Letters

COMMENTS
AND

RESPONSES

Prevention of Ventilator-Associated Pneumonia

TO THE EDITOR: Collard and colleagues review (1) on the preven-

tion of ventilator-associated pneumonia illustrates the failure of evidence-based medicine. First, concerning stress ulcer prophylaxis, the authors recommendation to use sucralfate rather than H2-antagonists (on the basis of lower rates of pneumonia with sucralfate [2, 3]) since neither sucralfate nor H2-antagonists prevent gastrointestinal bleeding. The authors choice was based on their preference rather than on evidence-based medicine. Second, the authors interpretation of the evidence is questionable: They stated that aspiration of subglottic secretions is most effective in patients who require prolonged mechanical ventilation. However, 2 of the 3 trials cited indicate that aspiration of subglottic secretions reduces early-onset but not late-onset ventilator-associated pneumonia. Third, the authors did not apply so-called evidence-based medicine when it did not t their beliefs. Evidence of efcacy is by far greater and less debated for selective digestive tract decontamination than for stress ulcer prophylaxis, but Collard and colleagues did not recommend the former because of a non evidence-based fear of growing antibiotic resistance. Fourth, the authors search of pertinent data on ventilator circuit management and ventilator-associated pneumonia was incomplete. They ignored a randomized, controlled trial showing a similar rate of ventilator-associated pneumonia when heat and moisture exchangers are changed every 48 hours instead of every 24 hours (4). In addition, they erroneously described the intervention group in the study by Davis and colleagues (5); heat and moisture exchangers in this group were changed every 72 hours, not every 5 days. Fifth, this review illustrates the absurd aspects of evidence-based medicine: The authors retained 2 trials reporting increased rates of endotracheal tube occlusion with heat and moisture exchangers solely because those trials were randomized. However, these studies are also more than 10 years old and studied obsolete devices. A thorough analysis of the literature indicates no increased risk for tube occlusion with heat and moisture exchangers in comparison with heated humidiers (zero tube occlusion in more than 1000 patients is sufcient evidence, even outside of a randomized, controlled trial). Finally, the authors ignored randomized, controlled trials showing that noninvasive ventilation reduced rates of ventilator-associated pneumonia. Jean-Damien Ricard, MD, PhD Didier Dreyfuss, MD Louis Mourier Hospital 92700 Colombes, France
References
1. Collard HR, Saint S, Matthay MA. Prevention of ventilator-associated pneumonia: an evidence-based systematic review. Ann Intern Med. 2003;138:494-501. [PMID: 12639084] 2. Cook DJ, Reeve BK, Guyatt GH, Heyland DK, Grifth LE, Buckingham L, et al. Stress ulcer prophylaxis in critically ill patients. Resolving discordant meta-analyses. JAMA. 1996;275:308-14. [PMID: 8544272]

3. Messori A, Trippoli S, Vaiani M, Gorini M, Corrado A. Bleeding and pneumonia in intensive care patients given ranitidine and sucralfate for prevention of stress ulcer: meta-analysis of randomised controlled trials. BMJ. 2000;321:1103-6. [PMID: 11061729] 4. Djedaini K, Billiard M, Mier L, Le Bourdelles G, Brun P, Markowicz P, et al. Changing heat and moisture exchangers every 48 hours rather than 24 hours does not affect their efcacy and the incidence of nosocomial pneumonia. Am J Respir Crit Care Med. 1995;152:1562-9. [PMID: 7582295] 5. Davis K Jr, Evans SL, Campbell RS, Johannigman JA, Luchette FA, Porembka DT, et al. Prolonged use of heat and moisture exchangers does not affect device efciency or frequency rate of nosocomial pneumonia. Crit Care Med. 2000;28:1412-8. [PMID: 10834688]

IN RESPONSE: Systematic reviews aim to summarize existing data identied through specied selection criteria and, at times, make recommendations on the basis of these data. We believe that welldone systematic reviews illustrate the strength of evidence-based medicine, not the absurdity. Recommendations in our article were made not on the basis of our preferences but after careful evaluation of the evidence. Ricard and Dreyfuss contest our review of 2 meta-analyses comparing sucralfate with H2-antagonists, claiming the studies reached opposite conclusions (1, 2). This is incorrect. Both studies found sucralfate to be associated with a decreased incidence of ventilatorassociated pneumonia compared with H2-antagonists. Whether clinicians need to routinely prevent stress-related upper gastrointestinal bleeding in critically ill patientsa question not addressed by our reviewis controversial. Ricard and Dreyfuss question our conclusion that aspiration of subglottic secretions may benet patients who require mechanical ventilation for more than 72 hours because of studies showing that this technique reduces early-onset ventilatorassociated pneumonia (pneumonia developing within 4 to 5 days of intubation). Many patients who require mechanical ventilation for longer than 72 hours develop early-onset ventilator-associated pneumonia, and it is these patients who appear to benet most from aspiration of subglottic secretions. Ricard and Dreyfusss statement that we did not apply socalled evidence-based medicine when it did not t [our] beliefs is groundless; there is ample evidence to suggest that the use of selective decontamination of the digestive tract may increase antimicrobial resistance (3, 4). Although we appreciate the additional information Ricard and Dreyfuss provided about the use of heat and moisture exchangers, the assertion that we misstated the methods in the study by Davis and colleagues (5) is incorrect. Davis and colleagues randomly assigned patients to groups in which heat and moisture exchangers were changed every 24 or 120 hours. We stand by our calculation of 120 hours as 5 days. Finally, Ricard and Dreyfuss state that we ignored randomized, controlled trials showing that noninvasive ventilation reduces ventilator-associated pneumonia. Although our search strategy identied several studies of noninvasive ventilation (as well as studies of many other strategies), none met our inclusion criteria. We attempted to provide an unbiased, comprehensive review of the literature on strategies to prevent ventilator-associated pneumonia. Although evidence-based medicine certainly has its aws, we hope that our review helps clinicians identify practices that may help reduce the occurrence of this highly morbid condition.

486 2004 American College of Physicians

Letters
Harold R. Collard, MD University of Colorado Health Sciences Center Denver, CO 80262 Sanjay Saint, MD, MPH Ann Arbor Veterans Affairs Medical Center and the University of Michigan Health System Ann Arbor, MI 48109 Michael A. Matthay, MD University of California, San Francisco San Francisco, CA 94143
References
1. Messori A, Trippoli S, Vaiani M, Gorini M, Corrado A. Bleeding and pneumonia in intensive care patients given ranitidine and sucralfate for prevention of stress ulcer: meta-analysis of randomised controlled trials. BMJ. 2000;321:1103-6. [PMID: 11061729] 2. Cook DJ, Reeve BK, Guyatt GH, Heyland DK, Grifth LE, Buckingham L, et al. Stress ulcer prophylaxis in critically ill patients. Resolving discordant meta-analyses. JAMA. 1996;275:308-14. [PMID: 8544272] 3. Kollef MH. Selective digestive decontamination should not be routinely employed. Chest. 2003;123:464S-8S. [PMID: 12740230] 4. Ebner W, Kropec-Hubner A, Daschner FD. Bacterial resistance and overgrowth due to selective decontamination of the digestive tract. Eur J Clin Microbiol Infect Dis. 2000;19:243-7. [PMID: 10834811] 5. Davis K Jr, Evans SL, Campbell RS, Johannigman JA, Luchette FA, Porembka DT, et al. Prolonged use of heat and moisture exchangers does not affect device efciency or frequency rate of nosocomial pneumonia. Crit Care Med. 2000;28:1412-8. [PMID: 10834688]

sive patients with type 2 diabetes mellitus, we must not forget that these patients often have underlying coronary atherosclerosis or ventricular dysfunction that warrants ACE inhibitor therapy. Given that the average wholesale price of losartan is almost 10-fold higher than that of generic captopril (6), the most effective medication may also be the least expensive. Mori J. Krantz, MD Denver Healths Cardiac Risk Reduction Program Denver, CO 80204
References
1. Vijan S, Hayward RA. Treatment of hypertension in type 2 diabetes mellitus: blood pressure goals, choice of agents, and setting priorities in diabetes care. Ann Intern Med. 2003;138:593-602. [PMID: 12667032] 2. Dickstein K, Kjekshus J. Effects of losartan and captopril on mortality and morbidity in high-risk patients after acute myocardial infarction: the OPTIMAAL randomised trial. Optimal Trial in Myocardial Infarction with Angiotensin II Antagonist Losartan. Lancet. 2002;360:752-60. [PMID: 12241832] 3. Pitt B, Poole-Wilson PA, Segal R, Martinez FA, Dickstein K, Camm AJ, et al. Effect of losartan compared with captopril on mortality in patients with symptomatic heart failure: randomised trialthe Losartan Heart Failure Survival Study ELITE II. Lancet. 2000;355:1582-7. [PMID: 10821361] 4. Berry C, Norrie J, McMurray JJ. Are angiotensin II receptor blockers more efcacious than placebo in heart failure? Implications of ELITE-2. Evaluation of Losartan In The Elderly [Editorial]. Am J Cardiol. 2001;87:606-7, A9. [PMID: 11230847] 5. Pfeffer MA, McMurray JJ, Velazquez EJ, Rouleau JL, Kober L, Maggioni AP, et al. Valsartan, captopril, or both in myocardial infarction complicated by heart failure, left ventricular dysfunction, or both. N Engl J Med. 2003;349:1893-906. [PMID: 14610160] 6. 2002 Drug Topics Redbook. Montvale, NJ: Thomson Medical Economics; 2002.

Blood Pressure Control in Type 2 Diabetes Mellitus

IN RESPONSE: Dr. Krantz points out that ACE inhibitors may be TO THE EDITOR: In their clinical guideline addressing optimal anti-

hypertensive therapy in patients with type 2 diabetes mellitus (1), Vijan and Hayward pointed out that effective treatment of hypertension is more important than tight glucose control in reducing vascular complications, which claim the lives of 80% of diabetic persons. However, in choosing an optimal agent, Vijan and Hayward made no clinical distinction between angiotensin-converting enzyme (ACE) inhibitors and angiotensin-receptor blockers (ARBs). Patients with type 2 diabetes mellitus frequently experience acute myocardial infarction (MI) and subsequent congestive heart failure because of left ventricular systolic dysfunction. In large randomized trials of both acute MI (2) and congestive heart failure (3), treatment with the ACE inhibitor captopril showed a consistent favorable trend toward lower mortality rates compared with the ARB losartan. Some authors have even questioned the efcacy of ARBs relative to placebo in the management of congestive heart failure (4). A recent trial demonstrated that the ARB valsartan was as effective as the ACE inhibitor captopril in high-risk patients with MI, but adverse hemodynamic and renal events leading to withdrawal of the study drug were signicantly more common in patients assigned to ARB therapy (5). Although diabetic persons were underrepresented in these trials, the data support ACE inhibitors as the current standard of care for acute MI and congestive heart failure. Angiotensin-receptor blockers are certainly reasonable second-line agents when ACE inhibitors are not tolerated. In choosing pharmacologic interventions in hypertenwww.annals.org

preferred over ARBs for patients with systolic dysfunction or acute MI. We agree, although trials show no clear statistical differences in outcomes between the 2 drugs. However, we would note that our review was of the primary treatment of hypertension in diabetes, not the treatment of comorbid conditions. Many other situations may lead clinicians to alter the initial choice of agent; for example, in patients with angina pectoris, a -blocker or a calcium-channel blocker may sometimes be preferred, and for those with prostatic hypertrophy, some clinicians may prefer -blockers. A complete discussion of these conditions was beyond the scope of our review. Although ACE inhibitors may be preferred over ARBs in some clinical situations, the evidence for preferential benet of ARB treatment for hypertension in patients with diabetes is somewhat stronger. For example, the data on the effectiveness of treatment of renal disease in patients with type 2 diabetes mellitus are currently more robust for ARBs than for ACE inhibitors, particularly compared with other classes of antihypertensive agents (13). Similarly, the Losartan Intervention for Endpoint Reduction in Hypertension study suggested that ARBs are more effective than -blockers in patients with diabetes, hypertension, and left ventricular hypertrophy, while the United Kingdom Prospective Diabetes Study found no benet of ACE inhibitors over -blockers (4, 5). As noted in our review, the studies comparing drug classes are somewhat inconsistent in their conclusions, and the current literature does not include much evidence to suggest that ARBs are necessarily better or worse than ACE inhibitors. Therefore, the American College of Physicians felt that
16 March 2004 Annals of Internal Medicine Volume 140 Number 6 487

Letters
the greater experience with and lower cost of ACE inhibitors were reasons to select them as preferred agents, and we concur with this conclusion. In our view, data from head-to-head comparisons of the 2 classes are needed to make further distinctions. In the interim, other factorssuch as cost, side effect prole, and comorbid conditionsshould be used to guide the choice of drug for an individual patient. Sandeep Vijan, MD, MS Rodney A. Hayward, MD Veterans Affairs Health Services Research and Development Ann Arbor, MI 48113 patients with acute venous thromboembolism, but its more general use, particularly in older patients, could be hazardous. Gregory M. Peterson, PhD, MBA Shane L. Jackson, BPharm(Hons) University of Tasmania Hobart, Tasmania 7001, Australia
References
1. Kovacs MJ, Rodger M, Anderson DR, Morrow B, Kells G, Kovacs J, et al. Comparison of 10-mg and 5-mg warfarin initiation nomograms together with low-molecular-weight heparin for outpatient treatment of acute venous thromboembolism. A randomized, double-blind, controlled trial. Ann Intern Med. 2003;138:714-9. [PMID: 12729425] 2. Crowther MA, Ginsberg JB, Kearon C, Harrison L, Johnson J, Massicotte MP, et al. A randomized trial comparing 5-mg and 10-mg warfarin loading doses. Arch Intern Med. 1999;159:46-8. [PMID: 9892329]

References
1. Nielsen FS, Rossing P, Gall MA, Skott P, Smidt UM, Parving HH. Long-term effect of lisinopril and atenolol on kidney function in hypertensive NIDDM subjects with diabetic nephropathy. Diabetes. 1997;46:1182-8. [PMID: 9200654] 2. Lewis EJ, Hunsicker LG, Clarke WR, Berl T, Pohl MA, Lewis JB, et al. Renoprotective effect of the angiotensin-receptor antagonist irbesartan in patients with nephropathy due to type 2 diabetes. N Engl J Med. 2001;345:851-60. [PMID: 11565517] 3. Schnack C, Hoffmann W, Hopmeier P, Schernthaner G. Renal and metabolic effects of 1-year treatment with ramipril or atenolol in NIDDM patients with microalbuminuria. Diabetologia. 1996;39:1611-6. [PMID: 8960851] 4. Lindholm LH, Ibsen H, Dahlof B, Devereux RB, Beevers G, de Faire U, et al. Cardiovascular morbidity and mortality in patients with diabetes in the Losartan Intervention For Endpoint reduction in hypertension study (LIFE): a randomised trial against atenolol. Lancet. 2002;359:1004-10. [PMID: 11937179] 5. Efcacy of atenolol and captopril in reducing risk of macrovascular and microvascular complications in type 2 diabetes: UKPDS 39. UK Prospective Diabetes Study Group. BMJ. 1998;317:713-20. [PMID: 9732338]

Randomized Trial of Warfarin Nomograms

TO THE EDITOR: We read with interest the article by Kovacs and

colleagues (1) comparing warfarin initiation nomograms in a select group of patients. The authors stated that their results convincingly demonstrate that a nomogram based on a 10-mg starting dose is more effective than a 5-mg nomogram for outpatient treatment of acute venous thromboembolism. We believe the authors have overstated the superiority of the 10-mg loading regimen of warfarin and that caution is required in applying their results in practice. Why expose patients to an increased risk for bleeding to reduce the time taken to reach a therapeutic international normalized ratio (INR) by a mean of 1.4 days? More important than rapidly achieving a therapeutic INR early in therapy is a predictable and safe increase in INR. The authors did not report the proportion of patients with an elevated INR in the rst week of therapy while under the care of the trial group. The authors referred to the study by Crowther and associates (2) and possible reasons for the discrepancy between the 2 studies results. They did not, however, mention the most probable reason: Their patients were considerably younger (by a mean of 10 years) and heavier than those in Crowther and associates study. Few were older than 75 years of age. In addition, Crowther and associates study included more patients with cancer, and they did not exclude patients with a baseline INR above 1.4. The 10-mg nomogram approach could be ideal for younger, relatively healthy out488 16 March 2004 Annals of Internal Medicine Volume 140 Number 6

TO THE EDITOR: The article by Kovacs and colleagues (1) presented data supporting the strategy of starting warfarin therapy using a dosage of 10 mg/d for 2 days, with dose adjustment starting on day 3 using a 10-mg initiation nomogram. We feel this approach is unsafe and, if widely used, will result in untoward outcomes. Using the widely accepted model of Theofanous and Barile for warfarin pharmacokinetics (2), we modeled the use of this 10-mg nomogram with a validated Bayesian forecasting program (3). For an elderly patient who responds with an INR of 3.0 at noon on day 3 after 2 doses of 10 mg, the predicted steady-state dosage is 1.8 mg/d, a common dosage. When the 10-mg nomogram is applied, the INR values on days 4 and 5 will be 4.2 and 4.6, respectively. These INR values are high but tolerable. However, if there is any miscommunication and the patient takes 10 mg on day 3, the INR the next day will be greater than 7.0. Magnifying this problem is the fact that 8% of the population has a genotype for cytochrome 2C9 that metabolizes warfarin very slowly (4). A patient with this genotype could easily achieve an INR of 6.0 at noon on day 3 (steady-state dose 1.1 mg/d), and if no warfarin is given, the modeled INR values the next 2 days exceed 9.0! One has to ask, why the rush? Use of this protocol may shorten the time it takes to achieve a therapeutic INR, but the cost will be dear for warfarin-sensitive patients. If one starts with an aggressive dose of 10 mg, it should be given early in the morning on day 1 and the INR should be measured approximately 24 hours after the initial dose (5). The subsequent dose should then be determined by using a different nomogram or computer-assisted dosing (2). Starting with 2 doses of 10 mg is unwise; speed kills.

Richard White, MD William Dager, PharmD University of California, Davis Sacramento, CA 95817
References
1. Kovacs MJ, Rodger M, Anderson DR, Morrow B, Kells G, Kovacs J, et al. Comparison of 10-mg and 5-mg warfarin initiation nomograms together with low-molecular-weight heparin for outpatient treatment of acute venous thromboembolism. A randomized, double-blind, controlled trial. Ann Intern Med. 2003;138:714-9. [PMID: 12729425] 2. Theofanous TG, Barile RG. Multiple-dose kinetics of oral anticoagulants: methods of analysis and optimized dosing. J Pharm Sci. 1973;62:261-6. [PMID: 4686401]
www.annals.org

Letters
3. White RH, Hong R, Venook AP, Daschbach MM, Murray W, Mungall DR, et al. Initiation of warfarin therapy: comparison of physician dosing with computer-assisted dosing. J Gen Intern Med. 1987;2:141-8. [PMID: 3295148] 4. Daly AK, King BP. Pharmacogenetics of oral anticoagulants. Pharmacogenetics. 2003;13:247-52. [PMID: 12724615] 5. Gage BF, Fihn SD, White RH. Management and dosing of warfarin therapy. Am J Med. 2000;109:481-8. [PMID: 11042238]

ple, amiodarone), or who are known to carry a polymorphism in the cytochrome P450 2C9 system that slows the metabolism of S-warfarin (7). Brian F. Gage, MD, MSc Washington University St. Louis, MO 63110
References
1. Kovacs MJ, Rodger M, Anderson DR, Morrow B, Kells G, Kovacs J, et al. Comparison of 10-mg and 5-mg warfarin initiation nomograms together with low-molecular-weight heparin for outpatient treatment of acute venous thromboembolism. A randomized, double-blind, controlled trial. Ann Intern Med. 2003;138:714-9. [PMID: 12729425] 2. Gedge J, Orme S, Hampton KK, Channer KS, Hendra TJ. A comparison of a low-dose warfarin induction regimen with the modied Fennerty regimen in elderly inpatients. Age Ageing. 2000;29:31-4. [PMID: 10690692] 3. OConnell MB, Kowal PR, Allivato CJ, Repka TL. Evaluation of warfarin initiation regimens in elderly inpatients. Pharmacotherapy. 2000;20:923-30. [PMID: 10939553] 4. Roberts GW, Druskeit T, Jorgensen LE, Wing LM, Gallus AS, Miller C, et al. Comparison of an age adjusted warfarin loading protocol with empirical dosing and Fennertys protocol. Aust N Z J Med. 1999;29:731-6. [PMID: 10630656] 5. Harrison L, Johnston M, Massicotte MP, Crowther M, Moffat K, Hirsh J. Comparison of 5-mg and 10-mg loading doses in initiation of warfarin therapy. Ann Intern Med. 1997;126:133-6. [PMID: 9005747] 6. Gage BF, Fihn SD, White RH. Warfarin therapy for an octogenarian who has atrial brillation. Ann Intern Med. 2001;134:465-74. [PMID: 11255522] 7. Gage BF, Eby C, Milligan PE, Banet GA, Duncan JR, McLeod HL. Use of pharmacogenetics and clinical factors to predict the maintenance dose of warfarin. Thromb Haemost. 2004;91:87-94. [PMID: 14691573]

TO THE EDITOR: Kovacs and colleagues (1) recommended a 10-mg

warfarin nomogram for outpatient treatment of acute venous thromboembolism as being superior to a 5-mg regimen (1). I should point out that using the INR as measure of efcacy early in the course of therapy may give one false assurance (2). Factor VII is affected early and will prolong the INR. However, lowering levels of factor VII does not prevent thromboembolism. Lowering levels of factor IX and factor X (the next 2 factors to be altered) does, but this takes longer. I note that 3 of Kovacs and colleagues patients in the 10-mg group had venous thromboembolism by 90 days versus none in the 5-mg group. Although this does not reach statistical signicance (P 0.09), there may be a trend favoring the 5-mg dose with low-molecular-weight heparin given for a longer period. A larger study might have revealed a statistical difference, should one truly exist. I therefore would recommend caution before accepting the 10-mg regimen. Robert Zeller, MD Winthrop University Hospital Mineola, NY 11501
References
1. Kovacs MJ, Rodger M, Anderson DR, Morrow B, Kells G, Kovacs J, et al. Comparison of 10-mg and 5-mg warfarin initiation nomograms together with low-molecular-weight heparin for outpatient treatment of acute venous thromboembolism. A randomized, double-blind, controlled trial. Ann Intern Med. 2003;138:714-9. [PMID: 12729425] 2. Deykin D. Warfarin therapy. 1. N Engl J Med. 1970;283:691-4. [PMID: 4916914]

TO THE EDITOR: The debate rages on! Does loading patients with

TO THE EDITOR: In their randomized, controlled trial, Kovacs and colleagues (1) showed that initiating warfarin therapy by prescribing 10 mg on days 1 and 2 reduced the time required to achieve a therapeutic INR, compared with 5-mg doses. Their ndings parallel the results of a similar trial (2) and show that starting with a higher dose of warfarin elevates the INR more rapidly. However, the risks of using this dosing scheme probably outweigh the benets, particularly among the elderly. The primary benet, achieving a therapeutic INR 1.4 days earlier, is small. In contrast, hemorrhagic complications caused by excessive anticoagulation (3) can have permanent sequelae. Others have reported that the use of an initial 10-mg dose can lead to overdose in elderly patients (2, 4), cause bleeding (3), and rapidly deplete levels of protein C (5). Given these concerns, it would be reassuring if the authors provided the rates of adverse events (including minor bleeding and use of vitamin K) during the rst 2 to 3 weeks of treatment and also showed the distribution of peak INR values in the 2 cohorts, particularly in persons older than 60 years of age. Until more data are available, we advocate that patients beginning warfarin with 10-mg doses have their INR checked 15 to 30 hours after the initial dose (6). We further caution against using initial 10-mg doses in patients who are elderly or petite, who take any drug associated with increased sensitivity to warfarin (for examwww.annals.org

warfarin achieve the therapeutic goal more quickly than starting with a lower dose that approximates a typical maintenance dose? Previous studies suggested that starting with a lower daily dose (5 mg) rather than a loading dose (10 mg) allows more patients to achieve the target INR of 2.0 to 3.0 while avoiding the excessive anticoagulation (INR 3.0) commonly seen with warfarin loading doses (1, 2). The American College of Chest Physicians recommends starting with an average daily maintenance dose of 5 mg (3). However, Kovacs and colleagues (4) demonstrated that 10-mg warfarin initiation achieved the target INR 1.4 days sooner than the 5-mg dose without causing excessive anticoagulation in the rst 4 weeks of therapy. The results of this well-designed study fuel this debate, particularly since it is the largest comparison published to date (1, 2, 4). Before clinicians adopt the strategy proposed by Kovacs and colleagues, however, additional information is required. Warfarin has the potential for drug drug interactions with many commonly prescribed medications (for example, amiodarone, quinolone antibiotics, trimethoprimsulfamethoxazole, uconazole, and metronidazole). Each of the studies comparing these dosing strategies failed to address concomitant medications and the possibility that drug drug interactions may have confounded variables that affected the overall results (1, 2, 4). Providing information about concomitant medications and their effect on the outcomes of the study by Kovacs and colleagues would be useful. In addition, Kovacs and associates compared supratherapeutic INR values (INR 3.0) over a 4-week period but did not report the incidence of this outcome during the initial 7 days of therapy, when
16 March 2004 Annals of Internal Medicine Volume 140 Number 6 489

Letters
earlier comparisons demonstrated excessive anticoagulation with the 10-mg dosing strategy (1, 2, 4). Comparison of the early (within 7 days) excessive anticoagulation would allow more direct comparison with similar studies. Robert J. DiDomenico, PharmD University of Illinois at Chicago Chicago, IL 60612
References
1. Harrison L, Johnston M, Massicotte MP, Crowther M, Moffat K, Hirsh J. Comparison of 5-mg and 10-mg loading doses in initiation of warfarin therapy. Ann Intern Med. 1997;126:133-6. [PMID: 9005747] 2. Crowther MA, Ginsberg JB, Kearon C, Harrison L, Johnson J, Massicotte MP, et al. A randomized trial comparing 5-mg and 10-mg warfarin loading doses. Arch Intern Med. 1999;159:46-8. [PMID: 9892329] 3. Ansell J, Hirsh J, Dalen J, Bussey H, Anderson D, Poller L, et al. Managing oral anticoagulant therapy. Chest. 2001;119:22S-38S. [PMID: 11157641] 4. Kovacs MJ, Rodger M, Anderson DR, Morrow B, Kells G, Kovacs J, et al. Comparison of 10-mg and 5-mg warfarin initiation nomograms together with low-molecular-weight heparin for outpatient treatment of acute venous thromboembolism. A randomized, double-blind, controlled trial. Ann Intern Med. 2003;138:714-9. [PMID: 12729425] randomized, double-blind, controlled trial. Ann Intern Med. 2003;138:714-9. [PMID: 12729425] 2. Crowther MA, Ginsberg JB, Kearon C, Harrison L, Johnson J, Massicotte MP, et al. A randomized trial comparing 5-mg and 10-mg warfarin loading doses. Arch Intern Med. 1999;159:46-8. [PMID: 9892329] 3. Harrison L, Johnston M, Massicotte MP, Crowther M, Moffat K, Hirsh J. Comparison of 5-mg and 10-mg loading doses in initiation of warfarin therapy. Ann Intern Med. 1997;126:133-6. [PMID: 9005747] 4. Ageno W, Turpie AG. Exaggerated initial response to warfarin following heart valve replacement. Am J Cardiol. 1999;84:905-8. [PMID: 10532508]

TO THE EDITOR: In their randomized, double-blind trial, Kovacs

TO THE EDITOR: The results of the study by Kovacs and colleagues

(1) differ from those that we observed in 2 previous randomized trials (2, 3). In those trials, we demonstrated that a 10-mg warfarin loading dose did not hasten the achievement of a therapeutic INR but did cause more overanticoagulation (2, 3). The differences between Kovacs and colleagues results and ours are likely to cause confusion among clinicians who have adopted a lower starting dose of warfarin. The most likely explanation for the differences in the results of our studies is patient selection. Our patients were older, were mostly hospitalized, and in many cases were receiving warfarin for prophylaxis of thrombosis. In contrast, Kovacs and colleagues patients were, on average, 10 years younger than our patients and were all outpatients being treated for acute deep venous thrombosis. On the basis of Kovacs and colleagues ndings, a 5-mg initial dose may be insufcient for otherwise well, young outpatients with deep venous thrombosis. On the other hand, a 10-mg starting dose would excessive for many postoperative patients (4) and is likely to be inappropriate for older patients and those with comorbid conditions. Mark A. Crowther, MD, MSc McMaster University, St. Josephs Hospital Hamilton, Ontario L8N 4A6, Canada Linda Harrison, RN Hamilton Health Sciences Corp. Hamilton, Ontario L8V 1C3, Canada Jack Hirsh, MD, FCCP McMaster University, Henderson Research Centre Hamilton, Ontario L8V 1C3, Canada
References
1. Kovacs MJ, Rodger M, Anderson DR, Morrow B, Kells G, Kovacs J, et al. Comparison of 10-mg and 5-mg warfarin initiation nomograms together with low-molecular-weight heparin for outpatient treatment of acute venous thromboembolism. A
490 16 March 2004 Annals of Internal Medicine Volume 140 Number 6

and colleagues (1) concluded that loading doses of warfarin (10 mg on 2 consecutive days) are more effective than standard dosing (5 mg on 2 consecutive days) in achieving therapeutic INR values among outpatients with deep venous thrombosis who were receiving at least 5 days of therapy with low-molecular-weight heparin. However, the authors have not adequately established the safety of this approach, particularly with respect to avoiding prothrombotic complications of warfarin. Among their studys end points, the authors did not list 2 such sequelae: warfarin-induced skin necrosis and warfarin-induced venous limb gangrene. These complications typically occur when patients receive loading doses of warfarin and have special predisposing circumstances, such as hereditary abnormalities in the protein C anticoagulant pathway (for example, protein C or protein S deciency) or certain acquired hypercoagulability states (for example, heparin-induced thrombocytopenia or cancer-associated disseminated intravascular coagulation) (2, 3). Kovacs and colleagues did not determine whether any of their patients had such coexisting abnormalities. An important reason to avoid loading doses of warfarin even when it may be safe in most patientsis because it is unsafe in the minority of patients who have one of these risk factors for warfarin-induced necrosis, and the presence of these risk factors is generally unknown when warfarin therapy is started. Finally, before physicians decide whether to use loading doses on the basis of Kovacs and colleagues report, they should also be aware of the advice of Bristol-Myers Squibb, the manufacturer of Coumadin: It is recommended that COUMADIN therapy be initiated with a dose of 2 to 5 mg per day with dosage adjustments based on the results of PT [prothrombin time]/INR determinations (4). To use a loading dose nowadays invites the specic criticism that the manufacturers recommendationswhich are aimed at maximizing patient safety have been ignored. Theodore E. Warkentin, MD Hamilton General Hospital Hamilton, Ontario L8L 2X2, Canada
References
1. Kovacs MJ, Rodger M, Anderson DR, Morrow B, Kells G, Kovacs J, et al. Comparison of 10-mg and 5-mg warfarin initiation nomograms together with low-molecular-weight heparin for outpatient treatment of acute venous thromboembolism. A randomized, double-blind, controlled trial. Ann Intern Med. 2003;138:714-9. [PMID: 12729425] 2. Warkentin TE. Thrombotic complications of anticoagulant therapy. In: Colman RW, Hirsh J, Marder VJ, Clowes AW, George JN, eds. Hemostasis and Thrombosis. Basic Principles and Clinical Practice. 4th ed. Philadelphia: Lippincott Williams & Wilkins; 2000:1371-82. 3. Warkentin TE, Elavathil LJ, Hayward CP, Johnston MA, Russett JI, Kelton JG.
www.annals.org

Letters
The pathogenesis of venous limb gangrene associated with heparin-induced thrombocytopenia. Ann Intern Med. 1997;127:804-12. [PMID: 9382401] 4. Physicians Desk Reference. 56th ed. Montvale, NJ: Medical Economics; 2002: 1247.

TO THE EDITOR: Kovacs and colleagues (1) did not discuss the lack of reliability of the INR during the onset of warfarin therapy. The omission of this information is crucial because Kovacs and colleagues study was not powered to detect clinically meaningful outcomes such as recurrent venous thromboembolism or major hemorrhagic events. The INR system does not correlate with the antithrombotic activity of warfarin during the initiation phase because the prolongation of the prothrombin time during warfarin induction primarily reects the depletion of factor VII (2, 3). However, the ability of warfarin to block tissue factorinduced coagulation in animal models is related to the reduction of levels of factors X and II, but not of factor VII (4). Furthermore, when factor X or II levels are restored, the anticoagulant effect of warfarin is negated (4). Harrison and colleagues (3) also compared 5-mg and 10-mg loading-dose warfarin initiation nomograms. As in the study by Kovacs and colleagues, the 10-mg group achieved a therapeutic INR faster than the 5-mg group. In contrast to the study by Kovacs and colleagues, factor X and II levels were measured and the plasma elimination rate of these proteins did not differ between the 5- and 10-mg groups. Thus, a 10-mg warfarin loading dose may enhance the disassociation between the prolongation of the INR and the onset of anticoagulation and may lead to premature discontinuation of overlapping heparin therapy. This issue is especially important in the outpatient setting, where medical surveillance of thrombus extension may be less rigorous than in the hospital.

Kenneth L. McCall, PharmD, BCPS, CACP Texas Tech School of Pharmacy Amarillo, TX 79106
References
1. Kovacs MJ, Rodger M, Anderson DR, Morrow B, Kells G, Kovacs J, et al. Comparison of 10-mg and 5-mg warfarin initiation nomograms together with low-molecular-weight heparin for outpatient treatment of acute venous thromboembolism. A randomized, double-blind, controlled trial. Ann Intern Med. 2003;138:714-9. [PMID: 12729425] 2. Loeliger EA, Van Der Esch B, Mattern MJ, Hemker HC. The biological disappearance rate of prothrombin, factors VII, IX and X from plasma in hypothyroidism, hyperthyroidism, and during fever. Thromb Diath Haemorrh. 1964;10:267-77. [PMID: 14113369] 3. Harrison L, Johnston M, Massicotte MP, Crowther M, Moffat K, Hirsh J. Comparison of 5-mg and 10-mg loading doses in initiation of warfarin therapy. Ann Intern Med. 1997;126:133-6. [PMID: 9005747] 4. Zivelin A, Rao LV, Rapaport SI. Mechanism of the anticoagulant effect of warfarin as evaluated in rabbits by selective depression of individual procoagulant vitamin Kdependent clotting factors. J Clin Invest. 1993;92:2131-40. [PMID: 8227329]

Much press has been given to the fact that the INR may predominantly reect factor VII levels during initiation of oral anticoagulant therapy, but with concomitant primary therapy (that is, therapy with low-molecular-weight heparin), no clinical consequences have been reported. Perhaps our nomogram should not be employed if concomitant low-molecular-weight heparin therapy is not used, but there are no data to support this. Despite theoretical concerns using Bayesian modelling, we did not experience a greater rate of critically higher INRs in patients treated using the 10-mg nomogram, and there are no data showing that patients with cytochrome 2C9 require lower initiating doses despite requiring lower maintenance doses of warfarin. Speed may kill in automobiles, but no data prove that it kills with warfarin. In our paper, we already stated that we cannot exclude a safety difference between the 5-mg and 10-mg warfarin initiation nomograms. It is probable that a study designed to detect differences in bleeding and recurrence would require a prohibitive sample size. We agree with Crowther and colleagues that the main reason for the differences between our study and theirs may be the patient samples. Until now, it was not apparent that they used primarily an inpatient population (including postoperative orthopedic prophylaxis) in their study. When we tried their nomogram in our outpatients, it was our impression that therapeutic anticoagulation was delayed, which was one impetus for our study. We disagree with the concerns of Dr. Warkentin. Warfarininduced skin necrosis is extremely rare and can occur in persons with and without protein C deciency. No cases have been seen in our practices despite thousands of treated patients. Previous anecdotal case reports have linked this disorder to loading doses that were much higher than those used in our study or that were administered without concomitant heparin therapy. Moreover, we would argue that by achieving an INR in a quicker, more efcient fashion, the time exposure to heparin is shortened and, hence, the risk for heparin-induced thrombocytopenia is lower, especially when patients are treated only with low-molecular-weight heparin (1). Finally, venous limb gangrene has been associated with warfarin used unopposed in the treatment of heparin-induced thrombocytopenia (2), unlike in our study. Although it is probable that very old or very small patients need lower initiating doses, this is not clear, and we did not exclude these patients. Ideally, there would be a nomogram for every type of patient, but that would not be practical. We believe a simple, standard ordering method will lead to better outcomes than the on-call persons best judgment. We have demonstrated that it is unnecessary to measure INRs every day. We suggest that other clinicians try our nomogram using our eligibility criteria (that is, outpatients of any age with venous thromboembolism, with or without cancer, regardless of medications) and judge for themselves how it works in their practices. Michael J. Kovacs, MD, FRCPC London Health Sciences Centre London, Ontario N6A 4G5, Canada Philip S. Wells, MD, FRCPC, MSc Ottawa HospitalCivic Campus Ottawa, Ontario K1Y 4E9, Canada
16 March 2004 Annals of Internal Medicine Volume 140 Number 6 491

IN RESPONSE: We thank all of the correspondents for the interest in our article, which has initiated a strong and emotional debate. Given the strong shift, however, of treatment of acute thromboembolism to an outpatient setting using low-molecular-weight heparin, it is imperative to have a simple means to achieve therapeutic oral anticoagulation in an efcient, timely, and predictable fashion.
www.annals.org

Letters
References
1. Warkentin TE, Levine MN, Hirsh J, Horsewood P, Roberts RS, Gent M, et al. Heparin-induced thrombocytopenia in patients treated with low-molecular-weight heparin or unfractionated heparin. N Engl J Med. 1995;332:1330-5. [PMID: 7715641] 2. Warkentin TE, Elavathil LJ, Hayward CP, Johnston MA, Russett JI, Kelton JG. The pathogenesis of venous limb gangrene associated with heparin-induced thrombocytopenia. Ann Intern Med. 1997;127:804-12. [PMID: 9382401]

CLINICAL OBSERVATIONS
Angiotensin-Converting Enzyme Inhibitors for Secondary Erythrocytosis
TO THE EDITOR: Background: Secondary erythrocytosis is related to

Occupational Exposure to Bloodborne Pathogens

TO THE EDITOR: Dr. Seiberts painful story of occupational HIV

exposure (1) should resonate with all clinicians. In our program, which serves 2 teaching hospitals, we have evaluated more than 5000 employees with body uid exposures since 1988, using Centers for Disease Control and Prevention guidelines (2). Specic measures we have found useful to minimize health care worker anxiety and facilitate postexposure prophylaxis include 1) orienting hospital staff to report body uid exposures immediately to the occupational medicine clinic or the emergency department, 2) implementing triage protocols to minimize waiting times, 3) providing 24-hour consult coverage of occupational medicine by experienced physicians, 4) offering condential HIV testing for health care workers through the occupational medicine clinic, 5) providing starter packs of antiretroviral agents to minimize treatment delays, 6) using individualized follow-up to rapidly provide health care workers with laboratory results on their source patients and themselves, 7) monitoring health care workers for side effects during and after postexposure prophylaxis, and 8) facilitating condential follow-up testing for HIV and hepatitis virus infection if indicated. These interventions, along with accurate assessment of risk magnitude, timely source-patient testing, and appropriate consultation for questions of HIV drug resistance, have increased health care workers willingness to seek evaluation and treatment immediately after exposures. Denitive postexposure prophylaxis, if needed, is generally begun within 2 hours of exposure. The anxiety and risk associated with body uid exposures can be decreased by accessing a hospitals dedicated treatment program.

Amy J. Behrman, MD Hospital of the University of Pennsylvania, University of Pennsylvania Health System Philadelphia, PA 19104

David A. Allan, MD, PhD Presbyterian Medical Center, University of Pennsylvania Health System Philadelphia, PA 19104

References
1. Seibert C. Stuck. Ann Intern Med. 2003;138:765-6. [PMID: 12729434] 2. Updated U.S. Public Health Service Guidelines for the Management of Occupational Exposures to HBV, HCV, and HIV and Recommendations for Postexposure Prophylaxis. Centers for Disease Control and Prevention. 29 June 2001. Available at www.cdc.gov/mmwr/preview/mmwrhtml/rr5011a1.htm.
492 16 March 2004 Annals of Internal Medicine Volume 140 Number 6

an inappropriate secretion of erythropoietin. It has been reported in the setting of chronic hypoxia, neoplasia, and renal transplantation and more rarely in patients with cystic renal diseases. Angiotensinconverting enzyme (ACE) inhibitors and angiotensin II receptor antagonists have proved to be an alternative to phlebotomy and hydroxyurea in patients with secondary erythrocytosis related to renal transplantation (1), chronic pulmonary disease (2), and altitude (3). Objective: To report what we believe to be the rst successful use of ACE inhibitors in patients with erythrocytosis associated with cystic kidney diseases. Case Reports: Patient 1, a 55-year-old man, had unilateral renal cystic disease. Patient 2, a 21-year-old man, had bilateral renal cysts and a nonfunctioning cystic left kidney (Figure). At study inclusion, the patients had hematocrits of 0.53 and 0.63 and increases of 25% and 66% in erythrocyte mass, respectively. Renal function and blood pressure were normal in both patients. The patients underwent selective catheterization, and blood samples were obtained from renal veins. These showed an increased erythropoietin concentration in the vein of the cystic kidney compared with the contralateral kidney (4 times higher in patient 1 and 3 times higher in patient 2). In patient 2, the erythropoietin concentration in the cystic uid was 210 mU/mL. Patient 2 initially presented with gastrointestinal bleeding and had 2 phlebotomy sessions before receiving an ACE inhibitor. Lisinopril, 20 mg/d, led to a 7% and 12% decrease in hematocrit in patients 1 and 2, respectively (Figure). Patient 1 reported cough during ACE inhibitor treatment and was switched to an angiotensin II receptor antagonist (irbesartan, 300 mg/d). Hematocrit increased slightly but remained below 0.5. After 9 and 20 months of follow-up, respectively, erythrocytosis resolved in both patients. Discussion: Cystic renal diseases are a relatively rare cause of erythrocytosis. In this setting, erythrocytosis is attributed to inappropriate secretion of erythropoietin by cystic cells, as suggested by the high erythropoietin concentration found in the cystic uid in patient 1. However, it remains unclear why only a small subset of patients with cystic kidney diseases develop erythrocytosis. A critical volume of cysts may be required for signicant erythropoietin secretion. The latter may be related to renal ischemia or to abnormal regulation of erythropoietin synthesis in cystic cells. Moreover, increased responsiveness of erythroid progenitors to erythropoietin has been reported in some patients with secondary erythrocytosis (4). The sensitivity of erythroid progenitors to erythropoietin was found to be normal in patient 2 (data not shown). In both of our patients, treatment with ACE inhibitors led to a signicant decrease in hematocrit that began as early as one month after the start of treatment. Hematocrit reached a plateau in both patients at 3 months. Of note, in patient 1, the decrease in hematocrit was more marked with ACE inhibitor therapy than with angiotensin II receptor antagonist therapy, as previously reported in patients who developed erythrocytosis after renal transplantation (5). Angiotensin II enhances erythropoiesis through the angiotensin
www.annals.org

Letters
Figure. Computed tomography scan and evolution of
hematocrit.

Fadi Fakhouri, MD Jean-Pierre Gru nfeld, MD Olivier Hermine, MD, PhD Richard Delarue, MD Ho pital Necker 75015 Paris, France
References
1. Vlahakos DV, Marathias KP, Agroyannis B, Madias NE. Posttransplant erythrocytosis. Kidney Int. 2003;63:1187-94. [PMID: 12631334] 2. Vlahakos DV, Marathias KP, Kosmas EN. Losartan reduces hematocrit in patients with chronic obstructive pulmonary disease and secondary erythrocytosis [Letter]. Ann Intern Med. 2001;134:426-7. [PMID: 11242510] 3. Plata R, Cornejo A, Arratia C, Anabaya A, Perna A, Dimitrov BD, et al. Angiotensin-converting-enzyme inhibition therapy in altitude polycythaemia: a prospective randomised trial. Lancet. 2002;359:663-6. [PMID: 11879862] 4. Mrug M, Stopka T, Julian BA, Prchal JF, Prchal JT. Angiotensin II stimulates proliferation of normal early erythroid progenitors. J Clin Invest. 1997;100:2310-4. [PMID: 9410909] 5. Wang AY, Yu AW, Lam CW, Yu LM, Li PK, Goh J, et al. Effects of losartan or enalapril on hemoglobin, circulating erythropoietin, and insulin-like growth factor-1 in patients with and without posttransplant erythrocytosis. Am J Kidney Dis. 2002;39: 600-8. [PMID: 11877580] 6. Glicklich D, Burris L, Urban A, Tellis V, Greenstein S, Schechner R, et al. Angiotensin-converting enzyme inhibition induces apoptosis in erythroid precursors and affects insulin-like growth factor-1 in posttransplantation erythrocytosis. J Am Soc Nephrol. 2001;12:1958-64. [PMID: 11518790]

Successful Treatment of Diffuse Alveolar Hemorrhage with Activated Factor VII

TO THE EDITOR: Background: Life-threatening diffuse alveolar hem-

Top. Renal computed tomography showed a nonfunctioning cystic left kidney in patient 2. Bottom. Change in hematocrit after the introduction of an angiotensin-converting enzyme (ACE) inhibitor. * The initial decrease in hematocrit in patient 2 resulted from blood loss caused by gastrointestinal bleeding and 2 phlebotomy sessions. Patient 1 reported cough during treatment with the ACE inhibitor and was switched to an angiotensin II receptor antagonist.

I receptor that is present on early erythroid progenitors (4). However, the mechanisms by which the blockade of the reninangiotensin system has a negative impact on erythropoiesis remain only partially understood and may include an increase in apoptosis of erythroid progenitors; a decrease in plasma levels of insulin-like growth factor I; and an increase in the plasma level of N-acetyl-serylaspartyl-lysyl-proline (Ac-SDKP), a physiologic inhibitor of erythropoiesis (1, 6). The last mechanism is specic to ACE inhibitors and may explain the more pronounced inhibition of erythropoiesis with these drugs compared with angiotensin II receptor antagonists. Conclusion: Angiotensin-converting enzyme inhibitors may prove to be an optimal treatment for secondary erythrocytosis associated with cystic renal diseases.
www.annals.org

orrhage has been successfully treated with recombinant-activated human factor VII (rFVIIa) in 3 different conditions: 2 small-vessel vasculitisassociated diseases and bone marrow transplantation (1). Diffuse alveolar hemorrhage was conrmed by bronchoscopy demonstrating progressively bloodier aliquots of bronchoalveolar lavage uid. Objective: To investigate the therapeutic efcacy of rFVIIa for diffuse alveolar hemorrhage. Case Reports: In case 1, diffuse alveolar hemorrhage, with antineutrophil cytoplasmic autoantibodyassociated microscopic polyangiitis, occurred 3 weeks after a viral illness in a 53-year-old man. Progressive anemia and hypoxia prompted intubation within 24 hours of admission despite therapy with 2 plasma exchanges, intravenous methylprednisolone sodium succinate, 500 mg; mycophenolate; and cyclophosphamide. The patient was near death from hypoxia, and rFVIIa, 120 g/kg of body weight, was administered. This dose was repeated twice at 3-hour intervals. Within minutes of the rst dose, oxygenation improved. The patient was discharged on room air with normal renal function and was receiving immunosuppressive therapy. In case 2, diffuse alveolar hemorrhage occurred in a 25-year-old man with systemic lupus erythematosus, the antiphospholipid antibody syndrome, and nephritis after chest tube drainage of a bloody pleural effusion. His anemia worsened despite treatment with methylprednisolone sodium succinate, daily blood and platelet transfusions, and plasma exchange. He was intubated for diffuse alveolar
16 March 2004 Annals of Internal Medicine Volume 140 Number 6 493

Letters
Figure. Formation of the tenase complex and the prothrombinase complex. thrombosis in the antiphospholipid antibody syndrome, no pathologic clotting was observed with rFVIIa therapy (3). Factor VIIa initiates coagulation by interacting with factor X and factor IX (Figure) (4). Zymogen factor X and factor IX activation classically occur by interaction with an FVIIatissue factor complex at injury sites (5). However, zymogen factor IX activation by rFVIIa also occurs in a concentration-dependent fashion on activated platelet surfaces, that is, independently of tissue factor (Gabriel DA, Monroe MD, Li X, Roberts HR. The effect of a high concentration of recombinant factor VIIa (rhFVIIa) on the activation of plateletbound zymogen factor IX. 2002. Unpublished data). Tissue factor independent hemostasis by FVIIa probably results from direct activation of platelet-bound zymogen factor IX, which combines with factor VIIIa to form a tenase complex (that is, factor VIIIa/factor IXa/factor X). This complex in turn activates zymogen factor X to generate thrombin and clot. Therefore, pharmacologic concentrations of rFVIIa may also directly activate zymogen factor IX on platelets at bleeding sites in diseases that generally have little apparent tissue damage, for example, the Goodpasture syndrome. Thrombin modulators, such as antithrombin III and protein C, may limit deleterious clot propagation at rFVIIas site of action. Concentrations of thrombin modulators should be considered before using rFVIIa, especially in patients with liver disease. Conclusion: The rapid effect of rFVIIa prolonged life in 3 patients with diffuse alveolar hemorrhage of different causes. No adverse events were observed. David Henke, MD, MPH Ronald J. Falk, MD Don A. Gabriel, MD, PhD University of North Carolina at Chapel Hill Chapel Hill, NC 27599
Acknowledgments: The authors thank Dr. Harold Roberts for help with the Discussion section and Jean Brown and Melissa Lawrence for editorial and secretarial assistance. They also thank the clinical teams who cared for the patients. References
1. Fraser RS, Muller NL, Colman N, Pare ED, Bralow L, eds. Fraser and Pares Diagnosis of Diseases of the Chest. 4th ed. Philadelphia: WB Saunders; 1999. 2. Hogan SL, Nachman PH, Wilkman AS, Jennette JC, Falk RJ. Prognostic markers in patients with antineutrophil cytoplasmic autoantibody-associated microscopic polyangiitis and glomerulonephritis. J Am Soc Nephrol. 1996;7:23-32. [PMID: 8808106] 3. Brandt JT, Triplett DA, Alving B, Scharrer I. Criteria for the diagnosis of lupus anticoagulants: an update. On behalf of the Subcommittee on Lupus Anticoagulant/ Antiphospholipid Antibody of the Scientic and Standardisation Committee of the ISTH. Thromb Haemost. 1995;74:1185-90. [PMID: 8560433] 4. Roberts HR, Monroe DM, Oliver JA, Chang JY, Hoffman M. Newer concepts of blood coagulation. Haemophilia. 1998;4:331-4. [PMID: 9873750] 5. Lawson JH, Mann KG. Cooperative activation of human factor IX by the human extrinsic pathway of blood coagulation. J Biol Chem. 1991;266:11317-27. [PMID: 2040636]

Infused recombinant-activated human factor VII (rFVIIa) may initiate hemostasis by a tissue factor (TF) dependent mechanism or by a TFindependent mechanism. Classically, TF expressed at an injury site activates factor VIIa (FVIIa). Then, FVIIa/TF activates both factor IX (FIX) and factor X (FX). Factor IXa (FIXa) and factor Xa (FXa) lead to stable thrombin formation by the assembly of the tenase complex (factor VIIIa [FVIIIa]/FIXa/FX) and the prothrombinase complex (factor Va [FVa]/ FXa/factor II [FII]) on the surface of activated platelets. Infused rFVIIa may also initiate hemostasis by directly interacting with FIX on activated platelet surfaces. FIIa factor IIa; FXIa factor XIa.

hemorrhage. Therapy with rFVIIa, 90 g/kg, improved oxygenation in minutes. This dose was repeated for the next 2 days. Diffuse alveolar hemorrhage and bloody effusion did not recur after the rst dose. The patient was discharged on room air and was receiving immunosuppressive therapy. In case 3, diffuse alveolar hemorrhage requiring intubation occurred after unrelated matched allogeneic bone marrow transplantation in a 28-year-old man with acute leukemia. The hemorrhage was refractory to treatment with methylprednisolone sodium succinate, 500 mg/m2, and platelet transfusions. Diffuse alveolar hemorrhage occurred twice on day 49 after bone marrow transplantation. The patient received rFVIIa, 120 g/kg, and then 180 g/kg 6 hours later when diffuse alveolar hemorrhage recurred. Diffuse alveolar hemorrhage occurred again on day 54 and was treated with rFVIIa, 90 g/kg. Oxygenation improved within minutes of each use of rFVIIa. The patient was extubated but later died of leukemia. Discussion: Diffuse alveolar hemorrhage is associated with high morbidity and mortality. In bone marrow transplantation, high-dose steroids reduce mortality from approximately 90% to 70% (1). In small-vessel vasculitis, diffuse alveolar hemorrhage is the single best predictor of death (2). Diffuse alveolar hemorrhage develops in 12% to 30% of patients with microscopic polyangiitis, with an early mortality rate of approximately 30% (1). Pulmonary hemorrhage carries the greatest risk for death early in the course of antineutrophil cytoplasmic autoantibodypositive small-vessel vasculitis (2). Diffuse alveolar hemorrhage in systemic lupus erythematosus is associated with nephritis and the antiphospholipid antibody syndrome, conditions seen in case 2. Despite the increased risk for
494 16 March 2004 Annals of Internal Medicine Volume 140 Number 6

Capecitabine Induces Severe Angina-Like Chest Pain

TO THE EDITOR: Background: The cardiovascular side effects of

uoropyrimidines have been extensively reported (1). However, cardiotoxicity associated with capecitabine, a novel oral uoropyrimiwww.annals.org

Letters
dine prodrug increasingly used in advanced breast, colorectal, and gastric cancer, has been infrequently described. Objective: To describe 3 cases of severe angina-like chest pain possibly or probably related to capecitabine treatment. These cases were reported to the Netherlands Pharmacovigilance Centre. Case Reports: Patient 1, a 44-year-old woman in good clinical condition, was receiving chemotherapy for advanced ovarian cancer. When her disease progressed, she was treated with 1500 mg of capecitabine twice daily. On the third day of treatment, she developed severe, reversible angina-like chest pain. Treatment with capecitabine was discontinued, and symptoms of chest pain fully disappeared within 2 days. Patient 2, a 60-year-old man with metastatic colon cancer, received chemotherapy consisting of intravenous irinotecan, 250 mg/m2 once every 3 weeks, plus capecitabine, 1000 mg/m2 twice per day for 14 days. Two days after capecitabine therapy was started, he developed continuous, severe chest pain that lasted for 1 hour and radiated to the left arm. He was also nauseated. Treatment with capecitabine was stopped, and the patient was rechallenged 4 days later. Two days after the rechallenge, he again developed severe chest pain. Electrocardiographic monitoring showed signicant but reversible ST-segment abnormalities indicative of myocardial ischemia. Treatment with nitrates was initiated, and capecitabine was withdrawn. Chemotherapy was continued 3 days later with irinotecan, 125 mg/m2, plus a 1-hour bolus infusion of uorouracilleucovorin (500 mg/m2 of uorouracil and 20 mg/m2 of leucovorin). Chest pain did not recur. Patient 3, a 68-year-old man with colon cancer, was prescribed 1000 mg/m2 of capecitabine twice daily. He was supposed to receive the drug for 14 days every 3 weeks but developed severe chest pain and severe nausea on day 3. The patient was pale but hemodynamically stable. Electrocardiography and cardiac ultrasonography showed a large myocardial infarction and hypokinesia, and troponin I and creatine kinaseMB levels were substantially elevated. Several hours later, the patient developed major complex tachycardia and hypotension; he died within 3 days of onset of angina pectoris. In summary, all 3 patients developed chest pain 48 to 72 hours after initiation of capecitabine therapy. In the second patient, identical chest pain recurred after rechallenge with capecitabine, but symptoms were not observed during a subsequent bolus infusion of uorouracil during co-treatment with isosorbidemononitrate and nifedipine. Discussion: Several factors can predispose patients to angina-like chest pain induced by uoropyrimidines, especially previous chest radiation, history of cardiovascular disease, and concurrent treatment with other cardiotoxic agents (for example, anthracyclines) (2). However, we could not identify predisposing factors in any of our patients. In the database of the World Health Organization Monitoring Centre in Uppsala, Sweden, angina pectoris is reported statistically signicantly more frequently in association with capecitabine (reporting odds ratio, 8.35 [95% CI, 4.82 to 14.44]) and uorouracil (reporting odds ratio, 9.40 [CI, 7.76 to 11.38]) than with other drugs in the database. For uorouracil, cardiotoxicity is more common after 96-hour continuous infusion than after bolus administration (7.6% vs. 2%). Oral capecitabine given for 14 days is thought to have toxic effects similar to those associated with continuous infusion of uorouracil (3). Conclusion: In view of the increasing use of capecitabine and the severity of the possible associated symptoms, it is important to inform patients about the risk for angina-like chest pain. Symptoms will most likely occur within 2 to 3 days after capecitabine therapy is started. If chest pain is reported, capecitabine treatment must be stopped immediately and the patient should be monitored closely until the pain resolves. Patients who develop angina-like chest pain should not be retreated with capecitabine. Isa E.L.M. Kuppens, PharmD Henk Boot, MD, PhD Jos H. Beijnen, PharmD, PhD Jan H.M. Schellens, MD, PhD The Netherlands Cancer Institute 1066 CX Amsterdam, the Netherlands Jerry Labadie, MD LAREB 5237 MH sHertogenbosch, the Netherlands
References
1. Becker K, Erckenbrecht JF, Haussinger D, Frieling T. Cardiotoxicity of the antiproliferative compound uorouracil. Drugs. 1999;57:475-84. [PMID: 10235688] 2. Pai VB, Nahata MC. Cardiotoxicity of chemotherapeutic agents: incidence, treatment and prevention. Drug Saf. 2000;22:263-302. [PMID: 10789823] 3. Bertolini A, Flumano M, Fusco O, Muffatti A, Scarinci A, Pontiggia G, et al. Acute cardiotoxicity during capecitabine treatment: a case report. Tumori. 2001;87:200-6. [PMID: 11504378]

www.annals.org

16 March 2004 Annals of Internal Medicine Volume 140 Number 6 495

Pharmacologic Management of Migraine

TO THE EDITOR: Snow and colleagues (1) provided clinical guide-

lines for the management of migraine by primary care physicians. Although these guidelines generally follow those published by the U.S. Headache Consortium (2), several issues relating to their advice may result in suboptimal care in clinical practice. All recently published evidence-based guidelines for migraine management, including those of the U.S. Headache Consortium (2), the U.S. Primary Care Network (3), and the U.K. Migraine in Primary Care Advisors (4), provide evidence for and recommend a strategy of individualized care for migraine. The therapy is chosen on the basis of the historic pattern of attack characteristics and the severity of symptoms. Patients with consistent patterns of mild to moderate attacks may be effectively treated with aspirin or nonsteroidal anti-inammatory drugs (NSAIDs) but in all likelihood take such drugs without the aid of a medical provider. Patterns of migraine attacks that consistently become moderate to severe frequently do cause patients to seek medical attention, and almost inevitably there is a history of failed intervention with aspirin and over-thecounter NSAIDs. These attacks often require migraine-specic therapies, such as triptans or dihydroergotamine. The evidence for the efcacy of triptans is much more established than Snow and colleagues suggested, and the numbers of trials referenced for specic triptans in the guidelines were an underestimate. Many of the studies cited to support nontriptans did not use International Headache Society criteria, included small numbers of patients, and had ill-dened efcacy points. For example, studies cited on tolfenaminic acid and ibuprofen were conducted in the mid-1980s, well before adoption of International Headache Society criteria. Abundant data conrm the signicant penalty in efcacy when effective intervention is delayed. These studies, not reviewed by Snow and colleagues, uniformly show that delaying intervention until headache pain is moderate to severe substantially reduces 2- and 4-hour pain-free efcacy. Despite the paucity of modern studies on NSAIDs and substantial evidence supporting early intervention with triptans both in terms of efcacy and pharmacoeconomics, Snow and colleagues recommended NSAIDs as rst-line acute therapy for all migraine attacks. We feel that the tailoring of therapies to individual attack severity is more likely to result in a satisfactory outcome than routine use of a single therapy, for several reasons. Migraine is a heterogeneous condition, both among individual sufferers and within separate attacks in the same sufferer. Most patients with migraine use over-the-counter medications such as NSAIDs for their attacks and consult their physicians only after such therapies have failed. Withdrawal from medical care is a major problem in migraine management, and 50% of sufferers or fewer visit their physicians for care (5). The use of ineffective rst-line therapies may play a signicant role in dissatisfaction with, and lapsing from, care. The authors also briey discussed the merits of the step care and stratied care approaches to migraine care, concluding that there is no clear evidence as to which one is more effective. However, clinical and pharmacoeconomic evidence from a large clinical study shows that stratied care is clearly the optimal strategy (6). The Disability in Strategies of Care Study compared step care between attacks, step care within attacks (sometimes known as staged care), and stratied
E-496 American College of Physicians

care for the treatment of 6 migraine attacks with an aspirinmetoclopramide combination or zolmitriptan. Headache relief rates were signicantly superior in the stratied care group than in either of the step care groups. Many of the patients treated with a step care approach ultimately took a dose of zolmitriptan and were unnecessarily delayed from doing so. Overall, the total mean cost of care over 6 attacks was lower in the stratied care group than in the other 2 groups. Stratied care on its own does not guarantee the success of acute migraine therapies, and rescue medications may be necessary (2 4), although they are less likely to be required with a stratied approach. Snow and colleagues suggested the use of an opioid or butalbital at home for this purpose. Virtually no evidence supports butalbital in this role, yet substantial evidence incriminates both of these medications in medication overuse headache. In our opinion, without clearly dened limits on the use of rescue therapy, these are potentially dangerous options. More appropriate rescue medications may include a second dose of the rst-line therapy, or fast-acting drugs such as nasal spray and subcutaneous injection formulations of triptans or dihydroergotamine. Symptomatic medications are sometimes useful for migraine when headache or other symptoms are intractable, but these should be provided under carefully monitored conditions, preferably in the physicians ofce or in an emergency department. If the correct formulation of abortive medication is made available, antiemetics are not generally required for migraine, except as rescue medication if nausea or vomiting becomes intolerable. We applaud the initiative from the American Academy of Family Physicians and the American College of Physicians to provide quality guidelines for migraine management for the primary care physician. However, fundamentally, we feel that clinical need should direct scientic evidence. Most of the class A evidence cited was never designed to instruct physicians in how to best care for patients with headache, and it is ironic that these lavish studies are now being promoted as evidence-based practice measures. We feel that the reservations we have outlined are signicant and encourage the publication of revised guidelines to include more clinical input and the latest research into individualized care, stratied care, and the use of rescue medication. Andrew J. Dowson, MB The Kings Headache Service, Kings College Hospital London SE5 9RS, United Kingdom Roger K. Cady, MD Headache Care Center and Primary Care Network Springeld, MO 65804 Stewart J. Tepper, MD New England Center for Headache Stamford, CT 06902 Timothy R. Smith, MD Mercy Health Research St. Louis, MO 63017 Gary Shapero, MD Markham Headache and Pain Treatment Center Unionville, Ontario L3R OW7, Canada

References
1. Snow V, Weiss K, Wall EM, Mottur-Pilson C. Pharmacologic management of acute attacks of migraine and prevention of migraine headache. Ann Intern Med. 2002;137: 840-9. [PMID: 12435222] 2. Silberstein SD. Practice parameter: evidence-based guidelines for migraine headache (an evidence-based review): report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology. 2000;55:754-62. [PMID: 10993991] Available at www.neurology.org/cgi/reprint/55/6/754.pdf. 3. Bedell AW, Cady RK, Diamond ML, Farmer KU, Friesen CP, Johnson JM, et al. Patient-Centered Strategies for Effective Management of Migraine. Springeld, MO: Primary Care Network; 2000. 4. Dowson AJ, Lipscombe S, Sender J, Rees T, Watson D. New guidelines for the management of migraine in primary care. Curr Med Res Opin. 2002;18:414-39. [PMID: 12487508] 5. Lipton RB, Diamond S, Reed M, Diamond ML, Stewart WF. Migraine diagnosis and treatment: results from the American Migraine Study II. Headache. 2001;41:63845. [PMID: 11554951] 6. Lipton RB, Stewart WF, Stone AM, Lainez MJ, Sawyer JP. Stratied care vs step care strategies for migraine: the Disability in Strategies of Care (DISC) Study: a randomized trial. JAMA. 2000;284:2599-605. [PMID: 11086366]

Adrenal Insufciency from Inhaled Corticosteroids

TO THE EDITOR: Background: Inhaled corticosteroids are the main-

level measured 304 nmol/L. Results of a subsequent cosyntropin stimulation test were normal, suggesting a normal underlying hypothalamicpituitaryadrenal axis. Discussion: Inhaled corticosteroids are the treatment of choice for persistent asthma and have an excellent safety prole. However, rare systemic absorption of inhaled steroids can occur, leading to alterations in bone metabolism, adrenal suppression, or other clinical sequelae. Of these, adrenal suppression is of the most immediate concern since adrenal crisis from complete suppression has been reported (3, 4). This should be considered the most serious potential complication related to inhaled corticosteroid use. A meta-analysis by Lipworth (5) concluded that several agents cause dose-related adrenal suppression. This effect was most notable with uticasone and is probably related to the unique topical potency and bioavailability of this agent (5). Conclusion: We report this case to increase awareness that clinically important systemic absorption of inhaled corticosteroids can occur, especially when higher doses of the more topically potent agents are used. Physicians should be aware of this fact when prescribing these medications. This report also further reinforces the importance of tapering inhaled steroids to each patients lowest effective dose in accordance with consensus guidelines (6). Andrew White, MD Scripps Clinic La Jolla, CA 92037 Donald P. Woodmansee, MD Naval Medical Center San Diego San Diego, CA 92106
References
1. Derom E, Van Schoor J, Verhaeghe W, Vincken W, Pauwels R. Systemic effects of inhaled uticasone propionate and budesonide in adult patients with asthma. Am J Respir Crit Care Med. 1999;160:157-61. [PMID: 10390394] 2. Wong CA, Walsh LJ, Smith CJ, Wisniewski AF, Lewis SA, Hubbard R, et al. Inhaled corticosteroid use and bone-mineral density in patients with asthma. Lancet. 2000;355:1399-403. [PMID: 10791523] 3. Todd GR, Acerini CL, Buck JJ, Murphy NP, Ross-Russell R, Warner JT, et al. Acute adrenal crisis in asthmatics treated with high-dose uticasone propionate. Eur Respir J. 2002;19:1207-9. [PMID: 12108877] 4. Todd GR, Wright D, Ryan M. Acute adrenal insufciency in a patient with asthma after changing from uticasone propionate to budesonide. J Allergy Clin Immunol. 1999;103:956-7. [PMID: 10329837] 5. Lipworth BJ. Systemic adverse effects of inhaled corticosteroid therapy: a systematic review and meta-analysis. Arch Intern Med. 1999;159:941-55. [PMID: 10326936] 6. National Asthma Education and Prevention Program. Expert Panel Report: Guidelines for the diagnosis and management of asthma. Update on selected topics2002. J Allergy Clin Immunol. 2002;110:S210-8.

stay of therapy for persistent asthma and are quite safe. Recently, however, studies have demonstrated that these agents can be absorbed systemically, as measured by effects on the hypothalamic pituitaryadrenal axis and bone mineral density (1, 2). Very few patients experience adverse clinical events from this systemic absorption, and this phenomenon has been reported mainly in children (3, 4). Objective: To point out a rare but potentially serious side effect of inhaled steroids, we present a case of adrenal insufciency in an adult secondary to high-dose inhaled uticasone propionate. Case Report: A 47-year-old woman was referred to us after hospitalization for an asthma exacerbation and completion of a prednisone taper. Despite this, she was having persistent symptoms, and spirometry revealed reversible obstruction with an FEV1 of 68% predicted (normal value 80% predicted). Inhaled triamcinolone was switched to inhaled uticasone, 440 g twice daily, and salmeterol, 2 puffs twice daily. On follow-up visits, her symptoms persisted, and spirometry results had slightly worsened. Montelukast was added, and in an effort to prevent further oral steroid courses or hospitalizations, the uticasone dose was increased to 880 g administered twice daily (the highest recommended dose in the package insert), with the eventual goal of tapering to 440 g twice daily. Reevaluation at 1 month revealed only partial symptomatic and spirometric improvement. At her next visit (3 months after she had taken any prednisone and 8 weeks after higher-dose uticasone therapy had begun), the patient noted fatigue, leg muscle weakness, and multiple large bruises. Examination revealed normal vital signs and large ecchymoses on her upper extremities. Studies revealed an 8 a.m. cortisol level of less than 28 nmol/L (normal level, 110 to 607 nmol/L), normal serum electrolyte levels, and normal results on spirometry. Fluticasone was slowly tapered (as instructed in the prescribing information), and the patient was switched to budesonide. Her bruising and other symptoms gradually resolved, and a repeated 8 a.m. cortisol
www.annals.org

Unilateral Pseudogynecomastia: A Novel Work-Related Disease

TO THE EDITOR: Background: The distinction between gynecomastia, which affects breast glandular tissue (1, 2), and pseudogynecomastia, which affects only subcutaneous fat tissue (lipomastia), is essential (3, 4). Pseudogynecomastia can occur bilaterally and unilaterally and is mainly related to obesity.
Annals of Internal Medicine Volume Number E-497

Figure. Extension of soft tissue enlargement to the axilla and

to the patients back.

Objective: To describe a patient who developed pseudogynecomastia of his right breast caused by strenuous physical work over a 10-year period. Methods and Findings: On clinical examination, the patient, a 39-year-old man, had a signicantly enlarged right breast with a horizontal skin fold of 6 to 7 cm. The enlargement extended to the right axilla and to the back (Figure). Palpation of the right breast indicated only soft tissue without any induration. Results of the andrologic examination and tumor marker levels were nonpathologic. Mammography of the right breast displayed lipomatosis with pronounced brous tissue but without any enlargement of breast glandular tissue. Histopathologic examination demonstrated fat lobuli (approximately 30%) combined with a severe increase of brotic tissue (approximately 70%). However, breast glandular tissue could not be detected. Over a period of 10 years, the patient worked in a factory at a lathe, producing lamp reectors. Approximately 70 times per hour, he put a heavy iron bar under his right axilla, pressed it rmly against his right chest wall, and performed a powerful lateral movement. This required extreme physical strength and exposed the right chest area to repetitive mechanical traumatization. Two of the patients colleagues, who worked at the same machine for 3 to 5 years, presented with similar although less marked breast enlargement. Discussion: In our patient, long-term occupational mechanical

pressure to the right chest wall presumably led to traumatic irritation of the subcutaneous tissue and subsequent posttraumatic reactive brosis. Connective tissue brosis is caused by inammation, vascular dysfunction, and metabolic and long-term mechanical injury (5, 6). Pathogenetically, the connective tissue growth factor has recently received much attention as a possible key determinant of progressive brosis (5). It is upregulated in numerous brotic conditions, including those affecting the human skin (6). Our diagnosis of pseudogynecomastia caused by post-traumatic reactive brosis is supported by the histopathologic demonstration of strikingly increased brous tissue, in contrast to the adipose tissue commonly observed in pseudogynecomastia. Because 3 men, the patient and 2 coworkers, were similarly affected, there is a strong correlation between breast enlargement and the patients work. Conclusion: Our ndings suggest the existence of a new workrelated disorder: mechanically induced unilateral pseudogynecomastia caused by severe brosis. The patients working environment should be further investigated so that new preventive measures can be implemented if necessary for persons performing strenuous physical work. S. Hanneken, MD R. Kruse, MD P. Dall, MD T. Ruzicka, MD N.J. Neumann, MD University of Duesseldorf 40225 Duesseldorf, Germany
References
1. Heufelder AE, Leinung MC, Northcutt RC. Holiday gynecomastia [Letter]. Ann Intern Med. 1992;116:877. [PMID: 1567108] 2. Couderc LJ, Clauvel JP. HIV-infection-induced gynecomastia [Letter]. Ann Intern Med. 1987;107:257. [PMID: 3605910] 3. Braunstein GD. Gynecomastia. N Engl J Med. 1993;328:490-5. [PMID: 8421478] 4. Lippens C, Ardies P, Van Steen A. Pseudogynecomastia. J Belge Radiol. 1990;73: 528-9. [PMID: 2277017] 5. Leask A, Holmes A, Abraham DJ. Connective tissue growth factor: a new and important player in the pathogenesis of brosis. Curr Rheumatol Rep. 2002;4:136-42. [PMID: 11890879] 6. Hishikawa K, Oemar BS, Nakaki T. Static pressure regulates connective tissue growth factor expression in human mesangial cells. J Biol Chem. 2001;276:16797803. [PMID: 11278731]

E-498

Annals of Internal Medicine Volume Number

www.annals.org