Both this database and pharmacy dispensing records were checked to identify discrepancies.

The inpatient regimen was considered correct if it matched the outpatient regimen. For those patients not followed at the hospital HIV clinic, admission data were also checked to rule out transcription errors. Drug–drug interactions were checked for contraindicated or not recommended combinations using national and international MI-503 mw HIV websites [9–11]. If an error or interaction was detected, the pharmacist phoned the attending physician or nurse or added a footnote with a recommendation to the computerized prescription, so that the attending physician could see it the following day. The acceptance of the pharmacist’s recommendations was also reviewed during the following days. If the error was not corrected within 48 h of the recommendation, the prescription was classed as not accepted. Data were entered into an Access 2.0 database (Microsoft Corp., Redmond, WA, USA). For the descriptive analysis, qualitative variables were expressed as percentages and frequencies; quantitative variables were expressed as the mean (standard deviation [SD]). Fisher’s exact test was used to analyse contingency tables. Odds ratios (ORs) for risk factors associated with HAART-related problems

were analysed using a generalized estimating equation model. This multivariate model takes into account the correlation between different admissions belonging to the same patient. The statistical analysis was performed ZD1839 http://www.selleckchem.com/products/KU-60019.html using stata (StataCorp. 2007, Stata Statistical Software, Release 10; Stata Corporation, College Station, TX, USA). Over a 1-year period, we reviewed the prescriptions for 247 admissions of 189 HIV-infected patients who received antiretroviral therapy. Forty-one patients were admitted more than once during the study period. Table 1

summarizes the demographic characteristics of these patients. The distribution of admissions by service was as follows: infectious diseases unit, 135 (54.7%); other medical units, 58 (23.5%); surgery services, 38 (15.4%); intensive care units, nine (3.6%); and units with surgical and nonsurgical patients, seven (2.8%). A total of 60 antiretroviral drug-related problems were identified in 41 patients (21.7% of the admitted patients had at least one antiretroviral problem). The types of HAART-related errors found are shown in Table 2. The most common was drug–drug interaction (33.3%), not only between antiretroviral agents, but also between antiretrovirals and other drugs. Atazanavir was the drug most commonly involved in interactions. The second most common problem was incorrect dose (16.7%), and the third most common was dose omission (15%), followed by lack of dosage reduction in patients with renal or hepatic impairment (11.7%), omission of one or more antiretroviral medications (10%), addition of an alternative antiretroviral drug (8.3%) and incorrect schedule according to outpatient treatment (5%).