JMIR Publications

ABSTRACT

Background:

Assessing disease progression among uncontrolled hypertension (HTN) patients is important for identifying opportunities for intervention.

Objective:

To develop and validate two models, one to predict sustained, uncontrolled HTN (multiple blood pressure (BP) readings ≥ 140/90 mm Hg or a BP ≥ 180/120 mm Hg) and one to predict hypertensive crisis (BP ≥ 180/120 mm Hg) within 1 year of an index visit (outpatient/ambulatory encounter in which an uncontrolled BP reading was recorded).

Methods:

Data from 142,897 patients with uncontrolled HTN within Atrium Health Greater Charlotte in 2018 were used. EHR-based predictors were based on the 1-year period before a patient’s index visit. The dataset was randomly split into a training set (80%) and a validation set (20%). Four machine learning frameworks were considered: L2-regularized logistic regression, multilayer perceptron, gradient boosting machines, and random forest. Model selection was performed with 10-fold cross-validation. The final models were assessed on discrimination (C-statistic), calibration (e.g., integrated calibration index (ICI)), and net benefit (with decision curve analysis). Additionally, internal-external cross-validation (IECV) was performed at the county level to assess performance with new populations and summarized using random-effect meta-analyses.

Results:

In internal validation, C-statistic and ICI were 0.72 (95% CI: 0.71 - 0.72) and 0.015 (0.012 - 0.020) for the sustained, uncontrolled HTN model, and 0.81 (0.79 - 0.82) and 0.009 (0.007 - 0.011) for the hypertensive crisis model. The models had higher net benefit than the default policies (i.e., treat-all and treat-none) across different decision thresholds. In IECV, the pooled performance was consistent with internal validation results.

Conclusions:

An EHR-based model predicted hypertensive crisis reasonably well in internal and internal-external validations. The model can potentially be used to support population health surveillance and HTN management. Further studies are needed to improve the ability to predict sustained, uncontrolled HTN.