A DRG relative-weight-adjusted defect burden approach for linking inpatient medical record front-page quality and content quality: a single-centre retrospective analysis of 4003 standardized quality-control records

Medical record data underpin diagnosis-related group (DRG) payment, case-mix evaluation and hospital quality governance. Conventional defect rates give equal weight to low- and high-complexity cases. We evaluated whether a relative weight (RW)-weighted defect burden could complement conventional defect rates when linking inpatient medical record front-page quality, content quality and DRG-related case complexity. We conducted a single-centre retrospective case-level observational study using routinely collected inpatient medical record quality-control records from a tertiary general hospital in China. Because the audit form was revised after the first quarter of 2024, the primary analysis was restricted to the standardized period from the second quarter of 2024 to the first quarter of 2026. A total of 4435 linked records were screened; 432 records from the pre-standardization quarter were excluded from the primary analysis, leaving 4003 records. Quality-control results were linked with front-page administrative data and DRG grouping data. Outcomes included any quality defect, front-page key-field defect, process documentation defect, timeliness defect, severe defect and defect count. RW was categorized as < 1.0, 1.0 to < 2.0 and > = 2.0. The RW-weighted defect burden index was calculated as sum(defect count x RW)/sum(RW). Descriptive analysis, trend analysis, group comparisons, sensitivity analysis and multivariable logistic regression were performed. Among the 4003 included records, 2756 (68.85%) had at least one quality defect. Timeliness defects (40.19%), process documentation defects (27.80%), front-page key-field defects (11.49%) and severe defects (16.29%) were the main outcomes. Any defect increased across low-, medium- and high-RW strata (62.38%, 74.65% and 81.11%; P < 0.001), as did severe defects (13.68%, 19.17% and 20.74%; P < 0.001). The RW-weighted defect burden index increased from 1.143 to 1.582 and 1.741 across the three strata. After adjustment, medium RW (odds ratio [OR] 1.332, 95% confidence interval [CI] 1.079–1.645) and high RW (OR 1.770, 95% CI 1.270–2.467) remained associated with any quality defect. Emergency/resuscitation care, surgery, transfusion, CT/MRI examination, pathology examination and bacterial culture were also independently associated with defects. In sensitivity analysis, department-level rankings based on the RW-weighted index were highly correlated with rankings after winsorizing RW at the 95th percentile (Spearman rho = 0.992). Quality defects were more frequent in high-RW and complex-care records. The RW-weighted defect burden complemented conventional defect rates by incorporating DRG resource intensity, but it did not measure reimbursement error, coding accuracy or defect severity directly. It should therefore be interpreted with conventional defect rates, severe-defect rates and sensitivity checks. Linked front-page quality, content documentation, DRG grouping and RW-weighted burden may help hospitals prioritize medical record quality-control review in a risk-oriented manner. Not applicable.