Harm Among Reported Patient Safety Incidents Across Age Groups in Mexican Tertiary Care Hospitals

Patient safety incidents remain an important challenge for quality of care and hospital performance. Evidence from low- and middle-income countries is limited, particularly on whether harm among reported incidents differs across age groups. This study examined age-related differences in harm among reported inpatient patient safety incidents in Mexican public tertiary care hospitals.

The authors conducted a cross-sectional analysis using administrative incident-report data from 10 public tertiary care hospitals in Mexico. The study analyzed 180 eligible reported inpatient incidents identified between January and September 2019. Incident type was categorized as near miss, no-harm incident, adverse event, or sentinel event. Harm severity was classified using the National Coordinating Council for Medication Error Reporting and Prevention (NCC MERP) Index and recoded as no harm vs. harm. The authors estimated adjusted predicted probabilities of harm and 95% confidence intervals (CIs) by age group using logistic regression with robust standard errors, adjusting for work shift and primary diagnosis.

Of all reported incidents, 12.2% were near misses, 38.3% were no-harm incidents, 46.7% were adverse events, and 2.8% were sentinel events. Overall, 49.5% of reported incidents involved harm. The adjusted probability of harm among reported incidents was 49.5% (95% CI 40.7%-58.3%) overall, 77.8% (95% CI 65.1%-90.5%) among children younger than 5 years, 48.2% (95% CI 31.9%-64.5%) among those aged 5-18 years, and 22.2% (95% CI 8.6%-35.9%) among adults.

In this sample of reported inpatient incidents, harm was more frequent among younger children, particularly those younger than 5 years. These findings support age-stratified review of reported incidents, strengthening of pediatric safety processes, and improvement of hospital reporting and learning systems. The estimates describe harm within a sample of reported incidents and should not be interpreted as population-level incidence.