Recruitment
The target sample size for the pilot RCT was 100 participants, who were to be recruited over 8 months (January to August 2018) from five CPs (see Chapter 1, Progression criteria).
Recruitment activity took place over a period of 14 months (March 2018 to April 2019) and from across a total of 15 CPs, indicating that recruitment was slower than anticipated, as was the case in the feasibility study. Thirteen CPs were recruited from the start of the pilot RCT (a larger number of CPs were recruited from the outset of the pilot RCT than originally planned given the difficulties experienced with recruitment in the feasibility study); two additional CPs were recruited after the commencement of the pilot RCT (from May 2018) and the original recruitment period of 8 months was extended (by 6 months). Five general practices conducted database searches and mailed out study information packs to potentially eligible patients (eligible participants recruited via general practices were associated with five recruiting CPs). Figure 4 shows a breakdown of recruitment over the 14-month recruitment period.
FIGURE 4 Pilot RCT: target and actual recruitment rates. (Figure not reproduced here; see the original article.)
Overall, 7342 study information packs were distributed: 4892 via pharmacy-based recruitment methods and 2450 via GP-based recruitment methods. This resulted in 201 (2.7%) pharmacy customers/patients returning a consent form indicating their interest in the study, and 166 (82.6% of those contacted) agreeing to complete a screening interview. Table 16 details the distribution of study information packs, returned consent forms and resultant screened and recruited participants. Recruitment activity by pharmacy is shown in Table 17.
| Recruitment method | Packs given out (n) | Customers/patients who consented to contact, n (%) | Screened, n (%) | Randomised, n (%) |
|---|---|---|---|---|
| In the CP | 1480 | 44 (3.0) | 37 (2.5) | 8 (0.5) |
| Via home-delivered prescriptions | 404 | 20 (5.0) | 18 (4.5) | 5 (1.2) |
| CP system searches (nine conducted) | 3008 | 66 (2.2) | 50 (1.7) | 14 (0.5) |
| General practice searches (five conducted) | 2450 | 71 (2.9) | 61 (2.5) | 17 (0.7) |
| Total | 7342 | 201 (2.7) | 166 (2.3) | 44 (0.6) |
| Pharmacy | Packs given out (n) | Returned consent forms, n (%)a | Screened, n (%)a | Randomised, n (%)a | ||||
|---|---|---|---|---|---|---|---|---|
| Pharmacy | GP | Pharmacy | GP | Pharmacy | GP | Pharmacy | GP | |
| 1 | 435 | 600 | 23 (5.3) | 18 (3.0) | 21 (5.1) | 14 (2.3) | 1 (0.2) | 2 (0.3) |
| 2 | 570 | 0 | 18 (3.2) | – | 15 (2.6) | – | 5 (0.9) | – |
| 3 | 454 | 0 | 10 (2.2) | – | 8 (1.8) | – | 0 (0.0) | – |
| 4 | 141 | 0 | 4 (2.8) | – | 2 (1.4) | – | 1 (0.7) | – |
| 5 | 285 | 0 | 5 (1.8) | – | 5 (1.8) | – | 3 (1.1) | – |
| 6 | 449 | 0 | 6 (1.3) | – | 5 (1.1) | – | 1 (0.2) | – |
| 7 | 45 | 400 | 4 (8.9) | 7 (1.8) | 4 (8.9) | 7 (1.8) | 2 (4.4) | 2 (0.5) |
| 8 | 408 | 450 | 13 (3.2) | 7 (1.6) | 7 (1.7) | 5 (1.1) | 1 (0.2) | 1 (0.2) |
| 9 | 828 | 600 | 11 (1.3) | 24 (4.0) | 11 (1.3) | 21 (3.3) | 2 (0.2) | 8 (1.3) |
| 10 | 467 | 0 | 13 (2.8) | – | 10 (2.1) | – | 2 (0.4) | – |
| 11 | 216 | 0 | 8 (3.7) | – | 7 (3.2) | – | 3 (1.4) | – |
| 12 | 487 | 0 | 9 (1.8) | – | 5 (1.0) | – | 2 (0.4) | – |
| 13 | 0 | 400 | – | 15 (3.8) | – | 14 (3.5) | – | 4 (1.0) |
| 14 | 106 | 0 | 5 (4.7) | – | 4 (3.8) | – | 4 (3.8) | – |
| 15 | 1 | 0 | 1 (100.0) | – | 1 (100.0) | – | 0 (0.0) | – |
| Total | 4892 | 2450 | 130 (2.7) | 71 (2.9) | 105 (2.1) | 61 (2.5) | 27 (0.6) [20.8]b | 17 (0.7) [23.9]b |
Reasons for not fully completing the screening interview are given for all patients; these included uncontactable (n = 17), no longer interested (n = 14), found to be ineligible (n = 1) and other (n = 3). A total of 159 participants (95.8% of those who agreed to a screening interview) were found to be eligible based on the presence of a LTC.
Overall, 45 participants (27.1% of the 166 screened; 22.3% of the 201 contacted) were found to be eligible for inclusion in the study. The reasons for ineligibility can be found in Table 18. The most common reasons for exclusion were being non-depressed (35.9%) or currently experiencing a major depressive episode (31.4%), similar to the results of the feasibility study.
| Reason | Number of patients excluded (N = 156), n (%) |
|---|---|
| Non-depressed | 56 (35.9) |
| Major depression | 49 (31.4) |
| Current psychological treatment | 3 (1.9) |
| Serious mental illness present | 1 (0.6) |
| Alcohol or drug dependence | 4 (2.6) |
| Actively suicidal | 2 (1.3) |
| No relevant LTC | 6 (3.9) |
| No longer interested | 14 (9.0) |
| Not contactable | 16 (10.3) |
| Other | 5 (3.2) |
A total of 44 (97.8% of those eligible) participants were recruited to the study, from 13 out of 15 recruiting CPs (86.7%). The average number of participants recruited per pharmacy was 2.9, ranging from 0 to 10 participants.
Randomisation
Of the 44 participants who were recruited to the study, 24 (54.6%) were randomised to the intervention arm and 20 (45.5%) were randomised to the usual-care arm. The number of participants randomised by pharmacy, and to which treatment arm of the study, can be seen in Figure 5, which also details the number of potential participants screened. The flow of participants through the study is shown in Figure 6.
FIGURE 5 Pilot RCT: overall number of participants screened and randomised by pharmacies (by treatment arm). (Figure not reproduced here; see the original article.)
FIGURE 6 Pilot RCT: CONSORT flow diagram. a, Multiple reasons for exclusion may be given. (Figure not reproduced here; see the original article.)
As can be seen in Figure 5, the number of packs and resulting number of randomised participants varied largely between the CPs, especially as not all CPs benefited from associated recruitment via a general practice search. Table 17 provides additional information on recruitment rates and conversion rates (from the number of packs given out to the number of participants randomised into the trial) via the different recruitment methods (pharmacy based or general practice based) and indicates variability across CPs. Overall, pharmacy-based and general practice-based recruitment methods resulted in a similar number of participants being randomised: 0.6% for pharmacy-based methods and 0.7% for general practice-based methods.
Participant characteristics
The demographic characteristics of the randomised participants can be found in Table 19. The average age of the participants was 67.6 years, ranging from 20.8 to 89.1 years. There was a roughly equal split of male and female participants, and all except one participant classed themselves as of white ethnicity. The most common health problem reported was arthritis, which 30 participants stated that they experienced, followed by cardiovascular (n = 24) and respiratory (n = 19) conditions.
| Characteristic | Treatment arm | Overall (N = 44) | |
|---|---|---|---|
| Intervention (N = 24) | Usual care (N = 20) | ||
| Age (years) | |||
| Mean (SD) | 69.0 (14.7) | 65.8 (14.6) | 67.6 (14.6) |
| Median (minimum, maximum) | 71.6 (20.8, 89.2) | 70.5 (27.0, 89.1) | 71.1 (20.8, 89.1) |
| Gender, n (%) | |||
| Male | 11 (45.8) | 10 (50.0) | 21 (47.7) |
| Female | 13 (54.2) | 10 (50.0) | 23 (52.3) |
| During the last month, have you often been bothered by feeling down, depressed or hopeless? n (%) | |||
| Yes | 22 (91.7) | 19 (95.0) | 41 (93.2) |
| No | 2 (8.3) | 1 (5.0) | 3 (6.8) |
| During the past month, have you often been bothered by having little interest or pleasure in doing things? n (%) | |||
| Yes | 19 (79.2) | 19 (95.0) | 38 (86.4) |
| No | 5 (20.8) | 1 (5.0) | 6 (13.6) |
| On average, do you drink ≥ 3 units of alcohol each day? n (%) | |||
| Yes | 4 (16.7) | 2 (10.0) | 6 (13.6) |
| No | 20 (83.3) | 18 (90.0) | 38 (86.4) |
| Do not know | 0 (0.0) | 0 (0.0) | 0 (0.0) |
| Smoking status, n (%) | |||
| Non-smoker | 10 (41.7) | 7 (35.0) | 17 (38.6) |
| Current smoker | 2 (8.3) | 0 (0.0) | 2 (4.6) |
| Ex-smoker | 12 (50.0) | 13 (65.0) | 25 (56.8) |
| Health problems,a n (%) | |||
| Arthritis | 18 (75.0) | 12 (60.0) | 30 (68.2) |
| Diabetes mellitus | 6 (25.0) | 8 (40.0) | 14 (31.8) |
| Stroke | 1 (4.2) | 0 (0.0) | 1 (2.3) |
| Cancer | 1 (4.2) | 5 (25.0) | 6 (13.6) |
| Respiratory conditions | 12 (50.0) | 7 (35.0) | 19 (43.2) |
| Progressive conditions | 2 (8.3) | 2 (10.0) | 4 (9.1) |
| Cardiovascular conditions | 15 (62.5) | 9 (45.0) | 24 (54.6) |
| Other | 6 (25.0) | 10 (50.0) | 16 (36.4) |
| Did your education continue after the minimum school leaving age? n (%) | |||
| Yes | 12 (50.0) | 13 (65.0) | 25 (56.8) |
| No | 12 (50.0) | 7 (35.0) | 19 (43.2) |
| Do you have a degree or equivalent professional qualification? n (%) | |||
| Yes | 10 (41.7) | 8 (40.0) | 18 (40.9) |
| No | 14 (58.3) | 12 (60.0) | 26 (59.1) |
| Ethnicity, n (%) | |||
| White | 23 (95.8) | 20 (100.0) | 43 (97.7) |
| Asian or Asian British | 0 (0.0) | 0 (0.0) | 0 (0.0) |
| Black or black British | 1 (4.2) | 0 (0.0) | 1 (2.3) |
| Other ethnic group | 0 (0.0) | 0 (0.0) | 0 (0.0) |
| Number of children, n (%) | |||
| 0 | 4 (16.7) | 5 (25.0) | 9 (20.5) |
| 1 | 3 (12.5) | 1 (5.0) | 3 (6.8) |
| 2 | 10 (41.7) | 4 (20.0) | 14 (31.8) |
| 3 | 6 (25.0) | 4 (20.0) | 10 (22.7) |
| ≥ 4 | 1 (4.2) | 6 (30.0) | 7 (15.9) |
| Marital status, n (%) | |||
| Single | 3 (12.5) | 2 (10.0) | 5 (11.4) |
| Divorced/separated | 0 (0.0) | 4 (20.0) | 4 (9.1) |
| Widowed | 6 (25.0) | 5 (25.0) | 11 (25.0) |
| Cohabiting | 0 (0.0) | 0 (0.0) | 0 (0.0) |
| Civil partnership | 0 (0.0) | 0 (0.0) | 0 (0.0) |
| Married | 15 (62.5) | 9 (45.0) | 24 (54.6) |
Intervention delivery
Of the 24 participants randomised to the ESI, 18 commenced the ESI sessions (75.0%). The average number of ESI sessions completed was four, with nine participants completing all six sessions. A total of 79 ESI sessions were completed by these 18 participants, out of a possible 108 (73.1%). Of the 18 participants who commenced the ESI, 16 completed at least two ESI sessions (88.9%) (Table 20).
| Total number of ESI sessions completed | Number of participants |
|---|---|
| 6 | 9 |
| 5 | 0 |
| 4 | 3 |
| 3 | 3 |
| 2 | 1 |
| 1 | 2 |
| Did not commence ESI | 6 |
A total of 18 ESI facilitators completed the ESI training as part of the pilot RCT (these were different ESI facilitators to those completing the feasibility study ESI training). For the pilot RCT, a total of 21 ESI facilitators (from those trained during the feasibility study or pilot RCT) were available to deliver the ESI, of whom 11 supported at least one participant through the ESI (five ESI facilitators delivered the ESI in both the feasibility study and the pilot RCT). On average, the number of participants per ESI facilitator was 1.6 (ranging from one to three participants).
All six of the participants who did not start the ESI withdrew before commencing the ESI sessions.
Follow-up and withdrawal
A total of 41 of the 44 randomised participants completed a 4-month follow-up questionnaire, indicating a retention rate of 93.2%.
A total of eight participants withdrew from the ESI. Six participants withdrew from the ESI before commencing any sessions (25.0%). A further two participants withdrew from the ESI after having commenced ESI sessions (one participant withdrew after session 2 and one participant withdrew after session 1). The reasons given for stopping the ESI sessions were that the intervention was unsuitable and that they did not think the intervention was helping.
Two participants fully withdrew from the pilot RCT prior to the 4-month follow-up.
Standardised measures
Primary outcome measure
The intended primary outcome measure for the pilot RCT was the PHQ-9,63 measuring depression severity at 4 months. The PHQ-9 is scored between 0 and 27, where higher scores represent a higher level of depression.
In addition to the descriptive overview of the PHQ-9 scores presented in Table 21, a linear regression model adjusting for allocation, age, gender and baseline PHQ-9 score was undertaken to explore any potential effect of the ESI. This found that, at the 4-month follow-up, the average PHQ-9 score was 0.77 points higher (indicating a higher level of depression) among participants who received the ESI than in the usual-care arm (95% CI –2.05 to 0.97 points). The CI suggests a potential benefit of the intervention of up to 2.05 points or a potential detriment of up to 0.97 points. These results are exploratory only and should be interpreted with caution, as the sample size for this estimate is small (n = 41).
| Treatment arm | Overall | ||
|---|---|---|---|
| Intervention | Usual care | ||
| Baseline | |||
| Raw score | N = 24 | N = 20 | N = 44 |
| Mean (SD) | 11.1 (5.2) | 10.8 (5.0) | 11.0 (5.0) |
| Median (Q1, Q3) | 11 (7, 13.5) | 11.5 (6.5, 14) | 11.0 (7.0, 14) |
| Minimum, maximum | 4, 24 | 0, 20 | 0, 24 |
| Categorised, n (%) | |||
| No depression (0–4) | 1 (4.2) | 1 (5.0) | 2 (4.6) |
| Mild depression (5–9) | 10 (41.7) | 7 (35.0) | 17 (38.6) |
| Moderate depression (10–14) | 7 (29.2) | 8 (40.0) | 15 (34.1) |
| Moderately severe depression (15–19) | 4 (16.7) | 3 (15.0) | 7 (15.9) |
| Severe depression (20–27) | 2 (8.3) | 1 (5.0) | 3 (6.8) |
| 4-month follow-up | |||
| Raw score | N = 21 | N = 20 | N = 41 |
| Mean (SD) | 10.0 (6.3) | 8.7 (4.4) | 9.3 (5.4) |
| Median (Q1, Q3) | 9.0 (6.0, 13.0) | 8.0 (6.0, 11.0) | 9.0 (6.0, 12.0) |
| Minimum, maximum | 0, 22 | 0, 17 | 0, 22 |
| Categorised, n (%) | |||
| No depression (0–4) | 4 (19.1) | 3 (15.0) | 7 (17.1) |
| Mild depression (5–9) | 8 (38.1) | 10 (50.0) | 18 (43.9) |
| Moderate depression (10–14) | 4 (19.1) | 4 (20.0) | 8 (19.5) |
| Moderately severe depression (15–19) | 3 (14.3) | 3 (15.0) | 6 (14.6) |
| Severe depression (20–27) | 2 (9.5) | 0 (0.0) | 2 (4.9) |
To further explore this result, Figure 7a and b shows the breakdown of the change in PHQ-9 categories between baseline and the 4-month follow-up by treatment arm. This shows that the baseline PHQ-9 categories that saw the most change between the two time points were the moderate and moderately severe depression, and that those at the top and bottom of the PHQ-9 scale (those not depressed and those most depressed) had no change in PHQ-9 score regardless of which treatment arm of the trial they were in. This figure also indicates that all participants who were in the usual-care arm of the trial and had moderately severe depression showed a reduction in depression severity, despite not being offered the ESI as part of CHEMIST.
FIGURE 7 Pilot RCT: (a) the change in the PHQ-9 category from baseline (left) to the 4-month follow-up (right) for the intervention arm; and (b) the change in the PHQ-9 category from baseline (left) to the 4-month follow-up (right) for the usual-care arm. (Figure not reproduced here; see the original article.)
Results are shown only for those participants who provided PHQ-9 scores at both baseline and the 4-month follow-up.
Given the limited number of participants recruited per pharmacy (average 2.9 participants, range 0–10 participants), the pharmacy effect was not explored.
Secondary outcome measures
As in the feasibility study, the GAD-765 was also completed both at baseline and at the 4-month follow-up; the results are presented in Table 22. This measure was well completed, with the scores calculated for every returned questionnaire.
The BADS was sent to a total of 43 participants (the BADS was not sent to one participant in the intervention group as this was deemed by the study clinicians as inappropriate at that time). Thirty participants completed and returned the questionnaire (69.8%), with similar response rates across the intervention group (69.6%) and usual care group (70.0%).
Mediational and moderational analyses were not conducted (see Chapter 6, Qualitative study). However, data on age, LTC, depression severity and number of ESI sessions are reported in Table 19 (age and LTC), Table 21 (depression severity, PHQ-9) and Table 20 (number of ESI sessions). The number of episodes of depression (obtained via the MINI at baseline) was provided by 40 of the 44 randomised participants. On average, the participants had experienced 6.8 episodes (standard deviation 17.1), ranging from 0 to 100 episodes, although the data was skewed as the median number of episodes was 1.5. The mean was similar between the two groups: 6.9 episodes (standard deviation 20.7) for the intervention group and 6.6 episodes (standard deviation 11.2) for the usual care group. Twenty-nine participants who reported experiencing at least one episode of depression provided an age at which they experienced their first episode. This was slightly older in the intervention group (mean 34.4 years, standard deviation 23.6) than in the usual care group (28.9 years, standard deviation 16.0), and ranged from 9 to 85 years.
| Treatment arm | Overall | ||
|---|---|---|---|
| Intervention | Usual care | ||
| Baseline | |||
| Raw score | N = 24 | N = 20 | N = 44 |
| Mean (SD) | 8.1 (5.0) | 6.6 (4.7) | 7.4 (4.9) |
| Median (Q1, Q3) | 8.0 (4.5, 11.0) | 5.0 (3.0, 9.5) | 7.0 (4.0, 11.0) |
| Minimum, maximum | 0, 20 | 0, 18 | 0, 20 |
| Categorised, n (%) | |||
| No anxiety (0–5) | 6 (25.0) | 8 (40.0) | 14 (31.8) |
| Mild anxiety (5–10) | 8 (33.3) | 7 (35.0) | 15 (34.1) |
| Moderate anxiety (10–15) | 8 (33.3) | 3 (15.0) | 11 (25.0) |
| Severe anxiety (15–21) | 2 (8.3) | 2 (10.0) | 4 (9.1) |
| 4-month follow-up | |||
| Raw score | N = 21 | N = 20 | N = 41 |
| Mean (SD) | 6.0 (5.1) | 5.2 (4.8) | 5.6 (4.9) |
| Median (Q1, Q3) | 5.0 (3.0, 7.0) | 3.0 (2.0, 7.5) | 5.0 (2.0, 7.0) |
| Minimum, Maximum | 0, 18 | 0, 18 | 0, 18 |
| Categorised, n (%) | |||
| No anxiety (0–5) | 9 (42.9) | 11 (55.0) | 20 (48.8) |
| Mild anxiety (5–10) | 8 (38.1) | 6 (50.0) | 14 (34.2) |
| Moderate anxiety (10–15) | 2 (9.5) | 2 (10.0) | 4 (9.8) |
| Severe anxiety (15–21) | 2 (9.2) | 1 (10.0) | 3 (7.3) |
Competency and intervention fidelity
Of the 18 ESI facilitators who completed the ESI training as part of the pilot RCT, 10 passed the competency assessment on the first attempt and seven passed on the second attempt following additional telephone-based support. One ESI facilitator did not undertake the competency assessment for personal reasons and did not go on to deliver the ESI. All ESI facilitators who went on to support participants through the ESI received session-by-session clinical supervision.
Enhanced support intervention facilitators were advised during the ESI training workshop (or via the clinical supervisor for those ESI facilitators who completed their ESI training during the feasibility study) about the need to audio-record some of their ESI sessions to undertake the ESI fidelity assessment. ESI facilitators were approached to record ESI sessions once they had delivered the ESI to a minimum of one participant to enable them to gain experience and confidence in delivery of the ESI. A total of four out of a possible six ESI facilitators (the figure of six does not include one ESI facilitator who had previously refused to audio-record ESI sessions during the feasibility study, a decision supported by their manager) were approached to audio-record ESI sessions. Two ESI facilitators were not asked to record sessions as it was deemed not to be appropriate by the clinical supervisor (in one case owing to concerns raised by the ESI facilitator and the complexity of the presenting participant, and in the other case owing to concerns expressed by the ESI facilitator). Of the four ESI facilitators who were approached, two agreed to record their ESI sessions and were supplied with an encrypted digital recorder (the study team had access to only two digital recorders at any one time because of the significant increase in the cost of these). One of these ESI facilitators successfully recorded one ESI session (ESI session number 2); reasons for not recording additional sessions included participant cancellation of planned session and anxiety about recording sessions on the part of the ESI facilitator. The second ESI facilitator advised that they attempted to record ESI sessions on several occasions but reported technical problems with the digital recorder (e.g. the digital recorder had no charge, or despite putting the recorder on charge the recorder still did not work) that prevented them from recording any ESI sessions.
In total, one ESI session was recorded by one ESI facilitator during the course of the pilot RCT.
Serious adverse events
There was one SAE reported in the pilot RCT. A participant in the intervention arm experienced a fall and was subsequently hospitalised. This event was reviewed by an independent clinician (as per study protocol) and was judged to be unrelated to the study. This participant later withdrew fully from the study.
Chapter summary
The pilot RCT demonstrated that recruitment remained a significant barrier to the delivery of the study, with the average recruitment being 3.1 participants per month. The target sample size of 100 participants was not achieved, which limits the feasibility of progressing to a main RCT. However, the intended future primary outcome (PHQ-9) was scoreable for all returned follow-up questionnaires, as was the GAD-7. This indicates a high level of data quality and completeness. Similar to the feasibility study, the ESI was well attended, with 75% of participants allocated to the intervention taking up the ESI and 89% completing at least two sessions. This suggests that participants are willing to undertake the ESI. Although the primary analysis was not powered to detect any effect of the ESI, this does suggest that there may be a detrimental effect from the ESI (0.77, 95% CI –2.05 to 0.97). Assessment of fidelity to the ESI was difficult to conduct via audio-recordings of ESI sessions. Completion rates for the BADS was acceptable, although it is noted that this was collected at only one time point. The pilot RCT process evaluation will further explore some of these issues (see Chapter 9).
The pilot RCT findings informed the predetermined progression criteria that would assess the feasibility of conducting a definitive RCT as follows:
- Recruit and randomise 100 participants across five or six CPs by month 20 – a total of 44 participants were recruited and randomised across 13 out of 15 CPs over a 14-month recruitment period (last participant recruited in month 23).
- For 80% of participants randomised to the ESI to receive two or more ESI sessions within the 4-month post-randomisation period – 16 out of the 24 participants randomised to the ESI (66.7%) received two or more ESI sessions (88.9% of those who commenced the ESI received two or more ESI sessions) within the 4-month post-randomisation period. The remaining six participants did not commence the ESI.
- Collect valid (intended) primary outcome measure on 80% of participants at the 4-month post-randomisation follow-up – the retention rate was 93.2% at the 4-month post-randomisation follow-up, with 100% data completeness for the PHQ-9.
- A fidelity score of ‘acceptable’ (3) or above to be achieved in at least 90% of assessed audio-recorded ESI session (criterion rolled over from feasibility study) – one ESI session was recorded.