Skip to main content

Research Repository

Advanced Search

Assessing the multidimensional relationship between medication beliefs and adherence in older adults with hypertension using polynomial regression.

Dillon, Paul; Phillips, L. Alison; Gallagher, Paul; Smith, Susan M.; Stewart, Derek; Cousins, Gr�inne


Paul Dillon

L. Alison Phillips

Paul Gallagher

Susan M. Smith

Derek Stewart

Gr�inne Cousins


The Necessity-Concerns Framework (NCF) is a multidimensional theory describing the relationship between patients' positive and negative evaluations of their medication, which interplay to influence adherence. Most studies evaluating the NCF have failed to account for the multidimensional nature of the theory, placing the separate dimensions of medication 'necessity beliefs' and 'concerns' onto a single dimension (e.g. the Beliefs about Medicines Questionnaire-difference score model). The purpose of this study was to assess the multidimensional effect of patient medication beliefs (concerns and necessity beliefs) on medication adherence using polynomial regression with response surface analysis. Community-dwelling older adults over 65 years old (n = 1,211), presenting their own prescription for antihypertensive medication to 106 community pharmacies in the Republic of Ireland, rated their concerns and necessity beliefs to antihypertensive medications at baseline and their adherence to antihypertensive medication at 12 months via structured telephone interview. Confirmatory polynomial regression found the difference-score model to be inaccurate; subsequent exploratory analysis identified a quadratic model to be the best-fitting polynomial model. Adherence was lowest among those with strong medication concerns and weak necessity beliefs, and adherence was greatest for those with weak concerns and strong necessity beliefs (slope β = -0.77, p < 0.001; curvature β = -0.26, p = 0.004). However, novel nonreciprocal effects were also observed; patients with simultaneously high concerns and necessity beliefs had lower adherence than those with simultaneously low concerns and necessity beliefs (slope β = -0.36, p = 0.004; curvature β = -0.25, p = 0.003). The difference-score model fails to account for the potential nonreciprocal effects. The results therefore extend evidence supporting the use of polynomial regression to assess the multidimensional effect of medication beliefs on adherence.


DILLON, P., PHILLIPS, L.A., GALLAGHER, P., SMITH, S.M., STEWART, D., COUSINS, G. 2018. Assessing the multidimensional relationship between medication beliefs and adherence in older adults with hypertension using polynomial regression. Annals of behavioral medicine [online], 52(2), pages 146-156. Available from:

Journal Article Type Article
Acceptance Date Jan 11, 2018
Online Publication Date Jan 11, 2018
Publication Date Feb 28, 2018
Deposit Date Jan 15, 2018
Publicly Available Date Jan 12, 2019
Journal Annals of behavioral medicine
Print ISSN 0883-6612
Electronic ISSN 1532-4796
Publisher Oxford University Press
Peer Reviewed Peer Reviewed
Volume 52
Issue 2
Pages 146-156
Keywords Beliefs about Medicines Questionnaire (BMQ); Medication adherence; Necessity concerns framework; Polynominal regression
Public URL