Journal reference: Parker C, Coupland C, Hippisley-Cox J. Antipsychotic drugs and risk of venous thromboembolism: nested case-control study. BMJ 2010;341:c4245. 
Published: 21 September 2010
|Antipsychotic drugs are associated with a 32% increased risk of venous thromboembolism (VTE).
The increased risk is higher with:
The increase in absolute risk is low enough that this should not deter antipsychotic use for standard clinical indications for these drugs.
However, caution should be particularly taken in using low potency atypical antipsychotics (e.g., quetiapine) “off indication” in the elderly as the increased absolute risk is substantial (number needed to harm = 185).
Study design:case-control study
determine whether antipsychotic drugs are associated with an increased risk of venous thromboembolism (VTE) and if so, examine risks by type of antipsychotic, potency and dose
- Use of the QResearch database version 16 (anonymised primary care clinical records of over 11 million people registered at any time in the past 16 years with 525 UK general practices)
- the study population extracted out of the database were those:
- registered with the database between 1 Jan 1996 and 1 July 2007
- aged between 16 and 100
- first ever record of deep vein thrombosis (DVT) or pulmonary embolism (both postmortem diagnosis and diagnosis from computer record)
- incidence density sampling to identify up to 4 controls for each case
- they were matched by single year of age, calendar time, sex and practice
- the controls had no recorded diagnosis of VTE at the date of the case’s diagnosis of VTE
- exclusions (cases and controls):
- less than 24 months of data before the index date
- area census data on socioeconomic status missing (usually temporary residents)
- any previous prescriptions for warfarin
- extraction of demographic data from the database including year of birth, sex, socioeconomic status, BMI, smoking status
- individuals were coded for diagnoses of schizophrenia, bipolar disorder, dementia if these diagnoses were present
- individuals were coded for a variety of conditions that are known to increase risk for VTE (see “Data”, page 2) 
- extraction of full prescription data for antipsychotics; individuals were classified as either a current user, recent user (4-12 months), past user (13-24 months) or not exposed in preceding 24 months
- extraction of limited data (whether prescription given in the 24 months prior to index date) for some drugs that increase likelihood of VTE (see “Data”, page 2) 
- statistical analysis comparing cases to controls; multiple conditional logistic regression to estimate odds ratio with 95% confidence interval (CI)
Antipsychotics users risk of VTE:
- adjusted OR 1.32, CI 1.23-1.42
- Current antipsychotic users:
- OR 1.56, CI 1.39-1.75
- Recent users:
- OR 1.36, CI 1.20-1.54
- New users:
- OR 1.97, CI 1.66-2.33
- Past users:
- OR 1.04, CI 0.91-1.18
- greater risk for atypical drugs (though note confidence intervals do meet):
- atypical: OR 1.73, CI 1.37-2.17
- typical: OR 1.28, CI 1.18-1.38
Low vs high potency drugs:
- greater risk for low potency drugs:
- low potency: OR 1.99, CI 1.52-2.62
- high potency drugs: OR 1.28, CI 1.18-1.38
- greater risk for those receiving two or more drugs compared to those receiving one only:
- single drug: OR 1.29, CI 1.20-1.39
- multiple drugs: OR 1.99, CI 1.49-2.65
- highest risk for VTE were from quetiapine, haloperidol and chlorpromazine:
- quetiapine: OR 2.81, CI 1.75-4.50
- haloperidol: OR 2.17, CI 1.55-3.02
- chlorpromazine: OR 1.77, CI 1.27-2.48
- the authors claims to have carried out analyses on individuals after excluding those with schizophrenia, manic depression, or major risk factors for venous thromboembolism
- the results were all similar
There is an association between use of antipsychotic drugs and risk of venous thromboembolism in a large primary care population. The increased risk was more marked among new users and those prescribed atypical antipsychotic drugs.
A total of 453 QResearch practices (general practices in the United Kingdom) met the inclusion criteria with a study population of 7,267,673 patients
- 31,612 incident cases of VTE indentified → 6016 (19%) excluded as less than 24 months of data
- matched to 125,559 controls → 17,595 (14%) excluded as less than 24 months data
- resulting unmatched cases and controls also excluded leaving:
- 25,532 eligible cases (15,975 with DVT and 9,557 with PE)
- 89,491 controls
- Slight predmoninance of women (about 56%)
- Median age = 67
- Analysis of the characteristics of the cases vs. controls demonstrated:
- higher rate of obesity in cases compared to controls (22% vs 14%)
- very slightly more likely to be belonging to a more deprived socioeconomic group
- higher rates of almost all coded comorbidities expected to elevate the rate of VTE
- higher rates of use of almost all medications expected to elevate the rate of VTE or may be associated with conditions that elevate the rate of VTE
- There was missing data (smoking status and BMI) in about a quarter of records, though the investigators did model the potential effects this might have had by imputing values (they claim that there was almost no effect).
- The dataset necessarily only includes patients who have access to primary health care and a regular general practitioner and might exclude the most socially disadvantaged.
- The study is unable to distinguish whether the associations seen are due to the antipsychotics or the conditions requiring the use of the antipsychotics.
- A (very) large and representative cohort of patients in primary health care; the results can be validly generalised to other populations.
- Patient selection and recruitment bias is avoided by direct assess to primary care records of participating general practices.
- Very thorough and comprehensive adjustment of a large number of potential confounders.
- Sufficient participant numbers to allow for subgroup statistical analyses that give odds ratios with relatively narrow confidence intervals.
- An alternate analysis restricted to individuals with no major risk factors for VTE performed; similar results achieved. This improves our confidence that the demonstrated associations are real and independent of other known risk factors.
Biases and conflicts of interests
- Hippisley-Cox J (third author) declared that they are a codirector of QResearch, an organisation that is in joint partnership between the University of Nottingham and EMIS (the leading supplier of IT for 60% of UK general practice).
Clinical relevance to primary health care
This study clearly demonstrates an association of between the use of antipsychotic medications and the risk of venous thromboembolic disease. It seems very probable that this association exists independent of all other common risks factors for VTE. This is important as thrombotic risk is not usually considered an issue with antipsychotic prescribing.
It should be noted that although the increased risk of VTE for any use of antipsychotics in the preceding 24-months is mild (32%), NNH (number needed to harm) = 2640,  some subgroups were associated with a much higher risk. Risks were highest for:
- new users of antipsychotics
- low potency antipsychotics
- atypical antipsychotics
- and multiple antipsychotics.
In particular, quetiapine (a low potency, atypical antipsychotic) was found to be associated with almost a three-fold increased risk of VTE . As a comparison, the typical combined oral contraceptive pill (COCP) increases the risk of VTE by approximately three to five times baseline . As a rule of thumb, patients prescribed continuing atypical antipsychotics could be considered as being at about the same increased risk of VTE as being on the COCP.
One specific scenario associated with very elevated risk is worth mentioning. In users of quetiapine aged 65 years or older, only 185 patients (95% CI 95-445) need be treated for one additional case of VTE.
In the general population VTE is a rare event and appropriate management of conditions requiring antipsychotics should obviously take precedence. However, caution should be taken in those patients already at marked increased risk for VTE (e.g., post-operative patients and patients with cancer). A particular scenario to be wary is the use of atypical antipsychotics off indication in elderly patients (e.g., for delirium or for sedation).
Note: this study cannot differentiate the direct effect of antipsychotics and the condition for which it is treated. Elements of the results do point towards an effect of the medication itself (highest associated risk for new users and a modest dose-effect relationship) but causation should not be considered demonstrated by this study alone.
- Parker C, Coupland C, Hippisley-Cox J. Antipsychotic drugs and risk of venous thromboembolism: nested case-control study. BMJ 2010;341:c4245
- Dunn N. Oral contraceptives and venous thromboembolism. BMJ 2009; 339:b3164