The SOSA programme was found to increase the number of homes with the three key safety practices and gain QALYs while decreasing the number of injuries among children as well as reducing healthcare expenditure. Meanwhile, the SOSA programme cost was smaller than the healthcare savings, suggesting that the SOSA programme was a dominant intervention in that it saved money and was more effective. However, sensitivity analyses demonstrated considerable uncertainty regarding the result, with a 52% chance that the SOSA programme led to an improvement in homes with the three key safety practices, a 75% chance that there was a reduction in injuries and a 95% chance of increasing QALYs.

We have investigated the cost-effectiveness of the SOSA programme in a real-world setting, using a combination of routinely collected data as well as data collected directly from families. The microcosting approach used provided an accurate picture of the costs of providing the SOSA programme.

Although we have captured most healthcare expenditure, we originally intended to capture more detailed healthcare data from the medical records of a sample (n=100) of study participants but only recruited 22 parents, so our analysis is based on self-reported data. This meant that we did not capture prescription data for any injuries that occurred in the 2-year follow-up, and our analyses may, therefore, slightly underestimate healthcare expenditure. However, Cooper et al found that prescriptions costs were only a small amount of the total cost of an injury,4 with £0.16 being for prescriptions out of a total of £194.11 for a child who has a fall. Therefore, although we are missing this data, it is unlikely that this would change the main findings of our study.

Recall bias may have occurred in parents’ self-reported injury data as this data was collected at 3-monthly intervals, a time period previously shown to have injury recall rates of between 58% for clinic visits and 86% for emergency department visits or hospital admissions.21 But the small amount of data we extracted from medical records found parents were accurately reporting their child’s medically attended injuries, though numbers are too small to assess accuracy of reporting between the two arms.

Our analysis was only able to take account of short-term healthcare costs. This will underestimate the true cost of injuries to health services, education, social care, parents, children, and wider society and hence our estimate of the cost-effectiveness of the SOSA programme is likely to be an underestimate.

The COVID-19 pandemic also impacted our study, with many of the SOSA programme activities becoming remote as face-to-face activities were suspended during periods of social restrictions, decreasing the provision of materials. This impacted the SOSA programme cost (see table 2 where the last two financial quarters are below the previous quarter’s cost) and potentially the effectiveness of the SOSA programme.

The considerable uncertainty within the evaluation results may be of concern. However, the probabilistic sensitivity analysis demonstrates a 62% chance that the SOSA programme saved money even though the evaluation excludes longer-term health and social costs for more severe injuries, educational costs and productivity losses. This means that SOSA is more likely to be cost-effective than our evaluation finds.

Findings of the economic evaluation of SOSA are consistent with studies indicating cost-effectiveness of interventions that improve home safety through home visiting,25 and educational interventions promoting safe poison storage26 and fire escape planning.14 Family mentors were a new type of role, and therefore, an economic evaluation of their inclusion in a child home safety intervention has not previously been performed. There is evidence, however, that interventions provided by trained laypersons to reduce child maltreatment (and therefore injuries) are cost-effective.27 Previous research shows that promotion of safe poison storage is more cost-effective when provided in disadvantaged as opposed to more affluent areas, and the disadvantaged areas in which SOSA was delivered may partly explain its cost-effectiveness.26

Policy-makers and health and social care commissioners should note that SOSA is cost saving, returning £1.39 for every one pound spent on the SOSA programme, even when only short-term healthcare costs are included in the evaluation. Using costs associated with longer-term health, social care, education and productivity losses is only likely to make SOSA more cost saving. Commissioners should, therefore, consider commissioning the SOSA programme for families in disadvantaged areas. Further research, perhaps as part of larger studies and incorporating longer-term costs of injuries, would be helpful to produce more precise estimates of cost-effectiveness.