Bjurlin, M. A. | bioRxiv Science

bioRxiv2018

The Charlson Index is insufficient to control for comorbidities in a national trauma registry

BackgroundThe Charlson Comorbidity Index (CCI) is frequently used to control for confounding by comorbidities in observational studies, but its performance as such has not been studied. We evaluated the performance of CCI and an alternative summary method, logistic principal component analysis (LPCA), to adjust for comorbidities, using as an example the association between insurance and mortality.\n\nMaterials and MethodsUsing all admissions in the National Trauma Data Bank 2010-2015, we extracted mortality, payment method, and 36 ICD-9-derived comorbidities. We estimated ORs for the association between uninsured status and mortality before and after adjusting for CCI, LPCA, and separate covariates. We also calculated standardized mean differences (SMDs) of comorbidity variables before and after weighting the sample using inverse probability of treatment weights (IPTW) for CCI, LPCA, and separate covariates.\n\nResultsIn 4,936,880 admissions, most (68.3%) had at least one comorbidity. Considerable imbalance was observed in the unweighted sample (mean SMD=0.086, OR=1.17), which was almost entirely eliminated by IPTW on separate covariates (mean SMD=0.012, OR=1.36). The CCI performed similarly to the unweighted sample (mean SMD=0.080, OR=1.25), while 2 LPCA axes were better able to control for confounding (mean SMD=0.04, OR=1.31). Using covariate adjustment, the CCI accounted for 56.1% of observed confounding, whereas 2 LPCA axes accounted for 91.3%.\n\nConclusionsThe use of the CCI to adjust for confounding may result in residual confounding, and alternative strategies should be considered. LPCA may be a viable alternative to adjusting for each comorbidity when samples are small or positivity assumptions are violated.

epidemiology↗

bioRxiv2017

Lack of insurance is associated with lower probability of diagnostic imaging use among US trauma patients: An instrumental variable analysis and simulation

BackgroundUninsured trauma patients have higher mortality than their insured counterparts. One possible reason is disparities in utilization of appropriate diagnostic imaging, including computed tomography (CT), X-ray, ultrasound (US), and magnetic resonance imaging (MRI). We examined the association between lack of insurance and use of diagnostic imaging.\n\nMethodsData come from the National Trauma Databank 2010-2015. Patients were determined uninsured if payment mode was self-pay or missing. The primary outcome was any diagnostic imaging procedure, and secondary outcomes included CT, X-ray, US, or MRI. Risk ratios (RRs) were adjusted for demographics, comorbidities, injury characteristics, facility characteristics. We also used the 2010 Patient Protection and Affordable Care Act as an instrumental variable (IV), with linear terms for year to account for annual trends in imaging use. Monte carlo simulations to test effect of hypothetical violations to IV assumptions of relevance, no direct effect, and no confounding.\n\nResultsOf 4,373,554 patients, 953,281 (21.8%) were uninsured. After adjusting, uninsured patients had lower chance of any imaging (RR 0.98, 95% CI 0.98 to 0.98), x-ray (RR 0.99, 95% CI 0.99 to 1.00), and MRI (RR 0.82, 95% CI 0.81 to 0.83), and higher chance of ultrasound (RR 1.01, 95% CI 1.01 to 1.02). In IV analysis, uninsured status was associated with reduction in any imaging (RR 0.60, 95% CI 0.52 to 0.70), tomography (RR 0.52, 95% CI 0.44 to 0.62) ultrasound (RR 0.46, 95% CI 0.32 to 0.65), and MRI (RR 0.19, 95% CI 0.10 to 0.37) and increased likelihood of x-ray use (RR 1.74, 95% CI 1.31 to 2.32). Simulations indicated that a direct effect RD of -0.02 would be necessary to produce observed results under the null hypothesis.\n\nDiscussionOur study suggests an association between insurance status and use of imaging that is unlikely to be driven by confounding or violations of IV assumptions. Mechanisms for this remain unclear, but could include unconscious provider bias or institutional financial constraints. Further research is warranted to elucidate mechanisms and assess whether differences in diagnostic imaging use mediate the association between insurance and mortality.

epidemiology↗