Prediction of postural hypotension in the Community ageing Research 75+ (CARE75+) dataset

Problem

Postural hypotension (PH) is a major risk factor for falls, cognitive decline and mortality, although found in approximately 20% of primary care patients it is underdiagnosed. NICE hypertension guidelines advise measuring standing as well as sitting blood pressure (BP) in patients aged >80 years to detect PH; this is seldom undertaken due to time constraints. Prediction tools may support clinicians by identifying people, likely to have PH, for targeted postural BP testing. We previously derived and internally validated such a tool (the DROP score) in an older Italian cohort (InCHIANTI). The DROP score is summative based on 5 factors: taking digoxin, having Parkinson’s disease, hypertension, stroke or cardiovascular disease. The electronic Frailty Index (eFI) is now integrated into primary care computer systems. While frailty is associated with PH, the eFI has not been tested for its ability to predict PH.The Community Ageing Research 75+ (CARE75+) study is a longitudinal cohort study collecting data on ageing and frailty in community-dwelling older people in England. It includes detailed demographic and frailty data, eFI scores and sitting and standing BPs. We aim to i) validate the DROP score within this cohort; ii) explore associations of the eFI with PH, and iii) examine whether the eFI, with the DROP score or with other variables, can best predict PH in this representative UK cohort.

Approach

Cross sectional cohort analyses. We will calculate individual DROP scores and assess their predictive value for PH, using area under receiver operating characteristic (AUROC) curves. We will compare AUROC curves for DROP and eFI scores to explore their incremental predictive values for PH. We will examine the combined effects of DROP and eFI scores in predicting PH through logistic regression analyses, and seek to refine a prediction model incorporating these factors.

Findings

Baseline recruitment data are available for 1004 participants; mean (standard deviation) age 81.5 (4.6) years; 49.50% female. Mean seated systolic BP is 143.7 (21.9) mmHg and diastolic 75.0 (12.2) mmHg. There are 559 (58.4%) participants with a diagnosis of hypertension and 222 (21.1%) have PH, defined as a drop from sitting to standing BP measurements ≥20/10 mmHg. Using eFI categories, 97 (10.3%) are classed as severely frail, 233 (24.1%) moderately frail and 310 (32.2%) have mild frailty. Analyses of models are underway and full findings will be presented to conference.

Consequences

We anticipate reporting the external validity of the DROP score, and developing a refined prediction score likely to include the eFI. Demonstration of an eFI based prediction model for PH should facilitate identification of those primary care patients at highest risk of PH, and therefore with most to gain from the investment of time and resources in measuring sitting and standing BP.