Surgical Checklist Success in Latin America

The purpose was to determine the impact of the implementation of the World Health Organization (WHO) Surgical Safety Checklist in terms of morbidity and mortality in adult surgical patients in a tertiary healthcare institution in Chile.

After Institutional Review Board (IRB) approval (Facultad de Medicina, Pontificia Universidad Católica de Chile, Santiago, Chile), a retrospective analysis of all surgical encounters on patients age 15 and above from January 2005 to December 2012 at our center will be reviewed.

Encounter data will include up to 14 diagnostic and procedure International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes, demographic data, date of admission and discharge, emergency status, healthcare system used and in-hospital death. A 5-level "high risk" variable was created in order to account for surgical complexity and associated in-hospital mortality (level 1, surgeries with <1% in-hospital mortality; level 2, 1% to <5%; level 3, 5% to <10%; level 4, 10% to <15%; level 5, > or = 15%)6.

Surgical heterogeneity will be calculated by the Internal Herfindahl Index, which represents the diversity or comprehensiveness of the types of procedures performed at a facility.

Statistics:

Propensity score (PS) analysis will be used to control for differences in baseline characteristics. The PS is the conditional probability of receiving an exposure (e.g. checklist) given a set of measured covariates. To estimate the PS, a logistic regression model will be used in which "treatment" status (checklist performed vs. not performed) will be regressed on the baseline (pre-treatment) characteristics.

PS analysis will be implemented in two ways to control for confounding:

• PS matching: matching will be performed using a one-to-one nearest neighbor caliper matching without replacement with a caliper size of 0.2 standard deviations. Balances in the distribution of baseline covariates will be assessed by estimating absolute standardized differences of the covariates between the two groups before and after matching. Any imbalanced covariates (standardized difference >10%) after matching will be adjusted for in the final analysis. As the PS matched sample does not consist of independent observations, we will use a marginal regression model with robust standard errors.
• PS weighting: the entire sample will be weighted by the inverse probability of the treatment weights derived from the PS. If a subject has a higher probability of being in a group, it will be considered over-represented and therefore will be assigned a lower weight. Conversely, if the subject has a smaller probability, it will be considered as under-represented and will be assigned a higher weight. We then will fit a weighted linear regression model using an indicator variable representing checklist intervention status as the sole predictor, and mortality as our outcome variable.

Data will be expressed as mean (SD; standard deviation) or median (IQR, interquartile range) unless otherwise stated. A two-sided p value less than 0.05 will be considered significant. The analyses will be performed using STATA v.12.0 (StataCorp, College Station, TX).

Inclusion criteria

• All surgical patients

Exclusion criteria

• Obstetrical patients delivering vaginally
• Patients less than 15 years old