Skip to main content

Psychometric properties of intentional and unintentional non-adherence measurement tools

Cite this dataset

Fahrni, Mathumalar Loganathan; Saman, Kamaliah Md (2022). Psychometric properties of intentional and unintentional non-adherence measurement tools [Dataset]. Dryad.


Background: Distinguishing intentional from unintentional non-adherence (persistent forgetfulness) is key when contemplating interventions to optimise medication adherence.

Objective: To categorise patient-reported-outcome measures (PROM) into their propensity to detect intentional and unintentional non-adherence to medication, and to synthesise their psychometric properties.

Design: Review’s conduct and reporting were based on the PRISMA-2020 checklist and COSMIN-2018 guidelines. We searched publications from January 1980 –1 March 2022 in MEDLINE, EMBASE, CINAHL, International Pharmaceutical Abstracts and Cochrane Library. Articles describing the PROM development and validation of the scoring methods to categorise patients into non or adherent, and those which incorporated a proxy measure for adherence were included. We assessed the quality of studies using the COSMIN risk of bias tool and included studies which described a minimum of 2 psychometric properties. Open-ended PROM items, which were without a method of scoring, or translated versions were excluded. To ensure an exhaustive binning process, 2 reviewers independently evaluated and coded every criterion-item for possible inclusion into both an outcome and a process domain bin deemed fit.

Outcomes: Outcome domain-bins were intentional, unintentional or mixed. Process domains were behaviour (e.g., self-efficacy), barrier (e.g., impaired dexterity), or belief (e.g., perception). PROM were assigned an overall category of either intentional(I), unintentional (UI) or mixed (UI/I) depending on the collective leading domain of the individual criterion-items.

Results: 59 PROM were validated among patients aged 18 to 85 years in the USA and UK, Europe (Netherlands, Belgium, France, Spain, Switzerland), Middle East (Turkey) and Australasia (Malaysia, China, Australia).  44 PROM were primarily intentional (I=491 criterion-items) while 13 were mixed (I=79/ UI=50) and the remaining 2, unintentional (UI=5). The corresponding process domains were belief (n=383 items), barrier (n=192 items), and behaviour (165 items). Psychometrics evaluation revealed heterogeneity for: i) PROM definitions – severity cut-off scores, and scoring method (ordinal, dichotomised, visual analogue scale), and for ii) criterion validity –varied proxy measures  used (biomarkers, pill count, electronic monitors). One’s belief, behaviour, and barriers encountered were likely indicators for I, UI and mixed U/I non-adherence respectively.

Conclusion: Medication non-adherence that is primarily intentional, is highly prevalent. Eight  intentional and 2 partial intentionality PROM exhibited adequate psychometric properties. Fortification of patients’ knowledge and illness perception, as opposed to daily reminders alone, is necessary.


After the process of searching and selecting articles based on the inclusion and exclusion criteria for patient reported outcome measures of non-adherence, we employed an additional step of binning each criterion of every PROM. PROMs are essentially made up of sets of criteria and the scoring can be ordinal, intergral, or with interval ratio etc. The total score determines whether a patient is adherent or non.  Data were collated from the PROMs and each of their criterion was binned into 2 groups of domain - process and outcome. We binned criterion on "behaviour", "barrier" and "belief" into process domain-bins, and "unintentional", "intentional" and "mixed" into outcome domain-bins. Data was collated for n=59. The variables were tabulated and each PROM had numerical variables containing number of items in: "intentional", unintentional" and "mixed", "behaviour", "barrier", "belief" and string variables for "condition/ disease/ setting" and "age category".