰ In a larger study of 1,821 Finnish women, however, only 17% of women reported amenorrhea at 12 months.²¹ A survey study of approximately 16,000 Finnish women who used the LNG-IUS found that 75% reported that they “totally or occasionally missed menses” at any time during as long as 5 years of use.¹⁹

The discontinuation rate for the LNG-IUS is lower than for the ENG implant or DMPA. Still, changes in bleeding patterns are the most common reason for discontinuation. At 1 year of use, approximately 10% of women discontinue the LNG-IUS because of changes in the bleeding pattern (TABLE 5).

In the most comprehensive study of early removal of the LNG-IUS, the total discontinuation rate—for all reasons—increased to 13% at 2 years, 19% at 3 years, 25% at 4 years, and 35% at 5 years.¹⁹ Women who reported excessive bleeding were almost three times more likely to discontinue the LNG-IUS early than women who did not report such a problem (RR, 2.77; 95% CI: 2.5, 3.07). Women who experience spotting are almost twice as likely to discontinue early (RR, 1.89; 95% CI: 1.75, 2.05). Others have reported the cumulative discontinuation rate to be as low as 14.4% at 5 years (when measuring discontinuation because of changes in menstrual bleeding) and as high as 35% at 5 years (when considering the total discontinuation rate for all reasons).²¹

Amenorrhea lowers the discontinuation rate. In one analysis, women who reported that they “totally or occasionally missed periods” were half as likely to discontinue the LNG-IUS as those who didn’t make such a report (RR, 0.46; 95% CI: 0.43, 0.50).¹⁹

PERSPECTIVE AND GUIDANCE FOR YOUR PRACTICE

Irregular bleeding is common with the LNG-IUS in the first 3 to 6 months of use, but overall discontinuation is relatively low—probably because of the high likelihood that bleeding patterns improve over time. Still, irregular bleeding remains the most common reason for discontinuation. Realistic expectations about bleeding patterns and the lower likelihood for amenorrhea, in comparison with DMPA, are important variables to discuss with women who are considering the LNG-IUS.

Progestin-only pills

Progestin-only pills (POPs) have a failure rate that ranges from 1.1 to 9.6 for every 100 users in the first year.²² A POP is used most often by women in whom estrogen is contraindicated, including those who are breastfeeding.²³

Disadvantages. POPs require precise adherence and make irregular vaginal bleeding likely. Although 40% to 50% of women who take a POP have normal menstrual cycles, 40% have short, irregular cycles, and another 10% experience even more markedly irregular cycles—from spotting to amenorrhea.²²

Studies that precede the WHO Belsey system showed that 1) as many as 70% of women who use a POP reported “breakthrough bleeding-spotting” in one or more cycles and 2) 6% to 16% have “breakthrough bleeding or inter-menstrual spotting” in all cycles (TABLES 2 AND 3).^24,25

On average, 25% of women discontinue POPs because of changes in their menstrual cycle (TABLE 5).

PERSPECTIVE AND GUIDANCE FOR YOUR PRACTICE

The mechanism that results in irregular vaginal bleeding in women taking a POP is unclear; evidence suggests that incomplete suppression of ovulation and direct endometrial effects are possible. To the frustration of patients and clinicians, it isn’t possible to predict who will have irregular bleeding—i.e., there is no association between body weight, or age, and the risk of irregular bleeding. As with other progestin-only methods, irregular bleeding is the most common reason for discontinuing POPs.²⁶ A Cochrane review of POPs is under way.²⁷

Is unpredictable bleeding with progestin-only contraceptives treatable?

Bleeding and discontinuation rates associated with progestin-only contraceptives that are observed in clinical trials, especially rates used for FDA review and approval of a product, don’t always translate to real-life medicine. Typically, in such trials, no treatment for irregular or unacceptable bleeding patterns is permitted: If an effective treatment is available, overall acceptability and continuation of the contraceptive could, potentially, be boosted. This matter is most relevant with injectable, intrauterine, and implantable progestin-only methods.

Findings of one meta-analysis. A recent Cochrane review evaluated the literature until December 2006 on the treatment of vaginal bleeding irregularities induced by progestin-only contraceptives.²⁸ Twenty-three randomized controlled trials, encompassing 2,674 subjects, were included. Seventy percent of the trials that were included were determined to reflect a low or moderate risk of bias.

Treatment with estrogen alone reduced the number of days of an ongoing bleeding episode among DMPA and levonorgestrel implant (Norplant) users. Treatment often led to individuals’ discontinuation in a study, however, because of gastrointestinal upset. Combined oral contraceptives can treat amenorrhea with success among DMPA users.

Antiprogestins such as mifepristone cause a reduction in bleeding among women using the levonorgestrel implant, but are not of benefit for ENG implant users.

Last, use of NSAIDs to treat irregular bleeding has shown variable efficacy. Additional small studies cited in the Cochrane review suggest that the following treatments were more effective than placebo for terminating an episode of bleeding among women using progestin-only contraception: the antiprogestin mifepristone for DMPA and POP users; mifepristone plus an estrogen for ENG implant users; and doxycycline for ENG implant users.²⁸

Overall, some women benefit from attempts at treatment. The authors of the Cochrane review caution that their findings do not support the routine clinical use of any of the regimens included in the trials, particularly for obtaining a long-term effect.²⁸

Newer trials, different findings? A more recent double-blind, randomized trial, in which the subjects were 100 Thai women, showed that irregular bleeding with DMPA ceased completely in 88% of those treated with tranexamic acid, 250 mg QID for 5 days, compared with 8% of women in whom bleeding ceased after treatment with placebo.²⁹

Another recent randomized trial found that mifepristone, combined with ethinyl estradiol or doxycycline, was significantly more effective than placebo in ending an episode of bleeding in ENG implant users. No improvement was seen, however, in subsequent bleeding patterns, and improvement with treatment, compared with placebo, amounted to a decrease of only about 2 days.³⁰

Noticeably missing from the literature are large trials that evaluate the use of combined hormonal contraceptives for bleeding irregularities in women using long-acting progestin-only contraceptives. True, some women use these methods because of a contraindication to estrogen-containing methods, but, in reality, most women who use these methods do so because of their high efficacy and ease of use.

PERSPECTIVE AND GUIDANCE FOR YOUR PRACTICE

For women who use the ENG implant or LNG-IUS and have no contraindication to estrogen-containing contraceptives, we often provide a short (1 or 2 months) course of a combined hormonal contraceptive when they find bleeding irregularities bothersome.

Because the serum progestin level provided with these methods is extremely low, adding a low-dose combined oral contraceptive, contraceptive patch, or contraceptive vaginal ring is not that different than using any of the combined hormonal contraceptives. A woman will not become pregnant if she forgets to take the pill or the ring falls out because she still has the progestin-only method in place. And if the short course of a combined hormonal contraceptive helps her continue the more effective method, then the overall goal of avoiding unintended pregnancy is better accomplished.

Large trials to evaluate the use of combined hormonal methods in such circumstances would, of course, be of great benefit.

Good counseling → informed choice → adherence and continuation

With all forms of progestin-only contraception, unpredictable bleeding occurs often and is the most common reason for method discontinuation.

Counseling that explicitly discusses the high likelihood of unpredictable menstrual bleeding allows women to prioritize this issue in their choice of a contraceptive.

Informed choice leads to a better continuation rate for progestin-only methods.

Seeking understanding. We lack full understanding of exactly what it is about changes in bleeding patterns that matter to women. Have definitions of bleeding and spotting that researchers utilize missed quality of life concerns that are more relevant to women? Are women concerned about how many days are spent avoiding sexual activity? Do religious restrictions figure prominently for some? How dissatisfied are they with days of cramping or bloating without bleeding? What do women want to know when they consider the bleeding patterns for their contraceptive options?

The answers to these questions likely vary from patient to patient—and that observation leads us back to grasping the art of contraceptive counseling: Our counseling needs to be concise, relatable, and honest.

The author reports no financial relationships relevant to this article.

Few issues in obstetrics spark as much controversy as home birth—and where controversy rages, media attention follows.

Press reports of a 2008 policy statement on home birth issued by the American Medical Association (AMA) and the American College of Obstetricians and Gynecologists (ACOG) highlight the rift between the formal medical establishment and advocates of home birth.^1-3

On one side, the AMA and ACOG assert that the hospital or an accredited birthing center “is the safest setting for labor, delivery, and the immediate postpartum period.”¹ On the other side, advocates of home birth argue that having the option adds to women’s empowerment and choice.

Some people have accused the medical community of trying to corner the “baby birthing industry.”⁴ The title of a recent Baltimore Sun article sums up this sentiment: “Home birth battle: Doctors strong-arm women away from healthy alternative to hospital care.”⁵

Neither ACOG nor the AMA advocates criminalization of home deliveries, but their statements on home birth have generated considerable fear that they will.

This article explores the controversy, focusing on the literature on home birth, gaps in knowledge, the state of regulation, liaison with midwives, and other issues. It also offers suggestions on how to discuss labor and delivery with patients so that they clearly understand the risks involved and do not feel that they have “failed” at meaningful childbirth when they choose hospital delivery.

Did a rise in hospital births reduce maternal mortality?

Obstetric care changed dramatically in the mid-20th century. In 1940, 55.8% of deliveries occurred in the hospital, but that percentage rose to 99.4 by 1970 and hasn’t changed appreciably since.⁶

Some proponents of hospital delivery note that, in 1940, when 44% of births occurred outside the hospital, the maternal mortality rate was 608 deaths for every 100,000 live births, compared with 37 deaths for every 100,000 live births in 1960, when fewer than 4% of deliveries occurred outside the hospital.⁶ And in 2003, with only 1% of deliveries occurring in a home setting, the maternal mortality rate was even lower: 12 deaths for every 100,000 live births.⁷

Others argue that this sharp decrease in maternal mortality cannot be attributed solely to the change in location of the delivery (and subsequent availability of services and personnel), but reflects universal advancement in safe practices such as aseptic technique.⁸

What do the data show? All studies of home birth have serious methodologic flaws, thanks largely to the nature of the subject matter. A recent Cochrane review observes that there is only one randomized, controlled trial—with a sample size of only 11 women—from which to draw conclusions.⁹ The review concludes that “there is no strong evidence to favour either home or hospital birth for selected, low-risk pregnant women.”¹⁰

Most data come from abroad

Much of the literature on home birth comes from international sites because of the higher prevalence of home delivery in other countries. These data reveal that:

• Two percent of deliveries in the United Kingdom occur in the home.¹¹ The British National Institute for Health and Clinical Excellence recommended that all women be offered the option to have their baby at home or in the hospital, although, depending on the “trust” (a geographically based public-system cooperative that provides care), 8% to 76% of women weren’t given this choice formally.¹²
• One study conducted in Switzerland involved 489 women who opted for home birth and 385 who chose hospital birth. Of the former, 37 were referred to a specialist during pregnancy, and 70 were referred during labor. The groups had similar birth weights, gestational ages, and clinical conditions.¹³
• In the Netherlands, 30% of infants are born at home.¹⁴ If a woman has an uncomplicated pregnancy, she remains under midwifery care and can decide where to deliver. A study of 280,000 “low-risk” women under primary midwifery care found that 68.1% completed childbirth under that care, 3.6% were referred urgently, and 28.3% were referred without urgency.¹⁴ When referrals were considered as a whole, 11.2% involved urgency, primarily for fetal distress (50.2%) and postpartum hemorrhage (33%). Adverse neonatal outcomes were most common in urgently referred cases, followed by nonurgent referrals. The authors acknowledge the importance of transport time once a referral is initiated, stating that, “The Netherlands is a very densely populated country where the average distance to the hospital is relatively short.” (The same cannot be said of many parts of rural America.)
• A study involving home deliveries in Australia from 1985 to 1990 identified 50 perinatal deaths out of 7,002 planned home births.¹⁵ The perinatal death rate of infants weighing more than 2,500 g exceeded the national average (5.7 versus 3.6 for every 1,000 deliveries), with a relative risk (RR) of 1.6 (95% confidence interval [CI], 1.1–1.4). Intrapartum death not attributable to prematurity or fetal malformation was also higher (2.7 versus 0.9 for every 1,000 deliveries), with a RR of 3.0 (95% CI, 1.9–4.8). According to the authors, the main contributors to excess mortality were underestimation of the risks associated with post-term birth, twin pregnancy, and breech presentation, and a lack of response to fetal distress.

Is the evidence on water birth just too murky?

In the summer of 1999, a woman delivered a 7.7-lb infant after 42 weeks of gestation. The birth took place in the woman’s home in Japan, and the baby was delivered in a bathtub of warm water. The woman had had an uneventful pregnancy, and the baby appeared to be perfectly normal.

Four days later, the infant developed fever and jaundice and was admitted to the hospital, where she was treated with phototherapy. She improved, but her symptoms recurred 3 days later, and she began to vomit. Eight days after birth, she suffered cardiopulmonary arrest and died. An autopsy revealed the cause of death to be legionellosis—infection with Legionella pneumonia. The most likely source was the bathtub in which she was born.⁴³

Other case reports describe similar tragedies associated with water birth (among them, drowning, infection, and a snapped umbilical cord), but no randomized, clinical trial has systematically compared delivery in water with conventional land-based birth.

The death, morbidity, and lack of data so troubled members of the American Academy of Pediatrics that the Committee on Fetus and Newborn issued an advisory in 2005:

• The safety and efficacy of underwater birth for the newborn has not been established. There is no convincing evidence of benefit to the neonate but some concern for serious harm. Therefore, underwater birth should be considered an experimental procedure that should not be performed except within the context of an appropriately designed randomized clinical trial after informed parental consent.⁴⁴

This statement contrasts the conclusion of the most recent Cochrane review of the subject, which found that, “Immersion in water during the first stage of labour significantly reduces women’s perception of pain and use of epidural/spinal analgesia.”⁴⁵ The review also noted, however, that, “No trials could be located that assessed the immersion of women in water during the third stage of labour.”⁴⁵

No studies have explored immersion in water during the third stage of labor.

What’s in that water?

Amy Tuteur, MD, an ObGyn who publishes a popular blog (“The Skeptical OB”), focused on the topic of water birth earlier this year. “What’s in the water at waterbirth?” she asks.⁴⁶

To answer the question, Dr. Tuteur cites a 1999 study of 4,030 deliveries in water, which found that 35 infants suffered serious morbidity and three died—although it is unclear if any of the deaths were a direct result of water birth. “However, of the 32 survivors who were admitted to the NICU,” writes Dr. Tuteur, “13 had significant respiratory problems, including pneumonia, meconium aspiration, water aspiration, and drowning. Other complications attributable to water birth include five babies who had significant hemorrhage due to snapped umbilical cord. In all, 18 babies had serious complications directly attributable to waterbirth.”⁴⁷

Dr. Tuteur also points to the poor quality of the water in birthing pools, arguing that it is “essentially toilet water.”⁴⁶ “The water in a birth pool, conveniently heated to body temperature, the optimum temperature for bacterial growth, is a microbial paradise,” she writes.⁴⁶ She cites a study of 1,500 water births that included analysis of the water found in the birthing pools (before anyone entered the water) and identified:

• coliforms in 21% of samples
• enterococcus in 19% of samples
• Escherichia coli in 10% of samples
• Legionella pneumophila in 12% of samples
• Pseudomonas aeruginosa in 11% of samples.⁴⁸

After a special water filter was installed, contamination diminished but did not disappear completely.

Pools in the home setting were not the only ones implicated in contamination; some hospital pools also were affected.

What’s the bottom line?

The American College of Obstetricians and Gynecologists has yet to weigh in on the matter. Until it does, ObGyns may be wise to heed the words of Ruth Gilbert, MD, of the Centre for Paediatric Epidemiology and Biostatistics at the Institute of Child Health in London.

“Can delivery in water cause serious adverse outcomes?” she asks, rhetorically, it turns out.

“Undoubtedly, the answer is ‘yes.’”⁴⁹ —JANELLE YATES, SENIOR EDITOR

The data we do have are difficult to interpret

Among the limitations of studies of home birth are:

• lack of follow-up after the delivery
• varying definitions of perinatal mortality internationally
• lack of clarity regarding the identity and education of delivering providers
• the fact that there are often “too few neonatal deaths from which to extrapolate reliable rate calculations.”¹⁶

One meta-analysis found a rate of intrapartum transfer ranging from 7.4% to 16.5%, and a rate of primary cesarean delivery of 1.4% to 17.7% (it was 13.8% to 28.25% in the “comparison group”).¹⁶

A challenge inherent in many of these studies is identifying exactly what the comparison group is. In addition, some of the data are obtained from discharge summary records, which don’t always reflect the level of risk or acuity.

Oft-cited study has weaknesses

The study that many advocates of home birth cite was conducted in the United States and Canada and published in 2005.¹⁷ It evaluated “all 5,418 women expecting to deliver in 2000 supported by midwives with a common certification [certified professional midwives] and who planned to deliver at home when labour began.” The hospital transfer rate was 12.1%, in line with other studies. The risk of adverse outcomes was lower in the group that planned to have home delivery, compared with a “relatively low-risk hospital group.”

The study focused on:

• electronic fetal monitoring, used in 9.6% of deliveries in the home-birth group, versus 84.3% of the hospital group
• episiotomy, performed in 2.1% of home deliveries, compared with 33% of hospital births
• cesarean delivery, 3.7% of planned home deliveries, versus 19% of hospital births
• vacuum-assisted vaginal delivery, performed in 0.6% of planned home deliveries, versus 5.5% of hospital births
• neonatal death, at a rate of 2.0 deaths for every 1,000 intended home births. No comparison figure was cited.

One of the weaknesses of this study, as of others, was identification of a comparison group as a “low-risk” population without data to back up that designation. In addition, this study derived its data from birth certificates for 3,360,868 singleton, vertex births at 37 weeks or more of gestation. Data from birth certificates are limited as a basis for accurate risk assessment. Moreover, although the authors of this study asserted that they had no conflict of interest, the investigation was funded by The Foundation for the Advancement of Midwifery.

Study cited by advocates of hospital birth is also flawed

One of the studies many hospital and birthing center advocates cite was published in 2002.¹⁸ It involved an analysis of birth registry information on uncomplicated singleton pregnancies at 34 weeks or more of gestation in Washington state between 1989 and 1996. These pregnancies were either:

• delivered at home by a health professional (n=5,854)
• transferred to medical facilities after attempted home delivery (n=279)
• planned to be delivered in the hospital (n=10,593).

Infants whose mothers planned to deliver at home had a higher risk of neonatal death (RR, 1.99; 95% CI, 1.06–3.73) and a higher risk of having a 5-minute Apgar score of less than 3 (RR, 2.31; 95% CI, 1.29–4.16). After adjustment for a gestational-age cutoff of 37 weeks, these risks remained similar.

Nulliparous women, in particular, had a higher risk for prolonged labor (RR, 1.73; 95% CI, 1.28–2.34) and postpartum bleeding (RR, 2.76; 95% CI, 1.74–4.36).

The authors themselves point out a potential flaw in this study: the use of data from birth certificates. These data create “the potential for misclassifying unplanned home births as planned home births.” The difference in outcomes could be significant. For example, the neonatal death rate for unplanned home deliveries in North Carolina and Kentucky was 18 to 20 times higher than the rate for planned home births in these states.^19,20

A study from Missouri observes that neonatal mortality was elevated for both planned and unplanned home birth, compared with physician-attended hospital birth.²¹

Selection bias is a concern

Selection bias is an inherent difficulty in many of these studies. Except for one previously mentioned paper—a very small study—none of the investigations involve randomization. As a result, we cannot exclude the possibility that “women who choose to deliver at home or in a birth center are likely to be different in terms of expectations and approach from women choosing to deliver in hospitals.”²²

Risk level can escalate rapidly

What is potentially troubling about home birth is the fact that a low-risk pregnancy that was complication-free during antepartum care can become a high-risk pregnancy in a matter of minutes, necessitating urgent, appropriate obstetric care. Some classic examples of urgent events include cord prolapse, postpartum hemorrhage, bleeding from vasa previa, and shoulder dystocia.

Let’s focus on shoulder dystocia, which occurs in 1.4% of all vaginal deliveries. The authors of one study point out that “most of the traditional risk factors for shoulder dystocia have no predictive value, shoulder dystocia itself is an unpredictable event, and infants at risk for permanent injury are virtually impossible to predict.”²³ This may make delivery in the home a high-risk endeavor because of the inability to mobilize an obstetric team to assist with shoulder dystocia maneuvers or perform a Zavanelli delivery.

Another variable overlooked in most studies is the speed of transfer and the outcomes of pregnancies in which the women intended to deliver at home but ended up requiring urgent transfer. One study that did examine this scenario found that “women who had booked for a home birth, but later needed to transfer their care for a hospital birth, appeared to have the highest risk of intrapartum-related perinatal mortality.”²⁴

There is also some controversy regarding the delivery of women who are pregnant with twins, who have a fetus in breech presentation, or who have a history of cesarean delivery. One study examined outcomes for intended home delivery of 57 women who had a prior abdominal delivery.²⁵ Fifty of these women delivered vaginally in the home, and seven (12.3%) delivered in the hospital. One hospital transfer was urgent for fetal distress. One baby was stillborn, delivered at home.

A 10-year prospective study of vaginal birth after cesarean (VBAC) in birth centers found that more than 50% of uterine ruptures and 57% of perinatal deaths involved the 10% of women who had more than one prior cesarean delivery or who had reached a gestational age of more than 42 weeks.²⁶

Skill of the caregiver is important

The training and qualifications of the obstetric care provider are incredibly important. One study evaluated 4,361 home births attended by “apprentice-trained midwives from 1970 to 1985 and 4,107 home births attended by family physicians from 1969 to 1981.”²⁷ The perinatal mortality rate for the midwife-attended births was 14 for every 1,000 births, in contrast to the rate of 5 for every 1,000 physician-attended births.

Three types of midwife are credentialed in this country:

• certified nurse-midwife (CNM)
• certified midwife (CM)
• certified professional midwife (CPM).

The first two categories are certified by the American Midwifery Certification Board (AMCB). CNMs and CMs undergo rigorous training and examination, and this designation will require a graduate degree within the next few years. The CPM category, however, requires much less rigorous training. Its midwives are certified by the North American Registry of Midwives. The clinical requirements for certification as a CPM include:

• attending a “minimum of 20 births”
• managing at least 20 additional births, at least half of them in the home or another out-of-hospital setting
• performing a small number of prenatal, newborn, and postpartum exams.²⁸

A high school diploma is not required.

I suspect that concerns about this lax certification process contributed to ACOG’s decision to issue a statement from its executive board in 2006: “While ACOG supports women having a choice in determining their providers of care, ACOG does not support the provision of care by lay midwives or other midwives who are not certified by the American College of Nurse-Midwives (ACNM) or AMCB.”²⁹

A number of midwifery advocates have made a legislative push to expand licensure for CPMs in this country, and the debate continues on a state-by-state basis.³⁰

Economics and other variables affect delivery decision

Some advocates of home birth note that the “average uncomplicated vaginal birth costs 68% less in a home than in a hospital.”³¹ Others try to organize support for women who want to give birth at home, such as the Home Birth Hotline, a voluntary, UK-based organization.³²

Some articles suggest that patient satisfaction is of significant importance in the decision about where to deliver. One noted that women who delivered where they had planned had higher overall satisfaction when that place was in the home (P<.01).³³

A randomized, controlled trial (n=3,510) simulated home delivery in a hospital, with “home delivery” patients having midwifery care in a room “similar to one in one’s own home” and the others having “consultant-led care” in rooms in the delivery suite that contained equipment to resuscitate both mother and baby, as well as monitors and other technology.³⁴ This study found no significant differences in measured outcomes, but “generally higher levels of satisfaction” among the women who had simulated home delivery.

A study from “remote and rural Scotland” found that most women “expressed a preference to give birth in hospital and have consultant-led care because they felt safer.”³⁵

Does the rhetoric surrounding home birth “empower” women?

Another frequently overlooked issue is the passionate rhetoric used to describe home birth—and the effect of that passion on women whose birth plan doesn’t play out as expected. Words such as “choice” and “empowerment” are often used. Regrettably, there is considerable mistrust of the medical system.

One woman describes how her planned home delivery, “influenced by the feminist literature,” went awry.³⁸ After a long labor, she wrote, she “just wanted the baby out, safe and healthy. It no longer mattered how it happened….I couldn’t get rid of the underlying feeling that I had ‘failed’ in some way….”³⁸

Because of her strong desire for home delivery, this woman was deeply affected when the delivery became difficult: “I did not have the authority to proclaim whether or not various medical interventions were necessary, or whether my case actually did constitute a medical emergency….Faced with these ‘options’—safe birth or potential death—how could I be said to be making a ‘choice’?…The obstetrician has more power than the woman because s/he has more knowledge.”³⁸

Despite having come to this realization, and delivering a healthy baby, she still experienced “a sense of disappointment and anger” and “traumatic flashbacks.”

I worry that patients may become so caught up in the rhetoric of their own power and choice that, when uncontrollable events occur, the happiness of a healthy delivery is overshadowed by deep disappointment.

Heated debate isn’t helpful

An unfortunate rift seems to have developed between some members of the midwifery community and some physicians. ACOG and the ACNM have a longstanding policy that: “In those circumstances in which obstetrician/gynecologists and certified nurse-midwives/certified midwives collaborate in the care of women, the quality of those practices is enhanced by a working relationship characterized by mutual respect and trust.”³⁹

Whether individual physicians agree with the practice of planned home birth or not, the health and welfare of the patient must be paramount. The American Public Health Association and the ACNM support home birth.^40,41

When obstetric emergencies do arise in the home setting, necessitating emergent transfer, it is critical that the transfer be managed in a way that ensures the best outcome.

One disturbing article describes both “disarticulations” that occur “when there is no correspondence of information or action between the midwife and the hospital staff” and “fractured articulations” that arise from “partial and incomplete correspondence.”⁴² A number of midwives were interviewed who no longer feel comfortable bringing patients to certain hospitals because of the negative response they received from health-care providers, sometimes to the detriment of the patient.

Can we improve the situation?

First, we need to choose our words carefully when we counsel women about labor and delivery, in recognition of the buzzwords used by advocates of home birth (“empowerment,” “choice”) and the sense of failure and distress some women feel when they eventually require heightened medical intervention.

Perhaps we should dispense with the term “failure,” as in failure to progress, failure to dilate, and so on, to avoid implying that this “failure” is the woman’s fault. And instead of saying that a patient’s pelvis is “adequate,” implying that another woman’s pelvis isn’t, we could use a term that sounds less judgmental.

We can also make the hospital environment more nurturing and supportive of women’s choices for labor, as long as safety isn’t compromised. And when we receive a transfer of a patient whose home delivery has gone awry, we should openly, efficiently, and professionally communicate with the home-delivery provider to best benefit the patient, regardless of our feelings on the subject.

Home birth isn’t going away

That’s my take on the literature. There are certainly data supporting the safety of home birth for the vast majority of women who choose it, but there is also a significant number of women who will experience unpredictable events that could be fatal if blood products or surgery isn’t rapidly available. For that reason, and in light of the very high stakes involved, I wonder: Why take that chance?

The author reports no financial relationships relevant to this article.

Few issues in obstetrics spark as much controversy as home birth—and where controversy rages, media attention follows.

Press reports of a 2008 policy statement on home birth issued by the American Medical Association (AMA) and the American College of Obstetricians and Gynecologists (ACOG) highlight the rift between the formal medical establishment and advocates of home birth.^1-3

On one side, the AMA and ACOG assert that the hospital or an accredited birthing center “is the safest setting for labor, delivery, and the immediate postpartum period.”¹ On the other side, advocates of home birth argue that having the option adds to women’s empowerment and choice.

Some people have accused the medical community of trying to corner the “baby birthing industry.”⁴ The title of a recent Baltimore Sun article sums up this sentiment: “Home birth battle: Doctors strong-arm women away from healthy alternative to hospital care.”⁵

Neither ACOG nor the AMA advocates criminalization of home deliveries, but their statements on home birth have generated considerable fear that they will.

This article explores the controversy, focusing on the literature on home birth, gaps in knowledge, the state of regulation, liaison with midwives, and other issues. It also offers suggestions on how to discuss labor and delivery with patients so that they clearly understand the risks involved and do not feel that they have “failed” at meaningful childbirth when they choose hospital delivery.

Did a rise in hospital births reduce maternal mortality?

Obstetric care changed dramatically in the mid-20th century. In 1940, 55.8% of deliveries occurred in the hospital, but that percentage rose to 99.4 by 1970 and hasn’t changed appreciably since.⁶

Some proponents of hospital delivery note that, in 1940, when 44% of births occurred outside the hospital, the maternal mortality rate was 608 deaths for every 100,000 live births, compared with 37 deaths for every 100,000 live births in 1960, when fewer than 4% of deliveries occurred outside the hospital.⁶ And in 2003, with only 1% of deliveries occurring in a home setting, the maternal mortality rate was even lower: 12 deaths for every 100,000 live births.⁷

Others argue that this sharp decrease in maternal mortality cannot be attributed solely to the change in location of the delivery (and subsequent availability of services and personnel), but reflects universal advancement in safe practices such as aseptic technique.⁸

What do the data show? All studies of home birth have serious methodologic flaws, thanks largely to the nature of the subject matter. A recent Cochrane review observes that there is only one randomized, controlled trial—with a sample size of only 11 women—from which to draw conclusions.⁹ The review concludes that “there is no strong evidence to favour either home or hospital birth for selected, low-risk pregnant women.”¹⁰

Most data come from abroad

Much of the literature on home birth comes from international sites because of the higher prevalence of home delivery in other countries. These data reveal that:

• Two percent of deliveries in the United Kingdom occur in the home.¹¹ The British National Institute for Health and Clinical Excellence recommended that all women be offered the option to have their baby at home or in the hospital, although, depending on the “trust” (a geographically based public-system cooperative that provides care), 8% to 76% of women weren’t given this choice formally.¹²
• One study conducted in Switzerland involved 489 women who opted for home birth and 385 who chose hospital birth. Of the former, 37 were referred to a specialist during pregnancy, and 70 were referred during labor. The groups had similar birth weights, gestational ages, and clinical conditions.¹³
• In the Netherlands, 30% of infants are born at home.¹⁴ If a woman has an uncomplicated pregnancy, she remains under midwifery care and can decide where to deliver. A study of 280,000 “low-risk” women under primary midwifery care found that 68.1% completed childbirth under that care, 3.6% were referred urgently, and 28.3% were referred without urgency.¹⁴ When referrals were considered as a whole, 11.2% involved urgency, primarily for fetal distress (50.2%) and postpartum hemorrhage (33%). Adverse neonatal outcomes were most common in urgently referred cases, followed by nonurgent referrals. The authors acknowledge the importance of transport time once a referral is initiated, stating that, “The Netherlands is a very densely populated country where the average distance to the hospital is relatively short.” (The same cannot be said of many parts of rural America.)
• A study involving home deliveries in Australia from 1985 to 1990 identified 50 perinatal deaths out of 7,002 planned home births.¹⁵ The perinatal death rate of infants weighing more than 2,500 g exceeded the national average (5.7 versus 3.6 for every 1,000 deliveries), with a relative risk (RR) of 1.6 (95% confidence interval [CI], 1.1–1.4). Intrapartum death not attributable to prematurity or fetal malformation was also higher (2.7 versus 0.9 for every 1,000 deliveries), with a RR of 3.0 (95% CI, 1.9–4.8). According to the authors, the main contributors to excess mortality were underestimation of the risks associated with post-term birth, twin pregnancy, and breech presentation, and a lack of response to fetal distress.

Is the evidence on water birth just too murky?

In the summer of 1999, a woman delivered a 7.7-lb infant after 42 weeks of gestation. The birth took place in the woman’s home in Japan, and the baby was delivered in a bathtub of warm water. The woman had had an uneventful pregnancy, and the baby appeared to be perfectly normal.

Four days later, the infant developed fever and jaundice and was admitted to the hospital, where she was treated with phototherapy. She improved, but her symptoms recurred 3 days later, and she began to vomit. Eight days after birth, she suffered cardiopulmonary arrest and died. An autopsy revealed the cause of death to be legionellosis—infection with Legionella pneumonia. The most likely source was the bathtub in which she was born.⁴³

Other case reports describe similar tragedies associated with water birth (among them, drowning, infection, and a snapped umbilical cord), but no randomized, clinical trial has systematically compared delivery in water with conventional land-based birth.

The death, morbidity, and lack of data so troubled members of the American Academy of Pediatrics that the Committee on Fetus and Newborn issued an advisory in 2005:

• The safety and efficacy of underwater birth for the newborn has not been established. There is no convincing evidence of benefit to the neonate but some concern for serious harm. Therefore, underwater birth should be considered an experimental procedure that should not be performed except within the context of an appropriately designed randomized clinical trial after informed parental consent.⁴⁴

This statement contrasts the conclusion of the most recent Cochrane review of the subject, which found that, “Immersion in water during the first stage of labour significantly reduces women’s perception of pain and use of epidural/spinal analgesia.”⁴⁵ The review also noted, however, that, “No trials could be located that assessed the immersion of women in water during the third stage of labour.”⁴⁵

No studies have explored immersion in water during the third stage of labor.

What’s in that water?

Amy Tuteur, MD, an ObGyn who publishes a popular blog (“The Skeptical OB”), focused on the topic of water birth earlier this year. “What’s in the water at waterbirth?” she asks.⁴⁶

To answer the question, Dr. Tuteur cites a 1999 study of 4,030 deliveries in water, which found that 35 infants suffered serious morbidity and three died—although it is unclear if any of the deaths were a direct result of water birth. “However, of the 32 survivors who were admitted to the NICU,” writes Dr. Tuteur, “13 had significant respiratory problems, including pneumonia, meconium aspiration, water aspiration, and drowning. Other complications attributable to water birth include five babies who had significant hemorrhage due to snapped umbilical cord. In all, 18 babies had serious complications directly attributable to waterbirth.”⁴⁷

Dr. Tuteur also points to the poor quality of the water in birthing pools, arguing that it is “essentially toilet water.”⁴⁶ “The water in a birth pool, conveniently heated to body temperature, the optimum temperature for bacterial growth, is a microbial paradise,” she writes.⁴⁶ She cites a study of 1,500 water births that included analysis of the water found in the birthing pools (before anyone entered the water) and identified:

• coliforms in 21% of samples
• enterococcus in 19% of samples
• Escherichia coli in 10% of samples
• Legionella pneumophila in 12% of samples
• Pseudomonas aeruginosa in 11% of samples.⁴⁸

After a special water filter was installed, contamination diminished but did not disappear completely.

Pools in the home setting were not the only ones implicated in contamination; some hospital pools also were affected.

What’s the bottom line?

The American College of Obstetricians and Gynecologists has yet to weigh in on the matter. Until it does, ObGyns may be wise to heed the words of Ruth Gilbert, MD, of the Centre for Paediatric Epidemiology and Biostatistics at the Institute of Child Health in London.

“Can delivery in water cause serious adverse outcomes?” she asks, rhetorically, it turns out.

“Undoubtedly, the answer is ‘yes.’”⁴⁹ —JANELLE YATES, SENIOR EDITOR

The data we do have are difficult to interpret

Among the limitations of studies of home birth are:

• lack of follow-up after the delivery
• varying definitions of perinatal mortality internationally
• lack of clarity regarding the identity and education of delivering providers
• the fact that there are often “too few neonatal deaths from which to extrapolate reliable rate calculations.”¹⁶

One meta-analysis found a rate of intrapartum transfer ranging from 7.4% to 16.5%, and a rate of primary cesarean delivery of 1.4% to 17.7% (it was 13.8% to 28.25% in the “comparison group”).¹⁶

A challenge inherent in many of these studies is identifying exactly what the comparison group is. In addition, some of the data are obtained from discharge summary records, which don’t always reflect the level of risk or acuity.

Oft-cited study has weaknesses

The study that many advocates of home birth cite was conducted in the United States and Canada and published in 2005.¹⁷ It evaluated “all 5,418 women expecting to deliver in 2000 supported by midwives with a common certification [certified professional midwives] and who planned to deliver at home when labour began.” The hospital transfer rate was 12.1%, in line with other studies. The risk of adverse outcomes was lower in the group that planned to have home delivery, compared with a “relatively low-risk hospital group.”

The study focused on:

• electronic fetal monitoring, used in 9.6% of deliveries in the home-birth group, versus 84.3% of the hospital group
• episiotomy, performed in 2.1% of home deliveries, compared with 33% of hospital births
• cesarean delivery, 3.7% of planned home deliveries, versus 19% of hospital births
• vacuum-assisted vaginal delivery, performed in 0.6% of planned home deliveries, versus 5.5% of hospital births
• neonatal death, at a rate of 2.0 deaths for every 1,000 intended home births. No comparison figure was cited.

One of the weaknesses of this study, as of others, was identification of a comparison group as a “low-risk” population without data to back up that designation. In addition, this study derived its data from birth certificates for 3,360,868 singleton, vertex births at 37 weeks or more of gestation. Data from birth certificates are limited as a basis for accurate risk assessment. Moreover, although the authors of this study asserted that they had no conflict of interest, the investigation was funded by The Foundation for the Advancement of Midwifery.

Study cited by advocates of hospital birth is also flawed

One of the studies many hospital and birthing center advocates cite was published in 2002.¹⁸ It involved an analysis of birth registry information on uncomplicated singleton pregnancies at 34 weeks or more of gestation in Washington state between 1989 and 1996. These pregnancies were either:

• delivered at home by a health professional (n=5,854)
• transferred to medical facilities after attempted home delivery (n=279)
• planned to be delivered in the hospital (n=10,593).

Infants whose mothers planned to deliver at home had a higher risk of neonatal death (RR, 1.99; 95% CI, 1.06–3.73) and a higher risk of having a 5-minute Apgar score of less than 3 (RR, 2.31; 95% CI, 1.29–4.16). After adjustment for a gestational-age cutoff of 37 weeks, these risks remained similar.

Nulliparous women, in particular, had a higher risk for prolonged labor (RR, 1.73; 95% CI, 1.28–2.34) and postpartum bleeding (RR, 2.76; 95% CI, 1.74–4.36).

The authors themselves point out a potential flaw in this study: the use of data from birth certificates. These data create “the potential for misclassifying unplanned home births as planned home births.” The difference in outcomes could be significant. For example, the neonatal death rate for unplanned home deliveries in North Carolina and Kentucky was 18 to 20 times higher than the rate for planned home births in these states.^19,20

A study from Missouri observes that neonatal mortality was elevated for both planned and unplanned home birth, compared with physician-attended hospital birth.²¹

Selection bias is a concern

Selection bias is an inherent difficulty in many of these studies. Except for one previously mentioned paper—a very small study—none of the investigations involve randomization. As a result, we cannot exclude the possibility that “women who choose to deliver at home or in a birth center are likely to be different in terms of expectations and approach from women choosing to deliver in hospitals.”²²

Risk level can escalate rapidly

What is potentially troubling about home birth is the fact that a low-risk pregnancy that was complication-free during antepartum care can become a high-risk pregnancy in a matter of minutes, necessitating urgent, appropriate obstetric care. Some classic examples of urgent events include cord prolapse, postpartum hemorrhage, bleeding from vasa previa, and shoulder dystocia.

Let’s focus on shoulder dystocia, which occurs in 1.4% of all vaginal deliveries. The authors of one study point out that “most of the traditional risk factors for shoulder dystocia have no predictive value, shoulder dystocia itself is an unpredictable event, and infants at risk for permanent injury are virtually impossible to predict.”²³ This may make delivery in the home a high-risk endeavor because of the inability to mobilize an obstetric team to assist with shoulder dystocia maneuvers or perform a Zavanelli delivery.

Another variable overlooked in most studies is the speed of transfer and the outcomes of pregnancies in which the women intended to deliver at home but ended up requiring urgent transfer. One study that did examine this scenario found that “women who had booked for a home birth, but later needed to transfer their care for a hospital birth, appeared to have the highest risk of intrapartum-related perinatal mortality.”²⁴

There is also some controversy regarding the delivery of women who are pregnant with twins, who have a fetus in breech presentation, or who have a history of cesarean delivery. One study examined outcomes for intended home delivery of 57 women who had a prior abdominal delivery.²⁵ Fifty of these women delivered vaginally in the home, and seven (12.3%) delivered in the hospital. One hospital transfer was urgent for fetal distress. One baby was stillborn, delivered at home.

A 10-year prospective study of vaginal birth after cesarean (VBAC) in birth centers found that more than 50% of uterine ruptures and 57% of perinatal deaths involved the 10% of women who had more than one prior cesarean delivery or who had reached a gestational age of more than 42 weeks.²⁶

Skill of the caregiver is important

The training and qualifications of the obstetric care provider are incredibly important. One study evaluated 4,361 home births attended by “apprentice-trained midwives from 1970 to 1985 and 4,107 home births attended by family physicians from 1969 to 1981.”²⁷ The perinatal mortality rate for the midwife-attended births was 14 for every 1,000 births, in contrast to the rate of 5 for every 1,000 physician-attended births.

Three types of midwife are credentialed in this country:

• certified nurse-midwife (CNM)
• certified midwife (CM)
• certified professional midwife (CPM).

The first two categories are certified by the American Midwifery Certification Board (AMCB). CNMs and CMs undergo rigorous training and examination, and this designation will require a graduate degree within the next few years. The CPM category, however, requires much less rigorous training. Its midwives are certified by the North American Registry of Midwives. The clinical requirements for certification as a CPM include:

• attending a “minimum of 20 births”
• managing at least 20 additional births, at least half of them in the home or another out-of-hospital setting
• performing a small number of prenatal, newborn, and postpartum exams.²⁸

A high school diploma is not required.

I suspect that concerns about this lax certification process contributed to ACOG’s decision to issue a statement from its executive board in 2006: “While ACOG supports women having a choice in determining their providers of care, ACOG does not support the provision of care by lay midwives or other midwives who are not certified by the American College of Nurse-Midwives (ACNM) or AMCB.”²⁹

A number of midwifery advocates have made a legislative push to expand licensure for CPMs in this country, and the debate continues on a state-by-state basis.³⁰

Economics and other variables affect delivery decision

Some advocates of home birth note that the “average uncomplicated vaginal birth costs 68% less in a home than in a hospital.”³¹ Others try to organize support for women who want to give birth at home, such as the Home Birth Hotline, a voluntary, UK-based organization.³²

Some articles suggest that patient satisfaction is of significant importance in the decision about where to deliver. One noted that women who delivered where they had planned had higher overall satisfaction when that place was in the home (P<.01).³³

A randomized, controlled trial (n=3,510) simulated home delivery in a hospital, with “home delivery” patients having midwifery care in a room “similar to one in one’s own home” and the others having “consultant-led care” in rooms in the delivery suite that contained equipment to resuscitate both mother and baby, as well as monitors and other technology.³⁴ This study found no significant differences in measured outcomes, but “generally higher levels of satisfaction” among the women who had simulated home delivery.

A study from “remote and rural Scotland” found that most women “expressed a preference to give birth in hospital and have consultant-led care because they felt safer.”³⁵

Does the rhetoric surrounding home birth “empower” women?

Another frequently overlooked issue is the passionate rhetoric used to describe home birth—and the effect of that passion on women whose birth plan doesn’t play out as expected. Words such as “choice” and “empowerment” are often used. Regrettably, there is considerable mistrust of the medical system.

One woman describes how her planned home delivery, “influenced by the feminist literature,” went awry.³⁸ After a long labor, she wrote, she “just wanted the baby out, safe and healthy. It no longer mattered how it happened….I couldn’t get rid of the underlying feeling that I had ‘failed’ in some way….”³⁸

Because of her strong desire for home delivery, this woman was deeply affected when the delivery became difficult: “I did not have the authority to proclaim whether or not various medical interventions were necessary, or whether my case actually did constitute a medical emergency….Faced with these ‘options’—safe birth or potential death—how could I be said to be making a ‘choice’?…The obstetrician has more power than the woman because s/he has more knowledge.”³⁸

Despite having come to this realization, and delivering a healthy baby, she still experienced “a sense of disappointment and anger” and “traumatic flashbacks.”

I worry that patients may become so caught up in the rhetoric of their own power and choice that, when uncontrollable events occur, the happiness of a healthy delivery is overshadowed by deep disappointment.

Heated debate isn’t helpful

An unfortunate rift seems to have developed between some members of the midwifery community and some physicians. ACOG and the ACNM have a longstanding policy that: “In those circumstances in which obstetrician/gynecologists and certified nurse-midwives/certified midwives collaborate in the care of women, the quality of those practices is enhanced by a working relationship characterized by mutual respect and trust.”³⁹

Whether individual physicians agree with the practice of planned home birth or not, the health and welfare of the patient must be paramount. The American Public Health Association and the ACNM support home birth.^40,41

When obstetric emergencies do arise in the home setting, necessitating emergent transfer, it is critical that the transfer be managed in a way that ensures the best outcome.

One disturbing article describes both “disarticulations” that occur “when there is no correspondence of information or action between the midwife and the hospital staff” and “fractured articulations” that arise from “partial and incomplete correspondence.”⁴² A number of midwives were interviewed who no longer feel comfortable bringing patients to certain hospitals because of the negative response they received from health-care providers, sometimes to the detriment of the patient.

Can we improve the situation?

First, we need to choose our words carefully when we counsel women about labor and delivery, in recognition of the buzzwords used by advocates of home birth (“empowerment,” “choice”) and the sense of failure and distress some women feel when they eventually require heightened medical intervention.

Perhaps we should dispense with the term “failure,” as in failure to progress, failure to dilate, and so on, to avoid implying that this “failure” is the woman’s fault. And instead of saying that a patient’s pelvis is “adequate,” implying that another woman’s pelvis isn’t, we could use a term that sounds less judgmental.

We can also make the hospital environment more nurturing and supportive of women’s choices for labor, as long as safety isn’t compromised. And when we receive a transfer of a patient whose home delivery has gone awry, we should openly, efficiently, and professionally communicate with the home-delivery provider to best benefit the patient, regardless of our feelings on the subject.

Home birth isn’t going away

That’s my take on the literature. There are certainly data supporting the safety of home birth for the vast majority of women who choose it, but there is also a significant number of women who will experience unpredictable events that could be fatal if blood products or surgery isn’t rapidly available. For that reason, and in light of the very high stakes involved, I wonder: Why take that chance?

The author reports no financial relationships relevant to this article.

Few issues in obstetrics spark as much controversy as home birth—and where controversy rages, media attention follows.

Press reports of a 2008 policy statement on home birth issued by the American Medical Association (AMA) and the American College of Obstetricians and Gynecologists (ACOG) highlight the rift between the formal medical establishment and advocates of home birth.^1-3

On one side, the AMA and ACOG assert that the hospital or an accredited birthing center “is the safest setting for labor, delivery, and the immediate postpartum period.”¹ On the other side, advocates of home birth argue that having the option adds to women’s empowerment and choice.

Some people have accused the medical community of trying to corner the “baby birthing industry.”⁴ The title of a recent Baltimore Sun article sums up this sentiment: “Home birth battle: Doctors strong-arm women away from healthy alternative to hospital care.”⁵

Neither ACOG nor the AMA advocates criminalization of home deliveries, but their statements on home birth have generated considerable fear that they will.

This article explores the controversy, focusing on the literature on home birth, gaps in knowledge, the state of regulation, liaison with midwives, and other issues. It also offers suggestions on how to discuss labor and delivery with patients so that they clearly understand the risks involved and do not feel that they have “failed” at meaningful childbirth when they choose hospital delivery.

Did a rise in hospital births reduce maternal mortality?

Obstetric care changed dramatically in the mid-20th century. In 1940, 55.8% of deliveries occurred in the hospital, but that percentage rose to 99.4 by 1970 and hasn’t changed appreciably since.⁶

Some proponents of hospital delivery note that, in 1940, when 44% of births occurred outside the hospital, the maternal mortality rate was 608 deaths for every 100,000 live births, compared with 37 deaths for every 100,000 live births in 1960, when fewer than 4% of deliveries occurred outside the hospital.⁶ And in 2003, with only 1% of deliveries occurring in a home setting, the maternal mortality rate was even lower: 12 deaths for every 100,000 live births.⁷

Others argue that this sharp decrease in maternal mortality cannot be attributed solely to the change in location of the delivery (and subsequent availability of services and personnel), but reflects universal advancement in safe practices such as aseptic technique.⁸

What do the data show? All studies of home birth have serious methodologic flaws, thanks largely to the nature of the subject matter. A recent Cochrane review observes that there is only one randomized, controlled trial—with a sample size of only 11 women—from which to draw conclusions.⁹ The review concludes that “there is no strong evidence to favour either home or hospital birth for selected, low-risk pregnant women.”¹⁰

Most data come from abroad

Much of the literature on home birth comes from international sites because of the higher prevalence of home delivery in other countries. These data reveal that:

• Two percent of deliveries in the United Kingdom occur in the home.¹¹ The British National Institute for Health and Clinical Excellence recommended that all women be offered the option to have their baby at home or in the hospital, although, depending on the “trust” (a geographically based public-system cooperative that provides care), 8% to 76% of women weren’t given this choice formally.¹²
• One study conducted in Switzerland involved 489 women who opted for home birth and 385 who chose hospital birth. Of the former, 37 were referred to a specialist during pregnancy, and 70 were referred during labor. The groups had similar birth weights, gestational ages, and clinical conditions.¹³
• In the Netherlands, 30% of infants are born at home.¹⁴ If a woman has an uncomplicated pregnancy, she remains under midwifery care and can decide where to deliver. A study of 280,000 “low-risk” women under primary midwifery care found that 68.1% completed childbirth under that care, 3.6% were referred urgently, and 28.3% were referred without urgency.¹⁴ When referrals were considered as a whole, 11.2% involved urgency, primarily for fetal distress (50.2%) and postpartum hemorrhage (33%). Adverse neonatal outcomes were most common in urgently referred cases, followed by nonurgent referrals. The authors acknowledge the importance of transport time once a referral is initiated, stating that, “The Netherlands is a very densely populated country where the average distance to the hospital is relatively short.” (The same cannot be said of many parts of rural America.)
• A study involving home deliveries in Australia from 1985 to 1990 identified 50 perinatal deaths out of 7,002 planned home births.¹⁵ The perinatal death rate of infants weighing more than 2,500 g exceeded the national average (5.7 versus 3.6 for every 1,000 deliveries), with a relative risk (RR) of 1.6 (95% confidence interval [CI], 1.1–1.4). Intrapartum death not attributable to prematurity or fetal malformation was also higher (2.7 versus 0.9 for every 1,000 deliveries), with a RR of 3.0 (95% CI, 1.9–4.8). According to the authors, the main contributors to excess mortality were underestimation of the risks associated with post-term birth, twin pregnancy, and breech presentation, and a lack of response to fetal distress.

Is the evidence on water birth just too murky?

In the summer of 1999, a woman delivered a 7.7-lb infant after 42 weeks of gestation. The birth took place in the woman’s home in Japan, and the baby was delivered in a bathtub of warm water. The woman had had an uneventful pregnancy, and the baby appeared to be perfectly normal.

Four days later, the infant developed fever and jaundice and was admitted to the hospital, where she was treated with phototherapy. She improved, but her symptoms recurred 3 days later, and she began to vomit. Eight days after birth, she suffered cardiopulmonary arrest and died. An autopsy revealed the cause of death to be legionellosis—infection with Legionella pneumonia. The most likely source was the bathtub in which she was born.⁴³

Other case reports describe similar tragedies associated with water birth (among them, drowning, infection, and a snapped umbilical cord), but no randomized, clinical trial has systematically compared delivery in water with conventional land-based birth.

The death, morbidity, and lack of data so troubled members of the American Academy of Pediatrics that the Committee on Fetus and Newborn issued an advisory in 2005:

• The safety and efficacy of underwater birth for the newborn has not been established. There is no convincing evidence of benefit to the neonate but some concern for serious harm. Therefore, underwater birth should be considered an experimental procedure that should not be performed except within the context of an appropriately designed randomized clinical trial after informed parental consent.⁴⁴

This statement contrasts the conclusion of the most recent Cochrane review of the subject, which found that, “Immersion in water during the first stage of labour significantly reduces women’s perception of pain and use of epidural/spinal analgesia.”⁴⁵ The review also noted, however, that, “No trials could be located that assessed the immersion of women in water during the third stage of labour.”⁴⁵

No studies have explored immersion in water during the third stage of labor.

What’s in that water?

Amy Tuteur, MD, an ObGyn who publishes a popular blog (“The Skeptical OB”), focused on the topic of water birth earlier this year. “What’s in the water at waterbirth?” she asks.⁴⁶

To answer the question, Dr. Tuteur cites a 1999 study of 4,030 deliveries in water, which found that 35 infants suffered serious morbidity and three died—although it is unclear if any of the deaths were a direct result of water birth. “However, of the 32 survivors who were admitted to the NICU,” writes Dr. Tuteur, “13 had significant respiratory problems, including pneumonia, meconium aspiration, water aspiration, and drowning. Other complications attributable to water birth include five babies who had significant hemorrhage due to snapped umbilical cord. In all, 18 babies had serious complications directly attributable to waterbirth.”⁴⁷

Dr. Tuteur also points to the poor quality of the water in birthing pools, arguing that it is “essentially toilet water.”⁴⁶ “The water in a birth pool, conveniently heated to body temperature, the optimum temperature for bacterial growth, is a microbial paradise,” she writes.⁴⁶ She cites a study of 1,500 water births that included analysis of the water found in the birthing pools (before anyone entered the water) and identified:

• coliforms in 21% of samples
• enterococcus in 19% of samples
• Escherichia coli in 10% of samples
• Legionella pneumophila in 12% of samples
• Pseudomonas aeruginosa in 11% of samples.⁴⁸

After a special water filter was installed, contamination diminished but did not disappear completely.

Pools in the home setting were not the only ones implicated in contamination; some hospital pools also were affected.

What’s the bottom line?

The American College of Obstetricians and Gynecologists has yet to weigh in on the matter. Until it does, ObGyns may be wise to heed the words of Ruth Gilbert, MD, of the Centre for Paediatric Epidemiology and Biostatistics at the Institute of Child Health in London.

“Can delivery in water cause serious adverse outcomes?” she asks, rhetorically, it turns out.

“Undoubtedly, the answer is ‘yes.’”⁴⁹ —JANELLE YATES, SENIOR EDITOR

The data we do have are difficult to interpret

Among the limitations of studies of home birth are:

• lack of follow-up after the delivery
• varying definitions of perinatal mortality internationally
• lack of clarity regarding the identity and education of delivering providers
• the fact that there are often “too few neonatal deaths from which to extrapolate reliable rate calculations.”¹⁶

One meta-analysis found a rate of intrapartum transfer ranging from 7.4% to 16.5%, and a rate of primary cesarean delivery of 1.4% to 17.7% (it was 13.8% to 28.25% in the “comparison group”).¹⁶

A challenge inherent in many of these studies is identifying exactly what the comparison group is. In addition, some of the data are obtained from discharge summary records, which don’t always reflect the level of risk or acuity.

Oft-cited study has weaknesses

The study that many advocates of home birth cite was conducted in the United States and Canada and published in 2005.¹⁷ It evaluated “all 5,418 women expecting to deliver in 2000 supported by midwives with a common certification [certified professional midwives] and who planned to deliver at home when labour began.” The hospital transfer rate was 12.1%, in line with other studies. The risk of adverse outcomes was lower in the group that planned to have home delivery, compared with a “relatively low-risk hospital group.”

The study focused on:

• electronic fetal monitoring, used in 9.6% of deliveries in the home-birth group, versus 84.3% of the hospital group
• episiotomy, performed in 2.1% of home deliveries, compared with 33% of hospital births
• cesarean delivery, 3.7% of planned home deliveries, versus 19% of hospital births
• vacuum-assisted vaginal delivery, performed in 0.6% of planned home deliveries, versus 5.5% of hospital births
• neonatal death, at a rate of 2.0 deaths for every 1,000 intended home births. No comparison figure was cited.

One of the weaknesses of this study, as of others, was identification of a comparison group as a “low-risk” population without data to back up that designation. In addition, this study derived its data from birth certificates for 3,360,868 singleton, vertex births at 37 weeks or more of gestation. Data from birth certificates are limited as a basis for accurate risk assessment. Moreover, although the authors of this study asserted that they had no conflict of interest, the investigation was funded by The Foundation for the Advancement of Midwifery.

Study cited by advocates of hospital birth is also flawed

One of the studies many hospital and birthing center advocates cite was published in 2002.¹⁸ It involved an analysis of birth registry information on uncomplicated singleton pregnancies at 34 weeks or more of gestation in Washington state between 1989 and 1996. These pregnancies were either:

• delivered at home by a health professional (n=5,854)
• transferred to medical facilities after attempted home delivery (n=279)
• planned to be delivered in the hospital (n=10,593).

Infants whose mothers planned to deliver at home had a higher risk of neonatal death (RR, 1.99; 95% CI, 1.06–3.73) and a higher risk of having a 5-minute Apgar score of less than 3 (RR, 2.31; 95% CI, 1.29–4.16). After adjustment for a gestational-age cutoff of 37 weeks, these risks remained similar.

Nulliparous women, in particular, had a higher risk for prolonged labor (RR, 1.73; 95% CI, 1.28–2.34) and postpartum bleeding (RR, 2.76; 95% CI, 1.74–4.36).

The authors themselves point out a potential flaw in this study: the use of data from birth certificates. These data create “the potential for misclassifying unplanned home births as planned home births.” The difference in outcomes could be significant. For example, the neonatal death rate for unplanned home deliveries in North Carolina and Kentucky was 18 to 20 times higher than the rate for planned home births in these states.^19,20

A study from Missouri observes that neonatal mortality was elevated for both planned and unplanned home birth, compared with physician-attended hospital birth.²¹

Selection bias is a concern

Selection bias is an inherent difficulty in many of these studies. Except for one previously mentioned paper—a very small study—none of the investigations involve randomization. As a result, we cannot exclude the possibility that “women who choose to deliver at home or in a birth center are likely to be different in terms of expectations and approach from women choosing to deliver in hospitals.”²²

Risk level can escalate rapidly

What is potentially troubling about home birth is the fact that a low-risk pregnancy that was complication-free during antepartum care can become a high-risk pregnancy in a matter of minutes, necessitating urgent, appropriate obstetric care. Some classic examples of urgent events include cord prolapse, postpartum hemorrhage, bleeding from vasa previa, and shoulder dystocia.

Let’s focus on shoulder dystocia, which occurs in 1.4% of all vaginal deliveries. The authors of one study point out that “most of the traditional risk factors for shoulder dystocia have no predictive value, shoulder dystocia itself is an unpredictable event, and infants at risk for permanent injury are virtually impossible to predict.”²³ This may make delivery in the home a high-risk endeavor because of the inability to mobilize an obstetric team to assist with shoulder dystocia maneuvers or perform a Zavanelli delivery.

Another variable overlooked in most studies is the speed of transfer and the outcomes of pregnancies in which the women intended to deliver at home but ended up requiring urgent transfer. One study that did examine this scenario found that “women who had booked for a home birth, but later needed to transfer their care for a hospital birth, appeared to have the highest risk of intrapartum-related perinatal mortality.”²⁴

There is also some controversy regarding the delivery of women who are pregnant with twins, who have a fetus in breech presentation, or who have a history of cesarean delivery. One study examined outcomes for intended home delivery of 57 women who had a prior abdominal delivery.²⁵ Fifty of these women delivered vaginally in the home, and seven (12.3%) delivered in the hospital. One hospital transfer was urgent for fetal distress. One baby was stillborn, delivered at home.

A 10-year prospective study of vaginal birth after cesarean (VBAC) in birth centers found that more than 50% of uterine ruptures and 57% of perinatal deaths involved the 10% of women who had more than one prior cesarean delivery or who had reached a gestational age of more than 42 weeks.²⁶

Skill of the caregiver is important

The training and qualifications of the obstetric care provider are incredibly important. One study evaluated 4,361 home births attended by “apprentice-trained midwives from 1970 to 1985 and 4,107 home births attended by family physicians from 1969 to 1981.”²⁷ The perinatal mortality rate for the midwife-attended births was 14 for every 1,000 births, in contrast to the rate of 5 for every 1,000 physician-attended births.

Three types of midwife are credentialed in this country:

• certified nurse-midwife (CNM)
• certified midwife (CM)
• certified professional midwife (CPM).

The first two categories are certified by the American Midwifery Certification Board (AMCB). CNMs and CMs undergo rigorous training and examination, and this designation will require a graduate degree within the next few years. The CPM category, however, requires much less rigorous training. Its midwives are certified by the North American Registry of Midwives. The clinical requirements for certification as a CPM include:

• attending a “minimum of 20 births”
• managing at least 20 additional births, at least half of them in the home or another out-of-hospital setting
• performing a small number of prenatal, newborn, and postpartum exams.²⁸

A high school diploma is not required.

I suspect that concerns about this lax certification process contributed to ACOG’s decision to issue a statement from its executive board in 2006: “While ACOG supports women having a choice in determining their providers of care, ACOG does not support the provision of care by lay midwives or other midwives who are not certified by the American College of Nurse-Midwives (ACNM) or AMCB.”²⁹

A number of midwifery advocates have made a legislative push to expand licensure for CPMs in this country, and the debate continues on a state-by-state basis.³⁰

Economics and other variables affect delivery decision

Some advocates of home birth note that the “average uncomplicated vaginal birth costs 68% less in a home than in a hospital.”³¹ Others try to organize support for women who want to give birth at home, such as the Home Birth Hotline, a voluntary, UK-based organization.³²

Some articles suggest that patient satisfaction is of significant importance in the decision about where to deliver. One noted that women who delivered where they had planned had higher overall satisfaction when that place was in the home (P<.01).³³

A randomized, controlled trial (n=3,510) simulated home delivery in a hospital, with “home delivery” patients having midwifery care in a room “similar to one in one’s own home” and the others having “consultant-led care” in rooms in the delivery suite that contained equipment to resuscitate both mother and baby, as well as monitors and other technology.³⁴ This study found no significant differences in measured outcomes, but “generally higher levels of satisfaction” among the women who had simulated home delivery.

A study from “remote and rural Scotland” found that most women “expressed a preference to give birth in hospital and have consultant-led care because they felt safer.”³⁵

Does the rhetoric surrounding home birth “empower” women?

Another frequently overlooked issue is the passionate rhetoric used to describe home birth—and the effect of that passion on women whose birth plan doesn’t play out as expected. Words such as “choice” and “empowerment” are often used. Regrettably, there is considerable mistrust of the medical system.

One woman describes how her planned home delivery, “influenced by the feminist literature,” went awry.³⁸ After a long labor, she wrote, she “just wanted the baby out, safe and healthy. It no longer mattered how it happened….I couldn’t get rid of the underlying feeling that I had ‘failed’ in some way….”³⁸

Because of her strong desire for home delivery, this woman was deeply affected when the delivery became difficult: “I did not have the authority to proclaim whether or not various medical interventions were necessary, or whether my case actually did constitute a medical emergency….Faced with these ‘options’—safe birth or potential death—how could I be said to be making a ‘choice’?…The obstetrician has more power than the woman because s/he has more knowledge.”³⁸

Despite having come to this realization, and delivering a healthy baby, she still experienced “a sense of disappointment and anger” and “traumatic flashbacks.”

I worry that patients may become so caught up in the rhetoric of their own power and choice that, when uncontrollable events occur, the happiness of a healthy delivery is overshadowed by deep disappointment.

Heated debate isn’t helpful

An unfortunate rift seems to have developed between some members of the midwifery community and some physicians. ACOG and the ACNM have a longstanding policy that: “In those circumstances in which obstetrician/gynecologists and certified nurse-midwives/certified midwives collaborate in the care of women, the quality of those practices is enhanced by a working relationship characterized by mutual respect and trust.”³⁹

Whether individual physicians agree with the practice of planned home birth or not, the health and welfare of the patient must be paramount. The American Public Health Association and the ACNM support home birth.^40,41

When obstetric emergencies do arise in the home setting, necessitating emergent transfer, it is critical that the transfer be managed in a way that ensures the best outcome.

One disturbing article describes both “disarticulations” that occur “when there is no correspondence of information or action between the midwife and the hospital staff” and “fractured articulations” that arise from “partial and incomplete correspondence.”⁴² A number of midwives were interviewed who no longer feel comfortable bringing patients to certain hospitals because of the negative response they received from health-care providers, sometimes to the detriment of the patient.

Can we improve the situation?

First, we need to choose our words carefully when we counsel women about labor and delivery, in recognition of the buzzwords used by advocates of home birth (“empowerment,” “choice”) and the sense of failure and distress some women feel when they eventually require heightened medical intervention.

Perhaps we should dispense with the term “failure,” as in failure to progress, failure to dilate, and so on, to avoid implying that this “failure” is the woman’s fault. And instead of saying that a patient’s pelvis is “adequate,” implying that another woman’s pelvis isn’t, we could use a term that sounds less judgmental.

We can also make the hospital environment more nurturing and supportive of women’s choices for labor, as long as safety isn’t compromised. And when we receive a transfer of a patient whose home delivery has gone awry, we should openly, efficiently, and professionally communicate with the home-delivery provider to best benefit the patient, regardless of our feelings on the subject.

Home birth isn’t going away

That’s my take on the literature. There are certainly data supporting the safety of home birth for the vast majority of women who choose it, but there is also a significant number of women who will experience unpredictable events that could be fatal if blood products or surgery isn’t rapidly available. For that reason, and in light of the very high stakes involved, I wonder: Why take that chance?