BACKGROUND: Pneumonia is more common in smokers compared with non-smokers. A high one-year prevalence of lung cancer following hospitalization for pneumonia was demonstrated in heavy smokers.

AIM: To assess the association between hospitalization for pneumonia among ever-smokers and subsequent lung cancer risk.

DESIGN: Retrospective analysis.

METHODS: The study cohort included all ever-smokers aged 55-80 hospitalized for pneumonia between the years 2010-2015 covered by a large medical insurer in Israel. Controls were matched to cases by age in a 4:1 ratio. The primary outcome was the association between hospitalization for pneumonia and subsequent one-year incidence of lung cancer, adjusted for gender, smoking status (past/current) and pack years. Pre-specified sensitivity analyses excluded heavy smokers (smoking history of more than 30 pack years) and patients diagnosed with lung cancer within 30 days of hospitalization, as they probably had clinical or radiological findings suggestive of lung cancer, making them ineligible for screening.

RESULTS: Lung cancer was identified in 275 of 12,807 (2.1%) patients following hospitalization for pneumonia and in 44 of 51,228 (0.1%) controls (adjusted odds ratio 22.46, 95% CI 16.29-30.96, p < 0.001). Among patients hospitalized for pneumonia, one-year lung cancer incidence remained high after excluding heavy smokers and patients diagnosed within 30 days of the index date (1.3% and 1.4%, respectively).

CONCLUSIONS: Hospitalization for pneumonia is associated with high one-year incidence of lung cancer in ever-smokers, supporting the important role of the widely used practice of performing follow up imaging post pneumonia to exclude occult malignancy.

OBJECTIVE: To illustrate the problem of subpopulation miscalibration, to adapt an algorithm for recalibration of the predictions, and to validate its performance.

MATERIALS AND METHODS: In this retrospective cohort study, we evaluated the calibration of predictions based on the Pooled Cohort Equations (PCE) and the fracture risk assessment tool (FRAX) in the overall population and in subpopulations defined by the intersection of age, sex, ethnicity, socioeconomic status, and immigration history. We next applied the recalibration algorithm and assessed the change in calibration metrics, including calibration-in-the-large.

RESULTS: 1 021 041 patients were included in the PCE population, and 1 116 324 patients were included in the FRAX population. Baseline overall model calibration of the 2 tested models was good, but calibration in a substantial portion of the subpopulations was poor. After applying the algorithm, subpopulation calibration statistics were greatly improved, with the variance of the calibration-in-the-large values across all subpopulations reduced by 98.8% and 94.3% in the PCE and FRAX models, respectively.

DISCUSSION: Prediction models in medicine are increasingly common. Calibration, the agreement between predicted and observed risks, is commonly poor for subpopulations that were underrepresented in the development set of the models, resulting in bias and reduced performance for these subpopulations. In this work, we empirically evaluated an adapted version of the fairness algorithm designed by Hebert-Johnson et al. (2017) and demonstrated its use in improving subpopulation miscalibration.

CONCLUSION: A postprocessing and model-independent fairness algorithm for recalibration of predictive models greatly decreases the bias of subpopulation miscalibration and thus increases fairness and equality.