Estimating the effect of antihypertensive treatments: evidence from Network meta-analysis and six-state Markov modelling

Mahsa Nazari, University of Bern

Hypertension is a major cause of cardiovascular disease (CVD) and mortality and is highly prevalent among older adults. Antihypertensive treatment can reduce CVD incidence and associated mortality. There are evidence suggesting that hypertension treatment should be personalized taking into account age and absolute CVD risk which may change the benefit-harm ratio of treatment. Several treatment options are available to treat hypertensive patients. We need however methods to compare their effects one against each other, notably to rank them. Network meta-analysis (NMA) is an extension of conventional pairwise meta-analysis (MA) that allows researchers to synthesize evidence from a network of trials that compared the benefits and harms of multiple treatments. NMA can provide estimates for comparisons that have never compared in a study by using indirect evidence. NMA allows also to rank treatments according to benefits and harms. Currently four NMAs exist on antihypertensive treatments for preventing CVDs, but they did not stratify patients by age and absolute CVD risk, assess harms or rank treatments. From an epidemiological prospective, another issue is to better understand how antihypertensive treatment influence blood pressure (BP) changes across the lifetime. One solution lies in analysing longitudinal data by using multi-state continues-time Markov model. The states in the model refer to different BP levels categories. For example, we might be interested in transition from a high-normal state when systolic BP is between 130 and 140 mmHg to Grade 3 hypertension state when systolic BP is higher than 180 mmHg. With this modelling strategy, we can investigate how a subject moves between BP categories over time and the impact of multiple factors on the transition from one category to another, called transition intensity. Few studies have modelled the transitions between BP categories with Markov model, and none has assessed the influence of drug treatment and how it differs by age and absolute CVD risk.

Keywords: Hypertension, cardiovascular disease, Network meta-analysis and multi-state Markov model

Short Biography

Mahsa Nazari, is a PhD student in Public Health (Epidemiology and Biostatistics) at the Institute of Primary Health Care (BIHAM) at the University of Bern since March 2019. She was trained as statistician at the University of Esfahan (BSc) and Biostatistician at Shahid Beheshti university of Medical Sciences (MSc) at Tehran. Ms. Nazari worked as data analyser. Her PhD project focuses on assessing the efficacy of treatments for people with hypertension: evidence from Network meta-analysis and six-state Markov modelling.

« back to overview