Antibiotic selection for purulent skin and soft-tissue infections in ambulatory care: a decision-analytic approach

Abstract

Objective: Community-associated methicillin-resistant Staphylococcus aureus (CA-MRSA) has caused a nationwide epidemic of skin and soft-tissue infections in ambulatory pediatrics. Antibiotic treatment recommendations suggest incorporating local epidemiology for the prevalence of CA-MRSA. We sought to identify the antibiotic strategy with the highest probability of activity and to identify threshold values for epidemiologic variables including bacterial prevalence and antibiotic resistance.

Methods: We used decision analysis to evaluate 3 empiric antibiotic strategies: clindamycin, trimethoprim/sulfamethoxazole (T/S), and cephalexin. We calculated the probability of activity against the bacteria causing the infection (CA-MRSA, methicillin-sensitive S. aureus and group A Streptococcus [GAS]) by incorporating estimates of prevalence and antibiotic resistance to determine the optimal strategy. Sensitivity analysis was used to identify thresholds for prevalence and antibiotic resistance where 2 strategies were equal.

Results: Clindamycin (0.95) and T/S (0.89) had substantially higher probability of activity than cephalexin (0.28) using baseline estimates for bacterial prevalence and antibiotic resistance. Cephalexin was the optimal antibiotic only when CA-MRSA prevalence was <10%. The probability of activity for clindamycin and T/S was highly sensitive to changes in the values for bacterial prevalence (both CA-MRSA and GAS) and CA-MRSA resistance to clindamycin.

Conclusions: Empiric treatment of skin and soft-tissue infections with either clindamycin or T/S maximizes the probability that the antibiotic will be active when CA-MRSA prevalence is >10%. Deciding between T/S and clindamycin requires consideration of antibiotic resistance and prevalence of GAS. This model can be customized to local communities and illustrates the importance of ongoing epidemiologic surveillance in primary care settings.

Figure 1

Schematic of decision tree for each antibiotic strategy.

Figure 2

One-way sensitivity analysis for the prevalence of CA-MRSA and the probability of antibiotic activity for cephalexin, clindamycin and trimethoprim/sulfamethoxazole (T/S).

Figure 3

One-way sensitivity analysis for the probability of CA-MRSA resistance to clindamycin and the probability of antibiotic activity for clindamycin and trimethoprim/sulfamethoxazole (T/S).

Figure 4

Two-way sensitivity analysis for the prevalence of GAS and the probability of CA-MRSA resistance to clindamycin for treating with clindamycin or trimethoprim/sulfamethoxazole (T/S). Shaded areas reflect which antibiotic has the highest probability of activity.

Figure 4

Two-way sensitivity analysis for the prevalence of CA-MRSA and the probability of CA-MRSA resistance to clindamycin for treating with clindamycin or trimethoprim/sulfamethoxazole (T/S). Shaded areas reflect which antibiotic has the highest probability of activity.