Correspondence

Primary Pulmonary Hypertension and Anorectic Drugs

N Engl J Med 1999; 340:480-482February 11, 1999

Article

To the Editor:

The article by Abenhaim et al.1 on the association of anorectic drugs and primary pulmonary hypertension contributed to public concern about the use of these drugs and the subsequent withdrawal of dexfenfluramine from the market. The association between anorectic drugs and primary pulmonary hypertension rested largely on the high odds ratio for the development of pulmonary hypertension with the use of anorectic drugs, the demonstration that the incidence of primary pulmonary hypertension increased with the duration of exposure, and the suggestion that obvious confounding factors were considered.

I am writing to report discrepancies between the article and an earlier, more detailed report submitted to the Australian regulatory authorities.2 The Journal article states that “only exposure to antihypertensive drugs, oral contraceptives, thyroid extracts, and anorexic agents . . . was analyzed.” This statement is not consistent with the fact that approximately 60 different drug groups were analyzed in the earlier report, 6 of which had P values of less than 0.1 (Table 1Table 1Differences in Drug Treatment between Case Patients and Controls with a P Value of 0.09 or Less before the Index Date.).

Specifically, four points were not mentioned in the Journal article. First, the rate of use of psychoanaleptic drugs was higher in the control group. The use of psychoanaleptic drugs usually precludes the use of anorectic drugs, a factor that undermines the validity of the control group and exaggerates the odds ratio for the association of anorectic-drug use with primary pulmonary hypertension.

Second, the rate of use of diuretics was higher among the case patients. The associated crude odds ratio of 3.72 increases to 6.0 if the seven case patients identified in the Journal article as taking compound preparations that “typically contained diuretics” are included. An odds ratio of 6.0 suggests that primary pulmonary hypertension is associated more frequently with diuretic use than with the use of dexfenfluramine (crude odds ratio, 4.0) or total anorectic-drug use (crude odds ratio, 5.8; adjusted odds ratio, 6.3).

Third, the rate of use of antiasthmatic drugs was higher among the case patients, a difference that approached statistical significance (P=0.07), as was the rate of use of antithrombotic drugs, a difference that was statistically significant (P=0.02).

Finally, the Journal article included compound preparations in the analysis of anorectic drugs but not thyroid hormones. The inclusion of thyroid hormone use yields a crude odds ratio of 3.5, suggesting an association between thyroid hormone use and primary pulmonary hypertension and contradicting one of the conclusions of the article.

It remains to be explained why the odds ratios for psychoanaleptic drugs, diuretics, and thrombolytic agents were not discussed in the Journal article and how their use, together with the incorrect assessment of the odds ratio for thyroid hormone use, modifies the assessment of risk. Readers should suspend judgment on the strength and nature of the association between anorectic drugs and primary pulmonary hypertension until the unpublished data in the report have been fully analyzed and presented.

W.J. Louis, M.D.
Austin and Repatriation Medical Centre, Heidelberg VIC 3084, Australia

2 References

1. 1

   Abenhaim L, Moride Y, Brenot F, et al. Appetite-suppressant drugs and the risk of primary pulmonary hypertension. N Engl J Med 1996;335:609-616
   Full Text | Web of Science | Medline

2. 2

   Abenhaim L, Moride Y, Brenot F, et al. IPPHS summary report no. 1. March 1995.
   

Author/Editor Response

The authors reply:

To the Editor: Dr. Louis raises questions about discrepancies between the results published in the Journal 1 and those of an earlier report2 that he reviewed soon after its publication.3

The report of March 1995 to which Dr. Louis refers was a confidential preliminary document that we supplied to the health ministries of several countries, six weeks after collecting data from 220 centers across Europe. We clearly stated in that document that the report was far from complete. The article published in the Journal in August 1996 reported the results of the final report supplied to the health ministries in July 1996.4

In the earlier report, we presented descriptive statistics for more than 150 variables, including the subjects' occupations, nutritional status, use of all drugs, history of diseases, and lifestyles. Drugs with no a priori suspected relation to primary pulmonary hypertension were grouped in 61 categories. Six of them — the ones listed by Dr. Louis — had an unadjusted P value of less than 0.1; this number is exactly the number that would be expected to be observed by chance alone (according to the hypothesis of independence), since it represents 10 percent of the sample (6 of 61). We did not consider the reporting of these unadjusted results for unsuspected risk factors to be scientifically sound.

Nonetheless, in response to the issues raised by Dr. Louis, we calculated the odds ratios for anorectic drugs after controlling for all the factors that he selected and compared them with the results in the Journal article (Table 1Table 1Use of Appetite Suppressants, Thyroid Extracts, and Drugs Not Suspected of Being Associated with Primary Pulmonary Hypertension and Adjusted Odds Ratios for the Risk of Primary Pulmonary Hypertension Obtained from Different Models.). There were no meaningful differences. We also calculated the adjusted odds ratios and 95 percent confidence intervals for all the drugs mentioned by Dr. Louis. None of these factors were significantly associated with primary pulmonary hypertension when confounding was adjusted for by multivariate analysis. Therefore, we advise readers not to draw any conclusions about these drugs, which were identified by “data dredging.”

The low odds ratio for psychoanaleptic drugs (which were almost exclusively antidepressants) cannot be explained on the basis of an overrepresentation of these drugs in our control group, since the rate of use (7.3 percent for a period of approximately two years before the index date) is consistent with the prevalence of the treatment of depression in France in 1991. The low odds ratio for these drugs is likely to be due to chance alone, but this result might prompt studies to explore whether antidepressants actually prevent primary pulmonary hypertension in patients who take anorectic drugs.

Dr. Louis assumed that all the compound preparations contained a diuretic and a thyroid extract. The exact content of the compound preparations was difficult to assess. In our article, we briefly addressed this issue and concluded, from the data available, that all patients who reported using an “appetite-suppressing preparation” should be grouped with appetite-suppressant users, but not with users of diuretics or thyroid extracts. Since then, we have obtained additional information (Table 2Table 2Exposure to Drugs of Patients Who Reported Using Compound Preparations of Appetite Suppressants.), which indicates that there is no basis for a belief that all preparations contained a diuretic or a thyroid extract. With the inclusion of this information, the odds ratios for the use of diuretics and of thyroid extracts are 2.68 (95 percent confidence interval, 0.99 to 7.28) and 0.91 (95 percent confidence interval, 0.15 to 5.33), respectively. These results are only slightly different from the adjusted results shown in Table 1. Six of seven case patients were reported to have used fenfluramine, dexfenfluramine, or both before the index date, which is the date of the onset of symptoms (one was unclear about the date and also reported exposure to an amphetamine-like anorectic drug), and four to have used an amphetamine-like anorectic drug (all of whom also used fenfluramine derivatives). Adding this information to that in Table 3 of our article1 provides an adjusted odds ratio of 6.3 (95 percent confidence interval, 2.5 to 15.6) for the use of fenfluramine derivatives and of 1.3 (95 percent confidence interval, 0.4 to 4.7) for the use of other anorectic drugs for any length of time.

Although we advise caution when interpreting results for individual products in view of the different ways of collecting data, these new data actually reinforce our conclusion that the fenfluramines were the chief drugs involved in the development of primary pulmonary hypertension.

Lucien Abenhaim, M.D., Sc.D.
McGill University, Montreal, QC H3T 1E2, Canada

Stuart Rich, M.D.
Rush University, Chicago, IL 60612

Jacques Benichou, M.D., Ph.D.
University of Rouen, 76031 Rouen Cedex, France

Bernard Bégaud, M.D., Ph.D.
Victor Segalen University, 33076 Bordeaux Cedex, France

4 References

1. 1

   Abenhaim L, Moride Y, Brenot F, et al. Appetite-suppressant drugs and the risk of primary pulmonary hypertension. N Engl J Med 1996;335:609-616
   Full Text | Web of Science | Medline

2. 2

   Abenhaim L, Moride Y, Brenot F, et al. IPPHS summary report no. 1. March 1995.
   

3. 3

   Louis WJ. Draft report on IPPH Study. Received from Institut de Recherches Internationales Servier, France, August 18, 1995.
   

4. 4

   Abenhaim L, Moride Y, Brenot F, et al. The International Primary Pulmonary Hypertension Study (IPPHS): final report. July 26, 1996.
   

Citing Articles (3)

Citing Articles

1. 1

   Lisa L Ioannides-Demos, Joseph Proietto, Andrew M Tonkin, John J McNeil. (2006) Safety of Drug Therapies Used for Weight Loss and Treatment of Obesity. Drug Safety 29:4, 277-302
   CrossRef

2. 2

   Hai Wang, Yuan Tang, Ying-Li Zhang. (2005) Hypoxic Pulmonary Hypertension (HPH) and Iptakalim, a Novel ATP-sensitive Potassium Channel Opener Targeting Smaller Arteries in Hypertension. Cardiovascular Drug Reviews 23:4, 293-316
   CrossRef

3. 3

   Saadia Eddahibi, Marc Humbert, Elie Fadel, Bernadette Raffestin, Michèle Darmon, Frédérique Capron, Gerald Simonneau, Philippe Dartevelle, Michel Hamon, Serge Adnot. (2001) Serotonin transporter overexpression is responsible for pulmonary artery smooth muscle hyperplasia in primary pulmonary hypertension. Journal of Clinical Investigation 108:8, 1141-1150
   CrossRef