Generalizability of predictive models for Clostridioides difficile infection, severity and recurrence at an urban safety-net hospital.

INTRODUCTION: Predictive models for Clostridioides difficile infection can identify high-risk patients and aid clinicians in preventing infection. Issues of generalizability regarding current predictive models have been acknowledged but, to the authors' knowledge, have never been quantified. METHODS: C. difficile infection, severity and recurrence predictive models were created using multi-variate logistic regression through case-control sampling from an urban safety-net hospital. Models were validated using five-fold cross-validation, and inverse probability weights (IPW) based on two different catchment area definitions were used to improve external validity. Akaike Information Criterion (AIC), area under the receiver operating characteristic curve (AUROC), and sensitivity and specificity with bootstrapped confidence intervals (CI) were used to assess and compare model fit and performance. RESULTS: Changes in performance before and after weighting were small across all models, although differences were more apparent after weighting the recurrence model (AUROC values of 0.78, 0.76 and 0.71 for the unweighted and two weighted models, respectively). Overall, the infection model performed the best (AUROC 0.82, 95% CI 0.78-0.85), followed by the recurrence model (AUROC 0.78, 95% CI 0.69-0.86) and then the severity model (AUROC 0.70, 95% CI 0.63-0.78). CONCLUSIONS: The performance of the models after weighting did not change drastically, suggesting that the models predicting C. difficile infection, severity and recurrence may not be impacted by patient selection factors. However, other researchers may wish to consider addressing these catchment forces using IPW.

Evaluating a neonatal intensive care unit MRSA surveillance programme using agent-based network modelling.

BACKGROUND: Surveillance for meticillin-resistant Staphylococcus aureus (MRSA) in neonatal intensive care units (NICUs) is a commonplace infection prevention strategy, yet the optimal frequency with which to monitor the unit is unknown. AIM: To compare various surveillance frequencies using simulation modelling. METHODS: One hundred NICU networks of 52 infants were simulated over a six-month period to assess MRSA transmission. Unit-wide surveillance occurred every N weeks where N={1,2,3,4}, and was compared with the current NICU policy of dynamic surveillance (i.e. weekly when at least one positive screen, otherwise every three weeks). For each surveillance period, colonized infants received a decolonization regimen (56% effective) and were moved to isolation rooms, if available. FINDINGS: As the surveillance frequency increased, the mean number of MRSA-colonized infants decreased, from a high of 2.9 (four-weekly monitoring) to a low of 0.6 (weekly monitoring) detected per episode. The mean duration of colonization decreased from 307 h (four-weekly monitoring) to 61 h (weekly monitoring). Meanwhile, the availability of isolation rooms followed an inverse relationship: as surveillance frequency increased, the availability of isolation rooms decreased (61% isolation success rate for four-weekly monitoring vs 49% success rate for weekly monitoring). The dynamic policy performed similar to a biweekly programme. CONCLUSIONS: An effective MRSA surveillance programme needs to balance resource availability with potential for harm due to longer colonization periods and opportunity for development of invasive disease. While more frequent monitoring led to greater use of a decolonization regimen, it also reduced the likelihood of isolation rooms being available.

Density of day cares in relation to reported pertussis incidence in Philadelphia.

OBJECTIVES: Reported pertussis incidence has increased markedly in recent years. In addition to the documented increase in under-immunization and waning immunity, the increase may be related to the more frequent use of child care services by parents over the last few decades. Additionally, clustering of outbreaks may be related to neighborhood characteristics not previously identified. STUDY DESIGN: We conducted a citywide case-control study of children in Philadelphia aged birth through six years, between 2001 and 2013. Cases were reported as probable pertussis diagnoses to the Health Department. Controls were sampled from the city's immunization information system and matched to the cases by date of birth. METHODS: Multilevel logistic regression was used to isolate the independent contributions of individual and neighborhood risk factors and the corresponding relative odds of pertussis. The density of day cares in each neighborhood served as the main exposure and reported incident cases of confirmed and probable pertussis was the main outcome. RESULTS: Between 2001 and 2013, 410 cases of confirmed and probable pertussis were included with four controls matched per case yielding a final sample of 2050 children from 45 Philadelphia neighborhoods. There was a 30% increase in the risk of pertussis based solely on the neighborhood where the children resided (median odds ratio 1.3, 95% credible interval 1.1, 1.6). The density of day cares in each neighborhood was unrelated to the distribution of pertussis cases. CONCLUSIONS: Pertussis clustering was observed at the neighborhood level in Philadelphia, but was unrelated to the neighborhood's day care density. From a Health Department perspective, the highest risk neighborhoods should be targeted for vaccine campaigns and further research to identify the etiologic risk factors.

Characteristics of late-onset sepsis in the NICU: does occupancy impact risk of infection?

OBJECTIVE: Infants in neonatal intensive care units (NICUs) are vulnerable to a variety of infections, and occupancy in the unit may correlate with risk of infection. STUDY DESIGN: A retrospective cohort of infants admitted to the NICUs between 1997 and 2014. Survival analysis was used to model the relative hazard of sepsis infection in relation to two measures of occupancy: 1) the average census and 2) proportion of infants <32 weeks gestation in the unit. RESULT: There were 446 (2.3%) lab-confirmed cases of bacterial or fungal sepsis, which steadily declined over time. For each additional percentage of infants <32 weeks gestation, there was an increased hazard of 2% (hazard ratio 1.02, 95% confidence interval: 1.00, 1.03) over their NICU hospitalization. Census was not associated with risk for infection. CONCLUSION: During times of a greater proportion of infants <32 weeks gestation in the NICU, enhanced infection-control interventions may be beneficial to further reduce the incidence of infections.