Hennessy, S., Strom, B.L., Berlin, J.A. et al. J Gen Intern Med (1995) 10: 380. doi:10.1007/BF02599836
OBJECTIVES: To measure the incidence of cutaneous hypersensitivity reactions to cotrimoxazole in the setting of primaryPneumocystis carinii pneumonia (PCP) prophylaxis; to measure the incidence of severe reactions; and to identify predictors for these outcomes.
DESIGN: Retrospective cohort study.
SETTING: One university-based outpatient HIV clinic and one university-affiliated internal medicine and infectious disease medical practice.
PATIENTS: Two hundred thirty-six HIV-infected individuals receiving cotrimoxazole for primary PCP prophylaxis.
MAIN OUTCOME MEASURE: Occurrence of a cutaneous hypersensitivity reaction, defined as rash, fever, or pruritus that resulted in permanent discontinuation of cotrimoxazole. Severe reactions were defined as those resulting in hospital admission or systemic treatment with a corticosteroid. Cox regression was used to calculate relative rates (RRs) and 95% confidence intervals (CIs) for a number of clinical and laboratory variables.
MEASUREMENTS AND MAIN RESULTS: Forty-eight (20%) subjects developed cutaneous hypersensitivity reactions, with six (12.5%) of these being severe. In the unadjusted analysis, the following factors demonstrated at least borderline association: male gender [RR (95% CI)=0.46 (0.21–0.99)], higher CD4 percentage [RR (95% CI)=0.95 (0.90–1.00)], syphilis history [RR (95% CI)=0.37 (0.13–1.04)], and higher total protein [RR (95% CI)=0.70 (0.45–1.09)]. Adjustment for potential confounding by measured variables did not meaningfully change these results.
CONCLUSIONS: Cutaneous hypersensitivity reactions to cotrimoxazole in the setting of primary PCP prophylaxis are common. Although male gender, higher CD4 percentage, syphilis history, and higher total protein have at least borderline associations with these reactions, routinely collected clinical and laboratory variables do not appear to be sufficiently associated with the reactions to permit development of a clinically useful prediction rule.