The Birmingham Lung Improvement Studies (BLISS) prognostic score for predicting respiratory hospitalisations among primary care COPD patients:development, internal and external validation in population cohorts

ObjectivesDevelopment, internal, external validation of a prognostic score to predict two-year risk of respiratory admission among individuals with COPD.Design, Setting and participantsModel development and internal validation using 1894 patients from the Birmingham COPD primary care (BLISS) cohort of new and existing COPD patients. External validation using the Evaluation of COPD Longitudinally to Identify Predictive Surrogate Endpoints COPD (ECLIPSE) international cohort (1749 moderate to very-severe COPD patients), and 27,340 COPD patients from the UK primary care Clinical Practice Research Datalink (CPRD) Aurum database linked with Hospital Episode Statistics.MethodsThe model was developed from 23 candidate predictors using multivariable logistic regression with bootstrapping for internal validation and adjustment for overfitting and optimism. Discrimination and calibration were assessed at each stage. Net benefit of the score (clinical utility) was examined across a range clinically relevant risk thresholds compared with using individual score components. Sub-group and sensitivity analyses were conducted in the CPRD. The BLISS score was directly compared with the Bertens’ score in the ECLIPSE cohort. Clinical implementation was explored with relevant stakeholders.Main outcome measureOne or more respiratory admissions within 2 years of cohort entry for development, internal validation and external validation in the CPRD; severe exacerbation within 2 years for external validation in the ECLIPSE cohort. ResultsSix predictors were retained (age, CAT score, respiratory admissions in the previous 12 months, BMI, diabetes, FEV1 % predicted) to form the BLISS score for estimating an individual’s 2-year risk of respiratory admission. The score had similar discrimination performance on internal validation (optimism-adjusted c statistic 0.73 (95% CI 0.70, 0.77)) and external validation (ECLIPSE: c = 0.73 (95% CI 0.71 to 0.76); CPRD: c=0.71 (95% CI 0.70 to 0.72)), and good calibration performance in the BLISS (slope = 0.87 (0.73 to 1.02), CPRD (0.89 (0.85, 0.93)) and ECLIPSE cohorts (0.90 (0.87, 0.93)). Stratified analysis in the CPRD cohort showed it was robust in different population sub-groups. Net benefit analyses demonstrated superiority of the BLISS score over individual predictors and the Bertens’ score (c=0.68 (0.65 to 0.71); calibration slope 0.68 (0.56 to 0.81)).ConclusionsThe BLISS score showed good performance in estimating individual risk of respiratory admission (within two years) in cohorts containing patients from different settings, geographical locations and severities. Four of the included six variables are readily available in primary care records and two, partially available but easy to collect. Impact evaluations are now needed to fully study its use in clinical care. Study registration: ECLIPSE ClinicalTrials.gov Identifier NCT00292552Funding: The BLISS cohort was funded by the UK National Institute for Health Research and ECLIPSE by GlaxoSmithKline (study code SCO104960). External validation using the CPRD was funded by the University of Birmingham Research Development Fund.