Pharmacist-oriented evaluation of inpatient management of primary hypertension using a problem-based care framework and exploratory predictive modeling

Pharmacy Practice

  • Fayiz Adel Aldheisat1Department of Clinical Pharmacy and Therapeutics, Applied Science Private University (ASU), Amman 11931, Jordan.
  • Heba A Khader2Department of Clinical Pharmacy and Pharmacy Practice, Faculty of Pharmaceutical Sciences, The Hashemite University, Zarqa 13133, Jordan.
  • Luai Z. Hasoun3Biopharmaceutics and Clinical Pharmacy Department, Faculty of Pharmacy, Al-Ahliyya Amman University, Amman-19328, Jordan.
  • Zekrayat J.H. Medras4College of Pharmacy, Amman Arab University, P.O. Box 2234, Amman 11953, Jordan.
  • Muath Alsarafandi5Faculty of Medicine, Islamic University of Gaza, P.O. Box 108, Gaza, State of Palestine.
  • Yazan Alsayed6Al-Rayhan Medical Center, Amman, Jordan.
  • Ahmad R. Alsayed1Department of Clinical Pharmacy and Therapeutics, Applied Science Private University (ASU), Amman 11931, Jordan.

Volume 24 Issue 3 Pages 1-18

DOI: 10.18549/PharmPract.2026.3.3685

Abstract

Background: Primary hypertension is a major global risk factor for cardiovascular disease and premature mortality. Despite the availability of diverse pharmacological and non-pharmacological interventions, clinical practice often faces challenges. Aim: To evaluate whether integrating structured clinical frameworks, specifically the Medical Problem-Oriented Plan (MPOP), with machine learning (ML) techniques can facilitate a systematic, data-driven assessment of care effectiveness and safety using real-world clinical datasets. Methods: This study investigated datasets of inpatients (n = 267 screened, 96 eligible). Clinical data were structured according to the MPOP framework to identify and categorise medical problems. Logistic regression, Bayesian logistic regression with Bernoulli likelihood, and alternative models (negative binomial, zero-inflated models, and gradient boosting) were applied to assess effectiveness and safety. Model performance was evaluated using accuracy, Area Under the Receiver Operating Characteristic curve (AUROC), precision–recall, and cross-validation metrics. Results: The effectiveness domain revealed frequent treatment problems: 41% of cases demonstrated suboptimal therapy effectiveness, and 29% reported at least one safety concern. Logistic regression models showed modest discrimination (AUROC = 0.63; accuracy = 61%), limited by small sample size and the presence of zero-inflated variables. Conclusion: A structured MPOP framework for pharmacists is feasible for assessing inpatient hypertension care and identifying medication issues related to effectiveness and safety. Predictive modeling showed promise for exploration, but it requires larger datasets before it can be used in real-world practice.

Keywords

  • primary hypertension
  • medical problem-oriented plan
  • medication-related problems
  • machine learning
  • predictive modelling
  • personalised therapeutics
Pharmacy Practice

Loading…