Sixty-nine patients were recruited to this pilot study. As data were pseudonymized for the study team with the IAPT staff member's details removed, it was not possible to compare those recruited by the eight PWPs taking part in the study and those on their caseloads that did not get recruited to the study. However, there were no significant differences in the baseline characteristics of those recruited and the population attending the services with all therapists over the study period (Table 1).
Table 1. Demographic and baseline clinical characteristics of pilot study participants and population of attendees at the IAPT services over the study period
Treatment: Study participants attended between two and 18 treatment sessions [median (IQR) = 5.00 (4–6)], all started treatment at low intensity, with the majority receiving guided self-help (41: 59.42%), then group-based therapy (22: 31.88%), computerized CBT (6: 8.70%) and other therapy (1: 1.45%). Eight participants (11.59%) were stepped up to high-intensity during treatment. The majority of participants (53.45%) were also prescribed psychotropic medications during their treatment in the IAPT services.
Post-treatment: Thirty-three participants (47.83%) reached remission, 51.15% of these had residual symptoms. Twenty-six participants (40.63%) met IAPT recovery criteria; see Fig. 1.
Figure 1. Participant flow throughout study.
Follow-up: Forty-eight participants (69.56%) completed a 1-year post-treatment follow-up appointment. Seven participants returned to the services for further treatment, six received psychological therapy elsewhere, and 16 were taking anti-depressant medications (ADM), of which five began ADM during follow-up. Seven of the 33 participant that reached remission post-treatment (21.21%) relapsed to depression, and a further four participants returned to caseness but did not meet full criteria for relapse (Table 2). Eleven of the 39 participants that had reliably improved at the end of treatment (36.67%) suffered a reliable deterioration on the PHQ-9 over follow-up, and 14 of the 36 participants that did not reach remission post-treatment (38.89%) remained depressed throughout follow-up; see Fig. 1.
Table 2. Relationship between baseline ACS score and each of the primary and secondary outcomes
Primary and secondary outcomes
Feasibility of data collection and follow-up. There was a 100% retention rate during treatment and 70% retention over the 1-year follow-up period, slightly exceeding our expectations of 20% drop-out over 6 months and a further 20% between 6 and 12 months. There was a difference in the loss-to-follow-up by post-treatment status; seven of the 36 not in remission were lost to follow-up, compared with 14 that reached remission or reliably improved. One concern of the clinical teams prior to commencing the study was that the addition of the ACS at each session could lead to a drop in the rate of data completion of the routine IAPT measures. IAPT services aim to have at least 98% of participants with paired PHQ-9 and anxiety measures (usually GAD-7) pre–post treatment; in this study just one participant did not have paired pre–post measures (1.45%), suggesting adequate data completion of the routine measures despite the addition of the ACS.
Treatment response: Scores on the ACS and PHQ-9 were moderately correlated pre-treatment (r = –0.52, p < .001) and strongly correlated post-treatment (r = –0.62, p < .001). There was a moderate correlation between the change in PHQ-9 scores and change in ACS scores pre-to-post treatment (r = 0.45, p < 001) (strength of correlations taken from conventions by Swinscow and Campbell, 2002).
ACS scores were related to outcomes at the end of treatment: the mean (SD) ACS score pre-treatment was lower for those that did not achieve remission [42 (9.06)] compared with those that did [48.85 (6.43)], p < .001 (see Table 2), standardized mean difference = –0.87. There was also a difference in the direction and degree of change in the ACS scores; the mean (SD) change in ACS score for those not reaching remission was –1.875 (6.65) points (the minus number indicating a decrease in ACS score, thus a decrease in attentional control pre-to-post treatment), compared with 2.43 (6.17) points for those in remission (p = .011). A positive change of any degree in ACS score pre-to-post treatment was associated with a greater probability of remission [OR (95% CI) = 3.80 (1.32–10.90), p =.013]. For every one point increase in ACS score pre-to-post treatment after controlling for PHQ-9 scores pre-treatment, the odds of reaching remission increased [OR (95% CI) = 1.27 (1.09–1.45), p = .002].
Mean ACS scores pre-treatment were higher for those that achieved IAPT recovery compared with those that did not (see Table 2). Pre-treatment ACS scores were predictive of IAPT recovery univariably but the association was not significant after controlling for pre-treatment PHQ-9 scores [OR (95% CI) = 1.08 (0.99–1.18), p = .081]. However, the change in ACS scores pre-to-post treatment was associated with IAPT recovery after controlling for pre-treatment PHQ-9 scores [OR (95% CI) = 1.18 (1.05–1.34), p = .006].
Residual symptoms. Seventeen of the 33 participants that reached remission post-treatment (51.51%) had residual symptoms, and their mean ACS score pre-treatment was lower than those that did not have residual symptoms (Table 2). All but one of those that relapsed over the study period had residual symptoms post-treatment.
Outcomes over follow-up. The mean pre-treatment ACS score in those that went on to relapse [42.43 (3.46)] was lower than the mean in those that did not relapse [49.43 (4.71)] (Table 2); standardized mean difference = –1.57. Pre-treatment ACS scores were associated with relapse and with reliably improving pre–post treatment then reliably deteriorating post-treatment to follow-up, after controlling for both baseline and end of treatment PHQ-9 scores (or residual symptoms), relapse [OR (95% CI) = 0.24 (0.06–0.96), p = .043]; reliable improvement then reliable deterioration [OR (95% CI) = 0.72 (0.54–0.97), p = .030]. There were no other variables measured at baseline that were significantly associated with both the outcome over follow-up and pre-treatment ACS score, and none that significantly improved the logistic regression models.
The follow-up outcomes among the 36 participants that did not reach remission post-treatment were not significantly associated with baseline ACS scores (p = .710), or changes in ACS score pre-to-post treatment (p = .310). Mean pre-treatment ACS scores were not significantly different between those that did and did not go on to remain depressed over the course of the study.
Inspection of plots of standardized residuals and deviance suggested no overly influential data points so no participants were removed from the multivariable models. There was no apparent problematic multi-collinearity, with tolerance and VIF values acceptable for all models (0.60–1.07 and 1.60–2.24, respectively).