Impact of chlorhexidine baths on suspected sepsis and bloodstream infections in hospitalized neonates in Zambia.

INTRODUCTION: Sepsis is the leading cause of infectious morbidity and mortality among hospitalized neonates. In high-resource pediatric and adult intensive care units, use of aqueous chlorhexidine (CHG) solution has been associated with reduced risk of bloodstream infections (BSI). OBJECTIVES: To assess the impact of bathing of neonates with 2% CHG on BSI, suspected sepsis, and mortality in a low-income country neonatal care unit. METHODS: We conducted a secondary analysis of data from the Sepsis Prevention in Neonates in Zambia (SPINZ) study, a prospective observational cohort study performed at a large public referral hospital in Lusaka, Zambia. The SPINZ study assessed the impact of an infection control bundle (consisting of alcohol hand rub, SMS hygiene reminders, enhanced environmental cleaning, and CHG baths for babies ≥1.5 kg) on sepsis, BSI, and all-cause mortality. Episodic shortages in study staffing resulted in some enrolled babies not receiving a CHG bath. Using Longitudinal Targeted Maximum Likelihood Estimation and Cox proportional hazards regression to adjust for observed confounding, we estimated the causal effect of receiving a CHG bath within the first 3 days of life on suspected sepsis, BSI, and death among inborn babies enrolled during the study implementation and intervention phases. RESULTS: The majority of inborn, enrolled babies ≥1.5 kg received a CHG bath within 3 days of NICU admission (864 of 1233, 70%). We found that CHG bathing reduced the hazard rate of BSI among inborn babies ≥1.5 kg by a factor of 0.58 (p = 0.10, 95% CI: 0.31, 1.11), corresponding to an absolute risk reduction of 9.6 percentage points within a week of admission (p = 0.002, 95% CI: 3.4-15.7 percentage points). We did not find a statistically significant effect of CHG bathing on culture-negative sepsis (p = 0.54) or death (p = 0.85). CONCLUSION: In our single center study, CHG bathing at admission was associated with a reduced risk of BSI due to a pathogenic organism after adjusting for potential confounding. Our results suggest that CHG may be an effective intervention for preventing neonatal sepsis in high-risk, low-income country settings.

Beyond prediction: A framework for inference with variational approximations in mixture models.

Variational inference is a popular method for estimating model parameters and conditional distributions in hierarchical and mixed models, which arise frequently in many settings in the health, social, and biological sciences. Variational inference in a frequentist context works by approximating intractable conditional distributions with a tractable family and optimizing the resulting lower bound on the log-likelihood. The variational objective function is typically less computationally intensive to optimize than the true likelihood, enabling scientists to fit rich models even with extremely large datasets. Despite widespread use, little is known about the general theoretical properties of estimators arising from variational approximations to the log-likelihood, which hinders their use in inferential statistics. In this paper we connect such estimators to profile M-estimation, which enables us to provide regularity conditions for consistency and asymptotic normality of variational estimators. Our theory also motivates three methodological improvements to variational inference: estimation of the asymptotic model-robust covariance matrix, a one-step correction that improves estimator efficiency, and an empirical assessment of consistency. We evaluate the proposed results using simulation studies and data on marijuana use from the National Longitudinal Study of Youth.