Modelling assumptions
| Treatment effectiveness | |
| Prevalence ratio for CMD state at 6 months post-treatment (score ≥9 on SSQ-14): FB vs usual care | 0.21 |
| % of service users entering remission each month post-treatment* | |
| FB | 28 |
| Usual care | 7 |
| Relapse | |
| % of remitters who relapse within 12 months (FB and usual care) | 53 |
| Implied % of remitters who relapse each month* | 6 |
| % of those who relapse who go back into remission within 12 months (FB and usual care) | 49 |
| Implied % of those who relapse who go back into remission each month* | 5 |
| Disability weights† | |
| CMD state (score ≥9 on SSQ-14) | 0.41 |
| Remission (score <9 on SSQ-14) | 0.15 |
| Mortality, % (FB and usual care) | 0.5 |
| Population monthly survival probability (both sexes) | 0.29 |
| Relative mortality risk | 1.71 |
| % of treatment cohort who die in each monthly cycle | 0.5 |
| Costs | |
| FB fixed costs | |
| Scale-up costs (US$) | |
| Phase 1: needs assessment | $64 751 |
| Phase 2: LHW training | $120 709 |
| Phase 3: implementation | $289 382 |
| Total scale-up cost | $474 842 |
| Central programme overhead costs (annual; US$) | |
| Staff | $251 640 |
| Running costs | $24 024 |
| Building occupied (annuitised cost of capital)‡ | $6617 |
| % of central operational costs attributable to FB | 40 |
| Total annual operational cost attributable to FB (total operational cost × 40%) | $112 913 |
| Other programme infrastructure—time input | |
| Patient mobilisation by LHWs (hours per clinic per month) | 9.20 |
| Oversite from district health promotion officers (hours per clinic per month) | 2.00 |
| FB variable costs | |
| Treatment-related activity | |
| Number of clinical assessments undertaken per treatment episode | 7.12 |
| LHW face-to-face treatment time with patient (hours per treated service user) | 3.35 |
| Additional non-patient contact time spent by LHW on administrative duties associated with each treatment episode (hours per treated patient) | 3.35 |
| Time allocated by LHW and LHW supervisors to case review (hours per treated service user) | 0.23 |
| LHW attendance at peer group meetings (hours per treated service user) | 0.44 |
| Usual care resource usage | |
| Health professional time (number of contacts per treated service user) | |
| Public hospital doctor§ | 0.03 |
| Public health clinic doctor§ | 0.03 |
| Psychiatrist§ | 0.03 |
| Community health worker¶ | 0.03 |
| Clinic nurse¶ | 0.07 |
| Counsellor§ | 0.18 |
| Health practitioner unit costs (based on annual salaries) | |
| LHW | $0.23 per hour |
| LHW supervisor | $0.23 per hour |
| Clinic nurse | $0.99 per hour |
| District health promotion officer | $0.99 per hour |
| Community health worker | $0.11 per contact |
| Psychiatrist | $3.75 per contact |
| Counsellor | $0.23 per contact |
| Public doctor (hospital and community) | $1.88 per contact |
| Clinical specialist | $3.75 per contact |
| Discount rate, % | 3 |
Created by the authors.
*To estimate a monthly % of the cohort who transition from remission to relapse or from relapse back into remission, we take the observed % (P) who have transitioned within the period elapsed (t; 12 months) using the reported values from the relevant papers cited in the main text and convert this to a rate of transition ‘r’ using the formula r=[−log(1−P)]/t. The rate is then converted to probability ‘Pr’ (%) using the formula Pr=1−exp(−rt). This method makes the simplifying assumption that the rate of transition from one state to another is constant through time.
†Disability weights (D) are transformed values of Zimbabwean population utility weights (U) applicable to self-reported health states for FB trial participants. Utility scores are located on a scale anchored at 1 (full health) and 0 (death), with negative scores allowed to account for health states viewed as being less preferable to death. The transformation is: D=1−U. This effectively characterises ‘disability’ as a health loss. For example, a disability weight=1 (1 minus U=0) describes a health loss/level of disability equivalent to death; a disability weight=0.1 (1 minus U=0.9) describes a relatively minor health loss/disability level. For modelling, we use the mean derived disability weight for participants at the trial baseline to weight the CMD state (eligible participants were required to have a CMD prior to randomisation); and the mean disability weight for participants identified to be in remission at follow-up (score <9 using the SSQ-14) to weight the remission state.
‡Based on purchase price of property housing central team annuitised assuming a discount rate of 3% and an asset lifetime of 80 years.
§Each contact assumed to use 60 min of health professional time in total, inclusive of patient contact and non-contact time.
¶Each contact assumed to use 30 min of health professional time in total, inclusive of patient contact and non-contact time.
CMD, common mental disorder; FB, Friendship Bench; LHW, lay health worker; SSQ-14, Shona Symptom Questionnaire.