European Journal of Clinical Pharmacology

, Volume 59, Issue 10, pp 767–773

Evaluation of the safety of bupropion (Zyban) for smoking cessation from experience gained in general practice use in England in 2000

Authors

    • Drug Safety Research Unit
    • University of Portsmouth
  • Lynda V. Wilton
    • Drug Safety Research Unit
    • University of Portsmouth
  • Saad A. W. Shakir
    • Drug Safety Research Unit
    • University of Portsmouth
Pharmacoepidemiology and Prescription

DOI: 10.1007/s00228-003-0693-0

Cite this article as:
Boshier, A., Wilton, L.V. & Shakir, S.A.W. Eur J Clin Pharmacol (2003) 59: 767. doi:10.1007/s00228-003-0693-0

Abstract

Background

Bupropion (Zyban) is the first new pharmacological treatment for smoking cessation to be introduced since nicotine replacement therapy. In smoking cessation trials, it has been associated with minimal side effects. A range of suspected adverse drug reactions (ADRs) were reported via the spontaneous reporting system following its use in smoking cessation.

Aim

To examine the safety of bupropion used in general medical practice in England as a treatment for cessation of smoking.

Objectives

To quantify the incidence of events that were reported for patients prescribed bupropion; and to identify any previously unrecognised ADRs.

Methods

A post-marketing observational cohort study was conducted using the methodology of prescription-event monitoring (PEM). Exposure data were derived from the first prescription dispensed for patients whose prescription details were processed by the Prescription Pricing Authority in August 2000. Outcome data were derived from ‘green form’ questionnaires (GFs) sent to general practitioners (GPs) at least 6 months following the first prescription issued. Incidence densities (IDs) were calculated for events reported per 1000 weeks of patient treatment and ID differences between time periods analysed. Events of interest were followed up by postal questionnaire sent to GPs. All-cause and condition-specific mortality up to 12 weeks after starting bupropion were compared through indirect standardisation between the PEM cohort and Cancer Prevention Study-II (USA) (CPS-II) data.

Results

GF response rate was 48.1%, with 11,735 GFs containing useful data – of these patients, 5695 (48.5%) were male (median age 47 years, range 16–88 years) and 6009 (51.2%) were female (median age 47 years, range 16–87 years). Age was recorded for 4092 (34.9%) of the cohort of 11,735 after follow-up. There were 566 events in 350 patients reported by GPs as ADRs to bupropion. GPs reported 10,200 reasons for stopping bupropion among 9056 patients. The highest ranked clinical events (by ID for weeks 1–6 of treatment) were; ‘insomnia’ (n=308), ‘nausea/vomiting’ (n=243) and ‘dizziness’ (n=185). Bupropion was taken in the first trimester of 12 pregnancies and the outcome ascertained in eight cases – five live births (no congenital abnormalities reported), two therapeutic terminations and one intrauterine death (no further details). The standardised mortality ratio (SMR) for all-cause mortality up to 12 weeks after starting bupropion was 0.77 (95% CI: 0.42, 1.28).

Conclusion

This study describes the safety profile of bupropion (Zyban) as used in the community; a small number of adverse events were reported that were not included on the SmPC. For many events, nicotine withdrawal was a confounding factor. SMR calculations did not provide evidence for a higher rate of mortality (either all-cause or condition-specific) in the PEM cohort relative to smokers from the CPS-II cohort in the USA. While reassuring, the SMR should be interpreted in context with results from other studies on bupropion when used for smoking cessation.

Keywords

Smoking cessationBupropion (Zyban)Prescription-event monitoring (PEM)

Introduction

Amfebutamone (bupropion) was first licensed for use in the United States in 1989 [1]. Sustained release (SR) bupropion is indicated for the treatment of depression at up to a maximum dose of 400 mg per day [2]. In July 2000, SR bupropion was launched in the UK as Zyban [3]. This was the first new oral pharmacological treatment for smoking cessation to be introduced since nicotine replacement therapy (NRT). The recommended dosage for Zyban in smoking cessation was initially 150 mg, rising to 300 mg maximum per day after 3 days [3]. Bupropion was available on a British National Health Service (NHS) prescription, though guidance to general practitioners (GPs) underlined the place of concomitant behavioural support in the treatment of nicotine dependence when a patient is motivated to stop smoking [4, 5].

Nicotine dependence brings with it heavy consequences for health – cigarette smoking is the most common cause of preventable mortality in the UK (estimated 120,000 deaths per year in those over 35 years of age) [6]. A consideration of the risks and benefits associated with any smoking cessation therapy should therefore take this observation into account.

Placebo-controlled clinical trials demonstrate the efficacy of bupropion in smoking cessation, as measured by higher rates of smoking cessation [7, 8]. Its mechanism of action in smoking cessation is not fully understood but is thought to depend on dopaminergic and/or noradrenergic properties – mediated by elevated levels of dopamine in the nucleus accumbens, a process also considered to be involved with nicotine addiction [9].

In clinical trials, bupropion SR in smoking cessation has been associated with minimal side effects [8], was well-tolerated [10], and no seizure events were reported [7, 10]. General populations, in contrast to those of clinical trials, represent patients bearing a wider range of physiological/pharmacokinetic characteristics, leading to a varied event profile. Spontaneous reports following the launch of Zyban in the UK numbered 1038 by 31 October 2000, including ten reports of suspected adverse drug reactions (ADRs) with fatal outcomes and reports of seizures (n=40), urticaria (n=110) and angioedema (n=94) [11].

Prescription-event monitoring (PEM) is a method of post-marketing safety surveillance of newly marketed medicines and was established in England in 1980 [12]. It observes drug safety under the conditions of general practice and is complementary to the ‘Yellow Card’ scheme of spontaneous reporting of ADRs to the UK Medicines & Healthcare Products Regulatory Agency (MHRA)/Committee on Safety of Medicines (CSM). This study aimed to examine the safety of bupropion used in general medical practice in England as a treatment for cessation of smoking; to quantify the incidence of a range of events that occurred in patients treated with Zyban; and to identify any previously unrecognised ADRs.

Materials and methods

Patients were identified by means of dispensed NHS prescription data supplied in confidence by the Prescription Pricing Authority (PPA) in England in August 2000. A ‘green form’ questionnaire (GF) was sent to the prescribing GP approximately 6 months after notification by the PPA of the date of the first dispensed prescription for each individual patient. In PEM, the GF requests information on patient age and sex; indication for prescribing; dose; duration of treatment (including start and stop dates); reasons for stopping the study drug; events that occurred after the drug was prescribed, including those considered by the reporting GP to have been an ADR.

Reported events were coded using the DSRU event dictionary, a hierarchical dictionary arranged by system-organ class with selective ‘lower level’ terms grouped together under broader ‘higher level’ terms. Those questionnaires with no information provided were classified as ‘void’ and excluded from the study cohort and subsequent analysis, as there is no means of determining whether uncompleted forms indicated no reported events. In this PEM study, the GFs also included questions (on the back) on number of cigarettes smoked; smoking status at the patient’s last consultation; and whether there existed previous history of ischaemic heart disease (IHD), hypertension, diabetes mellitus, chronic obstructive pulmonary disease (COPD), seizure, stroke or transient ischaemic attack (TIA) and psychiatric illness, to identify possible confounding factors.

Pregnancies (those that occurred during treatment and within 3 months of stopping bupropion) and any events of interest or particular concern with this drug, or considered medically important and where additional information was required, were followed up by sending additional postal questionnaires to the prescribing GP. If no clear cause of death could be established from a GF, the patient’s GP was contacted in order to ascertain the certified cause of death. Individual case reports were assessed for causality by a clinical research fellow at the DSRU, using the criteria of temporality, pharmacological plausibility, clinico-pathological nature and exclusion of other causes. Causality was graded as being probable, possible, unlikely or unassessable [13].

This study was conducted in accordance with the International Ethical Guidelines for Biomedical Research prepared by the Council for International Organisations of Medical Science in collaboration with the World Health Organisation [14]. The method of study 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 [15] and Multi-Centre Research Ethics Committees Guidance Notes 2000 [16].

Statistical analysis

The response rate for the study was reported, followed by a demographic description of the cohort. Prescribing indications (and contexts) were also described, along with adverse events attributed by GPs to bupropion and reasons for stopping treatment. Incidence densities (IDs) were calculated for all reported events during treatment within specified time periods and expressed as the number of first reports of an event per 1000 patient weeks of treatment. IDs for events occurring in the first 6 weeks of treatment (ID1–6), during weeks 7–9 of treatment (ID7–9) and for the overall treatment period (IDA) were calculated for patients for whom either the date of stopping the drug was known, or in those that continued to take the drug until the end of the study period. The difference between the two IDs (ID1–6–ID7–9) was calculated to test the hypothesis that the ID did not change over time. Where the arithmetic difference between ID1–6 and ID7–9 was statistically significant at the 1% level, this was considered to be a signal of a possible ADR.

A similar analysis was performed to examine the difference between ID1–9 (weeks 1–9 of treatment) and IDOFF (events occurring after treatment with bupropion had stopped). IDs for all events were also stratified according to the responses to the additional questions on the GF.

Finally, standardised mortality ratios for deaths due to any cause and those within specific ICD-9 (International Classification of Diseases, ninth revision) groupings [for IHD (code: 410–414) and cerebrovascular disease (code: 434, 437)] were calculated for patients in the cohort who had died up to 12 weeks after commencing bupropion therapy, using the technique of indirect standardisation. In the absence of comparable UK data, patients who had smoked from the USA Cancer Prevention (CPS)-II study were used as the comparator cohort [17].

It was anticipated that under-reporting of patient age might occur in this study, the proportions of person time at risk for those patients under 35 years, over 75 years and 10-yearly age bands between these ages in the part of the cohort with age-recorded were calculated. These proportions were then applied to person time at risk for the whole cohort, and number of patients in the corresponding age-bands estimated. GP surgeries were telephoned to attempt to increase the recording of age for patients who died in the 12 weeks after taking bupropion.

Results

GPs returned 48.1% (13,372/27,809) of GFs posted. Of these, 1637 (12.2%) were ‘void’ [reasons included: patient no longer registered with GP (n=658) and blank form returned (n=518)] giving a patient cohort of 11,735 where useful information was available. The median age for all patients in the total cohort was 47 years (range 16–88 years), with patient age unrecorded for 7653 (65.2%) of the cohort; 6009 (51.2%) of patients were recorded as being female. After telephone follow-up, 4092 (34.9%) of patients in the cohort of 11,735 had an age recorded. Most patients (82.8%) were prescribed bupropion for smoking cessation as per licensed indication on the SmPC for bupropion (Zyban) [3, 18], while the remaining indications included clinical contexts for prescribing such as COPD.

The clinical events reported most frequently (for which ID1–6–ID7–9 was greater than zero at the 1% significance level) are shown in Table 1, ranked in descending order according to number of events reported in the first month. The ranking of these events was similar when the treatment period (weeks 1–9) was compared with reports of events occurring after stopping bupropion therapy. However, the events of rash, anxiety, pruritus, constipation and dyspepsia were also observed to have been reported significantly more frequently during treatment than during the period after stopping. Again, most of these events were described in the SmPC [3, 18].
Table 1

Incidence densities (IDs) for events, ranked in order of number of events in weeks 1–6 of treatment (where ID1–6–ID7–9>0). N1–6=total number of reports of each event during weeks 1–6 of treatment, N7–9=total number of reports of each event during treatment in weeks 7–9, ID1–6=incidence density for each event during weeks 1–6 of treatment, ID7–9=incidence density for each event during weeks 7–9 of treatment, ID1–6–ID7–9=arithmetic difference between ID1–6 and ID7–9, 99% CI=99% confidence intervals for ID1–6–ID7–9, NA=total number of reports of each event during the total treatment period, IDA=incidence density for each event for the total treatment period, incidence risk (%)=proportion of events reported by study cohort (n=11,735)

Higher term description

N1–6

N7–9

ID1–6

ID7–9

ID1–6–ID7–9

99% CI

NA

IDA

Incidence risk (%)

      

Min

Max

   

Insomnia

308

18

6.1

1.5

4.6

3.4

5.9

348

4.6

2.97

Nausea, vomiting

243

16

4.8

1.3

3.5

2.4

4.7

277

3.7

2.36

Dizziness

185

7

3.7

0.6

3.1

2.2

4

204

2.7

1.74

Headache, migraine

173

13

3.4

1.1

2.4

1.4

3.4

196

2.6

1.67

Malaise, lassitude

163

10

3.2

0.8

2.4

1.5

3.3

190

2.5

1.62

Depression

123

10

2.4

0.8

1.6

0.7

2.5

145

1.9

1.24

Dry mouth

83

4

1.6

0.3

1.3

0.7

1.9

92

1.2

0.78

Sweating

80

7

1.6

0.6

1.0

0.3

1.7

91

1.2

0.78

Intolerance

69

6

1.4

0.5

0.9

0.2

1.5

81

1.1

0.69

Tremor

68

3

1.4

0.2

1.1

0.6

1.7

74

1.0

0.63

Pain chest, tight chest

63

5

1.3

0.4

0.8

0.2

1.5

76

1.0

0.65

Urticaria

58

3

1.2

0.2

0.9

0.4

1.4

66

0.9

0.56

Pain abdomen

47

4

0.9

0.3

0.6

0.1

1.2

55

0.7

0.47

Sensation abnormal

41

2

0.8

0.2

0.7

0.2

1.1

46

0.6

0.39

Agitation

37

1

0.7

0.1

0.7

0.3

1.0

38

0.5

0.32

Confusion

32

2

0.6

0.2

0.5

0.1

0.9

36

0.5

0.31

Panic attack

27

1

0.5

0.1

0.5

0.1

0.8

33

0.4

0.28

Events recorded on GFs that were coded as an ADR to bupropion numbered 566 in 350 (3.0%, n=11,735) patients, of which 108 (19.1%) of these events were reported to the CSM and 3 (0.5%) reported to the manufacturer (Table 2).
Table 2

‘Higher level’ event terms reported more than ten times as adverse drug reactions (ADRs) by general practitioners (GPs) to bupropion. (CSM=Committee on Safety of Medicines)

ADR

Total

Reported to CSM

Reported to manufacturer

Unspecified side effects

79

1

0

Nausea, vomiting

50

8

1

Insomnia

41

6

0

Dizziness

37

1

0

Headache, migraine

26

2

0

Malaise, lassitude

29

2

0

Rash

20

9

1

Depression

15

5

0

Tremor

15

4

1

Pain chest, tight chest

14

2

0

Sweating

14

4

0

Anxiety

13

3

0

Pruritus

11

6

0

The range of events reported as ADRs reflected those included on the initial SmPC for bupropion (Zyban) [3]. Depersonalisation was subsequently added to the SmPC for bupropion (Zyban), and this document also underlined the association between depressed mood and nicotine withdrawal [18]. Cardiovascular ADRs of interest that did not appear on the SmPC current at the time of the study [3] were palpitation (n=7), and the ‘lower level’ terms arrhythmia (n=1) and myocardial infarction (n=1). Other ADRs reported included events of angioneurotic oedema (n=2), bronchospasm (n=1) and convulsion (n=1). Two drug interactions were reported: one between bupropion and alcohol; and one between bupropion and tramadol, though details of the interactions were not specified.

In 9056 patients, 10,200 reasons were recorded by the GPs for stopping bupropion therapy. The most frequently reported clinical condition given as a reason for stopping was nausea/vomiting (n=204). However, 71.0% (n=7256) of all the events reported as a reason for stopping bupropion therapy were attributed to the events of – ‘course completed’ (n=3878); ‘condition improved’ (n=1192); ‘not effective’ (n=1093); and ‘not specified’ (n=1093). Clinical reasons for stopping bupropion reported more than 100 times are presented in Table 3 (reflecting the range of events listed in the SmPC [3, 18]) in addition to medically important reasons for stopping that do not appear on the SmPC [3, 18] (given in italics).
Table 3

Clinical reasons for stopping bupropion therapy that were reported more than 100 times and other medical reasons for stopping (in italics)

Reason

Number

Nausea, vomiting

204

Smoking cessation*

187

Insomnia

169

Malaise, lassitude

162

Dizziness

140

Non-compliance

128

Headache, migraine

125

Palpitation

29

Sensation abnormal

25

Diarrhoea

21

Ischaemic heart disease

10

Disorders of cardiac rhythm

5

Cerebrovascular accident

2

Transient ischaemic attack

2

Cardiac arrest

1

*The actual wording supplied by GPs on the GF was coded using DSRU dictionary terms. Thus, for the 187 reasons for stopping, coded as ‘smoking cessation’, the corresponding wording by GPs related to patients who ‘stopped’, ‘quit’ or ‘gave up’ smoking

Stratified analyses

The IDs for weeks 1–6 versus weeks 7–9 when stratified according to responses to the ‘additional questions’ located on the back of the GF revealed a greater number of events (ID1–6–ID7–9>0, P=0.01) such as insomnia or dizziness in the week 1–6 time period that could be attributed to nicotine withdrawal. This was also the case when there was stratification by number of cigarettes smoked.

Pregnancies

Twenty-one pregnancies occurred during or within 3 months of stopping bupropion therapy. After follow-up, there were 9 pregnancies where bupropion had been stopped before the last menstrual period and 12 where bupropion had been taken during the first trimester – the outcomes for the latter group of pregnancies were: five live births (no congenital abnormalities reported), two therapeutic terminations, one intrauterine death (no further details were provided) and, in four cases, the outcome was not ascertained.

Deaths

During the entire study period, 114 (0.97%) deaths were reported, for 25 patients the cause of death was not ascertained. There were 14 deaths reported during the 12 weeks after commencing bupropion therapy, and age was recorded for ten of these patients (median of 72.5 years, interquartile range: 46.5–76.0 years).

The reported cause of death was ascertained for 12 of the 14 patients (two reports of myocardial infarction and IHD respectively; one report each of atherosclerosis, aortic aneurysm, COPD, chest infection, cerebral thrombosis, CVA, inguinal abscess and carcinomatosis).

No significant differences were observed between patients with a recorded age and those where age was not recorded either on treatment time (two sample t-test, P=0.761) or mortality experience (Fisher’s exact test, P=0.374). Based on the proportion of person time at risk by age-band for the ‘age recorded’ part of the cohort, Table 4 gives person time at risk for the whole cohort after it was apportioned by age-band. The standardised mortality ratio (SMR) (all causes) obtained by indirect standardisation, using CPS-II data [17] as the comparator was 0.77 (95% confidence interval: 0.42–1.28).
Table 4

All-cause standardised mortality ratio (SMR) for deaths in the total prescription-event monitoring (PEM) cohort up to 12 weeks using the CPS-II data as a comparator. SMR=14/18.29=0.77 (95% confidence interval: 0.42–1.28)

Age band (years)

PEM TOTAL COHORT person time at risk (years)

CPS-II mortality rate/100,000 years

Expected deaths

Observed deaths

<35

546.02

-

-

-

35–44

651.44

160.90

1.05

-

45–54

683.87

428.62

2.93

-

55–64

543.32

1062.56

5.77

-

65–74

237.87

2574.22

6.12

-

75+

40.55

5957.30

2.42

-

Total

2703.06

1226.27

18.29

14.0

The SMR result for all-cause mortality was not statistically significant, and the same was true for SMRs for specific causes of death including: coronary heart disease (ICD-9: 410–414), SMR of 0.93 (95% confidence interval: 0.25–2.37); and stroke (ICD-9: 434, 437), SMR of 2.02 (95% confidence interval: 0.25–7.30).

Follow-up of specific events

Events of clinical interest that were followed up include those event terms given in Table 5. One event of convulsion (‘lower level’ term) was assessed as ‘probably’ caused by bupropion 14 days from starting treatment (there was concomitant use of fluoxetine, but no past history of epilepsy). Five patients with an event of convulsion (‘lower level’ term) assessed as ‘possible’ included two patients with a known past history of epilepsy and two patients with a past history of increased alcohol use. The remaining patient had no past history of epilepsy or increased alcohol use.
Table 5

Reports of specific events followed up with assessment of their causality, indicating where this was ‘possible’ or ‘probable’

‘Higher level’ term (‘lower level’ term)

N all events

N events reported*

Follow-up response**

Assessed as ‘probably’ related

Assessed as ‘possibly’ related

Sex

Median age (IQR) years

Male

Female

Disorders of heart rate (tachycardia)

13

7

4/7 (57.1%)

1

3

2

2

39.0 (23.5–50.8)

Disorders of rhythm (arrhythmia)

9

5

2/4 (50.0%)

0

1

1

0

50.0 (50.0–50.0)

Ischaemic heart disease (angina)

42

16

8/14 (57.1%)

0

4

4

0

53.5 (49.3–69.8)

Oedema (oedema face)

12

10

7/9 (77.8%)

2

4

1

5

57.0 (49.3–62.8)

Pain chest and/or tight chest (pain chest)

108

56

24/50 (48.0%)

1

8

2

7

51.0 (37.0–63.0)

Palpitation

63

48

24/36 (66.7%)

2

16

4

14

50.5 (38.0–57.8)

Cerebrovascular accident(cerebrovascular accident)

15

6

1/2 (50.0%)

0

0

-

-

-

Convulsion and/or epilepsy (convulsion)

12

11***

7/11 (63.6%)

1

5

3

3

57.0 (28.5–58.0)

Convulsion and/or epilepsy (status epilepticus)

1

1

0/1 (0.0%)

0

0

-

-

-

Depression (depression)

390

158

81/124 (65.3%)

7

37

15

29

51.0 (40.0–58.0)

Suicidal thoughts

2

2

0/2 (0.0%)

0

0

-

-

-

Urticaria

85

67

43/65 (66.2%)

4

30

11

23

43.5 (32.5–52.0)

Myalgia

64

23

14/23 (60.9%)

2

5

1

6

37.0 (33.0–65.0)

Anaphylaxis

1

1

1/1 (100%)

0

1

1

0

36.0 (36.0–36.0)

*Number of events that occurred during the period in which bupropion was thought to have been taken (includes patients definitely taking bupropion and those uncertain to be taking bupropion at the time of the event)

**Response rate for questionnaires returned with useful information (i.e. excluding blank questionnaires)

***Two of these 11 events occurred in the same patient

Further events assessed as ‘probably’ due to bupropion included: urticaria (n=4), occurring between 14 days and 24 days after commencement of bupropion therapy; myalgia (n=2), one patient described muscle pains and tremor after 2 days of taking bupropion and the other patient described aches 16 days after commencing bupropion which recurred on a re-challenge. Anaphylaxis (n=1), occurring 21 days after commencing bupropion, resulting in severe hypotension and urticaria was assessed as ‘possibly’ due to bupropion.

Discussion

This PEM study describes a population prescribed bupropion by GPs in England, observes the events that occurred after starting treatment and analysed these events for any possible signals of interest. The incidence of a range of events experienced in patients prescribed bupropion was quantified, and events not noted in the SmPC were indicated. PEM methodology facilitates collection of data on a wide range of patients with characteristics likely to be representative of the population at large. While data on co-morbidity and co-prescribing information are not systematically collected, they are obtained in the follow-up of selected events.

The ability to collect data on all events experienced by a patient during the study period permitted the search for new signals of events not suspected by GPs to be due to bupropion. The typical 6-month period of observation in PEM studies from dispensing a prescription to data-collection using the GF provides ample opportunity for patients to consult and thus for events to be recorded by the GP – though such visits could be subject to bias with events causing more discomfort or inconvenience more likely to lead to a consultation.

Response rate is a critical issue in most observational studies, in PEM this is on average 57.4% for GF over 83 previous studies. Selection bias should be considered as we do not at present know the characteristics of patients of doctors who do not respond and whether these patients experience similar rates of adverse events when compared with those patients of doctors who do return GF. The response rate for this study was slightly lower at 48.1%, and this may be a reflection of changing patterns of workload in primary care [19]. Furthermore, bupropion therapy may have been initiated through smoking cessation clinics, involving non-medical staff such as practice nurses – possibly a negative influence on the completion rate for GFs by GPs.

The time periods for which IDs were compared (ID1–6 versus ID7–9) were selected to account for possible confounding due to nicotine withdrawal symptoms [20] which occur at 3 weeks after smoking cessation but may last longer in 40% or more of smokers [21]. In fact, withdrawal symptoms have been varyingly reported to occur from 30 days post-cessation [22] to 6–9 weeks [7]. Likely delays in patients cashing a second prescription for bupropion made stratification of the treatment period desirable, as did the anticipated treatment period of 7–9 weeks [3, 18]. The events of malaise and/or lassitude and sensation abnormal were ranked (by ID1–6) 5th and 14th, respectively, among clinical events (where ID1–6–ID7–9>0) in weeks 1–6; however, these events are not listed on the SmPC [3, 18].

Comparison of the IDs during early and late treatment periods (ID1–6–ID7–9) revealed that events such as insomnia which are associated with nicotine withdrawal occurred significantly more often during the early treatment period. Further events included malaise and/or lassitude and sensation abnormal which do not appear on the SmPC for bupropion (Zyban) [3, 18]. It is possible that patients began to tolerate these symptoms to a greater extent as their therapy proceeded, or that there was negative reporting bias on the part of patients or GPs as time progressed. This might also explain why events such as dyspepsia and diarrhoea occurred less in the ‘off treatment’ versus week 1–9 treatment periods.

Nicotine withdrawal symptoms, e.g. nausea and/or vomiting and insomnia, were a commonly reported ADR, also being reflected among the highest ranking reasons for stopping bupropion. Events reported as ADRs and not listed on the SmPC [3, 18] included; malaise and/or lassitude (n=29), diarrhoea (n=8), unsteadiness (n=4), paranoia (n=2); and myocardial infarction (‘lower level’ term) (n=1).

The data on reasons for stopping a drug prematurely are one of the strengths of PEM and provide a good insight into how either a patient or prescriber decided to stop the therapy being monitored – no causality is implied. The range of cardiovascular events reported as reasons for stopping may simply reflect the cardiovascular morbidity one would expect in a population of patients who have smoked. Small numbers of events were described as ‘rare’ (<1:1000 incidence), e.g. syncope (n=6). Events such as myalgia (n=12) or joint pain (n=12) that can occur in the context of hypersensitivity reactions [3, 18] were reported; including single reports of anaphylaxis, bronchospasm (‘lower level’ term) and vasculitis.

Convulsion (‘lower level’ term) was reported as a reason for stopping on seven occasions and status epilepticus (‘lower level’ term) once. Eleven events of convulsion (‘lower level’ term) (in ten patients) were reported during all treatment periods over 11,735 patients, giving an incidence risk of 0.1%. This is in line with the number of events of this kind expected from the 1:1000 incidence of epileptic seizure described in the SmPC [3, 18] and trial data where patients did not have a predisposition to seizure [23]. Four of the eleven events of convulsion (‘lower level’ term) were associated with a past history of seizure, and this underlines the need to follow the guidance of the SmPC on this matter [18].

Regarding mortality rates in the PEM cohort, no adjustment was made for seasonal variation. Standardised mortality calculations did not provide evidence for a higher rate of mortality (either all-cause or condition-specific) in the PEM cohort relative to smokers from the CPS-II cohort in the USA [17]. Whilst our study did not show a statistically significant increase in observed mortality, relative to what would be expected, for patients 3 months after commencing bupropion; these results should be treated with caution – first due to under-reporting of patient age and, second, the only available comparator population was from the USA. However, in a report on smoking deaths in the UK [6], the comparability of the CPS-II population to British information was analysed leading to the conclusion that the CPS-II population was very comparable.

Conclusions

This study identified the safety profile of bupropion (Zyban) as used in the community. Follow-up of selected events provided important additional information such as for patients who experienced convulsions and any history of epilepsy. Standardised mortality calculations did not provide evidence for a higher rate of mortality (either all-cause or condition-specific) in the PEM cohort relative to smokers from the CPS-II cohort in the USA [17]. The SMR calculation, while reassuring, does not provide a definitive answer to the question and should be interpreted in context with results from other studies on bupropion when used for smoking cessation.

Acknowledgements

We would like to thank all the GPs and numerous other colleagues who have helped with this study. We also thank the Prescription Pricing Authority and the health authorities of England for their participation in this study. We are grateful to all the staff at the DSRU who contributed to this study, particularly Ms. K. Hughes for her assistance with the SMR analyses, Mr. Scott Harris, Mr. Shayne Freemantle and Mrs. Lesley Flowers. Last but not least, we thank Mrs. Gill Pearce for managing the data and wish her a very happy retirement. This study was conducted entirely by the DSRU, an independent charity that accepts funding on a non-profit basis. The DSRU acknowledges a non-contractual donation from the company marketing bupropion. Funding companies have no involvement with the conduct or conclusions of studies carried out at the DSRU. Saad Shakir was an employee of Glaxo Wellcome between 1994 and 1997.

Copyright information

© Springer-Verlag 2003