Orlistat, a gastrointestinal lipase inhibitor, and sibutramine, a serotonin-noradrenaline reuptake inhibitor, were licensed in the U.K. for the treatment of obesity in December 1998 and May 2001, respectively (1, 2). In clinical trials, orlistat was associated mainly with mild to moderate gastrointestinal adverse events, such as diarrhea, flatus, and oily or fatty stools (3), whereas the most commonly reported adverse events for sibutramine included headache, dry mouth, insomnia, and constipation (4). Hypertension and tachycardia have also been associated with starting sibutramine, and it is recommended that blood pressure and heart rate should be monitored in all patients (every 2 weeks for the first 3 months, monthly up to 6 months, and thereafter at least every 3 months) (2). Concerns over the cardiovascular adverse effects of sibutramine after it was marketed led to regulatory authority safety reviews (5, 6).
Post-marketing surveillance is important for assessing the safety of drugs because the data available at the time of the launch of a product are limited (7). Safety signals are generated from a number of sources including reporting of suspected adverse reactions in individuals to either regulatory authorities or the manufacturers (8), as well as reports in the published literature and epidemiological studies (9). The contribution of published case reports to signal detection has been the subject of recent debate (10, 11, 12).
Prescription-event monitoring (PEM)1 is an observational cohort technique conducted by the Drug Safety Research Unit to monitor the safety of newly licensed drugs prescribed in general practice in England (7). PEM is a hypothesis-generating form of pharmacovigilance; signals for suspected adverse drug reactions (ADRs) are generated by examining reasons for stopping, statistical analysis of events, and clinical interpretation of selected events.
Because of the escalating worldwide prevalence of obesity (13), understanding the safety profiles of orlistat and sibutramine is important, particularly as other anti-obesity agents have been the subject of safety concerns (14). In this study, we summarize the safety profiles of orlistat and sibutramine used in general practice during the immediate post-marketing period. In addition, we have examined published case reports to determine whether any equivalent or similar events were assessed as causally related to the study drug in the PEM studies.
Research Methods and Procedures
The observational cohort studies of orlistat and sibutramine were undertaken using PEM, a technique that has been described elsewhere (11). Patients were identified from dispensed National Health Service prescription data supplied in confidence by the Prescription Pricing Division of the National Health Service Business Services Authority (formerly Prescription Pricing Authority) in England, for prescriptions issued from December 1998 to November 1999 for orlistat and from October 2001 to June 2002 for sibutramine.
Information on patient demographics and outcomes was obtained by sending a questionnaire (green form) to the general practitioner (GP) who had issued the first prescription. For both drugs, green forms were sent at least 6 months after the first dispensed prescription for an individual patient. The green form requested information on the age of the patient, the dose and indication, the start and stop dates of treatment (duration), an opinion on the effectiveness of the treatment, reasons for stopping the drug (if it had been stopped), and any events during or after stopping treatment with the drug, including events suspected to be ADRs. The term “event,” in PEM, is defined as: any new diagnosis, any reason for referral to a consultant or admission to hospital, any unexpected deterioration (or improvement) in a concurrent illness, any suspected drug reaction, any alteration of clinical importance in laboratory values, or any other complaint that was considered of sufficient importance to enter in the patient's notes. The green form questionnaire for sibutramine also included additional questions regarding concurrent conditions (whether the patient had a history of hypertension, ischemic heart disease, and/or diabetes mellitus) and the previous use of other anti-obesity agents. Questionnaires that were returned with no information were classified as “void” and excluded from the cohort.
Events were coded using the Drug Safety Research Unit (DRSU) event dictionary, which is generally similar in structure to MedDRA (Medical Dictionary for Regulatory Activities). Terms used by the GPs are entered into the database (coded); these “Dr. summaries” are grouped into “lower level” terms, which are grouped under “higher level” terms within a system organ class.
Additional questionnaires were sent to the prescribing GP for any events considered medically important based on the safety profile of the drugs at the time the studies were undertaken. Patients who had taken orlistat or sibutramine during or 3 months before becoming pregnant were followed up. Deaths for which no cause was reported were also followed up to try to ascertain the cause of death. If there was no reply to the follow-up questionnaire, one reminder was sent. Individual case reports were assessed for causality using the following criteria: temporality, pharmacological plausibility, clinical and pathological characteristics of the event, and exclusion of other possible causes. Events were classified as probable, possible, unlikely, or not assessable (15).
Incidence densities (IDs) were calculated for the first reports of all of the reported events during treatment within specified time periods and expressed as the number of first reports of an event per 1000 patient-months of exposure. IDs were calculated for the first reports of events that occurred in the first month of treatment (ID1), during months 2 or 3 of treatment (ID2–3), and during the overall treatment period (IDA). The patient-months of exposure could be identified for patients for whom a date of stopping was given or who continued to take the drug throughout the study period. For both drugs, the differences between ID1 and ID2–3 were calculated along with the 99% confidence interval. A significant positive difference between ID1 and ID2–3 shows that the rates of events in month 1 were significantly greater than in months 2 or 3. These time periods were chosen because the prescribing guidelines for both drugs recommend that treatment should be stopped if <5% of the starting body weight is lost within 3 months of starting treatment (1, 2).
A cohort of at least 10,000 patients was considered desirable for both studies. This number would allow 95% certainty that events not observed in the cohort occur less frequently than 1 in 3333 cases (16).
Both studies were conducted in accordance with the International Guidelines for Biomedical Research prepared by the Council for International Organizations of Medical Sciences in collaboration with the World Health Organization (17). PEM also complies with the Guidelines on the Practice of Ethics Committees in Medical Research Involving Human Subjects, as issued by the Royal College of Physicians for records-based research (18). At the time of these studies, PEM was listed in the Multicenter Research Ethics Committee Guidelines for Researchers (Appendix C in reference 19) as a method of inquiry and survey conducted in the interest of the public that does not need reference to an ethics committee (19). These guidelines have now been superseded (see reference 20). The Drug Safety Research Unit is now included in the “Frequently Asked Questions” section of the General Medical Council booklet, “Confidentiality: Protecting and Providing Information,” as “a professional organization that monitors the safety of medicines to which doctors should provide relevant information from patients’ records wherever possible” (20).
Published case reports were identified by searching Medline (Ovid; 1966 to May 2006) and Embase (DialogDatastar; 1974 to May 2006), using the following search terms: “orlistat” or “Xenical” and “sibutramine” or “Meridia” or “Reductil.” These published case reports were examined to determine whether any equivalent or similar events were assessed as causally related to treatment in the PEM studies.
There were 16,021 patients in the orlistat cohort and 12,336 patients in the sibutramine cohort. The response rate was slightly higher for sibutramine than for orlistat, with 56.3% of green forms returned for sibutramine compared with 50.1% for orlistat. The percentages of forms returned that were classified as void in both cohorts were similar (Table 1).
Table 1. . Cohort details and patient demographics
|Number of green forms sent||35,304||23,836|
|Number of green forms returned (% of forms sent)||17,689 (50.1%)||13,418 (56.3%)|
|Number of voids (% of forms returned)||1668 (9.4%)||1082 (8.1%)|
|Number in cohort (% of forms sent)||16,021 (45.4%)||12,336 (51.8%)|
|Male (% of cohort)||3107 (19.4%)||2025 (16.4%)|
|Female (% of cohort)||12,831 (80.1%)||10,229 (82.9%)|
|Sex not specified (% of cohort)||83 (0.5%)||82 (0.7%)|
|Age not specified (% of cohort)||3604 (22.5%)||1382 (11.2%)|
|Median age (years) cohort (IQR)||45 (35–54)||45 (36–55)|
|Median age (years) males (IQR)||49 (39–56)||50 (40–58)|
|Median age (years) females (IQR)||44 (34–53)||44 (35–54)|
The age and gender profiles of patients in both cohorts were similar; the median age was 45 years, and ∼80% were female (Table 1). Orlistat is not intended for use in children (1), but 70 patients younger than 18 years of age were prescribed orlistat, one of whom was under 12 years of age. Contraindicated use of sibutramine due to age was found for 32 patients younger than 18 years and 729 older than 65 years (2). Age was not specified for a higher proportion of orlistat patients (3604; 22.5%) compared with sibutramine patients (1382; 11.2%), which may be explained by differences in green form design. The orlistat green form initially asked for date of birth, but changes in confidentiality guidance meant that during the study this was changed to age. The sibutramine green form requested the year of birth.
The additional questions on the green form for sibutramine showed that the most common comorbidity in patients using sibutramine was a history of hypertension (1549 patients; 18.0% of 8601 respondents). Ischemic heart disease (238; 2.8% of 8595) and diabetes mellitus (898; 10.4% of 8648) were reported to occur less frequently; 2266 (28.3%) of 7993 patients were reported to have used orlistat in the past.
Reasons for Stopping
Within 3 months of starting treatment, 4854 (30.3% of cohort) patients were known to have stopped orlistat and 5155 (41.8%) had stopped sibutramine. In PEM, GPs may specify more than one reason for stopping for an individual patient; where a reason for stopping was specified, there were 3677 reasons for stopping orlistat (3299 patients; 20.6% of cohort) and 4575 reasons for stopping sibutramine (3849 patients; 31.2%). The 10 most frequently reported reasons for stopping within 3 months of starting treatment are shown in ranked order in Table 2.
Table 2. . Clinical adverse event terms given as reasons for stopping orlistat and sibutramine within 3 months of starting treatment
|Diarrhea||332 (2.1)||Hypertension‡||203 (1.6)|
|Intolerance||84 (0.5)||Raised blood pressure§||179 (1.5)|
|Weight gain||75 (0.5)||Headache/migraine||131 (1.1)|
|Unspecified side effects*||64 (0.4)||Malaise/lassitude||119 (1.0)|
|Abdominal pain||53 (0.3)||Depression||94 (0.8)|
|Flatulence||44 (0.3)||Weight gain||91 (0.7)|
|Gastrointestinal, unspecified†||43 (0.3)||Insomnia||85 (0.7)|
|Fecal incontinence||39 (0.2)||Palpitation||75 (0.6)|
|Pregnancy||33 (0.2)||Unspecified side effects*||75 (0.6)|
|Headache||28 (0.2)||Constipation||72 (0.6)|
Diarrhea was the most common adverse event given as a reason for stopping orlistat within 3 months (332; 2.1% of cohort) and over the entire study period (525; 3.3%). Gastrointestinal events were given as a reason for stopping for 893 patients (5.6% of cohort) over the duration of the study.
Hypertension (within 3 months: 203 patients, 1.6% of cohort; entire study: 290, 2.4%) and raised blood pressure (within 3 months: 179, 1.5%; entire study: 285, 2.3%) were the most common clinical adverse events given as reasons for stopping sibutramine. Tachycardia was also of concern at the time of marketing (2), and raised pulse and/or tachycardia were given as reasons for stopping by 85 patients (0.7% cohort), of whom 76 (89.4%) stopped within 3 months of starting treatment.
Statistical Analysis of Events
Table 3 shows the 15 clinical conditions with the highest incidence densities in ranked order of ID1; included in the table are the clinical events for which ID1-ID2–3 was significantly different. With the exception of the non-specific adverse events “malaise/lassitude” and “intolerance,” and weight-related events, clinical conditions significantly associated with taking orlistat were mainly gastrointestinal adverse events, and those associated with sibutramine included headache/migraine, insomnia, dry mouth, “nausea/vomiting,” dizziness, and palpitations.
Table 3. . Incidence densities (IDs) for clinical conditions ranked in order of ID1per 1000 patient-months’ exposure and events significantly associated with starting treatment
|Orlistat|| || || || || || || || || |
| Weight loss**||230||109||16.96||5.15||11.81||8.67||14.96||521||6.14|
| Abdominal pain**||79||50||5.83||2.36||3.46||1.57||5.36||213||2.51|
| Upper respiratory tract infection||59||108||4.35||5.10||−0.75||−2.68||1.18||401||4.72|
| Weight gain||54||90||3.98||4.25||−0.27||−2.08||1.54||359||4.23|
| Rectal discharge**||47||15||3.47||0.71||2.76||1.37||4.14||77||0.91|
| Fecal incontinence**||44||23||3.24||1.09||2.16||0.77||3.55||86||1.01|
| Unspecified side effects*,**||42||35||3.10||1.65||1.44||0.02||2.87||113||1.33|
| Gastrointestinal, unspecified†,**||39||27||2.88||1.28||1.60||0.26||2.94||90||1.06|
|Sibutramine|| || || || || || || || || |
| Weight loss**||107||56||9.97||3.55||6.42||3.66||9.19||239||4.69|
| Raised blood pressure§||83||121||7.74||7.67||0.07||−2.76||2.90||322||6.32|
| Dry mouth**||69||24||6.43||1.52||4.91||2.76||7.06||106||2.08|
| Cardiovascular tests¶,**||59||43||5.50||2.73||2.77||0.64||4.91||126||2.47|
| Upper respiratory tract infection||42||100||3.91||6.34||−2.42||−4.68||−0.17||237||4.65|
During the entire observation period, there were 33 deaths (0.2% of cohort) in the orlistat cohort and 25 deaths (0.2%) in the sibutramine cohort, including those patients who died while taking the drug and those who died after stopping treatment. Causes of death were ascertained for 28 patients on orlistat and 14 on sibutramine. Cardiovascular or cerebrovascular events were the most common cause of death in both cohorts, with 18 and 9 deaths in the orlistat and sibutramine cohorts, respectively. Of these deaths, 6 in the orlistat cohort and 3 in the sibutramine cohort were known to have occurred while the patients were undergoing treatment. There were no instances where the GP attributed the cause of death to either drug.
The orlistat and sibutramine studies reported the outcomes of 109 and 67 pregnancies, respectively. In the orlistat cohort, 35 patients had stopped the drug before the last menstrual period, 2 had initially taken the drug in the second trimester, and in 5, the timing of exposure was uncertain. For sibutramine, 14 patients had stopped the drug before the last menstrual period. The remaining 67 pregnancies in the orlistat cohort and 53 in the sibutramine cohort were exposed to the drug in the first trimester (Table 4). Three babies in the orlistat cohort were born with congenital abnormalities: one with congenital dislocation of the hips, one with translocation of the great vessels, and a baby who developed colic was subsequently found to have malrotation that needed surgery. Another baby had shoulder dystocia at birth, which resulted in a fractured clavicle, and neonatal jaundice requiring phototherapy. In the sibutramine cohort, one baby was born with hypospadias and small patent ductus arteriosis, and there was a neonatal death in a baby who had alveolar capillary dysplasia causing severe pulmonary hypertension.
Table 4. . Pregnancy outcomes for women known to be exposed to drug in the first trimester
|Termination of pregnancy||9||4|
|Outcome not known||9||6|
Events Assessed as Probably/Possibly Related to Treatment
In the following section, we describe events assessed in PEM as probably/possibly related to treatment that were equivalent or similar to events published as case reports. All of the events assessed as possibly/probably related to orlistat are described in a separate publication (21).
While there have been a number of published case reports of serious hepatic adverse reactions in association with orlistat (22, 23, 24, 25), no cases of this severity were identified in PEM, but there were reports of elevated liver function tests (LFTs). Two cases were assessed as causally related: one, assessed as probably related, had a positive re-challenge, and the other had abnormal alanine transaminase and γ-glutamyl transferase levels that were assessed as possibly related based on temporality.
Case reports have described new onset hypertension (26) and the worsening of pre-existing hypertension through drug interactions with anti-hypertensive drugs (27). In the PEM study, 3 cases of hypertension were assessed as possibly related to orlistat based on temporality. Two of the patients were not known to be hypertensive before starting treatment; the other had a history of hypertension, but initial screening tests on starting orlistat were normal. The published case report of new-onset hypertension also involved edema (26), and worsening of pre-existing edema has also been described (28); in PEM, a case of mild pedal edema occurred in a patient with no other risk factors 6 weeks after starting orlistat and was assessed as possibly related based on the time to onset. There were also 3 reports of swollen ankles assessed as possibly related to orlistat. One patient had a history of swollen ankles and another had ischemic heart disease.
A lichenoid eruption has been described in the literature (29); in PEM, a report of lichen planus involving a white lesion on the lateral border of the tongue was reported. It occurred 4 months after starting orlistat and was assessed as possibly related to treatment, based on temporality and the lack of an alternative explanation.
There have been several reports of events related to mania in patients treated with sibutramine (30, 31, 32). In PEM, a report of mania on the second day of treatment was assessed as probably related to sibutramine, and there was positive de-challenge. The event was not pre-existing, and there was no known psychiatric history or any concurrent drug use. A report of panic attack has been published (33), and in PEM, two reports of panic attack were assessed as possibly related to sibutramine. Neither event was pre-existing, nor did either patient have a history of psychiatric illness, and in both instances there was a positive de-challenge. One of these patients also had dyspnea and palpitations assessed as possibly related. Psychotic episodes in patients treated with sibutramine have been described in the literature (34, 35). In PEM, an acute psychotic reaction occurred within 24 hours of taking sibutramine in a patient with a history of anxiety and depression. There was positive de-challenge, and the event was assessed as possibly related to sibutramine. A report of florid schizophrenic psychosis in a patient with a history of schizophrenia was assessed as possibly related to sibutramine. However, this was complicated by the patient having stopped taking risperidone before starting sibutramine. The patient was reported to be stable before starting sibutramine, and the possibility that it was due to sibutramine could not be ruled out.
A case report involving memory impairment has been published (36), and, in PEM, a patient was reported to have poorer concentration and memory one day after starting sibutramine, which was assessed as possibly related to treatment. The patient had a history of generalized anxiety disorder and bipolar affective disorder and was taking three other centrally-acting drugs.
A report of QT prolongation in a patient taking sibutramine has been published (37); and although this event was not identified in the PEM study, there were 3 cases of arrhythmia (2 possible, 1 probable) assessed as causally related to sibutramine. One patient with a history of ischemic heart disease was reported to have multiple premature ectopic beats, the other 2 cases were reported as an irregular pulse. Other events related to arrhythmia included probable supraventricular tachycardia in a patient with palpitations, and 10 subjects with palpitations assessed as possibly/probably related to treatment had symptoms suggestive of arrhythmia (including faintness, dizziness, chest pain, and/or dyspnea).
There is a published case report of reversible hepatotoxicity in association with sibutramine (38). In PEM, there were 5 cases of abnormal LFTs assessed as possibly related to sibutramine treatment. In 2 cases, the LFTs were reported to be normal before starting sibutramine, and both had positive de-challenges; however, they were both taking medications known to cause hepatic reactions. Another 2 cases had elevated LFTs before starting sibutramine, but they worsened during treatment. One of these cases was taking concomitant medications known to be associated with hepatic adverse reactions, and the other had a history of high alcohol consumption; in both cases, LFTs decreased after sibutramine was stopped. For the final case, there was no information on LFTs or hepatic function before starting sibutramine; however, there was a positive de-challenge.
A case report involving a severe bullous eruption with urticarial plaques has been described (39). There were no reports of severe bullous skin eruptions in the PEM study; however, a report of urticaria was assessed as possibly related to sibutramine. This event occurred 18 days after starting sibutramine, it was not pre-existing, and there was positive de-challenge; however, meloxicam was also being taken at the time.
This paper has presented data from the post-marketing PEM studies conducted during the immediate post-marketing period for orlistat and sibutramine prescribed in the primary care setting. These studies were performed using the same methodology and showed that the drugs were associated with different adverse event profiles, consistent with their distinct pharmacological mechanisms. Furthermore some of the events assessed as “signals” in PEM were equivalent or similar to events described in published case reports.
An advantage of PEM is that the studies were carried out on a nationwide scale without influencing the prescribing practice of doctors, as patients were identified from dispensed prescriptions. The studies provided information on the “real world” use of these products in general practice, as patients were included irrespective of comorbidities and concomitant use of other medications. The studies also allowed the identification of potential ADRs not attributed to the drug by the GP, since the GPs were requested to provide event data regardless of whether they considered the event to be related to use of drug.
The age and gender profiles of the two cohorts were similar, but it is not known how comparable the cohorts were with regard to other comorbidities and risk factors, as such information was not requested on the orlistat green form. The drugs were marketed at different times, and it is likely that the sibutramine cohort may include a number of patients who either had not tolerated orlistat or had not found it effective. This was demonstrated by 29% of sibutramine users having reported using orlistat in the past.
The response rates for both drugs were similar; however, for orlistat it was slightly lower than the PEM average for 90 study drugs, which is 55.9% (for green forms returned). The response rates, nevertheless, were greater than the proportion of ADRs that are reported spontaneously (40, 41) and than response rates for GP surveys in general (42). For data collection systems such as PEM that are dependent on a third party, response bias is possible, and under-reporting (including under-reporting of serious or fatal events) is possible. There may also be hidden biases if patients of non-responding GPs are different from the patients of GPs who returned the green forms.
The events and reasons for stopping within the first 3 months were examined because the prescribing guidelines for both products recommend discontinuing treatment if a specified weight loss has not been achieved in that time (1, 2). The adverse events identified from the reasons for stopping and statistical analysis of events, within this time period, were consistent with the distinct adverse event profiles of the 2 drugs (1, 2, 3, 4, 43, 44, 45, 46, 47, 48, 49, 50, 51). Orlistat was associated mainly with gastrointestinal adverse events, whereas in the sibutramine cohort, adverse events included hypertension, headache/migraine, palpitation, constipation, dizziness, insomnia, and dry mouth.
Adverse events significantly associated with starting treatment but not in the manufacturers’ Summary of Product Characteristics (SmPC) (excluding the non-specific side effects “malaise/lassitude” and “intolerance” and weight-related events) were nausea/vomiting and vaginitis/vulvitis for orlistat and faintness for sibutramine (however, light-headedness was listed in the SmPC) (1, 2). None of these events has been added to the SmPC since marketing (51, 52). Other studies have also found nausea to be one of the most frequently reported adverse events associated with orlistat (43, 44, 51).
The literature searches identified 8 published case reports for orlistat and 10 for sibutramine that had equivalent or similar events assessed as probably/possibly related in the PEM studies. However, it must be noted that, while some of the events are equivalent, others may represent different manifestations of disease. We have used these examples to put the results from PEM into context with other published evidence. A limitation is that only reports found in Medline and Embase have been included.
Both orlistat and sibutramine have had published reports of liver disorders associated with treatment (22, 23, 24, 25, 38). Although no severe hepatic events were identified in either PEM study, there were reports of abnormal LFTs assessed as causally related to treatment in both PEM studies. Reversible increases in liver enzymes were listed on the sibutramine SmPC at the time the study was undertaken (2); the orlistat SmPC has had increases in liver transaminases and alkaline phosphatase and hepatobiliary disorders added as very rare adverse reactions (51).
Case reports of hypertension in patients using orlistat have been published; these reports describe new-onset hypertension and peripheral edema (26) and worsening of controlled hypertension, possibly through interactions with anti-hypertensives (26); however, causality has been disputed (53). Hypertension was a known adverse effect of sibutramine before it was marketed (2), whereas for orlistat, this represents an unexpected event. Hypertension is confounded by obesity, making any association with orlistat difficult to assess based on the few cases and information available. However, for sibutramine, the effect on blood pressure, as well as heart rate, has been shown in clinical studies and meta-analyses (4, 43, 45, 46, 54). PEM also identified hypertension as the clinical adverse event most frequently associated with stopping sibutramine. Most of patients known to have stopped sibutramine due to hypertension/raised blood pressure or tachycardia/increased heart rate did so within the first 3 months, which is consistent with these effects tending to occur within the first 4 to 12 weeks of treatment (2). Although outside of the scope of this study, the long-term consequences of the effect of sibutramine on blood pressure still need to be determined, in particular, whether the effect of weight reduction on cardiovascular risks can offset the increased mortality that is associated with elevated blood pressure (55, 56).
The report of QT prolongation and cardiac arrest in a patient taking sibutramine occurred in an individual with a genetic susceptibility for membrane depolarization (37). In the PEM study, there were no reports of QT prolongation. However, causal associations were made between sibutramine and arrhythmias, as well as probable supraventricular tachycardia and symptoms possibly suggestive of arrhythmia in patients with palpitations. Paradoxical supraventricular tachycardia has been added to the SmPC since marketing (52).
Psychiatric events associated with sibutramine have been of concern, and we identified a number of cases reported in the literature (30, 31, 32, 33, 34, 35). Sibutramine is contraindicated in patients with psychiatric illness, as it may have the potential to cause manic episodes in patients with bipolar manic-depressive psychosis (2), and a case of such an event was published (30). It is of interest that the case of mania identified in PEM occurred in a patient without a known psychiatric history.
Event terms not found in published case reports, added to the SmPC since marketing, that had equivalent events assessed as probably/possibly related to orlistat in PEM are pruritus, rash, and urticaria, which are listed under hypersensitivity (51). For sibutramine, event terms include depression in patients both with and without a prior history, rash, angioedema, gastrointestinal hemorrhage, alopecia, cutaneous bleeding reactions, urinary retention, atrial fibrillation, vomiting, and impotence (52). In addition, transient short-term memory loss has been added to the SmPC since marketing (52) and has also been reported in PEM and in a published case report (36).
Case reports without equivalent events assessed as causally related in the PEM study for orlistat were lactic ketoacidosis (57), depression (58), vasculitis (59), oral contraceptive failure (60), and interactions with thyroxine (61), warfarin (62), metformin (63), olestra (64), and cyclosporine (65, 66, 67, 68, 69, 70, 71). In PEM, a drug interaction was reported between metformin and orlistat, but the nature of the interaction was not specified. There was also an interaction with an unspecified contraceptive leading to contraceptive failure (21). However, no further information was obtained to enable formal causality assessments. A warning has been added to the SmPC to recommend the use of an additional form of contraception in case of severe diarrhea (51). Although, in the PEM study, 2 cases of hypothyroidism were assessed as possibly related to orlistat, they did not involve an interaction with thyroxine, nor were the subjects taking thyroxine (21). There were no reports of depression assessed as causally related to treatment in the orlistat PEM study; however, it must be noted that the relationship between obesity and depression is complex (72).
For sibutramine, cardiomyopathy (73) and cyclosporine interaction (74) were found in published case reports. While a death due to cardiomyopathy was reported in the sibutramine PEM study, there was insufficient information on the case, and it was not known if the death occurred during treatment. There were no reports of interactions between cyclosporine and sibutramine.
Signals for suspected ADRs may arise from a variety of sources, including epidemiological investigations, spontaneous reporting, and published case reports (7). We focused on case reports because they have been the subject of research and debate relating to the way that signals are identified (8, 9, 10, 75, 76, 77, 78). An advantage of case reports is that they can identify rare adverse drug reactions that may not otherwise be detected in a controlled clinical setting and can alert physicians to potential adverse drug reactions (76). However, sometimes associations turn out to be false (75). Of the case reports we assessed, the majority were found in specialist journals, all of which were related to the adverse event. This means that they may be less likely to be seen by the primary care physician. Not all published case reports had events causally related in the PEM studies: the events may have been reported, but either there was no or insufficient follow-up information to enable causality assessments to be made or there could be no evidence for a causal association. For events not identified in PEM, it may be that they have not been reported or have not occurred in the patients in these cohorts.
Knowledge of the comparative safety profile of products used in the treatment of medical conditions is necessary for clinical decisions and patients’ choice. However, a comparison using statistical significance and power is not helpful when the pharmacological characteristics of the drugs are distinct. We feel that reviewing the safety profiles of orlistat and sibutramine in the way we did helps doctors, healthcare professionals, and patients in better understanding the comparative safety profiles of these two products and making better informed decisions.
In summary, this article has provided a comparison of the safety profiles of orlistat and sibutramine in general practice and has highlighted different adverse event profiles that are consistent with their distinct pharmacological mechanisms. The adverse events identified in these studies are in agreement with information from the SmPC, other studies, and published case reports. Furthermore, a number of published case reports had similar or equivalent events assessed as causally related in the PEM studies.