In 1753, Lind² described how he gave fresh fruit vs cider, vinegar, sulfuric acid, seawater, or barley water to 12 sailors aboard the HMS Salisbury in an effort to find a cure for scurvy. This landmark clinical trial has been hailed as an example of how clinical research can dramatically alter clinical practice. Yet practice did not change aboard British naval ships until almost 50 years after Lind’s treatise was published.³

For many reasons, randomized clinical trials may not immediately affect what physicians do. Sometimes, physicians believe that the trials were not well designed or well conducted, or that the results do not apply to their patients. I briefly discussed some limitations of evidence-based clinical decision-making in our September 2009 issue.⁴

Another reason is that the conclusions from some trials do not jibe with the experience of seasoned clinicians. That is why, this month, I have asked two physicians, a rheumatologist⁵ and a spine surgeon,⁶ to comment on how two studies^7,8 have influenced their clinical practice. Both studies concluded that vertebroplasty (injecting cement to shore up osteoporotic vertebrae) was no more beneficial than a sham procedure in patients with vertebral compression fractures. Neither physician is ready to completely abandon vertebroplasty on the basis of these two studies. Thus, it seems that published evidence may provide us guidance and fruit for discussion, but does not give us certainty.

In 1753, Lind² described how he gave fresh fruit vs cider, vinegar, sulfuric acid, seawater, or barley water to 12 sailors aboard the HMS Salisbury in an effort to find a cure for scurvy. This landmark clinical trial has been hailed as an example of how clinical research can dramatically alter clinical practice. Yet practice did not change aboard British naval ships until almost 50 years after Lind’s treatise was published.³

For many reasons, randomized clinical trials may not immediately affect what physicians do. Sometimes, physicians believe that the trials were not well designed or well conducted, or that the results do not apply to their patients. I briefly discussed some limitations of evidence-based clinical decision-making in our September 2009 issue.⁴

Another reason is that the conclusions from some trials do not jibe with the experience of seasoned clinicians. That is why, this month, I have asked two physicians, a rheumatologist⁵ and a spine surgeon,⁶ to comment on how two studies^7,8 have influenced their clinical practice. Both studies concluded that vertebroplasty (injecting cement to shore up osteoporotic vertebrae) was no more beneficial than a sham procedure in patients with vertebral compression fractures. Neither physician is ready to completely abandon vertebroplasty on the basis of these two studies. Thus, it seems that published evidence may provide us guidance and fruit for discussion, but does not give us certainty.

In 1753, Lind² described how he gave fresh fruit vs cider, vinegar, sulfuric acid, seawater, or barley water to 12 sailors aboard the HMS Salisbury in an effort to find a cure for scurvy. This landmark clinical trial has been hailed as an example of how clinical research can dramatically alter clinical practice. Yet practice did not change aboard British naval ships until almost 50 years after Lind’s treatise was published.³

For many reasons, randomized clinical trials may not immediately affect what physicians do. Sometimes, physicians believe that the trials were not well designed or well conducted, or that the results do not apply to their patients. I briefly discussed some limitations of evidence-based clinical decision-making in our September 2009 issue.⁴

Another reason is that the conclusions from some trials do not jibe with the experience of seasoned clinicians. That is why, this month, I have asked two physicians, a rheumatologist⁵ and a spine surgeon,⁶ to comment on how two studies^7,8 have influenced their clinical practice. Both studies concluded that vertebroplasty (injecting cement to shore up osteoporotic vertebrae) was no more beneficial than a sham procedure in patients with vertebral compression fractures. Neither physician is ready to completely abandon vertebroplasty on the basis of these two studies. Thus, it seems that published evidence may provide us guidance and fruit for discussion, but does not give us certainty.

Confronted with the unexpected results of two trials of vertebroplasty,^1,2 physicians are feeling some consternation, We had thought that percutaneous vertebroplasty helps patients with osteoporosis who sustain a painful vertebral insufficiency fracture. However, the trials found it to be no better than a sham procedure in terms of relieving pain.

See related commentary

How will these findings affect our practice? Should we abandon this popular procedure? Or are there other considerations that may mitigate these negative findings? And what should we tell our patients?

700,000 FRACTURES PER YEAR

Vertebral insufficiency fractures are the most common type of fracture in patients with osteoporosis. Every year in the United States, about 700,000 of them occur.

Nearly two-thirds are asymptomatic. The other one-third typically present with the acute onset of localized pain.

Vertebral insufficiency fractures often lead to chronic pain, impair the ability to walk and to perform daily activities, and accentuate thoracic kyphosis, which in turn can lead to pulmonary restrictive disease, and they raise the risk of death. Also, a patient who has a vertebral insufficiency fracture has a 20% risk of sustaining a new one within 1 year.³

Whether symptomatic or asymptomatic, finding a vertebral insufficiency fracture should prompt one to consider drug therapy for osteoporosis. In addition, until now, a patient who presented with the acute onset of back pain and whose evaluation revealed a vertebral insufficiency fracture would also be considered for a vertebral augmentation procedure, either vertebroplasty or kyphoplasty, to relieve the pain.

Vertebroplasty involves injecting polymethylmethacrylate cement percutaneously into the affected vertebral body. Kyphoplasty, a similar procedure, uses a balloon to create a cavity in the fractured vertebral body. After the balloon is withdrawn, the cavity is filled with cement.

TWO RANDOMIZED TRIALS OF SHAM VS REAL VERTEBROPLASTY

Two teams, Kallmes et al² and Buchbinder et al,¹ independently performed randomized controlled trials to see if vertebroplasty really relieves pain as well as has been reported in open studies, case series, and nonrandomized trials.^4–7

In both trials, patients were randomized to undergo either sham vertebroplasty or real vertebroplasty. The sham procedure closely approximated the real procedure, including inserting a needle, infiltrating a local anesthetic, bupivacaine (Marcaine), into the periosteum of the posterior lamina¹ or the pedicle of the target vertebrae,² and opening a vial of polymethylmethacrylate so that the patient would smell the product.

Inclusion criteria

Patients in both trials had to have evidence of a recent (acute) or nonhealed vertebral insufficiency fracture.

Pain was the primary outcome measured

In both trials, the investigators assessed the patients’ pain at baseline and again at several specified intervals, using validated tools.

Kallmes et al assessed pain intensity and functional measures at 1 month (the primary outcome measured), and also at 3, 14, and 90 days and at 1 year.

Buchbinder et al assessed pain at 1 week and at 1, 3, and 6 months. The primary outcome measured was pain at 3 months. Secondary outcomes included quality-of-life measures, pain at rest, and pain at night.

Surprising results

In both trials, the mean pain scores were better than at baseline at all time points after the procedure in both the real-procedure and the sham-procedure groups. Moreover, the effect did not differ between the two treatment groups in either study.

QUESTIONS COMPLICATE THE ISSUE

These two trials should make us consider whether this intervention is warranted. We should, however, also consider some limitations of these studies that raise questions about how the conclusions should or should not alter practice.

Does local anesthetic continue to relieve pain?

In both the sham and the real procedure, the bupivacaine injection may have helped relieve pain to some extent afterward, as its anesthetic effect may last longer than we would expect from its 3-hour half-life. The effect could certainly have contributed to improvements in pain levels at the earlier time points after the procedure.

Was there selection bias?

Both studies were highly rigorous and were done at hospitals that had extensive experience with vertebroplasty. However, they may have harbored selection bias, as many more patients were screened than were randomized.

Buchbinder et al¹ screened 468 patients. Of these, 30% declined to participate, and another 53% did not meet the eligibility criteria. In the end, only 78 patients were randomized.

Kallmes et al² screened 1,813 patients, 300 of whom declined and 1,382 of whom were excluded, leaving 131 patients to be randomized. The reasons for exclusion were not specifically reported in many cases.

In both studies, it would be interesting to know how many of those who declined proceeded to undergo a vertebral augmentation procedure.

In the study by Kallmes et al,² recruitment got off to a slow start. Thus, after three patients were recruited, the inclusion requirements were liberalized. The study was originally designed to include 250 patients, which would have given it a power of greater than 80% to detect differences in primary and secondary outcomes. The design was revised to include 130 patients. The statistical power was still 80%, but this was to detect a greater difference in the outcomes than originally projected.

Had the window of opportunity already closed?

Vertebroplasty may have a window of opportunity within which it is most effective. Sooner is probably better than later, but it would be good to identify this time frame.

Kaufmann et al⁹ reported that patients with older fractures needed slightly more analgesic drugs after the procedure. It has been shown previously that patients who are the most likely to respond to a vertebral augmentation procedure are those with fractures that occurred between 1 and 12 months prior to the procedure and who have evidence that the fracture was recent, ie, edema on magnetic resonance imaging (MRI) or increased uptake on a bone scan.¹⁰

Other studies suggested that intervention works best in patients who have had uncontrolled pain lasting less than 6 weeks.^8,11 (In the study by Buchbinder et al,¹ only 32% of the patients in either group reported pain lasting less than 6 weeks.)

The study by Kallmes et al included patients whose pain had begun within 1 year previously. However, if the duration of pain (ie, the age of the fracture) was uncertain, MRI was done to look for edema, which would indicate the fracture was fresh. It is thus unclear whether all patients in this study truly had an acute or subacute fracture, since all did not undergo confirmatory MRI.

Why did so many patients cross over from sham to real treatment?

Patients in the Kallmes trial² could cross over from one treatment group to the other as early as 1 month after the procedure. And, in fact, 43% of patients in the sham-treatment group did choose to cross over by 3 months. In contrast, after real vertebroplasty, significantly fewer—only 12% (P < .001)—crossed over to receive the sham procedure. The patients who crossed over from the sham-procedure group to receive vertebroplasty experienced an early improvement in pain, but this was not sustained at 1 or 3 months of follow-up.

The higher crossover rate in the shamprocedure group suggests they were dissatisfied with this intervention, although their outcomes were not significantly better after they got the real procedure. The patients who first received the sham treatment and elected to cross over to vertebroplasty had higher pain and disability scores at baseline. Thus, they may have had other, more chronic causes of pain or other factors affecting the likelihood of a response, particularly of a durable or sustained response.

How do the interventions compare with medical therapy?

Earlier studies showed that vertebroplasty relieves pain almost immediately.^4–6 But the benefit does not last: at 6 weeks and up to 12 months later there is no difference in either pain or functional capacity reported in patients receiving vertebroplasty vs conservative treatment.^4,6,7 It would thus appear that pain gradually diminishes over time after a vertebral insufficiency fracture, as the fracture heals.

The recent studies^1,2 raise the possibility that the pain relief is due to the local anesthetic, not the vertebroplasty itself. We do not know, however, if either vertebroplasty or the sham procedure is superior to conservative medical management. Prospective multicenter trials are under way to address this question.¹¹

Further complicating the issue, the two trials did not keep track of medical treatments patients were receiving concomitantly during the trial period. It is thus more difficult to compare the pain assessment outcomes following invasive procedures—real or sham.

Would kyphoplasty be better?

These studies addressed one procedure, vertebroplasty, and the results and conclusions should not be generalized to kyphoplasty. A prospective randomized trial of kyphoplasty is clearly warranted.

If kyphoplasty is found to be better than a sham procedure, then vertebroplasty should be re-examined in comparison with kyphoplasty. In any future studies, it will be important to select patients rigorously (eg, to include only patients with recent fractures), to match patients according to concomitant therapies, and to consider other potential superimposed causes of back pain in this elderly population, which has a high prevalence of back pain.

HOW SHOULD MY PRACTICE CHANGE? WHAT SHOULD I TELL PATIENTS?

Having considered the results, conclusions, and limitations of these two randomized trials, particularly in terms of recruitment, I cannot say that my practice has changed in terms of referring patients who have a vertebral compression fracture to an interventionalist. However, the education that I provide to patients has changed.

In my mind, the highest priority for a patient with a vertebral insufficiency fracture is to treat (or to reassess the current treatment of) the underlying systemic disease, ie, osteoporosis. This is especially true since most vertebral insufficiency fractures are asymptomatic.

On the other hand, a patient with a painful vertebral compression fracture needs prompt attention and consideration for interventional pain relief. Rapid pain relief is desirable. And in uncontrolled trials,^4–7 vertebroplasty and kyphoplasty rapidly relieved vertebral pain. However, it may be that an anesthetic injection is equivalent to vertebroplasty and could accomplish the goal of immediate pain relief just as well.

The pain relief from sham or real vertebroplasty may not be durable, and 3 to 12 months later the pain benefit may be no greater than if more conservative therapy had been pursued.

It is essential to determine the most appropriate window for treatment as well as the most appropriate candidates on whom to perform a procedure. The recently published studies^1,2 may have had significant patient selection bias and may not have optimized the window of opportunity for vertebral augmentation performance. There were many patients who declined the study, and some were excluded because of acute pain requiring hospitalization.

As a rheumatologist treating patients with osteoporosis, it is my responsibility to discuss with the patient and family the potential treatments available, to discuss the associated possible risks and benefits, to report on available evidence, and to refer patients to an appropriate interventional specialist if they desire. In light of the lack of superior pain reduction with vertebroplasty than with a sham procedure, many patients may opt for conservative therapy.

It is thus appropriate to determine the acuity of the fracture and to have a frank discussion with the patient about the options for pain management. Opiate drugs pose risks in elderly patients, particularly altered mentation, somnolence, interference with balance, and risk of falls. Vertebroplasty or anesthetic injection may rapidly relieve the pain and reduce the need for opiate therapy. Not yet subjected to the rigors of a randomized placebocontrolled trial, kyphoplasty may yet prove to be better than a sham intervention.

It is essential to determine if there is a role for vertebral augmentation in a select patient population—perhaps selected on the basis of the time that has elapsed since the fracture occurred (determined objectively), the severity of the fracture, and other factors. Perhaps a subset of patients would gain greater benefit from the procedure, whether it amounts solely to acute pain reduction or perhaps to a more durable response.

The recent studies by Kallmes et al² and Buchbinder et al¹ found vertebroplasty and sham vertebroplasty to be equally effective in reducing pain and improving function. However, given the limitations of each of these studies, particularly the low numbers of patients, it is difficult to establish that vertebral augmentation procedures should no longer be done. And vertebroplasty may still benefit correctly selected patients.

Confronted with the unexpected results of two trials of vertebroplasty,^1,2 physicians are feeling some consternation, We had thought that percutaneous vertebroplasty helps patients with osteoporosis who sustain a painful vertebral insufficiency fracture. However, the trials found it to be no better than a sham procedure in terms of relieving pain.

See related commentary

How will these findings affect our practice? Should we abandon this popular procedure? Or are there other considerations that may mitigate these negative findings? And what should we tell our patients?

700,000 FRACTURES PER YEAR

Vertebral insufficiency fractures are the most common type of fracture in patients with osteoporosis. Every year in the United States, about 700,000 of them occur.

Nearly two-thirds are asymptomatic. The other one-third typically present with the acute onset of localized pain.

Vertebral insufficiency fractures often lead to chronic pain, impair the ability to walk and to perform daily activities, and accentuate thoracic kyphosis, which in turn can lead to pulmonary restrictive disease, and they raise the risk of death. Also, a patient who has a vertebral insufficiency fracture has a 20% risk of sustaining a new one within 1 year.³

Whether symptomatic or asymptomatic, finding a vertebral insufficiency fracture should prompt one to consider drug therapy for osteoporosis. In addition, until now, a patient who presented with the acute onset of back pain and whose evaluation revealed a vertebral insufficiency fracture would also be considered for a vertebral augmentation procedure, either vertebroplasty or kyphoplasty, to relieve the pain.

Vertebroplasty involves injecting polymethylmethacrylate cement percutaneously into the affected vertebral body. Kyphoplasty, a similar procedure, uses a balloon to create a cavity in the fractured vertebral body. After the balloon is withdrawn, the cavity is filled with cement.

TWO RANDOMIZED TRIALS OF SHAM VS REAL VERTEBROPLASTY

Two teams, Kallmes et al² and Buchbinder et al,¹ independently performed randomized controlled trials to see if vertebroplasty really relieves pain as well as has been reported in open studies, case series, and nonrandomized trials.^4–7

In both trials, patients were randomized to undergo either sham vertebroplasty or real vertebroplasty. The sham procedure closely approximated the real procedure, including inserting a needle, infiltrating a local anesthetic, bupivacaine (Marcaine), into the periosteum of the posterior lamina¹ or the pedicle of the target vertebrae,² and opening a vial of polymethylmethacrylate so that the patient would smell the product.

Inclusion criteria

Patients in both trials had to have evidence of a recent (acute) or nonhealed vertebral insufficiency fracture.

Pain was the primary outcome measured

In both trials, the investigators assessed the patients’ pain at baseline and again at several specified intervals, using validated tools.

Kallmes et al assessed pain intensity and functional measures at 1 month (the primary outcome measured), and also at 3, 14, and 90 days and at 1 year.

Buchbinder et al assessed pain at 1 week and at 1, 3, and 6 months. The primary outcome measured was pain at 3 months. Secondary outcomes included quality-of-life measures, pain at rest, and pain at night.

Surprising results

In both trials, the mean pain scores were better than at baseline at all time points after the procedure in both the real-procedure and the sham-procedure groups. Moreover, the effect did not differ between the two treatment groups in either study.

QUESTIONS COMPLICATE THE ISSUE

These two trials should make us consider whether this intervention is warranted. We should, however, also consider some limitations of these studies that raise questions about how the conclusions should or should not alter practice.

Does local anesthetic continue to relieve pain?

In both the sham and the real procedure, the bupivacaine injection may have helped relieve pain to some extent afterward, as its anesthetic effect may last longer than we would expect from its 3-hour half-life. The effect could certainly have contributed to improvements in pain levels at the earlier time points after the procedure.

Was there selection bias?

Both studies were highly rigorous and were done at hospitals that had extensive experience with vertebroplasty. However, they may have harbored selection bias, as many more patients were screened than were randomized.

Buchbinder et al¹ screened 468 patients. Of these, 30% declined to participate, and another 53% did not meet the eligibility criteria. In the end, only 78 patients were randomized.

Kallmes et al² screened 1,813 patients, 300 of whom declined and 1,382 of whom were excluded, leaving 131 patients to be randomized. The reasons for exclusion were not specifically reported in many cases.

In both studies, it would be interesting to know how many of those who declined proceeded to undergo a vertebral augmentation procedure.

In the study by Kallmes et al,² recruitment got off to a slow start. Thus, after three patients were recruited, the inclusion requirements were liberalized. The study was originally designed to include 250 patients, which would have given it a power of greater than 80% to detect differences in primary and secondary outcomes. The design was revised to include 130 patients. The statistical power was still 80%, but this was to detect a greater difference in the outcomes than originally projected.

Had the window of opportunity already closed?

Vertebroplasty may have a window of opportunity within which it is most effective. Sooner is probably better than later, but it would be good to identify this time frame.

Kaufmann et al⁹ reported that patients with older fractures needed slightly more analgesic drugs after the procedure. It has been shown previously that patients who are the most likely to respond to a vertebral augmentation procedure are those with fractures that occurred between 1 and 12 months prior to the procedure and who have evidence that the fracture was recent, ie, edema on magnetic resonance imaging (MRI) or increased uptake on a bone scan.¹⁰

Other studies suggested that intervention works best in patients who have had uncontrolled pain lasting less than 6 weeks.^8,11 (In the study by Buchbinder et al,¹ only 32% of the patients in either group reported pain lasting less than 6 weeks.)

The study by Kallmes et al included patients whose pain had begun within 1 year previously. However, if the duration of pain (ie, the age of the fracture) was uncertain, MRI was done to look for edema, which would indicate the fracture was fresh. It is thus unclear whether all patients in this study truly had an acute or subacute fracture, since all did not undergo confirmatory MRI.

Why did so many patients cross over from sham to real treatment?

Patients in the Kallmes trial² could cross over from one treatment group to the other as early as 1 month after the procedure. And, in fact, 43% of patients in the sham-treatment group did choose to cross over by 3 months. In contrast, after real vertebroplasty, significantly fewer—only 12% (P < .001)—crossed over to receive the sham procedure. The patients who crossed over from the sham-procedure group to receive vertebroplasty experienced an early improvement in pain, but this was not sustained at 1 or 3 months of follow-up.

The higher crossover rate in the shamprocedure group suggests they were dissatisfied with this intervention, although their outcomes were not significantly better after they got the real procedure. The patients who first received the sham treatment and elected to cross over to vertebroplasty had higher pain and disability scores at baseline. Thus, they may have had other, more chronic causes of pain or other factors affecting the likelihood of a response, particularly of a durable or sustained response.

How do the interventions compare with medical therapy?

Earlier studies showed that vertebroplasty relieves pain almost immediately.^4–6 But the benefit does not last: at 6 weeks and up to 12 months later there is no difference in either pain or functional capacity reported in patients receiving vertebroplasty vs conservative treatment.^4,6,7 It would thus appear that pain gradually diminishes over time after a vertebral insufficiency fracture, as the fracture heals.

The recent studies^1,2 raise the possibility that the pain relief is due to the local anesthetic, not the vertebroplasty itself. We do not know, however, if either vertebroplasty or the sham procedure is superior to conservative medical management. Prospective multicenter trials are under way to address this question.¹¹

Further complicating the issue, the two trials did not keep track of medical treatments patients were receiving concomitantly during the trial period. It is thus more difficult to compare the pain assessment outcomes following invasive procedures—real or sham.

Would kyphoplasty be better?

These studies addressed one procedure, vertebroplasty, and the results and conclusions should not be generalized to kyphoplasty. A prospective randomized trial of kyphoplasty is clearly warranted.

If kyphoplasty is found to be better than a sham procedure, then vertebroplasty should be re-examined in comparison with kyphoplasty. In any future studies, it will be important to select patients rigorously (eg, to include only patients with recent fractures), to match patients according to concomitant therapies, and to consider other potential superimposed causes of back pain in this elderly population, which has a high prevalence of back pain.

HOW SHOULD MY PRACTICE CHANGE? WHAT SHOULD I TELL PATIENTS?

Having considered the results, conclusions, and limitations of these two randomized trials, particularly in terms of recruitment, I cannot say that my practice has changed in terms of referring patients who have a vertebral compression fracture to an interventionalist. However, the education that I provide to patients has changed.

In my mind, the highest priority for a patient with a vertebral insufficiency fracture is to treat (or to reassess the current treatment of) the underlying systemic disease, ie, osteoporosis. This is especially true since most vertebral insufficiency fractures are asymptomatic.

On the other hand, a patient with a painful vertebral compression fracture needs prompt attention and consideration for interventional pain relief. Rapid pain relief is desirable. And in uncontrolled trials,^4–7 vertebroplasty and kyphoplasty rapidly relieved vertebral pain. However, it may be that an anesthetic injection is equivalent to vertebroplasty and could accomplish the goal of immediate pain relief just as well.

The pain relief from sham or real vertebroplasty may not be durable, and 3 to 12 months later the pain benefit may be no greater than if more conservative therapy had been pursued.

It is essential to determine the most appropriate window for treatment as well as the most appropriate candidates on whom to perform a procedure. The recently published studies^1,2 may have had significant patient selection bias and may not have optimized the window of opportunity for vertebral augmentation performance. There were many patients who declined the study, and some were excluded because of acute pain requiring hospitalization.

As a rheumatologist treating patients with osteoporosis, it is my responsibility to discuss with the patient and family the potential treatments available, to discuss the associated possible risks and benefits, to report on available evidence, and to refer patients to an appropriate interventional specialist if they desire. In light of the lack of superior pain reduction with vertebroplasty than with a sham procedure, many patients may opt for conservative therapy.

It is thus appropriate to determine the acuity of the fracture and to have a frank discussion with the patient about the options for pain management. Opiate drugs pose risks in elderly patients, particularly altered mentation, somnolence, interference with balance, and risk of falls. Vertebroplasty or anesthetic injection may rapidly relieve the pain and reduce the need for opiate therapy. Not yet subjected to the rigors of a randomized placebocontrolled trial, kyphoplasty may yet prove to be better than a sham intervention.

It is essential to determine if there is a role for vertebral augmentation in a select patient population—perhaps selected on the basis of the time that has elapsed since the fracture occurred (determined objectively), the severity of the fracture, and other factors. Perhaps a subset of patients would gain greater benefit from the procedure, whether it amounts solely to acute pain reduction or perhaps to a more durable response.

The recent studies by Kallmes et al² and Buchbinder et al¹ found vertebroplasty and sham vertebroplasty to be equally effective in reducing pain and improving function. However, given the limitations of each of these studies, particularly the low numbers of patients, it is difficult to establish that vertebral augmentation procedures should no longer be done. And vertebroplasty may still benefit correctly selected patients.

Confronted with the unexpected results of two trials of vertebroplasty,^1,2 physicians are feeling some consternation, We had thought that percutaneous vertebroplasty helps patients with osteoporosis who sustain a painful vertebral insufficiency fracture. However, the trials found it to be no better than a sham procedure in terms of relieving pain.

See related commentary

How will these findings affect our practice? Should we abandon this popular procedure? Or are there other considerations that may mitigate these negative findings? And what should we tell our patients?

700,000 FRACTURES PER YEAR

Vertebral insufficiency fractures are the most common type of fracture in patients with osteoporosis. Every year in the United States, about 700,000 of them occur.

Nearly two-thirds are asymptomatic. The other one-third typically present with the acute onset of localized pain.

Vertebral insufficiency fractures often lead to chronic pain, impair the ability to walk and to perform daily activities, and accentuate thoracic kyphosis, which in turn can lead to pulmonary restrictive disease, and they raise the risk of death. Also, a patient who has a vertebral insufficiency fracture has a 20% risk of sustaining a new one within 1 year.³

Whether symptomatic or asymptomatic, finding a vertebral insufficiency fracture should prompt one to consider drug therapy for osteoporosis. In addition, until now, a patient who presented with the acute onset of back pain and whose evaluation revealed a vertebral insufficiency fracture would also be considered for a vertebral augmentation procedure, either vertebroplasty or kyphoplasty, to relieve the pain.

Vertebroplasty involves injecting polymethylmethacrylate cement percutaneously into the affected vertebral body. Kyphoplasty, a similar procedure, uses a balloon to create a cavity in the fractured vertebral body. After the balloon is withdrawn, the cavity is filled with cement.

TWO RANDOMIZED TRIALS OF SHAM VS REAL VERTEBROPLASTY

Two teams, Kallmes et al² and Buchbinder et al,¹ independently performed randomized controlled trials to see if vertebroplasty really relieves pain as well as has been reported in open studies, case series, and nonrandomized trials.^4–7

In both trials, patients were randomized to undergo either sham vertebroplasty or real vertebroplasty. The sham procedure closely approximated the real procedure, including inserting a needle, infiltrating a local anesthetic, bupivacaine (Marcaine), into the periosteum of the posterior lamina¹ or the pedicle of the target vertebrae,² and opening a vial of polymethylmethacrylate so that the patient would smell the product.

Inclusion criteria

Patients in both trials had to have evidence of a recent (acute) or nonhealed vertebral insufficiency fracture.

Pain was the primary outcome measured

In both trials, the investigators assessed the patients’ pain at baseline and again at several specified intervals, using validated tools.

Kallmes et al assessed pain intensity and functional measures at 1 month (the primary outcome measured), and also at 3, 14, and 90 days and at 1 year.

Buchbinder et al assessed pain at 1 week and at 1, 3, and 6 months. The primary outcome measured was pain at 3 months. Secondary outcomes included quality-of-life measures, pain at rest, and pain at night.

Surprising results

In both trials, the mean pain scores were better than at baseline at all time points after the procedure in both the real-procedure and the sham-procedure groups. Moreover, the effect did not differ between the two treatment groups in either study.

QUESTIONS COMPLICATE THE ISSUE

These two trials should make us consider whether this intervention is warranted. We should, however, also consider some limitations of these studies that raise questions about how the conclusions should or should not alter practice.

Does local anesthetic continue to relieve pain?

In both the sham and the real procedure, the bupivacaine injection may have helped relieve pain to some extent afterward, as its anesthetic effect may last longer than we would expect from its 3-hour half-life. The effect could certainly have contributed to improvements in pain levels at the earlier time points after the procedure.

Was there selection bias?

Both studies were highly rigorous and were done at hospitals that had extensive experience with vertebroplasty. However, they may have harbored selection bias, as many more patients were screened than were randomized.

Buchbinder et al¹ screened 468 patients. Of these, 30% declined to participate, and another 53% did not meet the eligibility criteria. In the end, only 78 patients were randomized.

Kallmes et al² screened 1,813 patients, 300 of whom declined and 1,382 of whom were excluded, leaving 131 patients to be randomized. The reasons for exclusion were not specifically reported in many cases.

In both studies, it would be interesting to know how many of those who declined proceeded to undergo a vertebral augmentation procedure.

In the study by Kallmes et al,² recruitment got off to a slow start. Thus, after three patients were recruited, the inclusion requirements were liberalized. The study was originally designed to include 250 patients, which would have given it a power of greater than 80% to detect differences in primary and secondary outcomes. The design was revised to include 130 patients. The statistical power was still 80%, but this was to detect a greater difference in the outcomes than originally projected.

Had the window of opportunity already closed?

Vertebroplasty may have a window of opportunity within which it is most effective. Sooner is probably better than later, but it would be good to identify this time frame.

Kaufmann et al⁹ reported that patients with older fractures needed slightly more analgesic drugs after the procedure. It has been shown previously that patients who are the most likely to respond to a vertebral augmentation procedure are those with fractures that occurred between 1 and 12 months prior to the procedure and who have evidence that the fracture was recent, ie, edema on magnetic resonance imaging (MRI) or increased uptake on a bone scan.¹⁰

Other studies suggested that intervention works best in patients who have had uncontrolled pain lasting less than 6 weeks.^8,11 (In the study by Buchbinder et al,¹ only 32% of the patients in either group reported pain lasting less than 6 weeks.)

The study by Kallmes et al included patients whose pain had begun within 1 year previously. However, if the duration of pain (ie, the age of the fracture) was uncertain, MRI was done to look for edema, which would indicate the fracture was fresh. It is thus unclear whether all patients in this study truly had an acute or subacute fracture, since all did not undergo confirmatory MRI.

Why did so many patients cross over from sham to real treatment?

Patients in the Kallmes trial² could cross over from one treatment group to the other as early as 1 month after the procedure. And, in fact, 43% of patients in the sham-treatment group did choose to cross over by 3 months. In contrast, after real vertebroplasty, significantly fewer—only 12% (P < .001)—crossed over to receive the sham procedure. The patients who crossed over from the sham-procedure group to receive vertebroplasty experienced an early improvement in pain, but this was not sustained at 1 or 3 months of follow-up.

The higher crossover rate in the shamprocedure group suggests they were dissatisfied with this intervention, although their outcomes were not significantly better after they got the real procedure. The patients who first received the sham treatment and elected to cross over to vertebroplasty had higher pain and disability scores at baseline. Thus, they may have had other, more chronic causes of pain or other factors affecting the likelihood of a response, particularly of a durable or sustained response.

How do the interventions compare with medical therapy?

Earlier studies showed that vertebroplasty relieves pain almost immediately.^4–6 But the benefit does not last: at 6 weeks and up to 12 months later there is no difference in either pain or functional capacity reported in patients receiving vertebroplasty vs conservative treatment.^4,6,7 It would thus appear that pain gradually diminishes over time after a vertebral insufficiency fracture, as the fracture heals.

The recent studies^1,2 raise the possibility that the pain relief is due to the local anesthetic, not the vertebroplasty itself. We do not know, however, if either vertebroplasty or the sham procedure is superior to conservative medical management. Prospective multicenter trials are under way to address this question.¹¹

Further complicating the issue, the two trials did not keep track of medical treatments patients were receiving concomitantly during the trial period. It is thus more difficult to compare the pain assessment outcomes following invasive procedures—real or sham.

Would kyphoplasty be better?

These studies addressed one procedure, vertebroplasty, and the results and conclusions should not be generalized to kyphoplasty. A prospective randomized trial of kyphoplasty is clearly warranted.

If kyphoplasty is found to be better than a sham procedure, then vertebroplasty should be re-examined in comparison with kyphoplasty. In any future studies, it will be important to select patients rigorously (eg, to include only patients with recent fractures), to match patients according to concomitant therapies, and to consider other potential superimposed causes of back pain in this elderly population, which has a high prevalence of back pain.

HOW SHOULD MY PRACTICE CHANGE? WHAT SHOULD I TELL PATIENTS?

Having considered the results, conclusions, and limitations of these two randomized trials, particularly in terms of recruitment, I cannot say that my practice has changed in terms of referring patients who have a vertebral compression fracture to an interventionalist. However, the education that I provide to patients has changed.

In my mind, the highest priority for a patient with a vertebral insufficiency fracture is to treat (or to reassess the current treatment of) the underlying systemic disease, ie, osteoporosis. This is especially true since most vertebral insufficiency fractures are asymptomatic.

On the other hand, a patient with a painful vertebral compression fracture needs prompt attention and consideration for interventional pain relief. Rapid pain relief is desirable. And in uncontrolled trials,^4–7 vertebroplasty and kyphoplasty rapidly relieved vertebral pain. However, it may be that an anesthetic injection is equivalent to vertebroplasty and could accomplish the goal of immediate pain relief just as well.

The pain relief from sham or real vertebroplasty may not be durable, and 3 to 12 months later the pain benefit may be no greater than if more conservative therapy had been pursued.

It is essential to determine the most appropriate window for treatment as well as the most appropriate candidates on whom to perform a procedure. The recently published studies^1,2 may have had significant patient selection bias and may not have optimized the window of opportunity for vertebral augmentation performance. There were many patients who declined the study, and some were excluded because of acute pain requiring hospitalization.

As a rheumatologist treating patients with osteoporosis, it is my responsibility to discuss with the patient and family the potential treatments available, to discuss the associated possible risks and benefits, to report on available evidence, and to refer patients to an appropriate interventional specialist if they desire. In light of the lack of superior pain reduction with vertebroplasty than with a sham procedure, many patients may opt for conservative therapy.

It is thus appropriate to determine the acuity of the fracture and to have a frank discussion with the patient about the options for pain management. Opiate drugs pose risks in elderly patients, particularly altered mentation, somnolence, interference with balance, and risk of falls. Vertebroplasty or anesthetic injection may rapidly relieve the pain and reduce the need for opiate therapy. Not yet subjected to the rigors of a randomized placebocontrolled trial, kyphoplasty may yet prove to be better than a sham intervention.

It is essential to determine if there is a role for vertebral augmentation in a select patient population—perhaps selected on the basis of the time that has elapsed since the fracture occurred (determined objectively), the severity of the fracture, and other factors. Perhaps a subset of patients would gain greater benefit from the procedure, whether it amounts solely to acute pain reduction or perhaps to a more durable response.

The recent studies by Kallmes et al² and Buchbinder et al¹ found vertebroplasty and sham vertebroplasty to be equally effective in reducing pain and improving function. However, given the limitations of each of these studies, particularly the low numbers of patients, it is difficult to establish that vertebral augmentation procedures should no longer be done. And vertebroplasty may still benefit correctly selected patients.

Cognitive dissonance describes how we respond to conflicting information that challenges our existing belief, the uncomfortable feeling we get when new evidence calls into question things that we “know” are true.

See related commentary

To the point: two recent clinical trials^1,2 have called into question the efficacy of vertebroplasty for treating osteoporotic vertebral compression fractures and have led many of us to question many of our assumptions, not only about vertebroplasty but also about evidencebased medicine.

Osteoporotic vertebral compression fractures are very common: more than 700,000 are estimated to occur in the United States annually. ³ They are costly and are associated with a risk of death.⁴ Fortunately, most heal without problems over 4 to 6 weeks with conventional treatment, ie, activity modification, analgesics, and bracing.

However, some patients do not seem to do so well and are debilitated by the pain of the fracture. Conventional fracture surgery carries very high risk and poor outcomes,⁵ and so has been reserved mostly for patients with neurologic deficits.

VERTEBROPLASTY GOES MAINSTREAM

Given these facts, investigators began looking for alternative treatments. One that rose to the fore was polymethylmethacrylate cement to stabilize the fracture. This technique, called vertebroplasty, involves injecting liquid cement through a needle into the vertebral body, where it hardens and is thought to restore stability.

Since the first description of vertebroplasty for treating symptomatic hemangiomas,⁶ many papers have been published about the procedure and about similar ones, now grouped under the general heading of vertebral augmentation. This includes kyphoplasty and other newer proprietary techniques. These procedures have been widely accepted, and their use is growing. They have shown good results in several prospective case series, and nonrandomized and randomized controlled studies have shown them to be more effective than conventional medical treatment.^7–25 For example, VERTOS, a small prospective randomized trial, showed that vertebroplasty was superior to conventional medical treatment.²⁵ When Wardlaw et al²⁴ showed that shortterm outcomes were better with kyphoplasty than with conventional medical therapy in a prospective randomized trial, many of us had moved past questioning whether vertebral augmentation is effective and were debating the relative merits of different methods and materials.

On a personal level, most of us became proponents of these procedures because we saw dramatic results—usually unequivocal. Most patients report significant improvement in pain immediately after the procedure, and many bedridden patients are able to leave the hospital within hours. In spine surgery, few procedures give such dramatic results with so few complications.

TWO NEW STUDIES UPSET ESTABLISHED BELIEF

This is why I am having such a hard time digesting the results of the trials by Buchbinder et al² and Kallmes et al,¹ published in the August 9, 2009, issue of the New England Journal of Medicine. Both were randomized controlled trials that used sham surgery rather than conventional medical treatment as the control. The sham procedure in each trial was the same as the intervention, with local anesthetic infiltration of the periosteum and mixing of the cement (so that the patients smelled its distinctive odor), but without placing the needle into the vertebra and injecting the cement.

In the study by Buchbinder et al,² the real treatment had no benefit in any primary or secondary end point. This study did not allow crossovers.

In the study by Kallmes et al,¹ more patients who received the real treatment reported clinically meaningful improvement in pain (a secondary end point), but the difference was not quite statistically significant (64% vs 48%, P = .06). In this trial, patients were allowed to cross over to the other study group after 1 month, and significantly more patients crossed over from the sham surgery group to the active treatment group than the other way around (43% vs 12%, P < .001).

My first instinct was to pick through the papers for flaws that would invalidate the results— and there were some problems. Both studies were initially planned to include more patients and therefore to have greater statistical power, but they were reassessed because of slow enrollment. In the study by Kallmes et al,¹ the difference in clinically meaningful improvement might have reached statistical significance if the trial had been larger. The study by Buchbinder et al² was a multicenter trial, but one center accounted for 53 (69%) of the 78 patients. Could this have biased the results?

The surgeon in me also seized for a while on the idea that since all of the interventions in both studies were done by interventional radiologists, the problem may have been in patient selection and that radiologists are not as astute as we are. However, even a surgeon’s ego cannot support this interpretation.

As I looked in more detail at the response I had written to these trials, I realized these criticisms were hardly fatal flaws, and the fact that two separate well-designed studies reached the same conclusion enhances their validity.

One concern that does bear some scrutiny is that the trials were too small to identify subgroups that may benefit from the procedure. In my experience, vertebral augmentation seems to have better results with certain types of fractures. Patients with a mobile pseudarthrotic cleft pattern of fracture seem to do much better than those with the more common nonmobile fracture.

THE POWERFUL PLACEBO EFFECT

Many commentaries on these two trials have discussed a famous study of a different procedure for a different condition. In this study, Moseley et al²⁶ evaluated the use of arthroscopy to treat osteoarthritis of the knee and found that sham arthroscopy was as effective as real arthroscopy and that both were better than conventional treatment.

I was not long out of my orthopedic residency when this trial was published and was very aware of the debate that preceded it, as I once had to prepare a talk about it for resident rounds. I remember that there was a lively debate in the orthopedic community over the efficacy of the procedure before the results of this trial were released.

In contrast, the vertebral augmentation controversy had become a debate about the relative efficacy and the economics of specific techniques, not about the effectiveness of the entire concept. The mainstream had accepted the validity of the procedure, which was not the case in the knee arthroscopy trial.

In both vertebroplasty studies, the activetreatment groups and the sham-treatment groups all showed significant and rapid improvement in pain and disability, and these results were maintained over the study period. Though most vertebral compression fractures do heal, the clinical improvement is usually gradual over a period of weeks. This raises the possibility that the sham treatment was actualy an active placebo.

There is some evidence to support this possibility. In a randomized trial of the efficacy of selective nerve root blocks for lumbar radiculopathy, Riew et al²⁷ showed that injection with a local anesthetic alone, although not as efficacious as a local anesthetic plus a corticosteroid at allowing patients to avoid surgery, showed an effect long after the expected duration of the anesthetic. The effect persisted even at 5 years of follow-up.²⁸

Is it possible that the local anesthetic in this trial and the vertebroplasty trials acted as some sort of “reset button” for pain sensation? This is an area that may bear further investigation.

WHERE DOES THIS LEAVE US?

So where does this leave us? On one hand, randomized controlled trials comparing vertebral augmentation with conventional medical therapy^24,25 showed augmentation to be beneficial. On the other hand, the studies by Kallmes et al¹ and Buchbinder et al² indicate vertebroplasty is no more effective than sham surgery.

It is very difficult for me to look at my own experience with vertebral augmentation and say that, on the basis of these trials, I am no longer going to offer it to my patients. I understand on an intellectual level that these trials call the efficacy of the procedure into question, but on a visceral level I cannot rationalize it. When faced with a patient who is barely ambulatory or in fact bed-bound due to pain, my experience tells me that vertebral augmentation has a very high chance of getting them ambulatory within hours. The trials of vertebroplasty would indicate this is a placebo effect or that local anesthetic alone is as effective, but I am not yet ready to make that leap.

Cognitive dissonance seems to rule.

Cognitive dissonance describes how we respond to conflicting information that challenges our existing belief, the uncomfortable feeling we get when new evidence calls into question things that we “know” are true.

See related commentary

To the point: two recent clinical trials^1,2 have called into question the efficacy of vertebroplasty for treating osteoporotic vertebral compression fractures and have led many of us to question many of our assumptions, not only about vertebroplasty but also about evidencebased medicine.

Osteoporotic vertebral compression fractures are very common: more than 700,000 are estimated to occur in the United States annually. ³ They are costly and are associated with a risk of death.⁴ Fortunately, most heal without problems over 4 to 6 weeks with conventional treatment, ie, activity modification, analgesics, and bracing.

However, some patients do not seem to do so well and are debilitated by the pain of the fracture. Conventional fracture surgery carries very high risk and poor outcomes,⁵ and so has been reserved mostly for patients with neurologic deficits.

VERTEBROPLASTY GOES MAINSTREAM

Given these facts, investigators began looking for alternative treatments. One that rose to the fore was polymethylmethacrylate cement to stabilize the fracture. This technique, called vertebroplasty, involves injecting liquid cement through a needle into the vertebral body, where it hardens and is thought to restore stability.

Since the first description of vertebroplasty for treating symptomatic hemangiomas,⁶ many papers have been published about the procedure and about similar ones, now grouped under the general heading of vertebral augmentation. This includes kyphoplasty and other newer proprietary techniques. These procedures have been widely accepted, and their use is growing. They have shown good results in several prospective case series, and nonrandomized and randomized controlled studies have shown them to be more effective than conventional medical treatment.^7–25 For example, VERTOS, a small prospective randomized trial, showed that vertebroplasty was superior to conventional medical treatment.²⁵ When Wardlaw et al²⁴ showed that shortterm outcomes were better with kyphoplasty than with conventional medical therapy in a prospective randomized trial, many of us had moved past questioning whether vertebral augmentation is effective and were debating the relative merits of different methods and materials.

On a personal level, most of us became proponents of these procedures because we saw dramatic results—usually unequivocal. Most patients report significant improvement in pain immediately after the procedure, and many bedridden patients are able to leave the hospital within hours. In spine surgery, few procedures give such dramatic results with so few complications.

TWO NEW STUDIES UPSET ESTABLISHED BELIEF

This is why I am having such a hard time digesting the results of the trials by Buchbinder et al² and Kallmes et al,¹ published in the August 9, 2009, issue of the New England Journal of Medicine. Both were randomized controlled trials that used sham surgery rather than conventional medical treatment as the control. The sham procedure in each trial was the same as the intervention, with local anesthetic infiltration of the periosteum and mixing of the cement (so that the patients smelled its distinctive odor), but without placing the needle into the vertebra and injecting the cement.

In the study by Buchbinder et al,² the real treatment had no benefit in any primary or secondary end point. This study did not allow crossovers.

In the study by Kallmes et al,¹ more patients who received the real treatment reported clinically meaningful improvement in pain (a secondary end point), but the difference was not quite statistically significant (64% vs 48%, P = .06). In this trial, patients were allowed to cross over to the other study group after 1 month, and significantly more patients crossed over from the sham surgery group to the active treatment group than the other way around (43% vs 12%, P < .001).

My first instinct was to pick through the papers for flaws that would invalidate the results— and there were some problems. Both studies were initially planned to include more patients and therefore to have greater statistical power, but they were reassessed because of slow enrollment. In the study by Kallmes et al,¹ the difference in clinically meaningful improvement might have reached statistical significance if the trial had been larger. The study by Buchbinder et al² was a multicenter trial, but one center accounted for 53 (69%) of the 78 patients. Could this have biased the results?

The surgeon in me also seized for a while on the idea that since all of the interventions in both studies were done by interventional radiologists, the problem may have been in patient selection and that radiologists are not as astute as we are. However, even a surgeon’s ego cannot support this interpretation.

As I looked in more detail at the response I had written to these trials, I realized these criticisms were hardly fatal flaws, and the fact that two separate well-designed studies reached the same conclusion enhances their validity.

One concern that does bear some scrutiny is that the trials were too small to identify subgroups that may benefit from the procedure. In my experience, vertebral augmentation seems to have better results with certain types of fractures. Patients with a mobile pseudarthrotic cleft pattern of fracture seem to do much better than those with the more common nonmobile fracture.

THE POWERFUL PLACEBO EFFECT

Many commentaries on these two trials have discussed a famous study of a different procedure for a different condition. In this study, Moseley et al²⁶ evaluated the use of arthroscopy to treat osteoarthritis of the knee and found that sham arthroscopy was as effective as real arthroscopy and that both were better than conventional treatment.

I was not long out of my orthopedic residency when this trial was published and was very aware of the debate that preceded it, as I once had to prepare a talk about it for resident rounds. I remember that there was a lively debate in the orthopedic community over the efficacy of the procedure before the results of this trial were released.

In contrast, the vertebral augmentation controversy had become a debate about the relative efficacy and the economics of specific techniques, not about the effectiveness of the entire concept. The mainstream had accepted the validity of the procedure, which was not the case in the knee arthroscopy trial.

In both vertebroplasty studies, the activetreatment groups and the sham-treatment groups all showed significant and rapid improvement in pain and disability, and these results were maintained over the study period. Though most vertebral compression fractures do heal, the clinical improvement is usually gradual over a period of weeks. This raises the possibility that the sham treatment was actualy an active placebo.

There is some evidence to support this possibility. In a randomized trial of the efficacy of selective nerve root blocks for lumbar radiculopathy, Riew et al²⁷ showed that injection with a local anesthetic alone, although not as efficacious as a local anesthetic plus a corticosteroid at allowing patients to avoid surgery, showed an effect long after the expected duration of the anesthetic. The effect persisted even at 5 years of follow-up.²⁸

Is it possible that the local anesthetic in this trial and the vertebroplasty trials acted as some sort of “reset button” for pain sensation? This is an area that may bear further investigation.

WHERE DOES THIS LEAVE US?

So where does this leave us? On one hand, randomized controlled trials comparing vertebral augmentation with conventional medical therapy^24,25 showed augmentation to be beneficial. On the other hand, the studies by Kallmes et al¹ and Buchbinder et al² indicate vertebroplasty is no more effective than sham surgery.

It is very difficult for me to look at my own experience with vertebral augmentation and say that, on the basis of these trials, I am no longer going to offer it to my patients. I understand on an intellectual level that these trials call the efficacy of the procedure into question, but on a visceral level I cannot rationalize it. When faced with a patient who is barely ambulatory or in fact bed-bound due to pain, my experience tells me that vertebral augmentation has a very high chance of getting them ambulatory within hours. The trials of vertebroplasty would indicate this is a placebo effect or that local anesthetic alone is as effective, but I am not yet ready to make that leap.

Cognitive dissonance seems to rule.

Cognitive dissonance describes how we respond to conflicting information that challenges our existing belief, the uncomfortable feeling we get when new evidence calls into question things that we “know” are true.

See related commentary

To the point: two recent clinical trials^1,2 have called into question the efficacy of vertebroplasty for treating osteoporotic vertebral compression fractures and have led many of us to question many of our assumptions, not only about vertebroplasty but also about evidencebased medicine.

Osteoporotic vertebral compression fractures are very common: more than 700,000 are estimated to occur in the United States annually. ³ They are costly and are associated with a risk of death.⁴ Fortunately, most heal without problems over 4 to 6 weeks with conventional treatment, ie, activity modification, analgesics, and bracing.

However, some patients do not seem to do so well and are debilitated by the pain of the fracture. Conventional fracture surgery carries very high risk and poor outcomes,⁵ and so has been reserved mostly for patients with neurologic deficits.

VERTEBROPLASTY GOES MAINSTREAM

Given these facts, investigators began looking for alternative treatments. One that rose to the fore was polymethylmethacrylate cement to stabilize the fracture. This technique, called vertebroplasty, involves injecting liquid cement through a needle into the vertebral body, where it hardens and is thought to restore stability.

Since the first description of vertebroplasty for treating symptomatic hemangiomas,⁶ many papers have been published about the procedure and about similar ones, now grouped under the general heading of vertebral augmentation. This includes kyphoplasty and other newer proprietary techniques. These procedures have been widely accepted, and their use is growing. They have shown good results in several prospective case series, and nonrandomized and randomized controlled studies have shown them to be more effective than conventional medical treatment.^7–25 For example, VERTOS, a small prospective randomized trial, showed that vertebroplasty was superior to conventional medical treatment.²⁵ When Wardlaw et al²⁴ showed that shortterm outcomes were better with kyphoplasty than with conventional medical therapy in a prospective randomized trial, many of us had moved past questioning whether vertebral augmentation is effective and were debating the relative merits of different methods and materials.

On a personal level, most of us became proponents of these procedures because we saw dramatic results—usually unequivocal. Most patients report significant improvement in pain immediately after the procedure, and many bedridden patients are able to leave the hospital within hours. In spine surgery, few procedures give such dramatic results with so few complications.

TWO NEW STUDIES UPSET ESTABLISHED BELIEF

This is why I am having such a hard time digesting the results of the trials by Buchbinder et al² and Kallmes et al,¹ published in the August 9, 2009, issue of the New England Journal of Medicine. Both were randomized controlled trials that used sham surgery rather than conventional medical treatment as the control. The sham procedure in each trial was the same as the intervention, with local anesthetic infiltration of the periosteum and mixing of the cement (so that the patients smelled its distinctive odor), but without placing the needle into the vertebra and injecting the cement.

In the study by Buchbinder et al,² the real treatment had no benefit in any primary or secondary end point. This study did not allow crossovers.

In the study by Kallmes et al,¹ more patients who received the real treatment reported clinically meaningful improvement in pain (a secondary end point), but the difference was not quite statistically significant (64% vs 48%, P = .06). In this trial, patients were allowed to cross over to the other study group after 1 month, and significantly more patients crossed over from the sham surgery group to the active treatment group than the other way around (43% vs 12%, P < .001).

My first instinct was to pick through the papers for flaws that would invalidate the results— and there were some problems. Both studies were initially planned to include more patients and therefore to have greater statistical power, but they were reassessed because of slow enrollment. In the study by Kallmes et al,¹ the difference in clinically meaningful improvement might have reached statistical significance if the trial had been larger. The study by Buchbinder et al² was a multicenter trial, but one center accounted for 53 (69%) of the 78 patients. Could this have biased the results?

The surgeon in me also seized for a while on the idea that since all of the interventions in both studies were done by interventional radiologists, the problem may have been in patient selection and that radiologists are not as astute as we are. However, even a surgeon’s ego cannot support this interpretation.

As I looked in more detail at the response I had written to these trials, I realized these criticisms were hardly fatal flaws, and the fact that two separate well-designed studies reached the same conclusion enhances their validity.

One concern that does bear some scrutiny is that the trials were too small to identify subgroups that may benefit from the procedure. In my experience, vertebral augmentation seems to have better results with certain types of fractures. Patients with a mobile pseudarthrotic cleft pattern of fracture seem to do much better than those with the more common nonmobile fracture.

THE POWERFUL PLACEBO EFFECT

Many commentaries on these two trials have discussed a famous study of a different procedure for a different condition. In this study, Moseley et al²⁶ evaluated the use of arthroscopy to treat osteoarthritis of the knee and found that sham arthroscopy was as effective as real arthroscopy and that both were better than conventional treatment.

I was not long out of my orthopedic residency when this trial was published and was very aware of the debate that preceded it, as I once had to prepare a talk about it for resident rounds. I remember that there was a lively debate in the orthopedic community over the efficacy of the procedure before the results of this trial were released.

In contrast, the vertebral augmentation controversy had become a debate about the relative efficacy and the economics of specific techniques, not about the effectiveness of the entire concept. The mainstream had accepted the validity of the procedure, which was not the case in the knee arthroscopy trial.

In both vertebroplasty studies, the activetreatment groups and the sham-treatment groups all showed significant and rapid improvement in pain and disability, and these results were maintained over the study period. Though most vertebral compression fractures do heal, the clinical improvement is usually gradual over a period of weeks. This raises the possibility that the sham treatment was actualy an active placebo.

There is some evidence to support this possibility. In a randomized trial of the efficacy of selective nerve root blocks for lumbar radiculopathy, Riew et al²⁷ showed that injection with a local anesthetic alone, although not as efficacious as a local anesthetic plus a corticosteroid at allowing patients to avoid surgery, showed an effect long after the expected duration of the anesthetic. The effect persisted even at 5 years of follow-up.²⁸

Is it possible that the local anesthetic in this trial and the vertebroplasty trials acted as some sort of “reset button” for pain sensation? This is an area that may bear further investigation.

WHERE DOES THIS LEAVE US?

So where does this leave us? On one hand, randomized controlled trials comparing vertebral augmentation with conventional medical therapy^24,25 showed augmentation to be beneficial. On the other hand, the studies by Kallmes et al¹ and Buchbinder et al² indicate vertebroplasty is no more effective than sham surgery.

It is very difficult for me to look at my own experience with vertebral augmentation and say that, on the basis of these trials, I am no longer going to offer it to my patients. I understand on an intellectual level that these trials call the efficacy of the procedure into question, but on a visceral level I cannot rationalize it. When faced with a patient who is barely ambulatory or in fact bed-bound due to pain, my experience tells me that vertebral augmentation has a very high chance of getting them ambulatory within hours. The trials of vertebroplasty would indicate this is a placebo effect or that local anesthetic alone is as effective, but I am not yet ready to make that leap.

Cognitive dissonance seems to rule.

Current Procedural Terminology (CPT) 2010, which took effect January 1, doesn’t bring many changes for ObGyn practice, but there’s been a major backpedaling in Medicare coverage of consultations that you must be aware of. In conjunction with this move by the Centers for Medicare & Medicaid Services (CMS), CPT has added a definition of “transfer of care” and established two possible reasons for providing a consultation. I’ll have more to report about these important developments later in this article.

Among the changes to billing codes for the work performed in ObGyn: rebundling of commonly performed urodynamics procedures and new codes for revision of a vaginal graft. There is also a new (and unpublished) code for administering the H1N1 influenza vaccine.

Last, CPT has revised the explanation of non–face-to-face prolonged services. Read on!

New codes bundle urodynamic studies—a product of joint CMS and CPT input

The biggest changes in coding for ObGyn procedures are urodynamics study codes. The American Medical Association (AMA) has 1) created three new codes that represent test bundles and, in the process, 2) deleted the stand-alone urodynamics codes 51772 (urethral pressure profile studies [UPP] [urethral closure pressure profile], any technique) and 51795 (voiding pressure studies; bladder voiding pressure, any technique).

These changes were made because the most commonly reported codes for a female patient were billed together 90% of the time (51726, 51772, 51795, and 51797); the AMA reasoned that the most frequent combinations were considered overvalued when billed separately—that is, there was no repeat of pre-test and post-test work when these combinations were performed and there was no duplication in the cost of supplies and staff time.

The new bundles were therefore considered to better reflect current medical practice, and the Relative Value Update Committee (RUC) recommended, and CMS accepted, the relative value units (RVU) for the combination codes to reflect the true physician work value and practice expense of the combined procedures.

New and revised codes are:

51726 Complex cystometrogram (i.e., calibrated electronic equipment)

51727 …with urethral pressure profile studies (i.e., urethral closure pressure profile), any technique

51728 …with voiding pressure studies (i.e., bladder voiding pressure), any technique

51729 …with voiding pressure studies (i.e., bladder voiding pressure) and urethral pressure profile studies (i.e., urethral closure pressure profile), any technique.

According to the clinical vignette submitted to the AMA for code 51727, this procedure will include a sustained Valsalva maneuver as part of the urethral closure pressure profile. CPT did, however, retain the add-on code +51797 (voiding pressure studies, intra-abdominal [i.e., rectal, gastric, intraperitoneal]) and has clarified that 51797 may be billed in addition to 51728 and 51729 if a rectal catheter is placed to determine if the patient is straining during the voiding event.

In other words, the add-on code may be reported only when the primary procedure includes a voiding pressure study.

RVU for these new procedures have also been revised (see the TABLE ). Notable is the seeming discrepancy in RVU between code 51726 (cystometrogram alone) and the bundled tests. This is the case because the practice expense for 51726 has not reached its final level (the practice expense RVU are being increased or decreased in increments over several years); for 2010 only, therefore, this code will have a higher total RVU value than the new codes (51727, 51728, 51729), despite having a lower physician work relative value.

The discrepancy will be corrected in 2011, when 51726 will have lower RVU than the other urodynamics combination test codes.

TABLE

Changes in 2010 to RVU for urodynamic studies

Laparoscopic revision of a vaginal graft

In 2006, the AMA added the code for a vaginal approach to revising a graft (57295, revision [including removal] of prosthetic vaginal graft; vaginal approach). Then, in 2007, it added a code for an abdominal approach (57296, revision [including removal] of prosthetic vaginal graft; open abdominal approach).

Now, you have a code for a laparoscopic approach, completing the code set for this procedure. As with 57295 and 57296, report the new code when the graft is either revised or removed entirely.

57426 Revision (including removal) of prosthetic vaginal graft, laparoscopic approach

Other, miscellaneous changes take effect

OBSTETRIC PANEL

Although code 80055 comprises a battery of tests that are performed routinely on obstetric patients, a new code, 86780, was created to report syphilis screening using a treponemal antibody method, in which IgM and IgG antibodies are measured. This test is not the same syphilis test that is now part of the 80055 panel. CPT has therefore cautioned that, when you use code 86780 instead of the standard syphilis test code 86592, you should not report the obstetrics panel but, instead, separately report each test performed.

REPRODUCTIVE MEDICINE

New code 89398 (unlisted reproductive medicine laboratory procedure) has been added, but CPT still directs billers to use the unlisted miscellaneous pathology test code 89240 to report cryopreservation of reproductive ovarian tissues.

BILLING FOR THE H1N1 INFLUENZA VACCINE

Because of the urgency of collecting data on the H1N1 influenza epidemic, CPT has revised code 90663 to include the H1N1 formulation of the flu vaccine product. In addition, CPT has created a new code, 90470, for administering the H1N1 flu vaccine, which became valid in September (but which isn’t included in the hard-copy version of CPT 2010). The new code is to be used for intramuscular injection or intranasal administration, and includes any time spent counseling.

In addition:

• Do not report established code 90471 (immunization administration [includes percutaneous, intradermal, subcutaneous, or intramuscular injections]; one vaccine [single or combination vaccine/toxoid]) when you administer the H1N1 flu vaccine
• Report the vaccine product code only when your practice has purchased the vaccine, or when the payer requires the code with a 0 charge to match the administration code.
• Medicare coding for administering the H1N1 flu vaccine is different than what I’ve just described. Do not use CPT codes for Medicare patients; instead, code H1N1 flu immunization as:

G9141 Influenza A (H1N1) immunization administration (includes the physician counseling the patient/family)

G9142 Influenza A (H1N1) vaccine, any route of administration

Medicare will not reimburse for the vaccine product because it is being given to its providers without cost. Some carriers may require that the new vaccine product code be listed with a 0 charge.

Prolonged inpatient E/M services

CPT has revised guidelines for prolonged services that do not involve direct face-to-face contact with a patient. Keep in mind, however, that, although these changes are welcome, many payers don’t reimburse separately for work that isn’t performed face to face.

These codes are no longer considered add-on codes; they can be reported on a different date than the related E/M service.

According to CPT, codes 99358 and 99359 are reported when the prolonged time:

• is greater than would be expected for normal pre-service and post-service work associated with the E/M service
• exceeds 30 minutes
• is related to an E/M service that has already occurred, or to one that will occur and represents ongoing patient management (for example, your review of extensive patient records that weren’t available at the time of the visit)
• is in addition to any telephone services codes (99441–99443)—but not with more specific codes, such as medical team conferences, online medical evaluation, or care plan oversight services, which have no upper limit to the time required to accomplish the service.

Consultation codes and clarifications

Two changes of note, from a CPT perspective, have been made in the area of consultations. CPT has:

• added a definition for a transfer of care
• defined two circumstances under which a consultation can be coded. These revisions come at the same time Medicare has made the decision to no longer pay for consultations other than tele-health consults (see following section).

For 2010, CPT defines transfer of care as

…the process whereby a physician who is providing management for some or all of a patient’s problems relinquishes this responsibility to another physician who agrees to accept this responsibility and who, from the initial encounter, is not providing consultative services.

The guidelines also explain that 1) a transferring physician is no longer responsible for caring for the problem for which the patient was referred and 2) the consultation codes should not be reported by the physician who accepts care.

Two alternative conditions must now apply for a consultation to be considered provided:

• A physician requested an opinion or advice for a specific condition or problem, or
• The consulting physician saw the patient first to determine whether to accept ongoing management of her entire care or of a specific condition or problem (i.e., transfer of care).

The second condition is new; it remains to be seen if payers will accept it as a valid reason to bill for consultation.

As with all billable services, you should ensure that the criteria required by the payer you are billing have been met. CPT also directs that the written request for consultation can be documented by either the requesting or the receiving physician—something that was unacceptable under Medicare guidelines.

Last, CPT has added instructions to clarify the type of consultation code to bill under certain circumstances:

• When the patient is admitted after an outpatient consultation but the physician does not see the patient on the unit on the date of admission, bill only for outpatient consultation
• When the patient is seen for an office visit, emergency room visit, or outpatient consult on the date of admission and the physician then sees the patient on the unit that day, bill only the inpatient consultation or initial hospital care code, whichever applies. All services that day are used to determine the final level of service.

Medicare tilts the playing field on consultations

Although CPT has retained all consultation codes, and although the hope is that commercial payers will continue to reimburse for such services in the near future, the big news is that Medicare has announced that it will no longer recognize (or reimburse for) codes for outpatient or inpatient consultations. (Note: This story is still unfolding, however. The changes announced by Medicare that I discuss below are still before Congress as this article goes to press. Although Medicare has, in fact, released the transmittal letter to all carriers instructing them about the changes, Senator Arlen Specter [D-Pa] has introduced an amendment to the Patient Protection and Affordable Care Act [H.R. 3590] to postpone the policy change for 1 year. If Congress has not passed this bill before the end of 2009, the changes go through as planned. Stay tuned for developments!)

Assuming the changes go through, here is what is expected of you in the circumstances of providing consultations and billing Medicare (Medicaid payers aren’t required to follow this policy change but may opt to do so).

Outpatients. Document, and report, the appropriate level of visit for a new or established Medicare patient using outpatient codes 99201–99215

Inpatients. If you are a non-admitting physician asked to see a patient for the first time, report the appropriate level of initial hospital care (codes 99221–99223). Note the following three points:

• Initial hospital care includes only three levels of service—not the five levels from which you choose for consultation codes
• The lowest level of history and exam for these initial visit codes is a detailed history and examination—no matter the level of medical decision-making. If the level of history or exam is documented lower than “detailed”—say, as “expanded problem-focused”—you are required to report the unlisted E/M code 99499.
• The admitting physician adds the new Healthcare Common Procedure Coding System (HCPCS) modifier –AI (that is, “‘A’ upper-case ‘i’”) to the initial visit code, so that Medicare can distinguish the admitting physician from others providing care for the patient.
• All subsequent visits with the inpatient continue to be billed with the subsequent care inpatient codes (99231–99233).

Fallout from this change? Medicare is studying the implications of its new policy on secondary payments—that is, when Medicare is the primary payer and there is a supplemental carrier, or when Medicare is the secondary payer. Note: Medicare strongly advises all providers to check with their primary payers, because 1) Medicare will not accept a consultation code when a primary insurer has paid on that code and 2) it’s doubtful that a commercial payer will accept a consultation code when Medicare has paid for a new or established patient service.

To add to the turmoil…

The CMS has announced that, as a result of the changes in Medicare policy on consultations, it is increasing the relative values for all new and established patient services and initial hospital care. CMS is doing this, however, by reducing the relative values of some consultation codes.

In addition, all surgical procedure codes that carry a 10- or 90-day global period will see an increase in work RVU because of the increase in E/M services that are a part of all global care. Keep in mind that payers who use the Resource-Based Relative Value Scale (RBRVS) to reimburse services will probably adopt the new values when contracts are up for renewal, although many will be unable to do so in the short term.

It also remains to be seen if any commercial payers adopt Medicare policy or continue to pay for consultations. This area might be a contract issue with payers.

Current Procedural Terminology (CPT) 2010, which took effect January 1, doesn’t bring many changes for ObGyn practice, but there’s been a major backpedaling in Medicare coverage of consultations that you must be aware of. In conjunction with this move by the Centers for Medicare & Medicaid Services (CMS), CPT has added a definition of “transfer of care” and established two possible reasons for providing a consultation. I’ll have more to report about these important developments later in this article.

Among the changes to billing codes for the work performed in ObGyn: rebundling of commonly performed urodynamics procedures and new codes for revision of a vaginal graft. There is also a new (and unpublished) code for administering the H1N1 influenza vaccine.

Last, CPT has revised the explanation of non–face-to-face prolonged services. Read on!

New codes bundle urodynamic studies—a product of joint CMS and CPT input

The biggest changes in coding for ObGyn procedures are urodynamics study codes. The American Medical Association (AMA) has 1) created three new codes that represent test bundles and, in the process, 2) deleted the stand-alone urodynamics codes 51772 (urethral pressure profile studies [UPP] [urethral closure pressure profile], any technique) and 51795 (voiding pressure studies; bladder voiding pressure, any technique).

These changes were made because the most commonly reported codes for a female patient were billed together 90% of the time (51726, 51772, 51795, and 51797); the AMA reasoned that the most frequent combinations were considered overvalued when billed separately—that is, there was no repeat of pre-test and post-test work when these combinations were performed and there was no duplication in the cost of supplies and staff time.

The new bundles were therefore considered to better reflect current medical practice, and the Relative Value Update Committee (RUC) recommended, and CMS accepted, the relative value units (RVU) for the combination codes to reflect the true physician work value and practice expense of the combined procedures.

New and revised codes are:

51726 Complex cystometrogram (i.e., calibrated electronic equipment)

51727 …with urethral pressure profile studies (i.e., urethral closure pressure profile), any technique

51728 …with voiding pressure studies (i.e., bladder voiding pressure), any technique

51729 …with voiding pressure studies (i.e., bladder voiding pressure) and urethral pressure profile studies (i.e., urethral closure pressure profile), any technique.

According to the clinical vignette submitted to the AMA for code 51727, this procedure will include a sustained Valsalva maneuver as part of the urethral closure pressure profile. CPT did, however, retain the add-on code +51797 (voiding pressure studies, intra-abdominal [i.e., rectal, gastric, intraperitoneal]) and has clarified that 51797 may be billed in addition to 51728 and 51729 if a rectal catheter is placed to determine if the patient is straining during the voiding event.

In other words, the add-on code may be reported only when the primary procedure includes a voiding pressure study.

RVU for these new procedures have also been revised (see the TABLE ). Notable is the seeming discrepancy in RVU between code 51726 (cystometrogram alone) and the bundled tests. This is the case because the practice expense for 51726 has not reached its final level (the practice expense RVU are being increased or decreased in increments over several years); for 2010 only, therefore, this code will have a higher total RVU value than the new codes (51727, 51728, 51729), despite having a lower physician work relative value.

The discrepancy will be corrected in 2011, when 51726 will have lower RVU than the other urodynamics combination test codes.

TABLE

Changes in 2010 to RVU for urodynamic studies

Laparoscopic revision of a vaginal graft

In 2006, the AMA added the code for a vaginal approach to revising a graft (57295, revision [including removal] of prosthetic vaginal graft; vaginal approach). Then, in 2007, it added a code for an abdominal approach (57296, revision [including removal] of prosthetic vaginal graft; open abdominal approach).

Now, you have a code for a laparoscopic approach, completing the code set for this procedure. As with 57295 and 57296, report the new code when the graft is either revised or removed entirely.

57426 Revision (including removal) of prosthetic vaginal graft, laparoscopic approach

Other, miscellaneous changes take effect

OBSTETRIC PANEL

Although code 80055 comprises a battery of tests that are performed routinely on obstetric patients, a new code, 86780, was created to report syphilis screening using a treponemal antibody method, in which IgM and IgG antibodies are measured. This test is not the same syphilis test that is now part of the 80055 panel. CPT has therefore cautioned that, when you use code 86780 instead of the standard syphilis test code 86592, you should not report the obstetrics panel but, instead, separately report each test performed.

REPRODUCTIVE MEDICINE

New code 89398 (unlisted reproductive medicine laboratory procedure) has been added, but CPT still directs billers to use the unlisted miscellaneous pathology test code 89240 to report cryopreservation of reproductive ovarian tissues.

BILLING FOR THE H1N1 INFLUENZA VACCINE

Because of the urgency of collecting data on the H1N1 influenza epidemic, CPT has revised code 90663 to include the H1N1 formulation of the flu vaccine product. In addition, CPT has created a new code, 90470, for administering the H1N1 flu vaccine, which became valid in September (but which isn’t included in the hard-copy version of CPT 2010). The new code is to be used for intramuscular injection or intranasal administration, and includes any time spent counseling.

In addition:

• Do not report established code 90471 (immunization administration [includes percutaneous, intradermal, subcutaneous, or intramuscular injections]; one vaccine [single or combination vaccine/toxoid]) when you administer the H1N1 flu vaccine
• Report the vaccine product code only when your practice has purchased the vaccine, or when the payer requires the code with a 0 charge to match the administration code.
• Medicare coding for administering the H1N1 flu vaccine is different than what I’ve just described. Do not use CPT codes for Medicare patients; instead, code H1N1 flu immunization as:

G9141 Influenza A (H1N1) immunization administration (includes the physician counseling the patient/family)

G9142 Influenza A (H1N1) vaccine, any route of administration

Medicare will not reimburse for the vaccine product because it is being given to its providers without cost. Some carriers may require that the new vaccine product code be listed with a 0 charge.

Prolonged inpatient E/M services

CPT has revised guidelines for prolonged services that do not involve direct face-to-face contact with a patient. Keep in mind, however, that, although these changes are welcome, many payers don’t reimburse separately for work that isn’t performed face to face.

These codes are no longer considered add-on codes; they can be reported on a different date than the related E/M service.

According to CPT, codes 99358 and 99359 are reported when the prolonged time:

• is greater than would be expected for normal pre-service and post-service work associated with the E/M service
• exceeds 30 minutes
• is related to an E/M service that has already occurred, or to one that will occur and represents ongoing patient management (for example, your review of extensive patient records that weren’t available at the time of the visit)
• is in addition to any telephone services codes (99441–99443)—but not with more specific codes, such as medical team conferences, online medical evaluation, or care plan oversight services, which have no upper limit to the time required to accomplish the service.

Consultation codes and clarifications

Two changes of note, from a CPT perspective, have been made in the area of consultations. CPT has:

• added a definition for a transfer of care
• defined two circumstances under which a consultation can be coded. These revisions come at the same time Medicare has made the decision to no longer pay for consultations other than tele-health consults (see following section).

For 2010, CPT defines transfer of care as

…the process whereby a physician who is providing management for some or all of a patient’s problems relinquishes this responsibility to another physician who agrees to accept this responsibility and who, from the initial encounter, is not providing consultative services.

The guidelines also explain that 1) a transferring physician is no longer responsible for caring for the problem for which the patient was referred and 2) the consultation codes should not be reported by the physician who accepts care.

Two alternative conditions must now apply for a consultation to be considered provided:

• A physician requested an opinion or advice for a specific condition or problem, or
• The consulting physician saw the patient first to determine whether to accept ongoing management of her entire care or of a specific condition or problem (i.e., transfer of care).

The second condition is new; it remains to be seen if payers will accept it as a valid reason to bill for consultation.

As with all billable services, you should ensure that the criteria required by the payer you are billing have been met. CPT also directs that the written request for consultation can be documented by either the requesting or the receiving physician—something that was unacceptable under Medicare guidelines.

Last, CPT has added instructions to clarify the type of consultation code to bill under certain circumstances:

• When the patient is admitted after an outpatient consultation but the physician does not see the patient on the unit on the date of admission, bill only for outpatient consultation
• When the patient is seen for an office visit, emergency room visit, or outpatient consult on the date of admission and the physician then sees the patient on the unit that day, bill only the inpatient consultation or initial hospital care code, whichever applies. All services that day are used to determine the final level of service.

Medicare tilts the playing field on consultations

Although CPT has retained all consultation codes, and although the hope is that commercial payers will continue to reimburse for such services in the near future, the big news is that Medicare has announced that it will no longer recognize (or reimburse for) codes for outpatient or inpatient consultations. (Note: This story is still unfolding, however. The changes announced by Medicare that I discuss below are still before Congress as this article goes to press. Although Medicare has, in fact, released the transmittal letter to all carriers instructing them about the changes, Senator Arlen Specter [D-Pa] has introduced an amendment to the Patient Protection and Affordable Care Act [H.R. 3590] to postpone the policy change for 1 year. If Congress has not passed this bill before the end of 2009, the changes go through as planned. Stay tuned for developments!)

Assuming the changes go through, here is what is expected of you in the circumstances of providing consultations and billing Medicare (Medicaid payers aren’t required to follow this policy change but may opt to do so).

Outpatients. Document, and report, the appropriate level of visit for a new or established Medicare patient using outpatient codes 99201–99215

Inpatients. If you are a non-admitting physician asked to see a patient for the first time, report the appropriate level of initial hospital care (codes 99221–99223). Note the following three points:

• Initial hospital care includes only three levels of service—not the five levels from which you choose for consultation codes
• The lowest level of history and exam for these initial visit codes is a detailed history and examination—no matter the level of medical decision-making. If the level of history or exam is documented lower than “detailed”—say, as “expanded problem-focused”—you are required to report the unlisted E/M code 99499.
• The admitting physician adds the new Healthcare Common Procedure Coding System (HCPCS) modifier –AI (that is, “‘A’ upper-case ‘i’”) to the initial visit code, so that Medicare can distinguish the admitting physician from others providing care for the patient.
• All subsequent visits with the inpatient continue to be billed with the subsequent care inpatient codes (99231–99233).

Fallout from this change? Medicare is studying the implications of its new policy on secondary payments—that is, when Medicare is the primary payer and there is a supplemental carrier, or when Medicare is the secondary payer. Note: Medicare strongly advises all providers to check with their primary payers, because 1) Medicare will not accept a consultation code when a primary insurer has paid on that code and 2) it’s doubtful that a commercial payer will accept a consultation code when Medicare has paid for a new or established patient service.

To add to the turmoil…

The CMS has announced that, as a result of the changes in Medicare policy on consultations, it is increasing the relative values for all new and established patient services and initial hospital care. CMS is doing this, however, by reducing the relative values of some consultation codes.

In addition, all surgical procedure codes that carry a 10- or 90-day global period will see an increase in work RVU because of the increase in E/M services that are a part of all global care. Keep in mind that payers who use the Resource-Based Relative Value Scale (RBRVS) to reimburse services will probably adopt the new values when contracts are up for renewal, although many will be unable to do so in the short term.

It also remains to be seen if any commercial payers adopt Medicare policy or continue to pay for consultations. This area might be a contract issue with payers.

Current Procedural Terminology (CPT) 2010, which took effect January 1, doesn’t bring many changes for ObGyn practice, but there’s been a major backpedaling in Medicare coverage of consultations that you must be aware of. In conjunction with this move by the Centers for Medicare & Medicaid Services (CMS), CPT has added a definition of “transfer of care” and established two possible reasons for providing a consultation. I’ll have more to report about these important developments later in this article.

Among the changes to billing codes for the work performed in ObGyn: rebundling of commonly performed urodynamics procedures and new codes for revision of a vaginal graft. There is also a new (and unpublished) code for administering the H1N1 influenza vaccine.

Last, CPT has revised the explanation of non–face-to-face prolonged services. Read on!

New codes bundle urodynamic studies—a product of joint CMS and CPT input

The biggest changes in coding for ObGyn procedures are urodynamics study codes. The American Medical Association (AMA) has 1) created three new codes that represent test bundles and, in the process, 2) deleted the stand-alone urodynamics codes 51772 (urethral pressure profile studies [UPP] [urethral closure pressure profile], any technique) and 51795 (voiding pressure studies; bladder voiding pressure, any technique).

These changes were made because the most commonly reported codes for a female patient were billed together 90% of the time (51726, 51772, 51795, and 51797); the AMA reasoned that the most frequent combinations were considered overvalued when billed separately—that is, there was no repeat of pre-test and post-test work when these combinations were performed and there was no duplication in the cost of supplies and staff time.

The new bundles were therefore considered to better reflect current medical practice, and the Relative Value Update Committee (RUC) recommended, and CMS accepted, the relative value units (RVU) for the combination codes to reflect the true physician work value and practice expense of the combined procedures.

New and revised codes are:

51726 Complex cystometrogram (i.e., calibrated electronic equipment)

51727 …with urethral pressure profile studies (i.e., urethral closure pressure profile), any technique

51728 …with voiding pressure studies (i.e., bladder voiding pressure), any technique

51729 …with voiding pressure studies (i.e., bladder voiding pressure) and urethral pressure profile studies (i.e., urethral closure pressure profile), any technique.

According to the clinical vignette submitted to the AMA for code 51727, this procedure will include a sustained Valsalva maneuver as part of the urethral closure pressure profile. CPT did, however, retain the add-on code +51797 (voiding pressure studies, intra-abdominal [i.e., rectal, gastric, intraperitoneal]) and has clarified that 51797 may be billed in addition to 51728 and 51729 if a rectal catheter is placed to determine if the patient is straining during the voiding event.

In other words, the add-on code may be reported only when the primary procedure includes a voiding pressure study.

RVU for these new procedures have also been revised (see the TABLE ). Notable is the seeming discrepancy in RVU between code 51726 (cystometrogram alone) and the bundled tests. This is the case because the practice expense for 51726 has not reached its final level (the practice expense RVU are being increased or decreased in increments over several years); for 2010 only, therefore, this code will have a higher total RVU value than the new codes (51727, 51728, 51729), despite having a lower physician work relative value.

The discrepancy will be corrected in 2011, when 51726 will have lower RVU than the other urodynamics combination test codes.

TABLE

Changes in 2010 to RVU for urodynamic studies

Laparoscopic revision of a vaginal graft

In 2006, the AMA added the code for a vaginal approach to revising a graft (57295, revision [including removal] of prosthetic vaginal graft; vaginal approach). Then, in 2007, it added a code for an abdominal approach (57296, revision [including removal] of prosthetic vaginal graft; open abdominal approach).

Now, you have a code for a laparoscopic approach, completing the code set for this procedure. As with 57295 and 57296, report the new code when the graft is either revised or removed entirely.

57426 Revision (including removal) of prosthetic vaginal graft, laparoscopic approach

Other, miscellaneous changes take effect

OBSTETRIC PANEL

Although code 80055 comprises a battery of tests that are performed routinely on obstetric patients, a new code, 86780, was created to report syphilis screening using a treponemal antibody method, in which IgM and IgG antibodies are measured. This test is not the same syphilis test that is now part of the 80055 panel. CPT has therefore cautioned that, when you use code 86780 instead of the standard syphilis test code 86592, you should not report the obstetrics panel but, instead, separately report each test performed.

REPRODUCTIVE MEDICINE

New code 89398 (unlisted reproductive medicine laboratory procedure) has been added, but CPT still directs billers to use the unlisted miscellaneous pathology test code 89240 to report cryopreservation of reproductive ovarian tissues.

BILLING FOR THE H1N1 INFLUENZA VACCINE

Because of the urgency of collecting data on the H1N1 influenza epidemic, CPT has revised code 90663 to include the H1N1 formulation of the flu vaccine product. In addition, CPT has created a new code, 90470, for administering the H1N1 flu vaccine, which became valid in September (but which isn’t included in the hard-copy version of CPT 2010). The new code is to be used for intramuscular injection or intranasal administration, and includes any time spent counseling.

In addition:

• Do not report established code 90471 (immunization administration [includes percutaneous, intradermal, subcutaneous, or intramuscular injections]; one vaccine [single or combination vaccine/toxoid]) when you administer the H1N1 flu vaccine
• Report the vaccine product code only when your practice has purchased the vaccine, or when the payer requires the code with a 0 charge to match the administration code.
• Medicare coding for administering the H1N1 flu vaccine is different than what I’ve just described. Do not use CPT codes for Medicare patients; instead, code H1N1 flu immunization as:

G9141 Influenza A (H1N1) immunization administration (includes the physician counseling the patient/family)

G9142 Influenza A (H1N1) vaccine, any route of administration

Medicare will not reimburse for the vaccine product because it is being given to its providers without cost. Some carriers may require that the new vaccine product code be listed with a 0 charge.

Prolonged inpatient E/M services

CPT has revised guidelines for prolonged services that do not involve direct face-to-face contact with a patient. Keep in mind, however, that, although these changes are welcome, many payers don’t reimburse separately for work that isn’t performed face to face.

These codes are no longer considered add-on codes; they can be reported on a different date than the related E/M service.

According to CPT, codes 99358 and 99359 are reported when the prolonged time:

• is greater than would be expected for normal pre-service and post-service work associated with the E/M service
• exceeds 30 minutes
• is related to an E/M service that has already occurred, or to one that will occur and represents ongoing patient management (for example, your review of extensive patient records that weren’t available at the time of the visit)
• is in addition to any telephone services codes (99441–99443)—but not with more specific codes, such as medical team conferences, online medical evaluation, or care plan oversight services, which have no upper limit to the time required to accomplish the service.

Consultation codes and clarifications

Two changes of note, from a CPT perspective, have been made in the area of consultations. CPT has:

• added a definition for a transfer of care
• defined two circumstances under which a consultation can be coded. These revisions come at the same time Medicare has made the decision to no longer pay for consultations other than tele-health consults (see following section).

For 2010, CPT defines transfer of care as

…the process whereby a physician who is providing management for some or all of a patient’s problems relinquishes this responsibility to another physician who agrees to accept this responsibility and who, from the initial encounter, is not providing consultative services.

The guidelines also explain that 1) a transferring physician is no longer responsible for caring for the problem for which the patient was referred and 2) the consultation codes should not be reported by the physician who accepts care.

Two alternative conditions must now apply for a consultation to be considered provided:

• A physician requested an opinion or advice for a specific condition or problem, or
• The consulting physician saw the patient first to determine whether to accept ongoing management of her entire care or of a specific condition or problem (i.e., transfer of care).

The second condition is new; it remains to be seen if payers will accept it as a valid reason to bill for consultation.

As with all billable services, you should ensure that the criteria required by the payer you are billing have been met. CPT also directs that the written request for consultation can be documented by either the requesting or the receiving physician—something that was unacceptable under Medicare guidelines.

Last, CPT has added instructions to clarify the type of consultation code to bill under certain circumstances:

• When the patient is admitted after an outpatient consultation but the physician does not see the patient on the unit on the date of admission, bill only for outpatient consultation
• When the patient is seen for an office visit, emergency room visit, or outpatient consult on the date of admission and the physician then sees the patient on the unit that day, bill only the inpatient consultation or initial hospital care code, whichever applies. All services that day are used to determine the final level of service.

Medicare tilts the playing field on consultations

Although CPT has retained all consultation codes, and although the hope is that commercial payers will continue to reimburse for such services in the near future, the big news is that Medicare has announced that it will no longer recognize (or reimburse for) codes for outpatient or inpatient consultations. (Note: This story is still unfolding, however. The changes announced by Medicare that I discuss below are still before Congress as this article goes to press. Although Medicare has, in fact, released the transmittal letter to all carriers instructing them about the changes, Senator Arlen Specter [D-Pa] has introduced an amendment to the Patient Protection and Affordable Care Act [H.R. 3590] to postpone the policy change for 1 year. If Congress has not passed this bill before the end of 2009, the changes go through as planned. Stay tuned for developments!)

Assuming the changes go through, here is what is expected of you in the circumstances of providing consultations and billing Medicare (Medicaid payers aren’t required to follow this policy change but may opt to do so).

Outpatients. Document, and report, the appropriate level of visit for a new or established Medicare patient using outpatient codes 99201–99215

Inpatients. If you are a non-admitting physician asked to see a patient for the first time, report the appropriate level of initial hospital care (codes 99221–99223). Note the following three points:

• Initial hospital care includes only three levels of service—not the five levels from which you choose for consultation codes
• The lowest level of history and exam for these initial visit codes is a detailed history and examination—no matter the level of medical decision-making. If the level of history or exam is documented lower than “detailed”—say, as “expanded problem-focused”—you are required to report the unlisted E/M code 99499.
• The admitting physician adds the new Healthcare Common Procedure Coding System (HCPCS) modifier –AI (that is, “‘A’ upper-case ‘i’”) to the initial visit code, so that Medicare can distinguish the admitting physician from others providing care for the patient.
• All subsequent visits with the inpatient continue to be billed with the subsequent care inpatient codes (99231–99233).

Fallout from this change? Medicare is studying the implications of its new policy on secondary payments—that is, when Medicare is the primary payer and there is a supplemental carrier, or when Medicare is the secondary payer. Note: Medicare strongly advises all providers to check with their primary payers, because 1) Medicare will not accept a consultation code when a primary insurer has paid on that code and 2) it’s doubtful that a commercial payer will accept a consultation code when Medicare has paid for a new or established patient service.

To add to the turmoil…

The CMS has announced that, as a result of the changes in Medicare policy on consultations, it is increasing the relative values for all new and established patient services and initial hospital care. CMS is doing this, however, by reducing the relative values of some consultation codes.

In addition, all surgical procedure codes that carry a 10- or 90-day global period will see an increase in work RVU because of the increase in E/M services that are a part of all global care. Keep in mind that payers who use the Resource-Based Relative Value Scale (RBRVS) to reimburse services will probably adopt the new values when contracts are up for renewal, although many will be unable to do so in the short term.

It also remains to be seen if any commercial payers adopt Medicare policy or continue to pay for consultations. This area might be a contract issue with payers.

Guanfacine extended release (GXR)—a selective α-2 adrenergic agonist FDA-approved for the treatment of attention-deficit/hyperactivity disorder (ADHD)—has demonstrated efficacy for inattentive and hyperactive/impulsive symptom domains in 2 large trials lasting 8 and 9 weeks.^1,2 GXR’s once-daily formulation may increase adherence and deliver consistent control of symptoms across a full day ( Table 1 ).

Table 1

Guanfacine extended release: Fast facts

Clinical implications

GXR exhibits enhancement of noradrenergic pathways through selective direct receptor action in the prefrontal cortex.³ This mechanism of action is different from that of other FDA-approved ADHD medications. GXR can be used alone or in combination with stimulants or atomoxetine for treating complex ADHD, such as cases accompanied by oppositional features and emotional dysregulation or characterized by partial stimulant response.

How it works

Guanfacine—originally developed as an immediate-release (IR) antihypertensive—reduces sympathetic tone, causing centrally mediated vasodilation and reduced heart rate. Although GXR’s mechanism of action in ADHD is not known, the drug is a selective α-2A receptor agonist thought to directly engage postsynaptic receptors in the prefrontal cortex (PFC), an area of the brain believed to play a major role in attentional and organizational functions that preclinical research has linked to ADHD.³

The postsynaptic α-2A receptor is thought to play a central role in the optimal functioning of the PFC as illustrated by the “inverted U hypothesis of PFC activation.”⁴ In this model, cyclic adenosine monophosphate (cAMP) levels build within the prefrontal cortical neurons and cause specific ion channels—hyperpolarization-activated cyclic nucleotide gated (HCN) channels—to open on dendritic spines of these neurons.⁵ Activation of HCN channels effectively reduces membrane resistance, cutting off synaptic inputs and disconnecting PFC network connections. Because α-2A receptors are located in proximity to HCN channels, their stimulation by GXR closes HCN channels, inhibits further production of cAMP, and reestablishes synaptic function and the resulting network connectivity.⁵ Blockade of α-2A receptors by yohimbine reverses this process, eroding network connectivity, and in monkeys has been demonstrated to impair working memory,⁶ damage inhibition/impulse control, and produce locomotor hyperactivity.

Direct stimulation by GXR of the postsynaptic α-2A receptors is thought to:

• strengthen working memory
• reduce susceptibility to distraction
• improve attention regulation
• improve behavioral inhibition
• enhance impulse control.⁷

Pharmacokinetics

GXR offers enhanced pharmaceutics relative to IR guanfacine. IR guanfacine exhibits poor absorption characteristics—peak plasma concentration is achieved too rapidly and then declines precipitously, with considerable inter-individual variation.

GXR’s once-daily formulation is implemented by a proprietary enteric-coated sustained release mechanism⁸ that is meant to:

• control absorption
• provide a broad but flat plasma concentration profile
• reduce inter-individual variation of guanfacine exposure.

Compared with IR guanfacine, GXR exhibits delayed time of maximum concentration (Tmax) and reduced maximum concentration (Cmax). Therapeutic concentrations can be sustained over longer periods with reduced peak-to-trough fluctuation,⁸ which tends to improve tolerability and symptom control throughout the day. The convenience of once-daily dosing also may increase adherence.

GXR’s pharmacokinetic characteristics do not change with dose, but high-fat meals will increase absorption of the drug—Cmax increases by 75% and area under the plasma concentration time curve increases by 40%. Because GXR primarily is metabolized through cytochrome P450 (CYP) 3A4, CYP3A4 inhibitors such as ketoconazole will increase guanfacine plasma concentrations and elevate the risk of adverse events such as bradycardia, hypotension, and sedation. Conversely, CYP3A4 inducers such as rifampin will significantly reduce total guanfacine exposure. Coadministration of valproic acid with GXR can result in increased valproic acid levels, producing additive CNS side effects.

Efficacy

GXR reduced both inattentive and hyperactive/impulsive symptoms in 2 phase III, forced-dose, parallel-design, randomized, placebo-controlled trials ( Table 2 ). In the first trial,¹ 345 children age 6 to 17 received placebo or GXR, 2 mg, 3 mg, or 4 mg once daily for 8 weeks. In the second study,² 324 children age 6 to 17 received placebo or GXR, 1 mg, 2 mg, 3 mg, or 4 mg, once daily for 9 weeks; the 1-mg dose was given only to patients weighing <50 kg (<110 lbs).

In both trials, doses were increased in increments of 1 mg/week, and investigators evaluated participants’ ADHD signs and symptoms once a week using the clinician administered and scored ADHD Rating Scale-IV (ADHD-RS-IV). The primary outcome was change in total ADHD-RS-IV score from baseline to endpoint.

In both trials, patients taking GXR demonstrated statistically signifcant improvements in ADHD-RS-IV score starting 1 to 2 weeks after they began receiving once-daily GXR:

• In the first trial, the mean reduction in ADHD-RS-IV total score at endpoint was –16.7 for GXR compared with –8.9 for placebo (P < .0001).
• In the second, the reduction was –19.6 for GXR and –12.2 for placebo (P=.004).

Placebo-adjusted least squares mean changes from baseline were statistically significant for all GXR doses in the randomized treatment groups in both studies.

Secondary efficacy outcome measures included the Conners’ Parent Rating Scale-Revised: Short Form (CPRS-R) and the Conners’ Teacher Rating Scale-Revised: Short Form (CTRS-R).

Significant improvements were seen on both scales. On the CPRS-R, parents reported significant improvement across a full day (as measured at 6 PM, 8 PM, and 6 AM the next day). On the CTRS-R—which was used only in the first trial—teachers reported significant improvement throughout the school day (as measured at 10 AM and 2 PM).

Treating oppositional symptoms. In a collateral study,⁹ GXR was evaluated in complex ADHD patients age 6 to 12 who exhibited oppositional symptoms. The primary efficacy measure was change from baseline to endpoint in the oppositional subscale of the Conners’ Parent Rating Scale-Revised: Long Form (CPRS-R:L) score.

All subjects randomized to GXR started on a dose of 1 mg/d—which could be titrated by 1 mg/week during the 5-week, dose-optimization period to a maximum of 4 mg/d—and were maintained at their optimal doses for 3 additional weeks. Among the 217 subjects enrolled, 138 received GXR and 79, placebo.

Least-squares mean reductions from baseline to endpoint in CPRS-R:L oppositional subscale scores were –10.9 in the GXR group compared with –6.8 in the placebo group (P < .001; effect size 0.590). The GXR-treated group showed a significantly greater reduction in ADHD-RS-IV total score from baseline to endpoint compared with the placebo group (–23.8 vs –11.4, respectively, P < .001; effect size 0.916).

Table 2

Randomized, controlled trials supporting GXR’s effectiveness
for treating ADHD symptoms

Tolerability

In the phase III trials, the most commonly reported drug-related adverse reactions (occurring in ≥2% of patients) were:

• somnolence (38%)
• headache (24%)
• fatigue (14%)
• upper abdominal pain (10%)
• nausea, lethargy, dizziness, hypotension/decreased blood pressure, irritability (6% for each)
• decreased appetite (5%)
• dry mouth (4%)
• constipation (3%).

Many of these adverse reactions appear to be dose-related, particularly somnolence, sedation, abdominal pain, dizziness, and hypotension/decreased blood pressure.

Overall, GXR was well tolerated; clinicians rated most events as mild to moderate. Twelve percent of GXR patients discontinued the clinical studies because of adverse events, compared with 4% in the placebo groups. The most common adverse reactions leading to discontinuation were somnolence/sedation (6%) and fatigue (2%). Less common adverse reactions leading to discontinuation (occurring in 1% of patients) included hypotension/decreased blood pressure, headache, and dizziness.

Open-label safety trial. Sallee et al¹⁰ conducted a longer-term, open-label, flexible-dose safety continuation study of 259 GXR-treated patients (mean exposure 10 months), some of whom also received a psychostimulant. Common adverse reactions (occurring in ≥5% of subjects) included somnolence (45%), headache (26%), fatigue (16%), upper abdominal pain (11%), hypotension/decreased blood pressure (10%), vomiting (9%), dizziness (7%), nausea (7%), weight gain (7%), and irritability (6%).¹⁰ In a subset of patients, the onset of sedative events typically occurred within the first 3 weeks of GXR treatment and then declined with maintenance to a frequency of approximately 16%. The rates of somnolence, sedation, or fatigue were lowest among patients who also received a psychostimulant ( Figure ).

Distribution of GXR doses before the end of this study was 37% of patients on 4 mg, 33% on 3 mg, 27% on 2 mg, and 3% on 1 mg, suggesting a preference for maintenance doses of 3 to 4 mg/d. The most frequent adverse reactions leading to discontinuation were somnolence (3%), syncopal events (2%), increased weight (2%), depression (2%), and fatigue (2%). Other adverse reactions leading to discontinuation (occurring in approximately 1% of patients) included hypotension/decreased blood pressure, sedation, headache, and lethargy. Serious adverse reactions in the longer-term study in >1 patient included syncope (2%) and convulsion (0.4%).

Figure: Incidence of somnolence, sedation, and fatigue in study patients receiving GXR
with or without psychostimulants

In the longer-term, open-label safety study,¹⁰ maximum decreases in systolic and diastolic blood pressure occurred in the first month of treatment; decreases were less pronounced over time. Syncope occurred in 1% of pediatric subjects but was not dose-dependent. Guanfacine IR can increase QT interval but not in a dose-dependent fashion.

Dosing

The approved dose range for GXR is 1 to 4 mg once daily in the morning. Initiate treatment at 1 mg/d, and adjust the dose in increments of no more than 1 mg/week, evaluating the patient weekly. GXR maintenance therapy is frequently in the range of 2 to 4 mg/d.

Because adverse events such as hypotension, bradycardia, and sedation are dose-related, evaluate benefit and risk using mg/kg range approximation. GXR efficacy on a weight-adjusted (mg/kg) basis is consistent across a dosage range of 0.01 to 0.17 mg/kg/d. Clinically relevant improvements are usually observed beginning at doses of 0.05 to 0.08 mg/kg/d. In clinical trials, efficacy increased with increasing weight-adjusted dose (mg/kg), so if GXR is well-tolerated, doses up to 0.12 mg/kg once daily may provide additional benefit up to the maximum of 4 mg/d.

Instruct patients to swallow GXR whole because crushing, chewing, or otherwise breaking the tablet’s enteric coating will markedly enhance guanfacine release.

Abruptly discontinuing GXR is associated with infrequent, transient elevations in blood pressure above the patient’s baseline (ie, rebound). To minimize these effects, GXR should be gradually tapered in decrements of no more than 1 mg every 3 to 7 days. Isolated missed doses of GXR generally are not a problem, but ≥2 consecutive missed doses may warrant reinitiation of the titration schedule.

Related resource

• Guanfacine extended release (Intuniv) prescribing information. www.intuniv.com/documents/INTUNIV_Full_Prescribing_Information.pdf.

Drug brand names

• Atomoxetine • Strattera
• Guanfacine extended release • Intuniv
• Guanfacine immediate release • Tenex
• Ketoconazole • Nizoral
• Rifampin • Rifadin, Rimactane
• Valproic acid • Depakene, Depakote

Disclosure

Dr. Sallee receives grant/research support from the National Institutes of Health. He is a consultant to Otsuka, Nextwave, and Sepracor and a consultant to and speaker for Shire. Dr. Sallee is a consultant to, shareholder of, and member of the board of directors of P2D Inc. and a principal in Satiety Solutions.

Guanfacine extended release (GXR)—a selective α-2 adrenergic agonist FDA-approved for the treatment of attention-deficit/hyperactivity disorder (ADHD)—has demonstrated efficacy for inattentive and hyperactive/impulsive symptom domains in 2 large trials lasting 8 and 9 weeks.^1,2 GXR’s once-daily formulation may increase adherence and deliver consistent control of symptoms across a full day ( Table 1 ).

Table 1

Guanfacine extended release: Fast facts

Clinical implications

GXR exhibits enhancement of noradrenergic pathways through selective direct receptor action in the prefrontal cortex.³ This mechanism of action is different from that of other FDA-approved ADHD medications. GXR can be used alone or in combination with stimulants or atomoxetine for treating complex ADHD, such as cases accompanied by oppositional features and emotional dysregulation or characterized by partial stimulant response.

How it works

Guanfacine—originally developed as an immediate-release (IR) antihypertensive—reduces sympathetic tone, causing centrally mediated vasodilation and reduced heart rate. Although GXR’s mechanism of action in ADHD is not known, the drug is a selective α-2A receptor agonist thought to directly engage postsynaptic receptors in the prefrontal cortex (PFC), an area of the brain believed to play a major role in attentional and organizational functions that preclinical research has linked to ADHD.³

The postsynaptic α-2A receptor is thought to play a central role in the optimal functioning of the PFC as illustrated by the “inverted U hypothesis of PFC activation.”⁴ In this model, cyclic adenosine monophosphate (cAMP) levels build within the prefrontal cortical neurons and cause specific ion channels—hyperpolarization-activated cyclic nucleotide gated (HCN) channels—to open on dendritic spines of these neurons.⁵ Activation of HCN channels effectively reduces membrane resistance, cutting off synaptic inputs and disconnecting PFC network connections. Because α-2A receptors are located in proximity to HCN channels, their stimulation by GXR closes HCN channels, inhibits further production of cAMP, and reestablishes synaptic function and the resulting network connectivity.⁵ Blockade of α-2A receptors by yohimbine reverses this process, eroding network connectivity, and in monkeys has been demonstrated to impair working memory,⁶ damage inhibition/impulse control, and produce locomotor hyperactivity.

Direct stimulation by GXR of the postsynaptic α-2A receptors is thought to:

• strengthen working memory
• reduce susceptibility to distraction
• improve attention regulation
• improve behavioral inhibition
• enhance impulse control.⁷

Pharmacokinetics

GXR offers enhanced pharmaceutics relative to IR guanfacine. IR guanfacine exhibits poor absorption characteristics—peak plasma concentration is achieved too rapidly and then declines precipitously, with considerable inter-individual variation.

GXR’s once-daily formulation is implemented by a proprietary enteric-coated sustained release mechanism⁸ that is meant to:

• control absorption
• provide a broad but flat plasma concentration profile
• reduce inter-individual variation of guanfacine exposure.

Compared with IR guanfacine, GXR exhibits delayed time of maximum concentration (Tmax) and reduced maximum concentration (Cmax). Therapeutic concentrations can be sustained over longer periods with reduced peak-to-trough fluctuation,⁸ which tends to improve tolerability and symptom control throughout the day. The convenience of once-daily dosing also may increase adherence.

GXR’s pharmacokinetic characteristics do not change with dose, but high-fat meals will increase absorption of the drug—Cmax increases by 75% and area under the plasma concentration time curve increases by 40%. Because GXR primarily is metabolized through cytochrome P450 (CYP) 3A4, CYP3A4 inhibitors such as ketoconazole will increase guanfacine plasma concentrations and elevate the risk of adverse events such as bradycardia, hypotension, and sedation. Conversely, CYP3A4 inducers such as rifampin will significantly reduce total guanfacine exposure. Coadministration of valproic acid with GXR can result in increased valproic acid levels, producing additive CNS side effects.

Efficacy

GXR reduced both inattentive and hyperactive/impulsive symptoms in 2 phase III, forced-dose, parallel-design, randomized, placebo-controlled trials ( Table 2 ). In the first trial,¹ 345 children age 6 to 17 received placebo or GXR, 2 mg, 3 mg, or 4 mg once daily for 8 weeks. In the second study,² 324 children age 6 to 17 received placebo or GXR, 1 mg, 2 mg, 3 mg, or 4 mg, once daily for 9 weeks; the 1-mg dose was given only to patients weighing <50 kg (<110 lbs).

In both trials, doses were increased in increments of 1 mg/week, and investigators evaluated participants’ ADHD signs and symptoms once a week using the clinician administered and scored ADHD Rating Scale-IV (ADHD-RS-IV). The primary outcome was change in total ADHD-RS-IV score from baseline to endpoint.

In both trials, patients taking GXR demonstrated statistically signifcant improvements in ADHD-RS-IV score starting 1 to 2 weeks after they began receiving once-daily GXR:

• In the first trial, the mean reduction in ADHD-RS-IV total score at endpoint was –16.7 for GXR compared with –8.9 for placebo (P < .0001).
• In the second, the reduction was –19.6 for GXR and –12.2 for placebo (P=.004).

Placebo-adjusted least squares mean changes from baseline were statistically significant for all GXR doses in the randomized treatment groups in both studies.

Secondary efficacy outcome measures included the Conners’ Parent Rating Scale-Revised: Short Form (CPRS-R) and the Conners’ Teacher Rating Scale-Revised: Short Form (CTRS-R).

Significant improvements were seen on both scales. On the CPRS-R, parents reported significant improvement across a full day (as measured at 6 PM, 8 PM, and 6 AM the next day). On the CTRS-R—which was used only in the first trial—teachers reported significant improvement throughout the school day (as measured at 10 AM and 2 PM).

Treating oppositional symptoms. In a collateral study,⁹ GXR was evaluated in complex ADHD patients age 6 to 12 who exhibited oppositional symptoms. The primary efficacy measure was change from baseline to endpoint in the oppositional subscale of the Conners’ Parent Rating Scale-Revised: Long Form (CPRS-R:L) score.

All subjects randomized to GXR started on a dose of 1 mg/d—which could be titrated by 1 mg/week during the 5-week, dose-optimization period to a maximum of 4 mg/d—and were maintained at their optimal doses for 3 additional weeks. Among the 217 subjects enrolled, 138 received GXR and 79, placebo.

Least-squares mean reductions from baseline to endpoint in CPRS-R:L oppositional subscale scores were –10.9 in the GXR group compared with –6.8 in the placebo group (P < .001; effect size 0.590). The GXR-treated group showed a significantly greater reduction in ADHD-RS-IV total score from baseline to endpoint compared with the placebo group (–23.8 vs –11.4, respectively, P < .001; effect size 0.916).

Table 2

Randomized, controlled trials supporting GXR’s effectiveness
for treating ADHD symptoms

Tolerability

In the phase III trials, the most commonly reported drug-related adverse reactions (occurring in ≥2% of patients) were:

• somnolence (38%)
• headache (24%)
• fatigue (14%)
• upper abdominal pain (10%)
• nausea, lethargy, dizziness, hypotension/decreased blood pressure, irritability (6% for each)
• decreased appetite (5%)
• dry mouth (4%)
• constipation (3%).

Many of these adverse reactions appear to be dose-related, particularly somnolence, sedation, abdominal pain, dizziness, and hypotension/decreased blood pressure.

Overall, GXR was well tolerated; clinicians rated most events as mild to moderate. Twelve percent of GXR patients discontinued the clinical studies because of adverse events, compared with 4% in the placebo groups. The most common adverse reactions leading to discontinuation were somnolence/sedation (6%) and fatigue (2%). Less common adverse reactions leading to discontinuation (occurring in 1% of patients) included hypotension/decreased blood pressure, headache, and dizziness.

Open-label safety trial. Sallee et al¹⁰ conducted a longer-term, open-label, flexible-dose safety continuation study of 259 GXR-treated patients (mean exposure 10 months), some of whom also received a psychostimulant. Common adverse reactions (occurring in ≥5% of subjects) included somnolence (45%), headache (26%), fatigue (16%), upper abdominal pain (11%), hypotension/decreased blood pressure (10%), vomiting (9%), dizziness (7%), nausea (7%), weight gain (7%), and irritability (6%).¹⁰ In a subset of patients, the onset of sedative events typically occurred within the first 3 weeks of GXR treatment and then declined with maintenance to a frequency of approximately 16%. The rates of somnolence, sedation, or fatigue were lowest among patients who also received a psychostimulant ( Figure ).

Distribution of GXR doses before the end of this study was 37% of patients on 4 mg, 33% on 3 mg, 27% on 2 mg, and 3% on 1 mg, suggesting a preference for maintenance doses of 3 to 4 mg/d. The most frequent adverse reactions leading to discontinuation were somnolence (3%), syncopal events (2%), increased weight (2%), depression (2%), and fatigue (2%). Other adverse reactions leading to discontinuation (occurring in approximately 1% of patients) included hypotension/decreased blood pressure, sedation, headache, and lethargy. Serious adverse reactions in the longer-term study in >1 patient included syncope (2%) and convulsion (0.4%).

Figure: Incidence of somnolence, sedation, and fatigue in study patients receiving GXR
with or without psychostimulants

In the longer-term, open-label safety study,¹⁰ maximum decreases in systolic and diastolic blood pressure occurred in the first month of treatment; decreases were less pronounced over time. Syncope occurred in 1% of pediatric subjects but was not dose-dependent. Guanfacine IR can increase QT interval but not in a dose-dependent fashion.

Dosing

The approved dose range for GXR is 1 to 4 mg once daily in the morning. Initiate treatment at 1 mg/d, and adjust the dose in increments of no more than 1 mg/week, evaluating the patient weekly. GXR maintenance therapy is frequently in the range of 2 to 4 mg/d.

Because adverse events such as hypotension, bradycardia, and sedation are dose-related, evaluate benefit and risk using mg/kg range approximation. GXR efficacy on a weight-adjusted (mg/kg) basis is consistent across a dosage range of 0.01 to 0.17 mg/kg/d. Clinically relevant improvements are usually observed beginning at doses of 0.05 to 0.08 mg/kg/d. In clinical trials, efficacy increased with increasing weight-adjusted dose (mg/kg), so if GXR is well-tolerated, doses up to 0.12 mg/kg once daily may provide additional benefit up to the maximum of 4 mg/d.

Instruct patients to swallow GXR whole because crushing, chewing, or otherwise breaking the tablet’s enteric coating will markedly enhance guanfacine release.

Abruptly discontinuing GXR is associated with infrequent, transient elevations in blood pressure above the patient’s baseline (ie, rebound). To minimize these effects, GXR should be gradually tapered in decrements of no more than 1 mg every 3 to 7 days. Isolated missed doses of GXR generally are not a problem, but ≥2 consecutive missed doses may warrant reinitiation of the titration schedule.

Related resource

• Guanfacine extended release (Intuniv) prescribing information. www.intuniv.com/documents/INTUNIV_Full_Prescribing_Information.pdf.

Drug brand names

• Atomoxetine • Strattera
• Guanfacine extended release • Intuniv
• Guanfacine immediate release • Tenex
• Ketoconazole • Nizoral
• Rifampin • Rifadin, Rimactane
• Valproic acid • Depakene, Depakote

Disclosure

Dr. Sallee receives grant/research support from the National Institutes of Health. He is a consultant to Otsuka, Nextwave, and Sepracor and a consultant to and speaker for Shire. Dr. Sallee is a consultant to, shareholder of, and member of the board of directors of P2D Inc. and a principal in Satiety Solutions.

Guanfacine extended release (GXR)—a selective α-2 adrenergic agonist FDA-approved for the treatment of attention-deficit/hyperactivity disorder (ADHD)—has demonstrated efficacy for inattentive and hyperactive/impulsive symptom domains in 2 large trials lasting 8 and 9 weeks.^1,2 GXR’s once-daily formulation may increase adherence and deliver consistent control of symptoms across a full day ( Table 1 ).

Table 1

Guanfacine extended release: Fast facts

Clinical implications

GXR exhibits enhancement of noradrenergic pathways through selective direct receptor action in the prefrontal cortex.³ This mechanism of action is different from that of other FDA-approved ADHD medications. GXR can be used alone or in combination with stimulants or atomoxetine for treating complex ADHD, such as cases accompanied by oppositional features and emotional dysregulation or characterized by partial stimulant response.

How it works

Guanfacine—originally developed as an immediate-release (IR) antihypertensive—reduces sympathetic tone, causing centrally mediated vasodilation and reduced heart rate. Although GXR’s mechanism of action in ADHD is not known, the drug is a selective α-2A receptor agonist thought to directly engage postsynaptic receptors in the prefrontal cortex (PFC), an area of the brain believed to play a major role in attentional and organizational functions that preclinical research has linked to ADHD.³

The postsynaptic α-2A receptor is thought to play a central role in the optimal functioning of the PFC as illustrated by the “inverted U hypothesis of PFC activation.”⁴ In this model, cyclic adenosine monophosphate (cAMP) levels build within the prefrontal cortical neurons and cause specific ion channels—hyperpolarization-activated cyclic nucleotide gated (HCN) channels—to open on dendritic spines of these neurons.⁵ Activation of HCN channels effectively reduces membrane resistance, cutting off synaptic inputs and disconnecting PFC network connections. Because α-2A receptors are located in proximity to HCN channels, their stimulation by GXR closes HCN channels, inhibits further production of cAMP, and reestablishes synaptic function and the resulting network connectivity.⁵ Blockade of α-2A receptors by yohimbine reverses this process, eroding network connectivity, and in monkeys has been demonstrated to impair working memory,⁶ damage inhibition/impulse control, and produce locomotor hyperactivity.

Direct stimulation by GXR of the postsynaptic α-2A receptors is thought to:

• strengthen working memory
• reduce susceptibility to distraction
• improve attention regulation
• improve behavioral inhibition
• enhance impulse control.⁷

Pharmacokinetics

GXR offers enhanced pharmaceutics relative to IR guanfacine. IR guanfacine exhibits poor absorption characteristics—peak plasma concentration is achieved too rapidly and then declines precipitously, with considerable inter-individual variation.

GXR’s once-daily formulation is implemented by a proprietary enteric-coated sustained release mechanism⁸ that is meant to:

• control absorption
• provide a broad but flat plasma concentration profile
• reduce inter-individual variation of guanfacine exposure.

Compared with IR guanfacine, GXR exhibits delayed time of maximum concentration (Tmax) and reduced maximum concentration (Cmax). Therapeutic concentrations can be sustained over longer periods with reduced peak-to-trough fluctuation,⁸ which tends to improve tolerability and symptom control throughout the day. The convenience of once-daily dosing also may increase adherence.

GXR’s pharmacokinetic characteristics do not change with dose, but high-fat meals will increase absorption of the drug—Cmax increases by 75% and area under the plasma concentration time curve increases by 40%. Because GXR primarily is metabolized through cytochrome P450 (CYP) 3A4, CYP3A4 inhibitors such as ketoconazole will increase guanfacine plasma concentrations and elevate the risk of adverse events such as bradycardia, hypotension, and sedation. Conversely, CYP3A4 inducers such as rifampin will significantly reduce total guanfacine exposure. Coadministration of valproic acid with GXR can result in increased valproic acid levels, producing additive CNS side effects.

Efficacy

GXR reduced both inattentive and hyperactive/impulsive symptoms in 2 phase III, forced-dose, parallel-design, randomized, placebo-controlled trials ( Table 2 ). In the first trial,¹ 345 children age 6 to 17 received placebo or GXR, 2 mg, 3 mg, or 4 mg once daily for 8 weeks. In the second study,² 324 children age 6 to 17 received placebo or GXR, 1 mg, 2 mg, 3 mg, or 4 mg, once daily for 9 weeks; the 1-mg dose was given only to patients weighing <50 kg (<110 lbs).

In both trials, doses were increased in increments of 1 mg/week, and investigators evaluated participants’ ADHD signs and symptoms once a week using the clinician administered and scored ADHD Rating Scale-IV (ADHD-RS-IV). The primary outcome was change in total ADHD-RS-IV score from baseline to endpoint.

In both trials, patients taking GXR demonstrated statistically signifcant improvements in ADHD-RS-IV score starting 1 to 2 weeks after they began receiving once-daily GXR:

• In the first trial, the mean reduction in ADHD-RS-IV total score at endpoint was –16.7 for GXR compared with –8.9 for placebo (P < .0001).
• In the second, the reduction was –19.6 for GXR and –12.2 for placebo (P=.004).

Placebo-adjusted least squares mean changes from baseline were statistically significant for all GXR doses in the randomized treatment groups in both studies.

Secondary efficacy outcome measures included the Conners’ Parent Rating Scale-Revised: Short Form (CPRS-R) and the Conners’ Teacher Rating Scale-Revised: Short Form (CTRS-R).

Significant improvements were seen on both scales. On the CPRS-R, parents reported significant improvement across a full day (as measured at 6 PM, 8 PM, and 6 AM the next day). On the CTRS-R—which was used only in the first trial—teachers reported significant improvement throughout the school day (as measured at 10 AM and 2 PM).

Treating oppositional symptoms. In a collateral study,⁹ GXR was evaluated in complex ADHD patients age 6 to 12 who exhibited oppositional symptoms. The primary efficacy measure was change from baseline to endpoint in the oppositional subscale of the Conners’ Parent Rating Scale-Revised: Long Form (CPRS-R:L) score.

All subjects randomized to GXR started on a dose of 1 mg/d—which could be titrated by 1 mg/week during the 5-week, dose-optimization period to a maximum of 4 mg/d—and were maintained at their optimal doses for 3 additional weeks. Among the 217 subjects enrolled, 138 received GXR and 79, placebo.

Least-squares mean reductions from baseline to endpoint in CPRS-R:L oppositional subscale scores were –10.9 in the GXR group compared with –6.8 in the placebo group (P < .001; effect size 0.590). The GXR-treated group showed a significantly greater reduction in ADHD-RS-IV total score from baseline to endpoint compared with the placebo group (–23.8 vs –11.4, respectively, P < .001; effect size 0.916).

Table 2

Randomized, controlled trials supporting GXR’s effectiveness
for treating ADHD symptoms

Tolerability

In the phase III trials, the most commonly reported drug-related adverse reactions (occurring in ≥2% of patients) were:

• somnolence (38%)
• headache (24%)
• fatigue (14%)
• upper abdominal pain (10%)
• nausea, lethargy, dizziness, hypotension/decreased blood pressure, irritability (6% for each)
• decreased appetite (5%)
• dry mouth (4%)
• constipation (3%).

Many of these adverse reactions appear to be dose-related, particularly somnolence, sedation, abdominal pain, dizziness, and hypotension/decreased blood pressure.

Overall, GXR was well tolerated; clinicians rated most events as mild to moderate. Twelve percent of GXR patients discontinued the clinical studies because of adverse events, compared with 4% in the placebo groups. The most common adverse reactions leading to discontinuation were somnolence/sedation (6%) and fatigue (2%). Less common adverse reactions leading to discontinuation (occurring in 1% of patients) included hypotension/decreased blood pressure, headache, and dizziness.

Open-label safety trial. Sallee et al¹⁰ conducted a longer-term, open-label, flexible-dose safety continuation study of 259 GXR-treated patients (mean exposure 10 months), some of whom also received a psychostimulant. Common adverse reactions (occurring in ≥5% of subjects) included somnolence (45%), headache (26%), fatigue (16%), upper abdominal pain (11%), hypotension/decreased blood pressure (10%), vomiting (9%), dizziness (7%), nausea (7%), weight gain (7%), and irritability (6%).¹⁰ In a subset of patients, the onset of sedative events typically occurred within the first 3 weeks of GXR treatment and then declined with maintenance to a frequency of approximately 16%. The rates of somnolence, sedation, or fatigue were lowest among patients who also received a psychostimulant ( Figure ).

Distribution of GXR doses before the end of this study was 37% of patients on 4 mg, 33% on 3 mg, 27% on 2 mg, and 3% on 1 mg, suggesting a preference for maintenance doses of 3 to 4 mg/d. The most frequent adverse reactions leading to discontinuation were somnolence (3%), syncopal events (2%), increased weight (2%), depression (2%), and fatigue (2%). Other adverse reactions leading to discontinuation (occurring in approximately 1% of patients) included hypotension/decreased blood pressure, sedation, headache, and lethargy. Serious adverse reactions in the longer-term study in >1 patient included syncope (2%) and convulsion (0.4%).

Figure: Incidence of somnolence, sedation, and fatigue in study patients receiving GXR
with or without psychostimulants

In the longer-term, open-label safety study,¹⁰ maximum decreases in systolic and diastolic blood pressure occurred in the first month of treatment; decreases were less pronounced over time. Syncope occurred in 1% of pediatric subjects but was not dose-dependent. Guanfacine IR can increase QT interval but not in a dose-dependent fashion.

Dosing

The approved dose range for GXR is 1 to 4 mg once daily in the morning. Initiate treatment at 1 mg/d, and adjust the dose in increments of no more than 1 mg/week, evaluating the patient weekly. GXR maintenance therapy is frequently in the range of 2 to 4 mg/d.

Because adverse events such as hypotension, bradycardia, and sedation are dose-related, evaluate benefit and risk using mg/kg range approximation. GXR efficacy on a weight-adjusted (mg/kg) basis is consistent across a dosage range of 0.01 to 0.17 mg/kg/d. Clinically relevant improvements are usually observed beginning at doses of 0.05 to 0.08 mg/kg/d. In clinical trials, efficacy increased with increasing weight-adjusted dose (mg/kg), so if GXR is well-tolerated, doses up to 0.12 mg/kg once daily may provide additional benefit up to the maximum of 4 mg/d.

Instruct patients to swallow GXR whole because crushing, chewing, or otherwise breaking the tablet’s enteric coating will markedly enhance guanfacine release.

Abruptly discontinuing GXR is associated with infrequent, transient elevations in blood pressure above the patient’s baseline (ie, rebound). To minimize these effects, GXR should be gradually tapered in decrements of no more than 1 mg every 3 to 7 days. Isolated missed doses of GXR generally are not a problem, but ≥2 consecutive missed doses may warrant reinitiation of the titration schedule.

Related resource

• Guanfacine extended release (Intuniv) prescribing information. www.intuniv.com/documents/INTUNIV_Full_Prescribing_Information.pdf.

Drug brand names

• Atomoxetine • Strattera
• Guanfacine extended release • Intuniv
• Guanfacine immediate release • Tenex
• Ketoconazole • Nizoral
• Rifampin • Rifadin, Rimactane
• Valproic acid • Depakene, Depakote

Disclosure

Dr. Sallee receives grant/research support from the National Institutes of Health. He is a consultant to Otsuka, Nextwave, and Sepracor and a consultant to and speaker for Shire. Dr. Sallee is a consultant to, shareholder of, and member of the board of directors of P2D Inc. and a principal in Satiety Solutions.

Nursemaid’s elbow—sudden subluxation of the radial head—usually results from forcible traction to a child’s pronated hand or wrist, with the elbow extended.¹ Ironically, this can occur when a parent tries to maneuver a child away from perceived danger, and the child experiences pain and acute loss of function of the affected limb.²

Nursemaid’s elbow reportedly occurs frequently among children,^3,4 and thus we would expect to encounter it often in primary care. However, this condition receives little attention in medical training or in the literature, and many physicians do not recognize it.^4-6

In this article we describe the epidemiology, underlying pathology, diagnosis, and treatment of nursemaid’s elbow, based on a systematic review of the current literature.

Methods

Literature search
Using PubMed and Embase, we conducted a literature search for articles published in Dutch, English, German, or French from 1966 through July 2007 on the topic of nursemaid’s elbow in children. We used as search terms all known synonyms for nursemaid’s elbow—eg, radial head subluxation, partial epiphyseal separation of the radial head, pulled elbow, babysitter’s elbow, curbstone fracture, etc. Publications cited in our initial search were also checked for relevance. Articles were reviewed and judged independently by 2 authors (M.K. and J.C.v.d.W.).

Articles we selected focused on proximal radial subluxation. We excluded articles on distal radial subluxation and luxation of the radius.

The 2 reviewers assessed the quality of articles on treatment using the validated Jadad score,⁷ wherein a maximum of 5 points may be awarded:

• 1 point if the study is described as randomized:
• 1 point if the study is described as double-blind:
• 1 point for a description of withdrawals or dropouts.

No cutoff limit for Jadad scores was planned as a criterion for exclusion. As it is not possible to treat nursemaid’s elbow in a double-blind fashion, 3 was the highest possible score in our study.

Results

Our literature search produced 368 potentially relevant papers; of these, 60 met our inclusion criteria. The reference lists of selected studies and reviews yielded an additional 25 acceptable papers, each covering various aspects of the topic (epidemiology, 19; pathology, 10; diagnosis, 10; treatment, 9;). Thirty-seven of the 85 selected papers were review articles.

Epidemiology
Most reports agree that nursemaid’s elbow is a frequent injury among children.^4,8-10 Unfortunately, published population-based incidence rates are scarce; only 1 article gives an occurrence rate in the total population—1.2%.¹¹ Most epidemiologic data are derived from case series, which show a predominance of injury among girls and to the left arm. Most cases occur at a median age of about 2 years.^2-5,8

Pathology
The many synonyms of nursemaid’s elbow reflect a once obscure understanding of its pathology. Among initial reports from the 1800s, the focus was on determining whether the injury occurred at the wrist or the elbow.¹² Subsequent studies showed that the mechanism of injury usually is a tug on the pronated arm^5,13-16 of young children (who have relatively lax tissue), thereby pulling the radius through the annular ligament,^13-15 which may partially tear and (with the meniscoid synovial fold) become entrapped between the radial head and the capitellum.¹⁵ Most commonly a parent or other caregiver is holding the child by the hand while walking and suddenly pulls the child away from a dangerous situation or merely drags the child up a curb or a step.¹

Diagnosis
We found no clinical studies that assessed the value of physical examination or history taking. The only studies relevant to diagnosis discussed radiography.

Nursemaid’s elbow is an easily recognized diagnosis based on the history and physical examination.¹⁷ Still, it seems many physicians do not recognize the condition.^4-6 Typically, a parent reports that the child cried out after a pull on the arm and then refused to use the arm, holding it slightly flexed and pronated.¹⁸ Pain may be felt only at the wrist or shoulder.^3,18 Occasionally, a snap or click is heard when the accident happens.⁵ The elbow can usually be flexed and extended, but the child resists supination of the forearm, which causes pain in the elbow. There is no swelling or bruising.¹⁹

Children are often referred for radiographic examination with the observation, “refuses to use arm; please x-ray from shoulder to wrist.”²⁰ Radiography is of little help, however, and exposes the child to a dose of ionizing radiation. Although some studies show small significant differences between nursemaid’s elbow and the normal elbow,^21-23 radiographic results generally are reported as normal.^4,6,8,24 (Some commentators assume this may occur if the radiology technician repositions the arm in an attempt to obtain a true anteroposterior projection of the elbow.^1,18,25) Restrict radiography, therefore, to cases with an unclear history or a history of trauma other than arm pull, to exclude more severe injuries.

The role of sonography is not yet clear, but it may turn out to be a fast and harmless technique for diagnosing uncertain cases.^20,25,26

Treatment
Although no articles have described the natural course of nursemaid’s elbow, most authors report that it resolves on its own when a child moves the arm in supination or pronation. It is so easily treated that parents of children with recurrent episodes have even been instructed by phone how to perform the reduction.²⁷

Most articles and textbooks recommend reducing nursemaid’s elbow by a rapid supination of the forearm, followed by flexion or extension.^9,28 However, some articles have described a pronation method.²⁹ We found 2 high-quality trials that compared the success rate of the supination method with the pronation method.^30,31 Researchers conducting 1 medium-quality trial assessed the difference in pain experienced with these 2 methods.³² And researchers conducting 1 low-quality trial tried to assess whether splinting after manipulation helps to prevent recurrences of nursemaid’s elbow.¹⁰

These trials indicate the pronation method is more successful. In addition, some studies report that the pronation method is less painful for the child and less frightening for a parent to watch.^29-31 Green et al confirmed this in their randomized trial.³²

Most compelling finding

The highest quality studies were those devoted to treatment,^28,30-33 and the clear conclusion from their findings—in contrast to what textbooks recommend—is that reduction with a pronation maneuver is more often successful than the supination method. Of course more studies will be needed before textbooks change their recommendations. But at least these studies provide helpful guidance now.

CORRESPONDENCE
Marjolein Krul, MD, Department of General Practice, Room Ff304, Erasmus MC-University Medical Center Rotterdam, PO Box 1738, 3000 DR Rotterdam, The Netherlands; [email protected]

Nursemaid’s elbow—sudden subluxation of the radial head—usually results from forcible traction to a child’s pronated hand or wrist, with the elbow extended.¹ Ironically, this can occur when a parent tries to maneuver a child away from perceived danger, and the child experiences pain and acute loss of function of the affected limb.²

Nursemaid’s elbow reportedly occurs frequently among children,^3,4 and thus we would expect to encounter it often in primary care. However, this condition receives little attention in medical training or in the literature, and many physicians do not recognize it.^4-6

In this article we describe the epidemiology, underlying pathology, diagnosis, and treatment of nursemaid’s elbow, based on a systematic review of the current literature.

Methods

Literature search
Using PubMed and Embase, we conducted a literature search for articles published in Dutch, English, German, or French from 1966 through July 2007 on the topic of nursemaid’s elbow in children. We used as search terms all known synonyms for nursemaid’s elbow—eg, radial head subluxation, partial epiphyseal separation of the radial head, pulled elbow, babysitter’s elbow, curbstone fracture, etc. Publications cited in our initial search were also checked for relevance. Articles were reviewed and judged independently by 2 authors (M.K. and J.C.v.d.W.).

Articles we selected focused on proximal radial subluxation. We excluded articles on distal radial subluxation and luxation of the radius.

The 2 reviewers assessed the quality of articles on treatment using the validated Jadad score,⁷ wherein a maximum of 5 points may be awarded:

• 1 point if the study is described as randomized:
• 1 point if the study is described as double-blind:
• 1 point for a description of withdrawals or dropouts.

No cutoff limit for Jadad scores was planned as a criterion for exclusion. As it is not possible to treat nursemaid’s elbow in a double-blind fashion, 3 was the highest possible score in our study.

Results

Our literature search produced 368 potentially relevant papers; of these, 60 met our inclusion criteria. The reference lists of selected studies and reviews yielded an additional 25 acceptable papers, each covering various aspects of the topic (epidemiology, 19; pathology, 10; diagnosis, 10; treatment, 9;). Thirty-seven of the 85 selected papers were review articles.

Epidemiology
Most reports agree that nursemaid’s elbow is a frequent injury among children.^4,8-10 Unfortunately, published population-based incidence rates are scarce; only 1 article gives an occurrence rate in the total population—1.2%.¹¹ Most epidemiologic data are derived from case series, which show a predominance of injury among girls and to the left arm. Most cases occur at a median age of about 2 years.^2-5,8

Pathology
The many synonyms of nursemaid’s elbow reflect a once obscure understanding of its pathology. Among initial reports from the 1800s, the focus was on determining whether the injury occurred at the wrist or the elbow.¹² Subsequent studies showed that the mechanism of injury usually is a tug on the pronated arm^5,13-16 of young children (who have relatively lax tissue), thereby pulling the radius through the annular ligament,^13-15 which may partially tear and (with the meniscoid synovial fold) become entrapped between the radial head and the capitellum.¹⁵ Most commonly a parent or other caregiver is holding the child by the hand while walking and suddenly pulls the child away from a dangerous situation or merely drags the child up a curb or a step.¹

Diagnosis
We found no clinical studies that assessed the value of physical examination or history taking. The only studies relevant to diagnosis discussed radiography.

Nursemaid’s elbow is an easily recognized diagnosis based on the history and physical examination.¹⁷ Still, it seems many physicians do not recognize the condition.^4-6 Typically, a parent reports that the child cried out after a pull on the arm and then refused to use the arm, holding it slightly flexed and pronated.¹⁸ Pain may be felt only at the wrist or shoulder.^3,18 Occasionally, a snap or click is heard when the accident happens.⁵ The elbow can usually be flexed and extended, but the child resists supination of the forearm, which causes pain in the elbow. There is no swelling or bruising.¹⁹

Children are often referred for radiographic examination with the observation, “refuses to use arm; please x-ray from shoulder to wrist.”²⁰ Radiography is of little help, however, and exposes the child to a dose of ionizing radiation. Although some studies show small significant differences between nursemaid’s elbow and the normal elbow,^21-23 radiographic results generally are reported as normal.^4,6,8,24 (Some commentators assume this may occur if the radiology technician repositions the arm in an attempt to obtain a true anteroposterior projection of the elbow.^1,18,25) Restrict radiography, therefore, to cases with an unclear history or a history of trauma other than arm pull, to exclude more severe injuries.

The role of sonography is not yet clear, but it may turn out to be a fast and harmless technique for diagnosing uncertain cases.^20,25,26

Treatment
Although no articles have described the natural course of nursemaid’s elbow, most authors report that it resolves on its own when a child moves the arm in supination or pronation. It is so easily treated that parents of children with recurrent episodes have even been instructed by phone how to perform the reduction.²⁷

Most articles and textbooks recommend reducing nursemaid’s elbow by a rapid supination of the forearm, followed by flexion or extension.^9,28 However, some articles have described a pronation method.²⁹ We found 2 high-quality trials that compared the success rate of the supination method with the pronation method.^30,31 Researchers conducting 1 medium-quality trial assessed the difference in pain experienced with these 2 methods.³² And researchers conducting 1 low-quality trial tried to assess whether splinting after manipulation helps to prevent recurrences of nursemaid’s elbow.¹⁰

These trials indicate the pronation method is more successful. In addition, some studies report that the pronation method is less painful for the child and less frightening for a parent to watch.^29-31 Green et al confirmed this in their randomized trial.³²

Most compelling finding

The highest quality studies were those devoted to treatment,^28,30-33 and the clear conclusion from their findings—in contrast to what textbooks recommend—is that reduction with a pronation maneuver is more often successful than the supination method. Of course more studies will be needed before textbooks change their recommendations. But at least these studies provide helpful guidance now.

CORRESPONDENCE
Marjolein Krul, MD, Department of General Practice, Room Ff304, Erasmus MC-University Medical Center Rotterdam, PO Box 1738, 3000 DR Rotterdam, The Netherlands; [email protected]

Nursemaid’s elbow—sudden subluxation of the radial head—usually results from forcible traction to a child’s pronated hand or wrist, with the elbow extended.¹ Ironically, this can occur when a parent tries to maneuver a child away from perceived danger, and the child experiences pain and acute loss of function of the affected limb.²

Nursemaid’s elbow reportedly occurs frequently among children,^3,4 and thus we would expect to encounter it often in primary care. However, this condition receives little attention in medical training or in the literature, and many physicians do not recognize it.^4-6

In this article we describe the epidemiology, underlying pathology, diagnosis, and treatment of nursemaid’s elbow, based on a systematic review of the current literature.

Methods

Literature search
Using PubMed and Embase, we conducted a literature search for articles published in Dutch, English, German, or French from 1966 through July 2007 on the topic of nursemaid’s elbow in children. We used as search terms all known synonyms for nursemaid’s elbow—eg, radial head subluxation, partial epiphyseal separation of the radial head, pulled elbow, babysitter’s elbow, curbstone fracture, etc. Publications cited in our initial search were also checked for relevance. Articles were reviewed and judged independently by 2 authors (M.K. and J.C.v.d.W.).

Articles we selected focused on proximal radial subluxation. We excluded articles on distal radial subluxation and luxation of the radius.

The 2 reviewers assessed the quality of articles on treatment using the validated Jadad score,⁷ wherein a maximum of 5 points may be awarded:

• 1 point if the study is described as randomized:
• 1 point if the study is described as double-blind:
• 1 point for a description of withdrawals or dropouts.

No cutoff limit for Jadad scores was planned as a criterion for exclusion. As it is not possible to treat nursemaid’s elbow in a double-blind fashion, 3 was the highest possible score in our study.

Results

Our literature search produced 368 potentially relevant papers; of these, 60 met our inclusion criteria. The reference lists of selected studies and reviews yielded an additional 25 acceptable papers, each covering various aspects of the topic (epidemiology, 19; pathology, 10; diagnosis, 10; treatment, 9;). Thirty-seven of the 85 selected papers were review articles.

Epidemiology
Most reports agree that nursemaid’s elbow is a frequent injury among children.^4,8-10 Unfortunately, published population-based incidence rates are scarce; only 1 article gives an occurrence rate in the total population—1.2%.¹¹ Most epidemiologic data are derived from case series, which show a predominance of injury among girls and to the left arm. Most cases occur at a median age of about 2 years.^2-5,8

Pathology
The many synonyms of nursemaid’s elbow reflect a once obscure understanding of its pathology. Among initial reports from the 1800s, the focus was on determining whether the injury occurred at the wrist or the elbow.¹² Subsequent studies showed that the mechanism of injury usually is a tug on the pronated arm^5,13-16 of young children (who have relatively lax tissue), thereby pulling the radius through the annular ligament,^13-15 which may partially tear and (with the meniscoid synovial fold) become entrapped between the radial head and the capitellum.¹⁵ Most commonly a parent or other caregiver is holding the child by the hand while walking and suddenly pulls the child away from a dangerous situation or merely drags the child up a curb or a step.¹

Diagnosis
We found no clinical studies that assessed the value of physical examination or history taking. The only studies relevant to diagnosis discussed radiography.

Nursemaid’s elbow is an easily recognized diagnosis based on the history and physical examination.¹⁷ Still, it seems many physicians do not recognize the condition.^4-6 Typically, a parent reports that the child cried out after a pull on the arm and then refused to use the arm, holding it slightly flexed and pronated.¹⁸ Pain may be felt only at the wrist or shoulder.^3,18 Occasionally, a snap or click is heard when the accident happens.⁵ The elbow can usually be flexed and extended, but the child resists supination of the forearm, which causes pain in the elbow. There is no swelling or bruising.¹⁹

Children are often referred for radiographic examination with the observation, “refuses to use arm; please x-ray from shoulder to wrist.”²⁰ Radiography is of little help, however, and exposes the child to a dose of ionizing radiation. Although some studies show small significant differences between nursemaid’s elbow and the normal elbow,^21-23 radiographic results generally are reported as normal.^4,6,8,24 (Some commentators assume this may occur if the radiology technician repositions the arm in an attempt to obtain a true anteroposterior projection of the elbow.^1,18,25) Restrict radiography, therefore, to cases with an unclear history or a history of trauma other than arm pull, to exclude more severe injuries.

The role of sonography is not yet clear, but it may turn out to be a fast and harmless technique for diagnosing uncertain cases.^20,25,26

Treatment
Although no articles have described the natural course of nursemaid’s elbow, most authors report that it resolves on its own when a child moves the arm in supination or pronation. It is so easily treated that parents of children with recurrent episodes have even been instructed by phone how to perform the reduction.²⁷

Most articles and textbooks recommend reducing nursemaid’s elbow by a rapid supination of the forearm, followed by flexion or extension.^9,28 However, some articles have described a pronation method.²⁹ We found 2 high-quality trials that compared the success rate of the supination method with the pronation method.^30,31 Researchers conducting 1 medium-quality trial assessed the difference in pain experienced with these 2 methods.³² And researchers conducting 1 low-quality trial tried to assess whether splinting after manipulation helps to prevent recurrences of nursemaid’s elbow.¹⁰

These trials indicate the pronation method is more successful. In addition, some studies report that the pronation method is less painful for the child and less frightening for a parent to watch.^29-31 Green et al confirmed this in their randomized trial.³²

Most compelling finding

The highest quality studies were those devoted to treatment,^28,30-33 and the clear conclusion from their findings—in contrast to what textbooks recommend—is that reduction with a pronation maneuver is more often successful than the supination method. Of course more studies will be needed before textbooks change their recommendations. But at least these studies provide helpful guidance now.

CORRESPONDENCE
Marjolein Krul, MD, Department of General Practice, Room Ff304, Erasmus MC-University Medical Center Rotterdam, PO Box 1738, 3000 DR Rotterdam, The Netherlands; [email protected]

CASE A 63-year-old woman came to our emergency department with her fiancé following an abrupt onset of confusion that began 1 hour earlier. The patient had been working outside in the yard when she approached her fiancé, repeatedly asking where she was and what she was doing. She remained conscious of her identity, however, and exhibited no other neurologic symptoms, such as muscle weakness, gait imbalance, sensory loss, vision changes, slurred speech, or facial droop. The fiancé did not witness any loss of consciousness, head trauma, or seizure-like activity.

Before the event, the patient was feeling well, without any fever, headache, emesis, or vertigo. She denied using tobacco, alcohol, or illicit drugs. Her medical history was unremarkable, including an absence of diabetes, hypertension, and hyperlipidemia. The only significant finding in her family history was a stroke her mother experienced at an advanced age. During our interview, the patient remained confused about where she was and what was happening. She was aware of her confusion and distressed by it.

On examination, the patient was alert and oriented to self and year. She appeared appropriately anxious about her situation. She was afebrile and slightly hypertensive. Her other vital signs were normal. She could not recall events immediately preceding her arrival at the emergency department, but could recall events of the day before and earlier. There was no evidence of trauma. Head, neck, cardiovascular, lung, and abdominal exams were within normal limits.

Her neurologic exam revealed intact cranial nerves, symmetric face, 5/5 muscle strength in all extremities, intact sensation, and normal gait. Grossly, visual fields were intact. There was no Babinski sign, clonus, or pronator drift. She had 3/3 immediate recall of named objects, but 0/3 recall at 5 minutes. Results for complete blood count, basic metabolic panel, and urinalysis were within normal limits, including a blood glucose level of 77 mg/dL and a low-density lipoprotein level of 161 mg/dL. The result for cardiac enzymes was negative. Noncontrast computed tomography of the head revealed a remote pontine lacunar infarct.

WHAT IS THE MOST LIKELY EXPLANATION FOR HER CONDITION?

Transient global amnesia

We admitted the patient for further evaluation with a presumptive diagnosis of transient global amnesia (TGA).

With a chief complaint of amnesia, the differential diagnosis is broad (TABLE 1).^1-3 In this case, a stroke was unlikely given the absence of neurologic deficits, specifically the lack of visual field defects. The elapsed time of her symptoms was too long for a transient ischemic attack or seizure. There was no supporting evidence for encephalitis, intracranial bleed, or hypoglycemia. While delirium could be considered, its characteristic features of inattention and a waxing and waning course were not present, nor was there any obvious underlying cause, such as infection or polypharmacy. The patient had no loss of self-identity that would suggest a psychogenic cause. The time course and the patient’s symptoms were congruent with the clinical criteria for TGA, and we confidently based our diagnosis on this.

TABLE 1
Rule out these disorders with acute anterograde amnesia^1-3

Type of memory loss as a clue to cause

Amnesia occurs when memory and learning in an alert person are impaired to a degree out of proportion to the person’s overall neurologic status. It may affect the formation of new memories (anterograde amnesia) or the recall of past memories (retrograde amnesia).

How memory works. Memory can be broken down into categories (TABLE 2).¹ Explicit memory requires a conscious effort to recall. An example is episodic memory, in which memories are framed within a context, such as recalling what was served for dinner the night before. Its function is critical to creating new memories. Other forms of explicit memory are semantic memory—memorized facts that are independent of a context—and working memory, in which focused attention is used to manipulate information. Implicit memory operates subconsciously. The prime example is procedural memory, involving the ability to learn new skills and perform them without total concentration.

Memory function affected in TGA. In TGA, episodic memory—critical in the laying down of new memories—is most affected. Episodic memory relies heavily on the hippocampus to function correctly. When it dysfunctions, a person cannot consolidate and retain new information, thus resulting in anterograde amnesia.¹

Retrograde amnesia generally requires dysfunction of the frontal lobe in addition to the temporal lobe.³ However, it may be present concurrently with anterograde amnesia when a lesion is isolated to the hippocampus; it is usually limited to more recent memories. That recent memories tend to be the more vulnerable is known as Ribot’s law. If retrograde amnesia is present, it usually resolves before anterograde amnesia.⁴

In TGA, procedural memory is unaffected. Thus, activities of daily living and instrumental activities of daily living remain intact—eg, the patient retains the necessary skills to drive a car.

TABLE 2
Categories of memory function¹

Most often the prognosis is good

TGA is an unusual manifestation of anterograde amnesia that is self-limited and tends not to recur.⁵ An episode typically lasts 1 to 8 hours.⁶ Although the disorder was first described in 1956, a set of clinical criteria (TABLE 3) was not defined until 1990.⁷ The highlights of these criteria are that self-identity is preserved and no evidence exists for neurologic deficit or seizure activity.⁶ The incidence of TGA is 3 to 10 in 100,000.⁵ TGA usually affects patients in their early 60s,² and men and women are affected equally.

Interestingly, more than half of patients with TGA report a precipitating event, usually involving physical activity or a Valsalva maneuver.⁶ Classically, the patient repeatedly asks the same questions. The most common associated symptoms are headache, dizziness, and nausea.^2,6

Generally, the patient’s prognosis is good, without long-term sequelae. Importantly, reassure patients and their families that there will be no memories of the event itself, as their memory-making ability was impaired.²

TABLE 3
Clinical criteria for transient global amnesia, as defined by Hodges and Warlow⁷

If episodes do recur

A small subset of people may have recurrent episodes. Recurrence rates over a 5-year span have been reported as 3% to 26%; however, this range includes cases and studies recorded before the diagnostic criteria were developed in 1990.⁶ Although the clinical criteria for TGA can be helpful in diagnosing the disorder, there is no standardized workup because TGA has no clear etiology or known underlying mechanism. Many causal theories exist, however, and have evidence to support them.

Possible underlying conditions. One proposed explanation is ischemia of the hippocampus. This raises questions of whether vascular risk factors place people at higher risk.⁸ Recent studies have not confirmed this theory, and patients with diabetes, hypertension, or hyperlipidemia appear not to be at higher risk of TGA. Still, it is interesting that TGA is a disease affecting older adults and that evidence of small-vessel ischemia is often discovered incidentally.^6,8

On the other hand, some experts take into account the high association of TGA with migraines documented in multiple studies, and therefore propose a spreading depression as the cause.⁵ Another hypothesis is a valvular insufficiency of the jugular veins that allows reflux, resulting in venous ischemia of the hippocampal area, especially during a Valsalva maneuver.⁹ Indeed, jugular valve insufficiency has been noted in up to two-thirds of TGA patients. However, if valvular insufficiency is truly the mechanism of disease, why do recurrence rates remain so low?¹⁰

MRI may be helpful. Given the many theories of TGA origin, several imaging mechanisms have been tried with mixed results: single photon emission computed tomography, magnetic resonance imaging (MRI) with diffusion-weighted imaging, and positron emission tomography.

The lack of reliable results makes it difficult to establish diagnostic criteria. Some generalized guidelines are as follows:

If there are any neurologic findings or concern about a transient ischemic attack or cerebrovascular accident, obtain an MRI. This should include diffusion-weighted imaging, which may reveal a transient lesion in the hippocampus.⁶ If the patient has recurrent episodes, or has episodes that last less than 1 hour, suspect the possibility of seizure and consider arranging for an electroencephalogram.^4,6 Likewise, recurrence may also be due to a patent foramen ovale (PFO) causing paradoxical emboli and transient ischemia of the hippocampus. In 1 study, the rate of PFO in the TGA arm was 55%; it was 50% in those with recurrent episodes.¹¹

Our patient’s outcome

In the 24 hours after admission, the patient’s anterograde amnesia gradually resolved. She was able to remember the medical staff caring for her and retain orientation to her situation. However, she was unable to regain memories of the events immediately surrounding the onset of amnesia. During her hospitalization, the patient underwent a thorough work-up, including carotid artery Doppler ultrasound and echocardiogram with agitated saline (bubble study), both of which yielded normal results. Her MR angiography showed patent cerebral vessels. As mentioned, an MRI of the head showed a remote lacunar infarct of her left upper pons and nonspecific subcortical white matter disease was noted, consistent with chronic small vessel disease. The patient was discharged with reassurance, and she has done well.

CORRESPONDENCE
Chris Bernheisel, MD, director, Family Medicine Inpatient Service, The University of Cincinnati, 2123 Auburn Ave, Suite 340, Cincinnati, OH 45219; [email protected]

CASE A 63-year-old woman came to our emergency department with her fiancé following an abrupt onset of confusion that began 1 hour earlier. The patient had been working outside in the yard when she approached her fiancé, repeatedly asking where she was and what she was doing. She remained conscious of her identity, however, and exhibited no other neurologic symptoms, such as muscle weakness, gait imbalance, sensory loss, vision changes, slurred speech, or facial droop. The fiancé did not witness any loss of consciousness, head trauma, or seizure-like activity.

Before the event, the patient was feeling well, without any fever, headache, emesis, or vertigo. She denied using tobacco, alcohol, or illicit drugs. Her medical history was unremarkable, including an absence of diabetes, hypertension, and hyperlipidemia. The only significant finding in her family history was a stroke her mother experienced at an advanced age. During our interview, the patient remained confused about where she was and what was happening. She was aware of her confusion and distressed by it.

On examination, the patient was alert and oriented to self and year. She appeared appropriately anxious about her situation. She was afebrile and slightly hypertensive. Her other vital signs were normal. She could not recall events immediately preceding her arrival at the emergency department, but could recall events of the day before and earlier. There was no evidence of trauma. Head, neck, cardiovascular, lung, and abdominal exams were within normal limits.

Her neurologic exam revealed intact cranial nerves, symmetric face, 5/5 muscle strength in all extremities, intact sensation, and normal gait. Grossly, visual fields were intact. There was no Babinski sign, clonus, or pronator drift. She had 3/3 immediate recall of named objects, but 0/3 recall at 5 minutes. Results for complete blood count, basic metabolic panel, and urinalysis were within normal limits, including a blood glucose level of 77 mg/dL and a low-density lipoprotein level of 161 mg/dL. The result for cardiac enzymes was negative. Noncontrast computed tomography of the head revealed a remote pontine lacunar infarct.

WHAT IS THE MOST LIKELY EXPLANATION FOR HER CONDITION?

Transient global amnesia

We admitted the patient for further evaluation with a presumptive diagnosis of transient global amnesia (TGA).

With a chief complaint of amnesia, the differential diagnosis is broad (TABLE 1).^1-3 In this case, a stroke was unlikely given the absence of neurologic deficits, specifically the lack of visual field defects. The elapsed time of her symptoms was too long for a transient ischemic attack or seizure. There was no supporting evidence for encephalitis, intracranial bleed, or hypoglycemia. While delirium could be considered, its characteristic features of inattention and a waxing and waning course were not present, nor was there any obvious underlying cause, such as infection or polypharmacy. The patient had no loss of self-identity that would suggest a psychogenic cause. The time course and the patient’s symptoms were congruent with the clinical criteria for TGA, and we confidently based our diagnosis on this.

TABLE 1
Rule out these disorders with acute anterograde amnesia^1-3

Type of memory loss as a clue to cause

Amnesia occurs when memory and learning in an alert person are impaired to a degree out of proportion to the person’s overall neurologic status. It may affect the formation of new memories (anterograde amnesia) or the recall of past memories (retrograde amnesia).

How memory works. Memory can be broken down into categories (TABLE 2).¹ Explicit memory requires a conscious effort to recall. An example is episodic memory, in which memories are framed within a context, such as recalling what was served for dinner the night before. Its function is critical to creating new memories. Other forms of explicit memory are semantic memory—memorized facts that are independent of a context—and working memory, in which focused attention is used to manipulate information. Implicit memory operates subconsciously. The prime example is procedural memory, involving the ability to learn new skills and perform them without total concentration.

Memory function affected in TGA. In TGA, episodic memory—critical in the laying down of new memories—is most affected. Episodic memory relies heavily on the hippocampus to function correctly. When it dysfunctions, a person cannot consolidate and retain new information, thus resulting in anterograde amnesia.¹

Retrograde amnesia generally requires dysfunction of the frontal lobe in addition to the temporal lobe.³ However, it may be present concurrently with anterograde amnesia when a lesion is isolated to the hippocampus; it is usually limited to more recent memories. That recent memories tend to be the more vulnerable is known as Ribot’s law. If retrograde amnesia is present, it usually resolves before anterograde amnesia.⁴

In TGA, procedural memory is unaffected. Thus, activities of daily living and instrumental activities of daily living remain intact—eg, the patient retains the necessary skills to drive a car.

TABLE 2
Categories of memory function¹

Most often the prognosis is good

TGA is an unusual manifestation of anterograde amnesia that is self-limited and tends not to recur.⁵ An episode typically lasts 1 to 8 hours.⁶ Although the disorder was first described in 1956, a set of clinical criteria (TABLE 3) was not defined until 1990.⁷ The highlights of these criteria are that self-identity is preserved and no evidence exists for neurologic deficit or seizure activity.⁶ The incidence of TGA is 3 to 10 in 100,000.⁵ TGA usually affects patients in their early 60s,² and men and women are affected equally.

Interestingly, more than half of patients with TGA report a precipitating event, usually involving physical activity or a Valsalva maneuver.⁶ Classically, the patient repeatedly asks the same questions. The most common associated symptoms are headache, dizziness, and nausea.^2,6

Generally, the patient’s prognosis is good, without long-term sequelae. Importantly, reassure patients and their families that there will be no memories of the event itself, as their memory-making ability was impaired.²

TABLE 3
Clinical criteria for transient global amnesia, as defined by Hodges and Warlow⁷

If episodes do recur

A small subset of people may have recurrent episodes. Recurrence rates over a 5-year span have been reported as 3% to 26%; however, this range includes cases and studies recorded before the diagnostic criteria were developed in 1990.⁶ Although the clinical criteria for TGA can be helpful in diagnosing the disorder, there is no standardized workup because TGA has no clear etiology or known underlying mechanism. Many causal theories exist, however, and have evidence to support them.

Possible underlying conditions. One proposed explanation is ischemia of the hippocampus. This raises questions of whether vascular risk factors place people at higher risk.⁸ Recent studies have not confirmed this theory, and patients with diabetes, hypertension, or hyperlipidemia appear not to be at higher risk of TGA. Still, it is interesting that TGA is a disease affecting older adults and that evidence of small-vessel ischemia is often discovered incidentally.^6,8

On the other hand, some experts take into account the high association of TGA with migraines documented in multiple studies, and therefore propose a spreading depression as the cause.⁵ Another hypothesis is a valvular insufficiency of the jugular veins that allows reflux, resulting in venous ischemia of the hippocampal area, especially during a Valsalva maneuver.⁹ Indeed, jugular valve insufficiency has been noted in up to two-thirds of TGA patients. However, if valvular insufficiency is truly the mechanism of disease, why do recurrence rates remain so low?¹⁰

MRI may be helpful. Given the many theories of TGA origin, several imaging mechanisms have been tried with mixed results: single photon emission computed tomography, magnetic resonance imaging (MRI) with diffusion-weighted imaging, and positron emission tomography.

The lack of reliable results makes it difficult to establish diagnostic criteria. Some generalized guidelines are as follows:

If there are any neurologic findings or concern about a transient ischemic attack or cerebrovascular accident, obtain an MRI. This should include diffusion-weighted imaging, which may reveal a transient lesion in the hippocampus.⁶ If the patient has recurrent episodes, or has episodes that last less than 1 hour, suspect the possibility of seizure and consider arranging for an electroencephalogram.^4,6 Likewise, recurrence may also be due to a patent foramen ovale (PFO) causing paradoxical emboli and transient ischemia of the hippocampus. In 1 study, the rate of PFO in the TGA arm was 55%; it was 50% in those with recurrent episodes.¹¹

Our patient’s outcome

In the 24 hours after admission, the patient’s anterograde amnesia gradually resolved. She was able to remember the medical staff caring for her and retain orientation to her situation. However, she was unable to regain memories of the events immediately surrounding the onset of amnesia. During her hospitalization, the patient underwent a thorough work-up, including carotid artery Doppler ultrasound and echocardiogram with agitated saline (bubble study), both of which yielded normal results. Her MR angiography showed patent cerebral vessels. As mentioned, an MRI of the head showed a remote lacunar infarct of her left upper pons and nonspecific subcortical white matter disease was noted, consistent with chronic small vessel disease. The patient was discharged with reassurance, and she has done well.

CORRESPONDENCE
Chris Bernheisel, MD, director, Family Medicine Inpatient Service, The University of Cincinnati, 2123 Auburn Ave, Suite 340, Cincinnati, OH 45219; [email protected]

CASE A 63-year-old woman came to our emergency department with her fiancé following an abrupt onset of confusion that began 1 hour earlier. The patient had been working outside in the yard when she approached her fiancé, repeatedly asking where she was and what she was doing. She remained conscious of her identity, however, and exhibited no other neurologic symptoms, such as muscle weakness, gait imbalance, sensory loss, vision changes, slurred speech, or facial droop. The fiancé did not witness any loss of consciousness, head trauma, or seizure-like activity.

Before the event, the patient was feeling well, without any fever, headache, emesis, or vertigo. She denied using tobacco, alcohol, or illicit drugs. Her medical history was unremarkable, including an absence of diabetes, hypertension, and hyperlipidemia. The only significant finding in her family history was a stroke her mother experienced at an advanced age. During our interview, the patient remained confused about where she was and what was happening. She was aware of her confusion and distressed by it.

On examination, the patient was alert and oriented to self and year. She appeared appropriately anxious about her situation. She was afebrile and slightly hypertensive. Her other vital signs were normal. She could not recall events immediately preceding her arrival at the emergency department, but could recall events of the day before and earlier. There was no evidence of trauma. Head, neck, cardiovascular, lung, and abdominal exams were within normal limits.

Her neurologic exam revealed intact cranial nerves, symmetric face, 5/5 muscle strength in all extremities, intact sensation, and normal gait. Grossly, visual fields were intact. There was no Babinski sign, clonus, or pronator drift. She had 3/3 immediate recall of named objects, but 0/3 recall at 5 minutes. Results for complete blood count, basic metabolic panel, and urinalysis were within normal limits, including a blood glucose level of 77 mg/dL and a low-density lipoprotein level of 161 mg/dL. The result for cardiac enzymes was negative. Noncontrast computed tomography of the head revealed a remote pontine lacunar infarct.

WHAT IS THE MOST LIKELY EXPLANATION FOR HER CONDITION?

Transient global amnesia

We admitted the patient for further evaluation with a presumptive diagnosis of transient global amnesia (TGA).

With a chief complaint of amnesia, the differential diagnosis is broad (TABLE 1).^1-3 In this case, a stroke was unlikely given the absence of neurologic deficits, specifically the lack of visual field defects. The elapsed time of her symptoms was too long for a transient ischemic attack or seizure. There was no supporting evidence for encephalitis, intracranial bleed, or hypoglycemia. While delirium could be considered, its characteristic features of inattention and a waxing and waning course were not present, nor was there any obvious underlying cause, such as infection or polypharmacy. The patient had no loss of self-identity that would suggest a psychogenic cause. The time course and the patient’s symptoms were congruent with the clinical criteria for TGA, and we confidently based our diagnosis on this.

TABLE 1
Rule out these disorders with acute anterograde amnesia^1-3

Type of memory loss as a clue to cause

Amnesia occurs when memory and learning in an alert person are impaired to a degree out of proportion to the person’s overall neurologic status. It may affect the formation of new memories (anterograde amnesia) or the recall of past memories (retrograde amnesia).

How memory works. Memory can be broken down into categories (TABLE 2).¹ Explicit memory requires a conscious effort to recall. An example is episodic memory, in which memories are framed within a context, such as recalling what was served for dinner the night before. Its function is critical to creating new memories. Other forms of explicit memory are semantic memory—memorized facts that are independent of a context—and working memory, in which focused attention is used to manipulate information. Implicit memory operates subconsciously. The prime example is procedural memory, involving the ability to learn new skills and perform them without total concentration.

Memory function affected in TGA. In TGA, episodic memory—critical in the laying down of new memories—is most affected. Episodic memory relies heavily on the hippocampus to function correctly. When it dysfunctions, a person cannot consolidate and retain new information, thus resulting in anterograde amnesia.¹

Retrograde amnesia generally requires dysfunction of the frontal lobe in addition to the temporal lobe.³ However, it may be present concurrently with anterograde amnesia when a lesion is isolated to the hippocampus; it is usually limited to more recent memories. That recent memories tend to be the more vulnerable is known as Ribot’s law. If retrograde amnesia is present, it usually resolves before anterograde amnesia.⁴

In TGA, procedural memory is unaffected. Thus, activities of daily living and instrumental activities of daily living remain intact—eg, the patient retains the necessary skills to drive a car.

TABLE 2
Categories of memory function¹

Most often the prognosis is good

TGA is an unusual manifestation of anterograde amnesia that is self-limited and tends not to recur.⁵ An episode typically lasts 1 to 8 hours.⁶ Although the disorder was first described in 1956, a set of clinical criteria (TABLE 3) was not defined until 1990.⁷ The highlights of these criteria are that self-identity is preserved and no evidence exists for neurologic deficit or seizure activity.⁶ The incidence of TGA is 3 to 10 in 100,000.⁵ TGA usually affects patients in their early 60s,² and men and women are affected equally.

Interestingly, more than half of patients with TGA report a precipitating event, usually involving physical activity or a Valsalva maneuver.⁶ Classically, the patient repeatedly asks the same questions. The most common associated symptoms are headache, dizziness, and nausea.^2,6

Generally, the patient’s prognosis is good, without long-term sequelae. Importantly, reassure patients and their families that there will be no memories of the event itself, as their memory-making ability was impaired.²

TABLE 3
Clinical criteria for transient global amnesia, as defined by Hodges and Warlow⁷

If episodes do recur

A small subset of people may have recurrent episodes. Recurrence rates over a 5-year span have been reported as 3% to 26%; however, this range includes cases and studies recorded before the diagnostic criteria were developed in 1990.⁶ Although the clinical criteria for TGA can be helpful in diagnosing the disorder, there is no standardized workup because TGA has no clear etiology or known underlying mechanism. Many causal theories exist, however, and have evidence to support them.

Possible underlying conditions. One proposed explanation is ischemia of the hippocampus. This raises questions of whether vascular risk factors place people at higher risk.⁸ Recent studies have not confirmed this theory, and patients with diabetes, hypertension, or hyperlipidemia appear not to be at higher risk of TGA. Still, it is interesting that TGA is a disease affecting older adults and that evidence of small-vessel ischemia is often discovered incidentally.^6,8

On the other hand, some experts take into account the high association of TGA with migraines documented in multiple studies, and therefore propose a spreading depression as the cause.⁵ Another hypothesis is a valvular insufficiency of the jugular veins that allows reflux, resulting in venous ischemia of the hippocampal area, especially during a Valsalva maneuver.⁹ Indeed, jugular valve insufficiency has been noted in up to two-thirds of TGA patients. However, if valvular insufficiency is truly the mechanism of disease, why do recurrence rates remain so low?¹⁰

MRI may be helpful. Given the many theories of TGA origin, several imaging mechanisms have been tried with mixed results: single photon emission computed tomography, magnetic resonance imaging (MRI) with diffusion-weighted imaging, and positron emission tomography.

The lack of reliable results makes it difficult to establish diagnostic criteria. Some generalized guidelines are as follows:

If there are any neurologic findings or concern about a transient ischemic attack or cerebrovascular accident, obtain an MRI. This should include diffusion-weighted imaging, which may reveal a transient lesion in the hippocampus.⁶ If the patient has recurrent episodes, or has episodes that last less than 1 hour, suspect the possibility of seizure and consider arranging for an electroencephalogram.^4,6 Likewise, recurrence may also be due to a patent foramen ovale (PFO) causing paradoxical emboli and transient ischemia of the hippocampus. In 1 study, the rate of PFO in the TGA arm was 55%; it was 50% in those with recurrent episodes.¹¹

Our patient’s outcome

In the 24 hours after admission, the patient’s anterograde amnesia gradually resolved. She was able to remember the medical staff caring for her and retain orientation to her situation. However, she was unable to regain memories of the events immediately surrounding the onset of amnesia. During her hospitalization, the patient underwent a thorough work-up, including carotid artery Doppler ultrasound and echocardiogram with agitated saline (bubble study), both of which yielded normal results. Her MR angiography showed patent cerebral vessels. As mentioned, an MRI of the head showed a remote lacunar infarct of her left upper pons and nonspecific subcortical white matter disease was noted, consistent with chronic small vessel disease. The patient was discharged with reassurance, and she has done well.

CORRESPONDENCE
Chris Bernheisel, MD, director, Family Medicine Inpatient Service, The University of Cincinnati, 2123 Auburn Ave, Suite 340, Cincinnati, OH 45219; [email protected]