Introduction

In June 2007, the Endocrinologic and Metabolic Drugs Advisory Committee of the Food and Drug Administration stated that rimonabant does not demonstrate an acceptable benefit/risk profile (1). This statement was largely based on the observed increased risk of neurological and psychiatric adverse effects. Subsequently, Sanofi-Aventis decided to withdraw the New Drug Application in the United States. In Europe, a review of the available data on psychiatric events by the European Medicines Agency resulted in new safety recommendations. These events attributed to the keen interest in the baseline risks of these and other types of adverse events in patients seeking pharmacological treatment for obesity. Therefore, the objective of this study was to assess whether the baseline risk of psychiatric and cardiovascular disease differs between patients with diabetes mellitus who start and those who do not start to use antiobesity drugs.

Methods and Procedures

Data were collected from the General Practice Research Database, a computerized database containing medical records from general practitioners in the United Kingdom. These records contain information about patient's demographics, specific characteristics (e.g., weight, height), symptoms, diagnoses (using Oxford Medical Information System and Read codes), and information about hospital admissions and drug prescriptions (using British National Formulary (BNF) codes). Information is available for almost 4 million person-years originating from about 450 general practices throughout the United Kingdom (2). The recorded information on drug exposure and diagnoses has been validated and proven to be of high quality (3,4). The study was approved by the Scientific and Ethical Advisory Group of the General Practice Research Database.

Between 1 June 1987 and 21 January 2002, all patients with diabetes mellitus (both type 1 and type 2) were included in the study cohort (n = 141,164) (refs. 5,6). A retrospective nested case–control study was performed within this cohort. Starters (cases) were defined as patients with a first prescription for any antiobesity drug (BNF-codes 04.05) within the study period. The date of first prescription for the antiobesity drug was defined as the index date. Nonstarters (controls) did not use antiobesity drugs before the index date. Within the same general practitioner practice, up to six controls were randomly sampled for each case from the study cohort, and they were assigned the same index date. Both starters and nonstarters were eligible for inclusion when they had at least one year of information available before the index date.

Information on characteristics of patient was extracted from the medical records. The following characteristics of patients were determined: number of cardiovascular risk factors at index date (risk factors include 45 years of age, male gender, smoker, type 2 diabetes, hypercholesterolemia, high blood pressure, BMI >25 kg/m²), in the year before index date prescriptions (BNF codes 04.03: antidepressants; BNF codes 04.01.02: anxiolytics, BNF codes 04.02 antipsychotics, BNF code 04 all psychotric medication) or diagnoses related to psychiatric disorders (e.g., depression, anxiety, psychosis), prescriptions or diagnoses related to cardiovascular events (search terms angina pectoris, arrhythmias, cardiac arrhythmias, cardiovascular disease, cardiovascular event, coronary heart disease, heart failure, ischemic heart disease, myocardial infarction, stroke; BNF-codes 02.01–02.09, except 02.08.01: parenteral anticoagulants), diagnoses related to hypertension, and prescriptions (BNF-codes 02.12: lipid-lowering drugs) or diagnoses related to dyslipidemia. In addition, information was collected on gender, BMI (at index date or based on most recent weight measurement and adult height measurement), smoking status (at index date), type of diabetes mellitus (based on both prescriptions (BNF-codes 06.01.01 (insulins) and 06.01.02 (oral antidiabetic drugs) and diagnoses), number of drugs prescribed at the index date, the number of previous dietary advices, and the total number of general practitioner visits in the year before the index date. Descriptive statistics were applied to examine and compare the characteristics of the patient population that started to use antiobesity medication and those that did not. We provided univariate estimates for all associations, because the aim of this study was not whether such an association was causal in nature.

Crude odds ratios (ORs) and 95% confidence intervals (CIs) were calculated.

Results

In our study cohort, 511 starters of antiobesity drugs (cases) and 3,065 nonstarters (controls) were identified. The antiobesity drugs which were used included mainly fluramines (72.2%) and orlistat (21.3%). The number of patients starting to use antiobesity drugs varied over time and can be partly explained by publications concerning safety issues in medical literature (Figure 1). The starters were younger and more frequently female (Table 1). Of the starters, 79.1% had a BMI 27 kg/m², compared to 34.8% of the nonstarters. In addition, 91.8% of the starters did not receive any dietary advice before starting treatment, whereas 5.9% received dietary advice only once before starting treatment. Although the absolute numbers were small, the use of an antiobesity drug was positively associated with receiving dietary advice.

Figure 1.

The number of patients starting to use antiobesity drugs over time.

Full figure and legend (59K)

Table 1 - Baseline characteristics of the study population.

Full table (116K)

Depression (in the year before index date) occurred more frequent among starters of antiobesity drugs (OR 2.2; 95% CI 1.7–2.8) as was anxiety (OR 1.6; 95% CI 1.1–2.4) (Table 1). Of the starters, 19.4% had been prescribed antidepressants in the year before index date compared to 9.7% of the controls. Diagnosis of any psychiatric disorder and the use of any psychotropic medication were both more prevalent among starters. Cardiovascular drug use and diagnoses were slightly, but not significantly, more frequent among the starters. However, in starters of antiobesity drugs, we observed in one out of four (25.2%) multiple cardiovascular risk factors (>4) (e.g., hypertension, dyslipidaemia), compared to 19.0% in the controls (OR 1.7; 95% CI 1.3–2.3).

Discussion

We found that patients with diabetes mellitus, who start to use antiobesity drugs, were more frequently female and younger compared to the patients not starting to use antiobesity drugs. They also had a higher BMI and were more likely to receive a dietary advice, although the absolute number of advices was very low. The baseline risk for psychiatric and cardiovascular disease was higher in starters of antiobesity drugs. The same was found for the use of antidepressants.

It would have been very informative to present a variable which can be used as a general psychiatric risk score (comparable to the Framingham risk score for cardiovascular disease), to use as a prediction model for psychiatric conditions. Unfortunately, no such general risk score for psychiatric disease is available. This might be due to the range of psychiatric disorders being very diverse, all with their own specific risk factors and indicators. In addition, it is known that underdiagnosis of psychiatric disorders is relatively common, which may lead to an underestimation of the psychiatric risk (7). Alternatively, the possible use of various psychotropic medication classes "off-label" our study might overestimate the true prevalence of even the targeted psychiatric problems at baseline. However, we expect that the amount of "off-label" use of psychotropic medication is relatively small compared to the use according to the indication.

Within the study period (1987–2002), a number of safety issues with antiobesity drugs occurred (8). The amphetamine-like compounds (fenfluramine, fluramine) were withdrawn from the market because they were associated with valvular heart disease and primary hypertension (9,10). Our study period covers the use of the newer drugs only to a limited extent. Nevertheless, we do not expect that the characteristics of our patients differ from the patients currently using these drugs. We assume that the characteristics of these patients are comparable to the population eligible for the use of more recently introduced antiobesity treatments associated with psychiatric events, such as rimonabant. The low number of dietary advices may partly be explained by underreporting in the General Practice Research Database, but reinforcement of nonpharmacological recommendations seems warranted (11).

In conclusion, the baseline risk for psychiatric and to a lesser extent cardiovascular disease is higher for patients with diabetes mellitus who start to use antiobesity drugs compared to patients with diabetes mellitus who do not start to use antiobesity drugs. Patients starting to use antiobesity drugs seem to be more vulnerable especially to psychiatric morbidity. This might be independent of the antiobesity drugs itself which may induce additional side effects. These finding urge us to be very careful in interpreting the benefits and risks of such antiobesity drugs, both in terms of preventing possible exposure of drugs associated with psychiatric events in susceptible patients and in the evaluation of causality when a possible drug induced problem occurs.

Disclosure

The authors declared no conflict of interest.

References

REFERENCES

1. Summary Minutes of the Endocrinologic and Metabolic Drugs Advisory Committee meeting: Food and Drug Administration; June 2007.
2. Wood L, Martinez C. The general practice research database: role in pharmacovigilance. Drug Saf 2004;27: 871–881. | Article | PubMed | ISI |
3. Jick SS, Kaye JA, Vasilakis-Scaramozza C et al. Validity of the general practice research database. Pharmacotherapy 2003;23: 686–689. | Article | PubMed | ISI |
4. Lawrenson R, Williams T, Farmer R. Clinical information for research; the use of general practice databases. J Public Health Med 1999;21: 299–304. | Article | PubMed | ChemPort |
5. Verdel BM, Souverein PC, Egberts AC, Leufkens HG. Difference in risks of allergic reaction to sulfonamide drugs based on chemical structure. Ann Pharmacother 2006;40: 1040–1046. | Article | PubMed | ChemPort |
6. van de, Garde EM, Hak E, Souverein PC et al. Statin treatment and reduced risk of pneumonia in patients with diabetes. Thorax 2006;61: 957–961. | Article | PubMed |
7. Prince M, Patel V, Saxena S et al. No health without mental health. Lancet 2007;370: 859–877. | Article | PubMed |
8. Ioannides-Demos LL, Proietto J, Tonkin AM, McNeil JJ. Safety of drug therapies used for weight loss and treatment of obesity. Drug Saf 2006;29: 277–302. | Article | PubMed | ChemPort |
9. 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. | Article | PubMed | ISI | ChemPort |
10. Jick H, Vasilakis C, Weinrauch LA et al. A population-based study of appetite-suppressant drugs and the risk of cardiac-valve regurgitation. N Engl J Med 1998;339: 719–724. | Article | PubMed | ChemPort |
11. Obesity: weight loss without drugs: a balanced diet avoiding high-calorie foods plus exercise. Prescrire Int 2007;16: 162–167.

Acknowledgments

We thank Patrick Souverein for his assistance with data collection and data analysis. This study was accepted for presentation in abstract form at the 7th Annual Meeting of the International Society of Pharmacovigilance in Bournemouth, United Kingdom, 21–24 October 2007. No financial support was received.