Background Certain broad medication classes have previously been associated with high rates of hospitalisation due to related adverse events in elderly Western Australians, based on clinical coding recorded on inpatient summaries. Similarly, some medications from the Beers Criteria, considered potentially inappropriate in older people, have been linked with an increased risk of unplanned hospitalisation in this population. Objective Our objective was to determine whether risk estimates of drug-related hospitalisations are altered in elderly patients taking 'high-risk drugs' (HRDs) when specific Beers potentially inappropriate medications (PIMS) are taken into consideration. Methods Using the pharmaceutical claims of 251,305 Western Australians aged >= 65 years (1993-2005) linked with other health data, we applied a case-time-control design to estimate odds ratios (ORs) for unplanned hospitalisations associated with anticoagulants, antirheumatics, opioids, corticosteroids and four major cardiovascular drug groups, from which attributable fractions (AFs), number and proportion of drug-related admissions were derived. The analysis was repeated, taking into account exposure to eight specific PIMs, and results were compared. Results A total of 1,899,699 index hospitalisations were involved. Of index subjects, 12-57 % were exposed to each HRD at the time of admission, although the proportions taking both an HRD and one of the selected PIMs were much lower (generally <= 2 %, but as high as 8 % for combinations involving temazepam and for most PIMs combined with hypertension drugs). Included PIMs (indomethacin, naproxen, temazepam, oxazepam, diazepam, digoxin, amiodarone and ferrous sulphate) all tended to increase ORs, AFs and drug-related hospitalisation estimates in HRD combinations, although this was less evident for opioids and corticosteroids. Indomethacin had the greatest overall impact on HRD ORs/AFs. Indomethacin (OR 1.40; 95 % confidence interval [CI] 1.27-1.54) and naproxen (OR 1.22; 1.14-1.31) were associated with higher risks of unplanned hospitalisation than other antirheumatics (overall OR 1.09; 1.06-1.12). Similarly, among cardiac rhythm regulators, amiodarone (OR 1.22; 1.13-1.32) was riskier than digoxin (OR 1.08; 1.04-1.13). For comparisons of drug-related hospitalisation estimates, temazepam yielded the greatest absolute increases, especially with hypertension drugs. Conclusions Indomethacin and temazepam should be prescribed cautiously in elderly patients, especially in drug combinations. Furthermore, it appears other antirheumatics should be favoured over indomethacin/naproxen and, in situations where both drugs may be appropriate, digoxin over amiodarone. Our methodology may help assess the safety of new medications in drug combinations in preliminary pharmacovigilance investigations.
