Main trial findings

In this chapter, we present the main findings of the RCT, including quantitative aspects of the process evaluation, a description of the study participants, the main outcome analyses and the exploratory analyses.

Process evaluation

In this section, measures of the implementation of the trial and the intervention will be presented.

Trial implementation

Figure 2 shows the recruitment and flow of participants through the trial, using the CONSORT format.73 Nearly 4000 primary care patients aged ≥ 60 years were initially approached by 17 general practices. The recruited participants were spread over the 17 general practices, as shown in Table 5. These practices were chosen to be in the lower two-fifths of the deprivation ranking distribution. The average deprivation level of the 17 participating practices (based on the IMD score at the MSOA level for practice postcode) was lower than that among the remaining 363 practices in the North East [IMD score at MSOA level: 27.0 (SD 10.4) vs. 38.5 (SD 11.5)].

FIGURE 2 The CONSORT flow chart of participants in the Do-Well RCT. WR, welfare rights. Reproduced from Howel et al.2 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution International license (CC BY 4.0), which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/. (Figure not reproduced here; see the original article.)

Of those patients approached, 1770 (45%) opted out of further involvement by returning the invitation slip to their general practice. GPs sent the remaining 2142 names (55% of those approached) and their contact details to the research team. The chief investigator wrote to all of these patients inviting them to participate in the research. Of the 2142 invited, 825 (39%) declined to participate further at that stage, 405 (19%) were not contactable, 116 (5%) initially did not refuse but later declined to participate and 41 (2%) were not contacted as the recruitment target had been reached before they were needed. Those remaining were 755 primary care patients aged ≥ 60 years from participating general practices in socially deprived areas of North East England, all of whom agreed to participate, provided informed consent and were randomised to either the intervention or the control group. The participants were recruited between 11 May 2012 and 28 February 2013.

However, those who agreed to participate (n = 755) were, on average, less socioeconomically disadvantaged [IMD score at LSOA level: 29.0 (SD 16.0)] than non-participants [n = 1387; IMD score: 33.5 (SD 17.9)], and were more likely to be female (53.5% vs. 50.5%).

During the trial, 121 participants withdrew from the study, 36 were lost to follow-up and 36 participants died (see Figure 2 for details). The losses were balanced across allocation groups. In total, 562 participants completed the 24-month follow-up (intervention group, n = 283; control group, n = 279) and were available for analysis.

Receipt of intervention (reach)

As shown in Figure 2, 381 trial participants (50%) were randomly allocated to receive the intervention. The intervention group participants were offered welfare rights advice shortly after recruitment: Figure 3a shows the distribution of the time, in days, between randomisation and the first visit by the WRA, while Figure 3b shows the distribution of time between randomisation and closing the welfare rights advice case, which occasionally took > 12 months. The number seen (as intended) within 2 weeks by their WRA was 5 (1.5%) and within 4 weeks was 37 (11%). The median number of days from study entry to first WRA visit was 58 days (IQR 40–89 days) and the range was 0–403 days. The length of time taken for WRAs to see intervention group participants for their initial assessment increased over time.

FIGURE 3 Distribution of (a) time (days) between randomisation and first WRA visit; and (b) time (days) from randomisation to welfare rights advice case being closed. (Figure not reproduced here; see the original article.)

Table 6 shows whether or not the advice was received or if benefits were awarded among the 381 participants in the intervention arm. Of these participants, 335 (88%) actually received the intervention as intended. For those randomised to but not receiving the intervention (n = 46, 12%), the most commonly cited reason for not receiving was that the participant declined the welfare rights advice consultation (n = 23); this represented 6% of all who were eligible and 50% of those not receiving the intervention (in the intervention group). In addition, some participants completely withdrew from the study before intervention delivery (n = 16), while a small number of participants could not be contacted in order to arrange the welfare rights advice appointment (n = 7).

In total, 84 of the 381 participants (22%) in the intervention arm were awarded an additional benefit, but this rises to 25% when expressed as a percentage of the 335 who received welfare rights advice. Table 7 shows the distribution of the benefits received for those who took up the offer of welfare rights advice. This information came from the records kept by the WRAs (CCS). There were some discrepancies observed between WRA records and those obtained when participants were interviewed at 24 months concerning new benefits that had been awarded. This is explored further in Chapter 5.

Table 7 shows that non-means-tested benefits were most commonly received. There were small numbers in each of the categories of different combinations of financial and non-financial benefits received. A detailed description of exactly which financial and non-financial benefits were received is given in Chapter 5.

Dose

The intervention was a standard service and our training and delivery checklist aimed to ensure that the fidelity and ‘dose’ of the intervention was consistent across participants. However, the intervention was also intentionally tailored to individual social welfare needs, and so consultations varied in duration according to the particular circumstances and requirements of each participant. Further details of the duration and number of consultations are given in Chapter 5.

Changes to the intervention

The main alterations to the Do-Well intervention were changes to the personnel delivering the intervention, a consequence of local authority budget cuts resulting from national austerity measures that were out of our control. In May 2012, the WRA from one local authority (G) withdrew from the study without giving a reason (see Table 1 for details of local authorities and general practices). In July 2012, the WRA from another local authority (K) withdrew from the study as a result of staff redundancies. In July 2014, the welfare rights advice team from a further local authority (J) went through a process of restructuring whereby welfare benefits posts were removed and a new team was formed which combined local welfare provision with supported housing. One officer continued dealing with welfare benefits cases but only had a remit for complex issues, and, therefore, was unable to continue with the study. In March 2015, the WRAs in another local authority (H) asked the study team to stop any further referrals, as they had moved into redundancy consultation notice104 and found that they were not be able to provide follow-up appointments. To mitigate the impact of these changes in welfare rights advice provision, we secured the services of two recently retired, qualified WRAs with extensive local authority experience, who were able to work freelance and who stepped in to continue providing the intervention in these local authorities, paid from our contingency budget.

During the course of the study in the eight local authorities, 19 WRAs provided the intervention: three in local authority F; two each in local authorities M and E; two local authority-based and one freelance in local authorities H, J and K; four in local authority L; and one freelance in local authority G.

Fidelity assessment

Seven recordings of WRA initial assessments with participants in the intervention group from five local authorities were made available for fidelity assessment: one local authority refused to allow its WRAs to undergo assessment and two local authorities were unable to complete the assessments within the agreed time frame. In one local authority this was because the welfare advisors were unable to find a client who would consent to the recording within the time frame for the fidelity assessment; in another, local authority budget cuts caused staff shortages that prevented an assessment taking place. In another area, only one WRA was assessed for fidelity owing to staff shortages and time constraints.

All consultations were found to be carried out systematically, and the approach to undertaking each consultation was found to be consistent and in accordance with the protocol for intervention delivery. All consultations were found to have appropriate assessment of financial and health status, and all relevant applications for eligible means- and non-means-tested awards and benefits were carried out.

Contamination

Comparison of the responses to the baseline and 24-month interviews indicated that some control group participants reported receiving new welfare benefits during that period. Despite this, we found no direct evidence that any of the control group participants independently sought welfare rights advice before it was offered following the 24-month interview.

Unanticipated consequences of the intervention

We found no evidence from our quantitative analyses of any unanticipated consequences of the intervention.

Baseline characteristics of trial participants

To illustrate how representative the Do-Well participants were of other residents of England, Table 8 compares the demographic characteristics of the Do-Well participants with those of members of the ELSA cohort in wave 4 who were aged ≥ 60 years. The Do-Well trial participants were recruited from practices in the lower two-fifths of the deprivation ranking distribution in areas of the North East, whereas the ELSA cohort was recruited from the whole population of England. There were clear differences between cohorts on only a couple of variables: higher proportions of Do-Well participants lived alone and ‘were paying rent or mortgage’ (as opposed to owning their own home or, occasionally, living rent free).

Tables 9 and 10 show the baseline characteristics of the 755 participants for categorical and numerical outcomes, respectively, by trial arm. The distributions were well balanced between the intervention and control arms. Roughly half of the participants were male, living alone and not paying for accommodation. They were predominantly white, were educated to secondary school level, were retired, were never or ex-smokers and had a satisfactory level of emotional support. The average age was 70 years, while typical Townsend Activities of Daily Living scales were relatively high and stressful life events scores were relatively low.

Table 11 shows the distribution of the primary and secondary outcomes at baseline. The participants had typical values towards the more favourable end of each scale for quality of life, affordability index, standard of living, depression and healthy diet; and towards the less favourable end of the scale for social interaction and physical activity (PASE). The distributions were well balanced between the intervention and control arms.

Table 12 shows the baseline characteristics of the 755 participants for categorical and numerical outcomes for responders and non-responders at 24 months. Those who dropped out of the trial before the 24-month assessment on average tended to be a little older, slightly more likely to be male and have a lower average CASP-19 score at baseline.

Outcome analysis

Intention-to-treat analysis of primary and secondary outcomes

Table 13 shows the distribution of the primary outcome, CASP-19 score, at baseline, 12 and 24 months by trial arm: the results are reported on complete cases and, in addition, after the application of multiple imputation. The primary comparison was at 24 months, when the mean CASP-19 scores were 42.9 in the intervention group and 42.4 in the control group: an adjusted mean difference of 0.3 (95% CI –0.8 to 1.5). The unadjusted means showed little difference between trial arms at any time point, and, when the distribution of covariates was taken into account, there was still little or no difference between mean CASP-19 scores. The differences in mean CASP-19 between trial arms remained very similar at all time points when multiple imputation was used to deal with missing primary outcomes for some participants.

The trial was powered to detect a difference of 1.5 units between means on the CASP-19 scale, and there is little evidence of a difference of this size. It should be emphasised that, although all members of the intervention group were offered welfare rights advice, fewer than one-quarter actually received any additional resources, either financial or non-financial, as a result of the advice. The fact that the majority did not receive any welfare benefits means that any potential change in quality of life as a result of receiving welfare rights advice is likely to have been heavily diluted. An exploration of whether or not the difference in means between trial arms varied significantly between subgroups of trial participants was carried out by testing for interactions between trial arm and sex, age group (dichotomised at median age: 68.6 years) and educational group (primary or secondary vs. tertiary): all interaction terms were not statistically significant (p-values of 0.94, 0.15 and 0.22, respectively) and, therefore, dropped from the regression model.

The difference at 24 months in means between the trial arms was small with narrow CIs for all of the secondary outcome scales (Table 14), other than for average hours of care received each week. The average hours per week were higher in the intervention arm (53.7 vs. 42.0; adjusted difference of 26.3 hours/week, 95% CI 0.8 to 56.1 hours/week). However, for this variable, as few reported receiving care and the amount received varied from 1 to 168 hours per week, the estimates of the difference were imprecise and consistent with either a large or virtually no increase in means.

Table 15 shows the comparison of the four categorical secondary outcomes at 24 months (living independently, mortality, increase in smoking status and social isolation). The proportions were similar across the trial arms, and no associations were statistically significant: the odds ratios were close to 1 and 95% CIs were narrow.

Exploratory analyses

A set of exploratory analyses on those in the intervention arm investigated whether the primary outcome (CASP-19) at 24-month follow-up varied between those in subgroups defined by welfare rights advice received or benefits awarded. These subgroups were not likely to be balanced on key covariates, because socioeconomic and health-related variables are key eligibility criteria considered when a welfare award is made. Table 16 shows the distribution of sociodemographic variables for those who did and did not receive any welfare benefits: there were no great differences, but those awarded welfare benefits tended to be slightly older and more likely to live alone and be female, and less likely to be well educated and be able to carry out activities of daily living: there was little difference in mean IMD scores, which might have been expected.

Table 17 shows that there was little difference in the crude mean total CASP-19 score at 24-month follow-up between those in the intervention group who did and those who did not receive welfare rights advice, with a small unadjusted difference in means favouring those not receiving welfare rights advice (42.8 for those who received welfare rights advice vs. 44.2 for those who did not). However, after adjustment for key covariates (see Table 17), the direction of difference did not change, but this was not statistically significant (adjusted difference –2.1, 95% CI –5.5 to 1.3).

When comparing the subgroups who did and did not receive benefits (as opposed to advice), the lower part of Table 17 shows that there was a lower average CASP-19 score at the 24-month follow-up in those receiving benefits, for both total score and subscales (mean total score 39.2 for those who received benefits vs. 43.8 for those who did not). This might be explained by the fact that those who received benefits tended to be worse off at baseline, in terms of health and socioeconomic variables, than those who did not receive welfare benefits, and this could be reflected in the CASP-19 scores. After adjusting for these covariates, the difference in mean total CASP-19 score was smaller (adjusted difference –0.7, 95% CI –2.8 to 1.4), and there was no indication of a statistically significant or clinically important improvement in average CASP-19 in those receiving welfare benefits.

Secondary outcomes were also compared between those who did and those who did not receive welfare rights advice. Table 18 shows that the adjusted differences in mean between the two subgroups were all small with narrow CIs.

A comparison of secondary outcomes between those who did and those who did not receive any benefits was also undertaken. Table 19 shows that the adjusted differences in means between the two subgroups were all small with narrow 95% CIs, except for the physical activity score (PASE). The adjusted mean PASE score was significantly higher in those who were not awarded any benefits (adjusted difference –13.9, 95% CI –27.0 to –0.9); this may reflect the better health of those who did not receive welfare benefits, rather than any effect of non-receipt of welfare benefits on the ability or choice to exercise more.

A final comparison was between the CASP-19 scores at 24 months for those in the intervention arm who had been awarded a financial welfare benefit and those in the control arm who were later awarded a financial benefit (after 24-month follow-up). As they were all eligible for financial welfare benefits, these participants should be similar in their socioeconomic and health profiles. The 55 in the intervention group who had been awarded financial benefits had a mean CASP-19 score of 39.2 (SD 9.1), whereas the 48 in the control group (who were later found to be eligible) had a mean CASP-19 score of 39.7 (SD 9.4). After adjustment for covariates, the mean CASP-19 score was 1.4 higher in the control group (95% CI –2.0 to 4.7). This comparison shows again that there was little or no evidence that CASP-19 varied between a subgroup who had received benefits and a subgroup who had not.

A further exploration of any association between CASP-19 and receipt of welfare benefits in the intervention group was to look at the weekly amount of financial benefits (in pounds sterling). Figure 4 shows the association between these two variables; it can be seen that there is no indication that CASP-19 at 24 months is higher for those receiving greater financial benefits. However, this does not adjust for the covariates used in previous regression models. When adjustment was made for these, the regression coefficient for CASP-19 on weekly amount was 0.02 (95% CI –0.08 to 0.04) also showing a lack of association.

FIGURE 4 Association between CASP-19 score at 24 months and weekly amount of financial benefit awarded. (Figure not reproduced here; see the original article.)

The final exploration looked at the association between the length of time that a person had been receiving additional benefits and the CASP-19 scores at 24 months (Figure 5). It might be expected that among those who had only started receiving welfare benefits shortly before the assessment at 24 months, health may have improved less, and this might be reflected in a smaller change in CASP-19 score. However, the vast majority of people had been receiving benefits for at least a year by the time of final assessment. The correlation between the two variables was 0.39 (95% CI 0.16 to 0.58), a weak positive association.

FIGURE 5 Association between length of time that participants were in receipt of additional benefits and CASP-19 score at 24-month follow-up. (Figure not reproduced here; see the original article.)