Abstract
BACKGROUND: Hospital discharge codes are relied upon for research, accounting/invoicing and health systems planning. Congenital heart disease (CHD), however, is uniquely difficult for non-cardiologists to code due to the rarity, variety and complexity of lesions. It is therefore important that the accuracy of hospital discharge codes is regularly checked to ensure that the prevalence and burden of CHD is being correctly estimated and recorded. METHODS AND RESULTS: We identified all inpatient admissions of adults with CHD to Royal Prince Alfred Hospital in Sydney, Australia from January 2018 to March 2021 (257 admissions, 106 unique patients). The associated discharge coding summaries were extracted and compared to the codes in the separately collected and audited Adult CHD database. Only a quarter of discharge coding summaries contained any diagnosis of CHD, and just one-tenth accurately recorded all appropriate CHD diagnoses. Patients with simple lesions were most likely to have a coded diagnosis of CHD, while those with moderate and complex lesions were much less likely. Moreover, patients admitted under a cardiovascular specialty were twice as likely to have a coded diagnosis of CHD, compared with those admitted under non-cardiovascular specialties (p = 0.006). Overall, less than half of patients had any hospital-coded diagnosis of CHD in any admission over the three-year study period. CONCLUSIONS: Hospital discharge coding dramatically underreports CHD, especially for patients with moderate and severe CHD lesions and for admissions under non-cardiovascular specialties. This suggests that discharge coding-based estimates of the burden of CHD on hospitals and health systems may be substantially underestimated.