A new prediction tool for managing patients with suspected high blood pressure in primary care could reduce by half the number of people needing to wear 24-hour ambulatory blood pressure monitors to confirm a diagnosis of hypertension.
Published today in the BMJ, a study led by Oxford and Birmingham University researchers, with Professor Jonathan Mant, head of the Primary Care Unit at Cambridge University, shows how applying a simple computer algorithm can correctly classify patients with hypertension in 97% of cases.
High blood pressure – also known as hypertension – affects more than 1 in 4 adults in England. While often preventable, it is a leading cause of cardiovascular disease and is the biggest risk factor for death and disability internationally.
Blood pressure levels can fluctuate throughout the day and can easily change as a result of stress, physical activity or even talking, so clinic-based readings are not always representative of an individual’s true blood pressure.
To get a true reading, 24-hour ambulatory monitoring is the gold standard for diagnosing hypertension and is currently recommended for all patients with raised blood pressure readings in the clinic. This involves wearing a monitor throughout the day and night, with readings taken automatically every 15–20 minutes, which can be uncomfortable for some people.
To help clinicians diagnose hypertension more effectively, the researchers analysed data from 2000 patients to understand the differences between clinic and 24-hour blood pressure readings, and developed an algorithm that can accurately predict true blood pressure.
This algorithm has now been evaluated on 887 patients with suspected hypertension recruited through 10 general practices and one hospital in England. Doctors participating in the study took three blood pressure readings during their consultation to decide whether the patient required 24-hour monitoring. The patient’s readings were also inputted into the prediction tool along with basic information from their electronic health record including their age, gender, BMI, history of hypertension and presence of cardiovascular disease.
The algorithm identified whether the patient had a normal blood pressure, a high blood pressure or required further investigation. While it correctly predicted patients with hypertension in almost all cases when compared with true blood pressure readings, although around a quarter of patients without hypertension could have received potentially unnecessary treatment.
Lead author, Dr James Sheppard, Senior Research Fellow in Oxford University’s Nuffield Department of Primary Care Health Sciences, said: “Getting an accurate blood pressure reading in a clinical setting can present quite a challenge, given how blood pressure changes throughout the day.”
“By using this simple online prediction tool, GPs, nurses or other healthcare assistants can quickly determine whether a patient requires treatment or further monitoring to diagnose hypertension. We have found this approach can accurately capture patients with hypertension requiring treatment, and significantly reduces the amount of work doctors need to do to diagnose them.”
The study was funded by the Medical Research Council.
The online prediction tool is freely available online here: https://sentry.phc.ox.ac.uk/proof-bp
The full paper ‘Prospective external validation of the Predicting Out-of-OFfice Blood Pressure (PROOF-BP) strategy for triaging ambulatory monitoring in the diagnosis and management of hypertension: observational cohort study‘, is published in the BMJ
Media queries: Lucy Lloyd, communications manager, Primary Care Unit