A digital quality-improvement program to reduce mortality among intubated children in a public pediatric intensive care unit in Colombia: a pre–post study with interrupted time-series analysis and a protocol for multisite replication

Introduction Mortality among intubated children in public pediatric intensive care units (PICUs) in lowand middle-income countries remains high, driven in part by gaps in workforce training, team coordination, and standardized processes. We evaluated a low-cost, multidisciplinary digital education and team-based quality-improvement (QI) intervention—a multilingual online curriculum, telesimulationbased team training, standardized airway and handoff workflow tools, and on-site visits—in the PICU of a public pediatric hospital in Cartagena, Colombia. Methods In this single-center pre–post QI study, we included all intubated patients (<18 years) admitted to the PICU of Casa del Niño during a baseline period (1 October 2023–30 April 2024), a washout period of near-complete nursing turnover (1 April–30 June 2025), and a post-intervention period (1 July–22 October 2025). The primary outcome, 30-day all-cause mortality, was analyzed by multivariable logistic regression adjusted for age, sex, and the logit of Pediatric Index of Mortality 3 (PIM3)-predicted mortality, calculated without diagnosis-risk terms for all patients because these were not recorded for later patients, supplemented by additional case-mix adjustment, an interrupted time-series analysis of all 272 intubated patients admitted between October 2023 and October 2025 (with the implementation period also modeled as a separate transition segment), and sensitivity analyses. Cost-effectiveness was explored as incremental cost per death and per disability-adjusted life-year (DALY) averted. Results Of 144 patients analyzed (95 pre-intervention; 49 post-intervention), age, sex, and median PIM3 were similar between periods, but mean PIM3-predicted mortality was twice as high before the intervention (10.7% versus 5.3%), reflecting 10 pre-intervention patients with bilaterally fixed pupils (none after). All-cause mortality was 28.4% before and 10.2% after the intervention (risk difference −18.2 percentage points, 95% confidence interval [CI] −29.6 to −4.1; relative risk 0.36, 95% CI 0.15–0.87; adjusted odds ratio 0.26, 95% CI 0.09–0.77). PIM3 under-predicted mortality in both periods (standardized mortality ratio 2.68 and 1.94, respectively). Findings were consistent in case-mix-adjusted analyses and in interrupted time-series analysis with implementation modeled as a separate transition segment (level-change odds ratio 0.24, 95% CI 0.07–0.82). Mortality was higher during implementation (35.9%) than at baseline (27.6%), which was not explained by measured severity of illness. Under the most conservative assumptions (the entire 5-year program cost attributed to the approximately two deaths averted when expected deaths were estimated with PIM3, with 3% discounting), the intervention cost up to approximately US$4,800 per DALY averted, below Colombia’s gross domestic product per capita but above half of it; less conservative scenarios gave lower estimates. Discussion Implementation of this program was associated with lower mortality among intubated children in one public Colombian PICU. Given the single-center, non-randomized design, small sample, and imperfect severity adjustment, these findings require confirmation in controlled multisite evaluations, for which we provide an implementation protocol.