Antipsychotic drug exposure and risk of venous thromboembolism: a systematic review and meta-analysis of observational studies
Corrado Barbui,1 MD; Valentino Conti, 2 StatD; Andrea Cipriani,1 PhD
1 Department of Public Health and Community Medicine, Section of Psychiatry,
University of Verona, Verona, Italy
2 Regional Centre for Pharmacovigilance, Lombardy Region, Milano, Italy
DRUG SAFETY
ELECTRONIC SUPPLEMENTARY MATERIAL 3
Cochrane Risk of Bias Summary Table
Cochrane Risk of Bias Table
Allenet 2012
Bias / Authors' judgement / Support for judgementSelection / Unclear risk / Exposed cohort representative of the average population of individuals attending a hospital consultation or being hospitalised in the US. Non-exposed cohort drawn from the same population. Ascertainment of exposure from a database of hospital records. No demonstration that the exposure variable preceded the outcome of interest.
Comparability / Low risk / Quote: "Adjusted on potential confounders: age, sex, myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia, chronic pulmonary disease, connective tissue disease, ulcerative disease, mild liver disease, diabetes without complications, diabetes with complications, hemiplegia, moderate or severe renal disease, malignancy, moderate or severe liver disease, metastatic solid tumour, AIDS, diagnoses of infection-sepsis, inflammatory bowel disease, psychotic disorders, hospital inpatient or outpatient status, obesity, pregnancy, hormone therapy and thrombophilia."
Exposure or Outcome / High risk / Outcome data extracted from the same source used for exposure data during a one year period. It is unclear if the outcome of interest was not present at start of study. Quote: "a major limitation to our results is a problem in the proof of causality between the prescription of AP and the diagnosis of PE because of the absence of a temporal sequence."
Hamanaka 2004
Bias / Authors' judgement / Support for judgementSelection / Unclear risk / Cases and controls are taken from a consecutive series of records of autopsies performed from January 1998 to December 2002 for investigation of the cause of sudden unexpected death. However, definition of cases and controls is implicit.
Comparability / High risk / Adjustment for age, sex and BMI, but no matching.
Exposure or Outcome / Low risk / Ascertainment of exposure with medical records.
Hippisley-Cox 2011
Bias / Authors' judgement / Support for judgementSelection / Low risk / Exposed cohort representative of the target population, quote: "We did a prospective open cohort study in a large population of primary care patients using the QResearch database (version 29). We included all general practices in England and Wales that had been using their computer systems with Egton Medical Information Systems (EMIS) for at least a year. Non-exposed cohort drawn from the same population. Ascertainment of exposure with the QResearch database. The outcome of interest was not present at start of study. This study included only incident cases and excluded patients at risk of VTE
Comparability / Low risk / Adjustment for a number of confounding factors (reported in Box 1 of the study report).
Exposure or Outcome / Low risk / Quote: "Our clinical outcome was incident diagnosis of venous thromboembolism including either deep vein thrombosis or pulmonary embolism, recorded either on the patients’ general practice record using the relevant Read diagnostic codes or on their linked Office of National Statistics cause of death record using the relevant International Classification of Diseases (ICD)-9 codes or ICD-10 diagnostic codes."
Ishiguro 2011
Bias / Authors' judgement / Support for judgementSelection / Low risk / Cases and controls extracted from the UK General Practice Research Database. All eligible cases with outcome of interest included, and the control series is derived from the same population as the cases.
Comparability / High risk / Quote: "Controls matched on age, gender, index date and general practice". However, adjustment for confounders is not mentioned.
Exposure or Outcome / Low risk / Ascertainment of exposure with the General Practice Research Database for both cases and controls.
Jonsson 2008
Bias / Authors' judgement / Support for judgementSelection / Low risk / Case definition: quote: "The cause of death as determined by the pathologist and coded using the International Classification of Diseases (ninth revision) (ICD-9) codes was retrieved from the forensic pathology database. All participants in whom pulmonary embolism was the cause of death (ICD-9 code: 415.1) were identified". Control series is derived from the same population as the cases.
Comparability / High risk / Adjustment for age and sex, no matching.
Exposure or Outcome / Low risk / Quote: "Use of antipsychotic drugs was based on the results of post-mortem analyses."
Jonsson 2009
Bias / Authors' judgement / Support for judgementSelection / Unclear risk / Quote: "We used the hospital registries to identify all patients with a first-time diagnosis of VTE, ie, deep vein thrombosis in a lower limb (ICD-8 code: 451.00 and ICD-10 codes: I80.1, 180.2, 180.3) and/or pulmonary embolism (ICD-8 code: 450.99 and ICD-10 code: 126)." Control series is derived from the same population as the cases. Cases are first occurrence of VTE, but it is not explicitly stated that controls have no history of VTE.
Comparability / Low risk / Quote: "10 population controls for each VTE case, matched on age, sex and county." Adjustment for discharge diagnoses of myocardial infarction, stroke, COPD, peripheral atherosclerosis in the legs, heart failure, diabetes and current use of statins, low dose acetylsalicylic acid, postmenopausal hormone replacement therapy and vitamin K antagonists.
Exposure or Outcome / Low risk / Ascertainment of exposure with the population-based prescription databases of North Jutland and Aarhus Counties for both cases and controls.
Kleijer 2010
Bias / Authors' judgement / Support for judgementSelection / Low risk / Data from a database including pharmacy dispensing records from community pharmacies linked to hospital discharge records of all 950,000 community-dwelling residents of 25 population-defined areas in the Netherlands from 1985 onwards. Quote: "Cases were those patients from the cohort with a primary hospital diagnosis of DVT (ICD-9 code, 453.2-9) or PE (ICD-9 code 415.1) at admission." "For each case, 4 controls were randomly selected from the exposure cohort."
Comparability / Low risk / Quote: "The controls were matched for age, sex, and duration of registration in the database." Adjustment for a pre-defined set of comorbidity and drug use variables.
Exposure or Outcome / Low risk / Ascertainment of exposure with the prescription databases for both cases and controls.
Lacut 2007
Bias / Authors' judgement / Support for judgementSelection / Low risk / Data from an observational study designed to test interactions between genetic and environmental risk factors for VTE. All patients hospitalized with a well-documented symptomatic VTE were eligible for enrolment. VTE had to be objectively confirmed. For each case, one control was selected from the roster of patients hospitalized in the same ward in the 12 months following the case’s event date (the control series is derived from the same population as the cases).
Comparability / Low risk / Quote: "Controls were matched with the cases by age and gender, and had no major acquired risk factors as previously defined." Adjustment for BMI, factor V Leiden and prothrombin G20210A gene variation.
Exposure or Outcome / Low risk / Quote: "Exposure was defined as current use of drugs at the time of admission. All drugs recorded had to be taken at admission for more than 1 week.". Same method of ascertainment for cases and controls.
Liperoti 2005
Bias / Authors' judgement / Support for judgementSelection / Low risk / Data collected in the nursing homes of 5 US States. Eligible candidates were residents 65 years or older. Exposed residents were new users of antipsychotic agents. Non-exposed cohort were drawn from the same population as the exposed cohort. Ascertainment of exposure with medical records. The outcome of interest was not be present at start of study. This study included only incident cases and excluded patients at risk of VTE
Comparability / Low risk / The following confounders were taken into account: body mass index, indicators of functional and cognitive status, history of deep venous thrombosis, history of hip fracture, chronic obstructive pulmonary disease, cancer, and use of anticoagulants, aspirin or antiplatelets, and estrogens. Additional confounders were those variables that altered the estimate of effect by more than 10% after being included in the multivariate model. These variables included age, sex, dementia, depression, peripheral vascular disease, cerebrovascular
disease, heart failure, and diabetes mellitus.
Exposure or Outcome / Low risk / Outcome was assessed using electronic medical records (hospitalization with a primary discharge diagnosis of VTE).
Masopust 2007
Bias / Authors' judgement / Support for judgementSelection / Unclear risk / Data from one ‘hospital’s electronic information system. Quote: "We retrospectively analysed all electronic case records of patients aged 1860 years who have been hospitalized with deep-vein thrombosis and/or pulmonary embolism (case group), or arterial hypertension (control group) at the Department of Internal Medicine, University Hospital in Hradec Kralove, from 1 January 1996 to 31 December 2004. Representativeness of the cases is unclear.
Comparability / High risk / No matching and no adjustment.
Exposure or Outcome / Low risk / Ascertainment of exposure with electronic case records for cases and controls.
Parker 2010
Bias / Authors' judgement / Support for judgementSelection / Low risk / Quote: "Cases were all patients aged between 16 and 100 with a first ever record of venous thromboembolism (deep vein thrombosis or pulmonary embolism) during the study period, including post-mortem diagnoses, identified from computer recorded Read codes (diagnostic codes) for venous thromboembolism recorded within the patient’s electronic record." The control series is derived from the same population as the cases. Controls had no diagnosis of venous thromboembolism in their record. This study included only incident cases and excluded patients at risk of VTE
Comparability / Low risk / Controls matched on age, calendar time, sex, and practice. Analyses were adjusted for socioeconomic status, comorbidity and drug use variables. Quote: "Adjustment was not initially made for smoking or BMI, because of missing data, but after we imputed missing values we repeated the analyses with these variables added to the models."
Exposure or Outcome / Low risk / Exposure ascertained on the basis of prescriptions on or before the index date. Same method of ascertainment for cases and controls.
Parkin 2003
Bias / Authors' judgement / Support for judgementSelection / Low risk / Cases were New Zealand men and women aged 15–59 years who died between 1 January 1990 and 31
December 1998, where the underlying cause was certified as codes 415.1, 451 or 453 of the International
Classification of Diseases (9th revision). For each case, 4 controls were selected from the group medical practice to which the case had belonged on the index date (controls derived from the same population).
Comparability / High risk / Cases and controls were matched for sex and year of birth. Analyses were adjusted for weight and combined oral contraceptive use and hormone replacement therapy within 3 months of the index date. No adjustment for other confounders.
Exposure or Outcome / Low risk / Exposure ascertained on the basis of prescriptions. Same method of ascertainment for cases and controls.
Ray 2002
Bias / Authors' judgement / Support for judgementSelection / Low risk / Data were extracted from linked health care administrative databases covering over 1.4 million senior residents within the province of Ontario. Non-exposed cohort was drawn from the same baseline population. Exposure was ascertained using the linked health care administrative databases. The primary outcome should have occurred after initiation of exposure drugs. This study included only incident cases and excluded patients at risk of VTE
Comparability / Low risk / Analyses were adjusted for age, sex, hospitalization within one year prior to study entry or current residence within a long-term care facility, estrogen, ASA or warfarin.
Exposure or Outcome / Low risk / Outcome was assessed using electronic medical records.
Thomassen 2001
Bias / Authors' judgement / Support for judgementSelection / Unclear risk / Cases were selected from three anticoagulation clinics in the Netherlands. It is unclear if the control series is derived from the same population as the cases.
Comparability / High risk / Quote: "The patients were asked to find a control subject of the same gender, about the same age, who wan not a biological relative, had no history of thrombosis, or malignant disorders and did not use coumarins". Analysis was not adjusted for potential confounders.
Exposure or Outcome / Unclear risk / Ascertainment of exposure based on interviews not blinded to case/control status.
Walker 1997
Bias / Authors' judgement / Support for judgementSelection / Low risk / Data extracted from the US Clozaril National Registry. All patients had recorded prescriptions of clozapine. For each patient, each day after registration was classified as current, recent or past. Past use was used as reference category.
Comparability / High risk / Analyses adjusted for ethnicity, sex and age. No other adjustments were carried out. Patients above 54 years were excluded in order not to include patients prescribed clozapine for Parkinson's disease.
Exposure or Outcome / Low risk / Underlying causes of death were validated.
Wolstein 2000
Bias / Authors' judgement / Support for judgementSelection / Unclear risk / Data from a continuing surveillance programme of severe adverse drug reactions (ADRs) in all inpatients of 35 psychiatric hospitals in Germany and Switzerland. It is not clearly reported how it was assured that the outcome of interest was not present at start of study.
Comparability / High risk / No details reported.
Exposure or Outcome / Low risk / Outcome data collected by specially instructed drug-monitors, according to a detailed protocol.
Zornberg 2000
Bias / Authors' judgement / Support for judgementSelection / Low risk / Data extracted from the UK General Practice Research Database. Quote: "Individuals with a first-time diagnosis of venous thromboembolism between Jan 1, 1990, and Oct 31, 1998, were identified by computer-recorded medical diagnoses. To be included, the diagnosis of VTE had to be confirmed by impedance plethysmography, venogram, ultrasonography, or doppler test, and PE had to be confirmed by ventilation-perfusion scan, computed
tomography, magnetic resonance imaging, or angiography. Controls were selected from the same baseline population. This study included only incident cases and excluded patients at risk of VTE
Comparability / Low risk / Controls were matched to each case on age, sex, general practice attended, years in GPRD, and index date. Analyses were adjusted for smoking status, body-mass index, exposure to oestrogens, antidepressant use, and hypertension.
Exposure or Outcome / Low risk / Data on medication exposure were derived from the computerised prescriptions. Same method for cases and controls.
References