Participant identification and recruitment
Participants were recruited as part of the PHOENIx trial (trial registration: ISRCTN10585019). All participants were homeless adults aged ≥18 years who had a non-fatal drug overdose in the preceding 6 months. Patients were recruited by independent researchers from a range of venues in Glasgow, Scotland, that included the streets, temporary accommodation, and homeless drop-in centres. Participants were randomised 1:1 to the intervention or control group. Data collected included: demographics; living situation; physical and mental health conditions; illicit drug use; diet; social activities; prescribed medications; physiological measures; blood-borne virus status; treatment burden (PETS); psychological distress (Patient Health Questionnaire 4); and quality of life (EQ-5D-5L). Of 128 participants recruited into the trial, 123 (96.1%) responded to the PETS. All data were collected in person at a face-to-face interview conducted by a research nurse.
Participants in the intervention group were offered weekly visits by the PHOENIx team for 6–9.5 months; these visits used a holistic approach to help patients with a wide range of issues, including physical and mental health problems, problem substance misuse, housing, and welfare. All patients (intervention and control) continued to receive care as usual. Usual care comprised all NHS primary and acute care services, community services (for example, community pharmacies), and social care services. All patients were followed up at 6 and then 9 months to repeat the health assessment, including the PETS. The analyses reported here are of the baseline assessment only (pre-intervention). The results of the PHOENIx trial will be made available in a forthcoming report.
Burden of treatment
Through consultation with its developer, the authors of the current study selected domain scales and slightly modified the original PETS measure (version 2.0), tailoring it for use in this population. PEH were not involved in this process. The original PETS measure was developed to assess treatment burden in patients with chronic health conditions requiring self-management.19 It has been extensively validated in patients with multimorbidity, including those with low income.19,20,23
Version 2.0 of the PETS consists of 60 items divided into 14 content domains (12 multi-item burden scales and two single-item indicators of medication bother): medicines (n = 7), medication side effect bother (n = 1), medication reliance bother (n = 1), medical information (n = 7), medical appointments (n = 6), monitoring health (n = 2), diet (n = 3), exercise and physical therapy (n = 4), interpersonal challenges (n = 4), medical and healthcare expenses (n = 5), difficulty with health services (n = 7), medical equipment (n = 2), role and social activity limitations due to self-management (n = 6), and physical and mental exhaustion due to self-management (n = 5). The modified version used in this study consisted of 52 items divided into 12 content domains. Differences between this version and the original are highlighted below.
The five items of the ‘medical and healthcare expenses’ domain were removed as the content is not relevant to a Scottish population where health care is free at point of access. The three items of the ‘diet’ domain were also removed, as this content was covered elsewhere in baseline data collection. As no single global treatment burden score is generated for the PETS, it is permissible for users to select for use only those domain scales that are most relevant to their study. One item from the ‘medical appointments’ domain was removed as it was deemed similar to a previous question (‘how easy/difficult is it to schedule medical appointments?’); this avoided repetition for the participants in an otherwise lengthy baseline assessment. Participants were also asked to categorise why it was difficult to attend medical appointments to gain information to identify barriers as to why participants may struggle to access their appointments. A single item was added to the ‘medical equipment’ domain to assess how easy/difficult it was to obtain medical equipment. This was added because PEH experience more barriers in accessing health care and this was seen as useful information to add when looking at treatment burden.
Scoring for the domains that were modified (that is, medical appointments and medical equipment) was adjusted slightly to account for the item changes, but followed the same general scoring procedure of the original PETS measure.19 Participants provided responses on either a four- or five-point categorical-ordered response scale, depending on the domain (for example, very easy to very difficult, or strongly agree to strongly disagree). As treatment burden can vary in terms of personal relevance, a ‘not applicable’ response option was available for some domains, and was coded as missing for the purposes of domain scoring. Raw scale scores were computed per guidelines defined by Eton et al19 then transformed to a standardised 0–100 scale with a higher score indicating higher treatment burden.
Aggregate summary scores for ‘workload and impact’ were also calculated.23 A workload summary score was calculated from the mean of the medical information, medications, medical appointments, and monitoring health domains; these are scales that indicate the difficulty of performing the work of self-management. An impact summary score was calculated from the mean of the role and social activity limitations and the physical and mental exhaustion domains as they are scales that indicate the impact of the workload of self-management on a person’s functioning and wellbeing.