Table 1

Studies on the risk of death associated with antipsychotic medicines
Study Population Outcome Propensity for bias/Study quality Procedures to minimize bias Follow-up Outcome rate in reference group Result
LEVEL I Evidence: Meta-Analyses
Studies that compared atypical antipsychotic (ATYP) treatment to placebo (PLA)
Katz [5] 895 Institutionalised dementia patients (age>=55) Death Low/High Yes 12 weeks 1.8% HR(RISP v PLA); 1.26 95% CI; 0.53-2.99
Schneider [19] 5,204 Dementia patients (age>55) Death Low/High Yes 6-26 weeks 2.3% OR(ATYP v PLA); 1.54 95% CI; 1.06-2.23 RD(ATYP - PLA); 0.01 95% CI; 0.004-0.02
2.8% OR(RISP v PLA); 1.30 95%CI; 0.76-2.23
Haupt [20] 1,721 Alzheimers Patients (mean age 82.3) Death Low/High Yes 4-12 weeks 3.1% RR(RISP v PLA); 1.21 95% CI; 0.71-2.06
LEVEL II Evidence: Randomised Controlled Trial
Studies that compared atypical antipsychotic (ATYP) treatment to placebo (PLA)
Ballard [21] 165 Institutionalised dementia patients (mean age 85) Death Low/High Yes 12 months 33% HR(RISP v PLA); 0.58 95% CI; 0.36-0.92
Studies that compared conventional antipsychotic (C) treatment to placebo (PLA)
DeDeyn [8] 344 Dementia patients (mean age 81 placebo, 82 haloperidol) Death Low/High Yes 12 weeks 3.8% OR(HAL v PLA); 1.68 95% CI; 0.72-3.92 [19]
LEVEL III Evidence: Observational Studies
Studies that compared atypical antipsychotic (ATYP) treatment to non-use (NU)
COHORT STUDIES
Gill [22]

9,100 matched pairs, Non-institutionalised dementia patients in a universal health fund in Ontario Canada

(age>=65)

Death Moderate/Moderate Yes (Propensity score matching, sensitivity analysis) 180 days 8.0% HR(ATYP v NU);1.32 95% CI; 1.12-1.54 RD(ATYP - NU); 1.1 per 100 95% CI; 0.1-2.1
Gill [22] 4,036 matched pairs, Institutionalised dementia patients in a universal health fund in Ontario Canada (age>=65) Death Moderate/Moderate Yes (Propensity score matching, sensitivity analysis) 180 days 15.1% HR(ATYP v NU);1.23 95% CI; 1.05-1.45 RD(ATYP - NU); 1.5 per 100 95% CI; -0.5-3.4
CASE-CONTROL STUDIES
Trifiro [23] 398 cases, 4,023 controls, dementia patients, Integrated Primary Care Information Database (Netherlands)(age>85) Death Moderate/Low Yes (matching on age and duration of dementia) 9 years NA OR(ATYP v NU); 2.2 95% CI; 1.2-3.9
Raivio [24] 254 institutionalized dementia patients (Finland) (age>70) Death Moderate/Low Yes (covariate adjustment) 2 years 49.6% HR(ATYP v NU);0.49 95% CI; 0.24-0.99
Studies that compared conventional antipsychotic (CONV) treatment to non-use (NU)
COHORT STUDIES
Ray [25] 1,282,995 Non-institutionalised dementia, medicaid-enrolled patients (Tennessee) (age 15-84) Sudden cardiac Death Moderate/Moderate Yes (covariate adjustment, sensitivity analysis) 1 year 11.3/10000 PY Moderate Dose >100mg: RR(CONV v NU); 2.39 95% CI; 1.77-3.22 Low Dose <100mg: RR(CONV v NU); 1.3 95% CI; 0.98-1.72
Kales [26] 10,615 Veterans enrolled in VA Serious Mental Illness Treatment Research and Evaluation Centre, Dementia Patients (US) (age>65) Death Moderate/Moderate Yes (Propensity score adjustment, sensitivity analysis, subgroup analysis) 1 year 25.2% RR(NU v CONV); 0.66 95% CI; 0.53-0.82
CASE-CONTROL STUDIES
Trifiro [23] 398 cases, 4,023 controls, dementia patients Integrated Primary Care Information Database (Netherlands) (age>85) Death Moderate/Low Yes (matching on age and duration of dementia) 9 years NA OR(CONV v NU); 1.8 95% CI; 1.4-2.3
Raivio [24] 254 institutionalized dementia patients (Finland) (age>70) Death Moderate/Low Yes (covariate adjustment) 2 years 49.6% HR(CONV v NU); 0.68 95% CI; 0.46-1.03
Studies that compared conventional antipsychotic (CONV) and atypical antipsychotic (ATYP) treatment
COHORT STUDIES
Gill [22] 9,100 matched pairs, Non-institutionalised dementia patients (age>=65) Death Moderate/Moderate Yes (Propensity score matching, sensitivity analysis) 180 days 10.7%

HR(CONV v ATYP); 1.23 95% CI; 1.00-1.50 RD(CONV - ATYP); 2.6 per 100

95% CI; 0.5-4.5

Gill [22] 4,036 matched pairs, Institutionalised dementia patients (age>=65) Death Moderate/Moderate Yes (Propensity score matching, sensitivity analysis) 180 days 17.8% HR(CONV v ATYP); 1.27 95% CI; 1.09-1.48 RD(CONV - ATYP); 2.2 per 100 95% CI; 0.0-4.4
Hollis [27] 16,634 Australian Department of Veterans Affairs Veterans/spouses (Australia) (age>65) Death Moderate/Moderate Yes (covariate adjustment) 2 years 246 per 1000 RR (HALO v OLA); 2.26 95% CI; 2.08-2.47 RR (CHL v OLA); 1.39 95% CI; 1.15-1.67
Hollis [27] 6,602 Institutionalised Australian Department of Veterans Affairs Veterans/spouses (Australia) (age>65) Death Moderate/Moderate Yes (covariate adjustment) 2 years 291 per 1000 RR (HALO v OLA); 1.67 95% CI; 1.50-1.84 RR (CHL v OLA); 1.75 95% CI; 1.31-2.34
Kales [26] 10,615 Veterans enrolled in VA Serious Mental Illness Treatment Research and Evaluation Centre, Dementia Patients (US) (age>65) Death Moderate/Moderate Yes (Covariate and propensity score adjustment, sensitivity analysis, subgroup analysis) 1 year 25.2% Covariate adjusted RR(ATYP v CONV); 0.93 95% CI; 0.75-1.16
Schneeweiss [28] 37,241 British Columbia Residents (Canada) (age>=65) Death Moderate/Moderate Yes (Covariate and propensity score adjustment, instrumental variable analysis) 180 days 9.6% Covariate adjusted HR(CONV v ATYP); 1.32 95% CI; 1.23-1.42 PS adjusted HR(CONV v ATYP); 1.39 95% CI; 1.30-1.49 IV RD(CONV – ATYP); 4.2 per 100, 95% CI; 1.2-7.3
Wang [29] 22,890 Drug Insurance Beneficiaries (Pennsylvania US) (age>=65) Death Moderate/Moderate Yes (Covariate and propensity score adjustment, instrumental variable analysis) 180 days 14.6% Covariate adjusted HR(CONV v ATYP); 1.37 95% CI; 1.27-1.49 Propensity score adjusted adjusted HR(CONV v ATYP); 1.37 95% CI; 1.27-1.49 IV RD(CONV - ATYP); 7.3 per 100 95% CI; 2.0-12.6
Liperoti [30] 9,729 Institutionalised dementia patients (age >=65) Death Moderate/Moderate Yes (covariate adjustment, sensitivity analysis, subgroup analysis) 180 days 40.0 per 100 person-years HR(CONV v ATYP); 1.26 95% CI; 1.13-1.42 HR (HALO v RISP); 1.31 95% CI; 1.13–1.53 HR(Phenothiazines V RISP); 1.17 95% CI; 1.00–1.38 HR(Other Conventional v RISP); 1.32 95% CI; 0.99–1.80
Pratt [31] 7,311 Institutionalised Australian Department of Veterans Affairs Veterans/spouses (Australia) (Age >65) Death Moderate/Moderate Yes (Propensity score adjustment, instrumental variable analysis) 1 year 37.4% Covariate adjusted RD(CONV v ATYP); 8.5 95% CI; 6.2-10.7 Propensity score adjusted RD(CONV v ATYP); 9.1 95% CI; 6.9-11.4 IV RD(CONV - ATYP); 10.1 per 100 95% CI; 6.6-13.7
CASE-CONTROL STUDIES
Trifiro [23] 398 cases, 4,023 controls, dementia patients Integrated Primary Care Information Database (Netherlands) (age>85) Death Moderate/Low Yes (matching on age and duration of dementia) Up to 9 years NA OR(ATYP v CONV); 1.3 95% CI; 0.7-2.4

Pratt et al.

Pratt et al. BMC Medical Research Methodology 2012 12:72   doi:10.1186/1471-2288-12-72

Open Data