Clinical outcomes and mortality before and after implementation of a pediatric sepsis protocol in a limited resource setting: A retrospective cohort study in Bangladesh

Abstract

Background: Pediatric sepsis has a high mortality rate in limited resource settings. Sepsis protocols have been shown to be a cost-effective strategy to improve morbidity and mortality in a variety of populations and settings. At Dhaka Hospital in Bangladesh, mortality from pediatric sepsis in high-risk children previously approached 60%, which prompted the implementation of an evidenced-based protocol in 2010. The clinical effectiveness of this protocol had not been measured. We hypothesized that implementation of a pediatric sepsis protocol improved clinical outcomes, including reducing mortality and length of hospital stay.

Materials and methods: This was a retrospective cohort study of children 1-59 months old with a diagnosis of sepsis, severe sepsis or septic shock admitted to Dhaka Hospital from 10/25/2009-10/25/2011. The primary outcome was inpatient mortality pre- and post-protocol implementation. Secondary outcomes included fluid overload, heart failure, respiratory insufficiency, length of hospital stay, and protocol compliance, as measured by antibiotic and fluid bolus administration within 60 minutes of hospital presentation.

Results: 404 patients were identified by a key-word search of the electronic medical record; 328 patients with a primary diagnosis of sepsis, severe sepsis, or septic shock were included (143 pre- and185 post-protocol) in the analysis. Pre- and post-protocol mortality were similar and not statistically significant (32.17% vs. 34.59%, p = 0.72). The adjusted odds ratio (AOR) for post-protocol mortality was 1.55 (95% CI, 0.88-2.71). The odds for developing fluid overload were significantly higher post-protocol (AOR 3.45, 95% CI, 2.04-5.85), as were the odds of developing heart failure (AOR 4.52, 95% CI, 1.43-14.29) and having a longer median length of stay (AOR 1.81, 95% CI 1.10-2.96). There was no statistically significant difference in respiratory insufficiency (pre- 65.7% vs. post- 70.3%, p = 0.4) or antibiotic administration between the cohorts (pre- 16.08% vs. post- 12.43%, p = 0.42).

Conclusions: Implementation of a pediatric sepsis protocol did not improve all-cause mortality or length of stay and may have been associated with increased fluid overload and heart failure during the study period in a large, non-governmental hospital in Bangladesh. Similar rates of early antibiotic administration may indicate poor protocol compliance. Though evidenced-based protocols are a potential cost-effective strategy to improve outcomes, future studies should focus on optimal implementation of context-relevant sepsis protocols in limited resource settings.

Fig 1. icddr,b Pediatric sepsis protocol adapted from the Surviving Sepsis Campaign.

Fig 2. Subject identification and inclusion in the analysis.

Fig 3. Mortality rate pre- and post-protocol implementation.

Error bars represent 95% confidence intervals and the p-value for mortality pre- and post-protocol implementation is shown.

Fig 4. Key protocol time points and events.

Bar graph shows the proportion of patients pre- and post-protocol implementation that received antibiotics within 60 minutes, the first fluid bolus received within 60 minutes, a blood transfusion, and an admission to the ICU. Error bars represent 95% confidence intervals and p-values for each comparison pre- and post-protocol are shown.