PT  - JOURNAL ARTICLE
AU  - Christopher Elles Clark
AU  - Daniel Thomas
AU  - Fiona C Warren
AU  - David J Llewellyn
AU  - Luigi Ferrucci
AU  - John L Campbell
TI  - Detecting Risk Of Postural hypotension (DROP): derivation and validation of a prediction score for primary care
AID  - 10.1136/bmjopen-2017-020740
DP  - 2018 Apr 01
TA  - BMJ Open
PG  - e020740
VI  - 8
IP  - 4
4099  - http://bmjopen.bmj.com/content/8/4/e020740.short
4100  - http://bmjopen.bmj.com/content/8/4/e020740.full
SO  - BMJ Open2018 Apr 01; 8
AB  - Objectives Falls are a common problem in older people. Postural hypotension contributes to falls but is often asymptomatic. In the absence of symptoms, postural hypotension is only infrequently checked for in clinical practice. We undertook this study to derive, validate and explore the prospective associations of a prediction tool to identify people likely to have unrecognised postural hypotension.Design and setting Cross-sectional and prospective multivariable cohort analysis.Participants 1317 participants of the Invecchiare in Chianti study, a population-based cohort representative of the older Italian population.Primary outcome measures Predictive value of score to suggest presence of postural hypotension.Methods Subjects were randomised 1:1 to derivation or validation cohorts. Within the derivation cohort, univariable associations for candidate predictors of postural hypotension were tested. Variables with p<0.1 entered multivariable linear regression models. Factors retaining multivariable significance were incorporated into unweighted and weighted Detecting Risk Of Postural hypotension (DROP) scores. These scores were tested in the validation cohort against prediction of postural hypotension, cognitive decline and mortality over 9 years of follow-up.Results Postural hypotension was present in 203 (15.4%) of participants. Factors predicting postural hypotension were: digoxin use, Parkinson’s disease, hypertension, stroke or cardiovascular disease and an interarm systolic blood pressure difference. Area under the curve was consistent at 0.65 for all models, with significant ORs of 1.8 to 2.4 per unit increase in score for predicting postural hypotension. For a DROP score ≥1, five cases need to be tested to identify one with postural hypotension.Increasing DROP scores predicted mortality (OR 1.8 to 2.8 per unit rise) and increasing rates of decline of Mini Mental State Examination score (analysis of variance p<0.001) over 9 years of follow-up.Conclusions The DROP score provides a simple method to identify people likely to have postural hypotension and increased risks to health who require further evaluation.