This article is more than 5 years old. Information may no longer be current.
New prediction model provides more accurate BP measurement
Click Here to Manage Email Alerts
A combination of patient characteristics and three clinic BP measurements from a single visit can more accurately predict a patient’s out-of-office BP compared with existing guidelines, researchers reported in Hypertension.
Ambulatory BP readings often can be different than clinic BP readings, leading to misdiagnosis of hypertension. According to James P. Sheppard, PhD, BSc(Hons), from the department of primary care health sciences, University of Oxford, “One phenomenon where readings are higher in the clinic than at home is referred to as the ‘white coat effect.’ This can lead to people being started on [BP]-lowering treatments they do not really need.
“A reverse effect is also seen — some patients have lower readings in the clinic than they would in normal life, meaning that they can miss out on treatment that they could potentially benefit from. Understanding and accounting for the scale of these home-clinic differences would improve diagnosis and treatment,” he said in a press release.
Factors in model
To better understand the home-clinic difference and to develop a model that can more accurately predict out-of-office hypertension, Sheppard and colleagues analyzed data from more than 2,000 patients across six studies. Among the factors assessed to build this prediction model were patient characteristics (age, sex, BMI, alcohol consumption, tobacco use, history of hypertension) and BP characteristics derived from multiple readings (difference between the first and last readings and the rate of change in BP).
The derivation cohort consisted of 991 patients from the Blood Pressure in Ethnic groups and Telemonitoring and Self-Management in the Control of Hypertension studies. The researchers then used 1,172 patients from the remaining four studies for the validation cohort. Patients in both cohorts were similar in age, sex, prevalence of systolic white coat hypertension and systolic masked hypertension.
In the derivation stage, goodness-of-fit was similar between models examining three or six clinic BP readings (derivation stage 1; adjusted coefficient of determination [R2], 0.5-0.52) and those using different BP definitions (derivation stage 2: adjusted R2, 0.5-0.52).
The final prediction model the researchers developed had good calibration across all datasets from the derivation and validation cohorts (Pearson correlation, 0.62-0.8 [systolic]; 0.48-0.8 [diastolic]; P < .001). However, in situations where there was a large masked or white coat effect, the model was less accurate.
Successful predictions
The model was successful at predicting out-of-office hypertension in the derivation cohort (area under the receiver operating curve for systolic model, 0.8; 95% CI, 0.78-0.83; for diastolic model, 0.82; 95% CI, 0.8-0.85) and in the validation cohort (area under the receiver operating curve for systolic model, 0.75; 95% CI, 0.72-0.79; for diastolic model, 0.87; 95% CI, 0.85-0.89).
“We compared the accuracy of our model to the current UK NICE guidelines and those in use in the USA, Canada and Europe. It correctly classified 93% of cases, compared to the next best, the NICE guidelines, which correctly classified 78% of patients,” Sheppard said.
An online calculator based on the prediction model can be found here. – by Tracey Romero
D isclosure: The study was funded by a Medical Research Council Strategic Skills postdoctoral fellowship held by Sheppard with additional support from a NIH Research Program Grant. Please see full study for a list of all other researchers’ relevant financial disclosures.
Click Here to Manage Email Alerts