To investigate cataract risk associated with the use of atypical antipsychotics.
Retrospective, nested case-control study.
A large health claims database (The British Columbia Ministry of Health Databases) from British Columbia, Canada, was used from January 2000 through December 2007. Cases were defined as clinically significant cataracts requiring surgery and were identified using cataract surgery procedure codes. For each case, 4 to 10 controls were selected randomly using a density-based sampling approach and were matched to cases by age and calendar time. Rate ratios were calculated for users of atypical and typical antipsychotics adjusting for known cataractogenic factors.
One hundred sixty-two thousand five hundred one cases of cataract surgery and 650 004 controls were included. The adjusted rate ratio for current users of atypical antipsychotics was 0.84 (95% confidence interval, 0.80 to 0.89) compared with nonusers. A greater number of prescriptions filled in the year before cataract surgery compared with the median number of filled prescriptions was associated with a lower cataract surgery rate (adjusted rate ratio, 0.70; 95% confidence interval, 0.65 to 0.75) than those with fewer prescriptions filled (adjusted rate ratio, 0.85; 95% confidence interval, 0.79 to 0.91).
A protective association between the use of atypical antipsychotics and risk of clinically significant cataracts requiring surgery was established. Potential biochemical and neurochemical mechanisms for this protective effect are discussed.
Cataracts represent the leading cause of treatable blindness in the world. Cataract surgery also has become one of the most common surgical procedures performed in North America, reducing significant visual morbidity. Cataracts also are responsible for the largest proportion of total direct medical costs ($6.8 billion; 41.9%) among the main causes of adult visual loss in the United States. Identifying factors that increase or decrease risk of cataract have important clinical implications from the patient level to resource allocation at the healthcare management level. Known risk factors include age, female sex, corticosteroid use, diabetes mellitus, hypertension, intraocular surgery, ocular trauma, uveitis, selective serotonin reuptake inhibitors used for an antidepressant effect, and smoking.
Typical antipsychotics, particularly the phenothiazines and with the exception of haloperidol, increase risk of cataract. The association of newer, atypical antipsychotics such as clozapine, olanzapine, risperidone, or quetiapine with cataracts, however, is less well studied. Initial reports of increased cataract in beagle dogs with supratherapeutic doses of quetiapine branded this risk in the safety profile of this medication, despite less convincing human evidence subsequently. Recommendations persist for biannual ophthalmic screening examinations with quetiapine use.
Although variable in methodology, other reports fail to demonstrate significant increased risk of cataracts associated with atypical antipsychotic use. Previous studies do not differentiate between typical and atypical antipsychotics, do not have surgical or cataract severity end points, lack a control group, or have a study design that limits convincing conclusions. For example, in the case-control study by Ruigómez and associates, controls were given a random index date from the main cohort and were matched to the cases by the index date. This approach in the selection of controls does not allow for equal probability of exposure to antipsychotics between the cases and controls. Thus, the exposure prevalence in the controls might have been much lower than the cases, leading to a higher odds ratio. In light of these studies, and given the increase in popularity of atypical antipsychotics because of reduced incidence of extrapyramidal side effects and broader indications beyond psychotic disorders, this pharmacoepidemiologic study was conducted to examine further their association with cataracts.
Ethics approval was obtained for this nested case-control study from the University of British Columbia’s Behavioural Ethics Board. All analyses were performed using SAS software version 9.2 (SAS Institute, Inc, Cary, North Carolina, USA) using 2-sided tests of significance at the P < .05 level.
The data source for this study has been described previously. In brief, we used the British Columbia Ministry of Health Databases as the main source of data for this study. All hospitalizations are captured through the Discharge Abstract Database (DAD). All information on patients’ physician visits are captured through the Medical Services Plan database. Prescription drug information was ascertained through PharmaNet and all prescription drugs dispensed in the province including drug name, dose, day supply and quantity dispensed are captured in PharmaNet. Access to MSP, DAD and PharmaNet were obtained from the British Columbia Ministry of Health and the PharmanNet Stewardship committee. All the data files are linkable through unique identifiers. The British Columbia Ministry of Health Databases is one of the largest longitudinal health claim database in Canada. The database has been used in several pharmacoepidemiologic studies.
The cohort comprised all subjects who had made at least 1 visit to an ophthalmologist in British Columbia from January 2000 through December 2007. Patients entered the cohort on the day of the first ophthalmologic visit and were followed up until any 1 of the following 4 criteria occurred: the diagnosis of a cataract was made, healthcare coverage was terminated, termination of the study period, or death.
All patients newly diagnosed with a cataract after cohort entry were identified. Because cataracts do not require immediate surgical intervention, cataracts were defined as being present in a subject who underwent the first cataract procedure after cohort entry. The date of the first cataract, as defined by a procedure, was deemed the index date.
For each case, a risk set of all eligible controls was formed. Specifically, this included patients with no previous physician services for a cataract diagnosis or a cataract procedure during the follow-up period, as identified by billing and procedure codes. This ensured no identifiable evidence of pre-existing cataracts in these patients. For each risk set, between 4 controls for adequate power and 10 controls where possible to improve the 95% confidence interval (CI) precision were selected randomly and were matched with the cases by age (± 1 year), month and year of cohort entry, and follow-up. This approach ensured that the cases and controls had equal probability of receiving an antipsychotic medication, controlling for physician prescribing trends that may have led to differential prescribing of an antipsychotic to the cases or controls. In most cases, more than 4 controls were not available. Both cases and controls were required to have had 1 year of prescription drug data that would allow assessment of prescription drug use during this period.
Descriptive statistics were examined for cases and controls. A conditional logistic regression model was constructed to compute adjusted rate ratios comparing the rate of cataracts among users of antipsychotics compared with nonusers of the drugs. A density-based sampling approach was used, allowing the rate ratio to approximate the odds ratio closely. We examined the number of cataract surgeries for both atypical antipsychotics (including clozapine, olanzapine, quetiapine, and risperidone) and typical antipsychotics (including chlorpromazine, prochlorpromazine, and haloperidol).
Antipsychotic drugs were defined as the class of medications traditionally formulated to treat the symptoms of psychosis, such as hallucinations, delusions, and disordered thought, primarily by blocking dopaminergic receptors in the brain and were differentiated into typical and atypical by the neuropharmacologic community based on receptor binding profile and affinity. We defined a current user of an antipsychotic as a subject who had used at least 1 prescription within 90 days of the index date. We also examined a duration-response analysis by looking at the number of prescriptions used that was more than or less than the mean number of prescriptions dispensed before the index date. As a quality measure for our study validity, we also examined the number of cataract surgeries associated with 2 unrelated prescription drugs. We quantified the association of cataract surgery with oral corticosteroids, because these drugs have been shown to increase the risk of cataracts. We also examined the association of cataract surgery with proton pump inhibitors (PPIs), an unrelated class of medications that have not been shown to increase the risk of cataracts. In the multivariate conditional logistic regression analysis, we adjusted for the following known risk factors for cataracts: age, sex, history of uveitis (International Classification of Diseases, Ninth Revision, code), hypertension (International Classification of Diseases, Ninth Revision, code), or vitrectomy (procedure code), and use of prescription drugs such as antidiabetic drugs, selective serotonin reuptake inhibitors, and oral steroids.
Table 1 compares the 162 501 cases of cataract surgery and 650 004 age-matched controls included in this study. The ages of cases using atypical, typical, and no antipsychotic use were 74.4 ± 11.8 years, 73.1 ± 9.7 years, and 73.2 ± 10.4 years, respectively. Atypical antipsychotic use among cases was 1.44% (n = 2340) and was 1.81% (n = 11 765) among controls. These relatively low values formed the basis for subsequent analyses that follow. The adjusted (for variables in Table 1 ) rate ratio for cataract surgeries performed in association with atypical antipsychotic use in the previous year was 0.80 (95% CI, 0.77 to 0.84). For current use, the adjusted rate ratio was 0.84 (95% CI, 0.80 to 0.89; Table 2 ). In the year before index, having only 1 prescription of antipsychotic filled was not associated with cataract surgery (adjusted rate ratio, 0.92; 95% CI, 0.83 to 1.01), whereas having been dispensed more than 1 prescription was associated with a lower cataract surgery rate (adjusted rate ratio, 0.78; 95% CI, 0.74 to 0.82; Table 3 ). Furthermore, a longer-duration response, as reflected by a higher number of prescriptions of atypical antipsychotics filled compared with the median in those with more than 1 prescription filled, corresponded to lower cataract surgery rates (adjusted rate ratio, 0.70; 95% CI, 0.65 to 0.75) compared with a lower number of prescriptions filled (adjusted rate ratio, 0.85; 95% CI, 0.79 to 0.91). The adjusted rate ratio for oral steroids (positive control) was 1.17 (95% CI, 1.14 to 1.19), and the adjusted rate ratio for PPIs (negative control) was 1.05 (95% CI, 1.03 to 1.07).
|No.||162 501||650 004|
|Age (y)||73.2 ± 10.5||73.2 ± 10.5|
|Follow-up (y)||1.7 ± 1.8||1.7 ± 1.8|
|Male gender, no. (%)||67 600 (41.6)||285 351 (43.9)|
|Cataract risk factors the year prior to index, no. (%)|
|Uveitis||1462 (0.9)||5200 (0.8)|
|Hypertension||64 025 (39.4)||226 201 (34.8)|
|Vitrectomy||81 (0.05)||130 (0.02)|
|Antidiabetics||20 800 (12.8)||81 250 (12.5)|
|Selective serotonin reuptake inhibitors||6987 (4.3)||27 300 (4.2)|
|Oral steroids||11 537 (7.1)||40 300 (6.2)|
|Cases, No. (%)||Controls, No. (%)||Crude Rate Ratio||Adjusted Rate Ratio a||95% Confidence Interval|
|No. of subjects||162 501||650 004|
|No use in the prior year (%)||98.56||98.19||1.00||1.00||Reference|
|Atypical antipsychotics||2340 (1.44)||11 765 (1.81)||0.79||0.80||0.77 to 0.84|
|Current b||1966 (1.21)||9360 (1.44)||0.84||0.84||0.80 to 0.89|
|Typical antipsychotics||650 (0.40)||2795 (0.43)||0.93||0.88||0.81 to 0.96|
|Current b||276 (0.17)||1300 (0.20)||0.89||0.84||0.74 to 0.96|
a Adjusted for variables in Table 1 .
|Atypical Antipsychotics||Cases, No. (%)||Controls, No. (%)||Crude Rate Ratio||Adjusted Rate Ratio a||95% Confidence Interval|
|1 prescription||503 (0.31)||2210 (0.34)||0.91||0.92||0.83 to 1.01|
|>1 prescription||1852 (1.14)||9555 (1.47)||0.78||0.78||0.74 to 0.82|
|Low continuous users b||1251 (0.77)||4225 (0.65)||0.84||0.85||0.79 to 0.91|
|High continuous users c||796 (0.49)||4550 (0.70)||0.69||0.70||0.65 to 0.75|