Risk factors for severe respiratory syncytial virus infection during the first year of life: development and validation of a clinical prediction model

Abstract

Background Novel immunisation methods against respiratory syncytial virus (RSV) are emerging, but knowledge of risk factors for severe RSV disease is insufficient for their optimal targeting. We aimed to identify predictors for RSV hospitalisation, and to develop and validate a clinical prediction model to guide RSV immunoprophylaxis for under 1-year-old infants. Methods In this retrospective cohort study using nationwide registries, we studied all infants born in 1997-2020 in Finland (n = 1 254 913) and in 2006-2020 in Sweden (n = 1 459 472), and their parents and siblings. We screened 1 510 candidate predictors and we created a logistic regression model with 16 predictors and compared its performance to a machine learning model (XGboost) using all 1 510 candidate predictors. Findings In addition to known predictors such as severe congenital heart defects (CHD, adjusted odds ratio (aOR) 2.89, 95% confidence interval 2.28-3.65), we identified novel predictors for RSVH, most notably esophageal malformations (aOR 3.11, 1.86-5.19) and lower complexity CHDs (aOR 1.43, 1.25-1.63). In validation data from 2018-2020, the C-statistic was 0.766 (0.742-0.789) in Finland and 0.737 (0.710- 0.762) in Sweden. The clinical prediction model's performance was similar to the machine learning model (C-statistic in Finland 0.771, 0.754-0.788). Calibration varied according to epidemic intensity. Model performance was similar across different strata of parental income. The infants in the 90th percentile of predicted RSVH probability hospitalisation had 3.3 times higher observed risk than the population's average. Assuming 60% effectiveness, immunisation in this top 10% of infants at highest risk would have a number needed to treat of 23 in Finland and 40 in Sweden in preventing hospitalisations. Interpretation The identified predictors and the prediction model can be used in guiding RSV immunoprophylaxis in infants.

Data Resource Profile: Nationwide registry data for high-throughput epidemiology and machine learning (FinRegistry)

“…For example, we are currently developing a clinical prediction model to assess the infant’s risk of severe respiratory syncytial virus-caused disease which aims at helping the administration of novel immunoprophylaxis methods against the disease. 23 Additionally, we are exploring methods to generate latent representations, or embeddings, across all FinRegistry data. These latent representations will help reduce the data dimensionality while identifying the major axes of variation in the underlying data.…”