Innovations in Practice: Further evidence on the effectiveness of the strengths and difficulties added value score as an outcome measure for child and adolescent services

Abstract

Background: The Strengths and Difficulties Added Value Score (SDQ AVS) uses a large epidemiological study to predict follow-up parental SDQ scores for the evaluation of routine outcomes. Method: We tested the prediction of the SDQ AVS derived from a national population survey separately on scores for the waiting list control and intervention groups in a randomised controlled trial. If the SDQ AVS is to be clinically useful, it needs to function as expected across different populations. Results: In the control arm, the SDQ AVS predicted an effect size of 0.15 (95% CI À0.01-0.30) compared to an expected effect size of 0, as the children in this arm received no treatment. In the experimental arm, the SDQ AVS predicted an effect size of 0.62 (95% CI 0.42-0.83) compared to the study effect size of 0.53. Change scores overestimated the effect size in both arms (control 0.50 95% CI 0.34-0.66, intervention 0.85 95% CI 0.66-1.04). Conclusion: Our findings suggest that the SDQ AVS adjusts for spontaneous improvement, regression to the mean and attenuation. Key Practitioner Message• The SDQ AVS aims to estimates change attributable to clinical intervention, by using a clinically relevant epidemiological sample as a proxy control group.• The SDQ AVS can be calculated using a standard equation from baseline SDQ total difficulties and impact scores.It is the difference between the predicted and is positive if the child is doing better than expected or negative if outcomes are poorer than predicted.• This is the second study to test the SDQ AVS separately in the intervention and control arms of a randomised controlled trial, as with the first it suggests that it offers a more accurate assessment of the impact of interventions than simple change scores among groups of children attending CAMHS

Application of the Strengths and Difficulties Questionnaire Added Value Score in evaluating the effectiveness of Functional Family Therapy within local authority social work services

“…When conducting evaluations of interventions, there are several factors that may cause the scores on follow‐up outcome measures to decrease, such as attenuation (i.e. the inclination for clients to report less on the second round of measurements), regression to the mean and perceived decrease in difficulties at follow‐up due to adolescents generally being referred to the service when their difficulties are most severe (Fugard et al., ; Rotheray et al., ). For these reasons, it is important to compare the results from an intervention study to a control sample in order to determine whether there is more or less improvement than would have been predicted without the FFT intervention.…”

A comparison of parent reported outcome with experience of services

“…Two studies, as described above (Ford et al, 2009;Rotheray et al, 2014) have demonstrated that DS greatly inflate the estimates of change while the SDQ AVS does adjust for attenuation, random fluctuation and regression to the mean as expected. The SDQ AVS, therefore, does seem to be the most rigorous way to assess change in our study.…”

“…It seems unlikely, however, that our choice of measures completely explains our failure to detect improvement as others using these measures have been able to do so. The mean SDQ AVS for clinics sending data to CORC have been positive (mean=0.19 see www.CORC.uk.net, Wolpert et al, 2012b;Fugard et al, 2014), and reanalysis of trial data suggests that the SDQ AVS functions as expected (Ford et al, 2009;Rotheray et al 2014). Similarly, the smaller study based on one of the services that participated in the current work (Byrne et al 1999), Australian clinics working secondary school-age children (Matthai et al, 2003) and routine data from an Inner London clinic on parenting (Hurst et al, 2014) detected clinically and statistically significant change over a similar time period using the SDQ.…”

“…Value Score that approximated to the effect size detected for the intervention in the original trial analysis (Ford et al, 2009: Rotheray et al, 2014. In both these studies, simple difference / change scores provided large effect sizes that greatly exceed that suggested by the SDQ AVS as the influence of attenuation, random fluctuation, and regression to the mean were not controlled.…”

Routine mental health outcome measurement in the UK

“…Another advantage -currently available only for the parent version -is that an "added-value score" (AVS) can be calculated using someone's initial scores and scores six months later. (Ford, Hutchings, Bywater, Goodman, & Goodman, 2009;Rotheray S et al, 2014) The AVS allows the impact of treatment to be estimated above and beyond spontaneous improvement and factors like regression to the mean. So far there have been two tests of the score's predictions using data from randomised controlled trials (RCTs) of parent training.…”