Evaluating the temporal and sociodemographic generalizability of the emergency heart failure mortality risk grade

Background — Emergency Heart Failure Mortality Risk Grade (EHMRG) are a series of models derived in 2004–2007 to predict mortality in acute heart failure (AHF) patients. Although EHMRG models have been validated in Ontario, it remains unknown whether EHMRG maintains temporal validity and performs robustly across different sociodemographic strata.

Objectives — To assess the temporal validity of EHMRG models and determine if EHMRG models perform differently across sociodemographic subgroups.

Methods — We selected a sample of 7,537 patients presenting to 10 emergency departments with an AHF diagnosis between (2017- 2019) in Ontario. We linked our cohort to national databases to obtain the Ontario marginalization index (ON-Marg). We calculated the EHMRG-7P (7-day mortality probability) and EHMRG30-ST (30-day mortality) scores for each patient and assessed their performance using discrimination metrics and calibration plots. We then compared EHMRG performance across several sociodemographic subgroups.

Results — In our temporal validation cohort, EHMRG-7P had a high c-statistic (0.794 [95% CI, 0.764–0.824]) and demonstrated good calibration, with minor overestimation of risk in the highest decile. EHMRG30-ST had a good c-statistic (0.777 [0.759–0.796]) but mildly overestimated risk in the five upper deciles. The discrimination of EHMRG models was robust across subgroups of sex, income quintiles, marital status, and the ON-Marg indices. However, the c-statistic was slightly lower for patients living in long-term-care homes.

Conclusions — EHMRG models retained good model discrimination. Risk was mildly overestimated at 30-days for higher-risk patients, which would lead to conservative recommendations tending towards hospital admission. EHMRG models were robust over most sociodemographic characteristics.