Why so much research has not solved it
Adherence has been studied for decades, and the Cochrane review alone brings together 182 randomised trials (Nieuwlaat et al., 2014). The problem is not a lack of studies but the kind of evidence they produce. Most interventions are complex, combining education, counselling, reminders and follow-up, so it is hard to tell which part made the difference. The review authors found no common characteristic among the few trials that improved both adherence and clinical outcomes, and concluded that it remains uncertain how adherence can be improved consistently (Nieuwlaat et al., 2014).
Later overviews reach a similar position. Some approaches, such as pharmacist- or nurse-led support and simpler regimens, show benefit, yet no single strategy works in every setting (Wilhelmsen & Eriksson, 2019). A 2025 umbrella review identified eight promising approaches but noted risk of bias and a lack of head-to-head comparisons (Mackridge et al., 2025). Meanwhile, the global prevalence of non-adherence did not change between 2010 and 2020 (Lee et al., 2022).
- Short follow-up: many trials stop before long-term persistence can be judged.
- Inconsistent measurement: estimates vary with the method used (Lee et al., 2022), and chemical testing thresholds are not standardised (Rabbitt et al., 2024).
- Settings: evidence comes mostly from high-income, secondary care settings, while the burden lies elsewhere (Rabbitt et al., 2024; Aminde et al., 2025).
- System barriers: an intervention aimed at behaviour cannot work if the medicine is not available (World Health Organization, 2025).
Where a doctoral project can add value
A PhD cannot solve non-adherence, but it can close a defined part of the gap. Three angles are realistic. The first is analytical: re-examining the trial literature by component and setting to identify which ingredients matter. The second is methodological: testing how well cheap adherence measures agree with objective testing in a real clinic. The third is applied: a pilot or feasibility trial of a clearly specified intervention in primary care, with a process evaluation that explains the result.
| Angle | Typical data | Main skill |
|---|---|---|
| Evidence synthesis | Published trials and reviews | Systematic review and meta-analysis |
| Measurement study | Self-report, refill and chemical test results from one cohort | Agreement statistics |
| Pilot or feasibility trial | Clinic records, blood pressure readings, interviews | Trial design and mixed methods |
| Routine data analysis | Prescription and refill records | Longitudinal modelling |
Patient voice matters in all of them. The barriers people describe, from cost to fear of side effects (Zhou et al., 2024), should shape what is tested, not only how results are explained.
