What can we learn from these analyses?

This evaluation is based on data on hospital use over the period February 2016 to March 2017 for residents who lived in one of 15 vanguard care homes in Wakefield. These analyses provide insights into hospital use by vanguard residents at an early stage of the vanguard’s enhanced support, in comparison with a broadly similar group of individuals who lived in ‘similar’ care homes in Wakefield not supported by the vanguard.

We found some evidence that people living in vanguard care homes had 27% fewer potentially avoidable emergency admissions than the matched control group, although the 95% confidence interval, which is likely to be underestimated due to reusing some residents’ records, ranged between 45% to 2% fewer such admissions. Therefore, this should be interpreted as promising but not definitive evidence of fewer potentially avoidable admissions in the intervention group. There was no conclusive evidence that people living in vanguard care homes had lower A&E attendances or overall emergency admissions.

This evaluation aimed to explore whether there were early signals of change in hospital use and was therefore conducted approximately one year after the introduction of the enhanced support, only capturing residents’ outcomes over an average of six months. Research has highlighted that implementing complex interventions needs time to take effect. This is so that teams introducing these changes can be supported to work through implementation challenges and learn from experience and evaluation. Information provided by Wakefield CCG indicates that the primary purpose and focus of the LES was to get GPs thinking about how they can work differently with care homes and to move towards the ‘one GP practice one care home’ model. However, Wakefield CCG identify that the move to one GP per care home model was not implemented during the time period of this study and the GP LES did not appear to have had much influence on the working patterns of the GP practices working with the care homes supported by the vanguard. This means that this evaluation may be identifying the additional effect of the MDTs and the voluntary sector over and above the support provided by GP practices.

Furthermore, residents may need to receive the intervention for a period of time before their outcomes are affected. This is consistent with our sensitivity analysis, which showed that when looking at residents that were in the study for at least three months and therefore were getting the enhanced support for longer, potentially avoidable emergency admissions were significantly lower (38% fewer such admissions, 16% to 54%, 95% confidence interval) for residents receiving the enhanced support. Here, the confidence interval is further away from 0% (ie the threshold for ‘no difference’), therefore we can have more confidence around these results.

The fact that we are seeing an effect specifically on potentially avoidable admissions may be due to the MDTs, who proactively planned and managed care of those care home residents considered at high risk who may otherwise have care needs which could lead to residents needing inappropriate hospital care. Furthermore, they provided training relating to, for example, falls and pressure sore prevention, which can otherwise result in potentially avoidable admissions.

However, we could not secure access to data on which residents were referred to MDTs. Therefore, we do not know how many of the vanguard residents included in this study were actually seen by the MDT or whether the lower rate of potentially avoidable admissions is associated with referral to the MDT.

Study strengths and limitations

This evaluation focused on hospital use, since it was an objective of the vanguard to respond to pressures on the health and care system by reducing the use of unplanned emergency care. It does not tell us how the care home vanguard affects other parts of the health care system, if it achieved against their aims to improve residents’ wellbeing or quality of life, or whether the care home vanguard affected staff satisfaction and capability. For a more complete picture of the enhanced support’s impact, this study should be viewed together with the results of local evaluations. For example, analyses carried out by Wakefield Public Health Intelligence on the impact of the enhanced support have shown promising results on ambulance call outs.

By linking data sets together, we were able to study the hospital records for a large number of care home residents, without the problems with non-response that can occur with other data collection methods such as surveys. However, due to limitations with the data we did not identify as many care home residents as expected, based on the number of beds (999 beds in the vanguard care homes). We are unable to determine how similar the characteristics of the missing residents are to the residents in the study.

We were unable to link MDT referrals to the hospital records, which prevented us from knowing who was seen by the MDT. Knowing who was seen by the MDT could have helped identify more care home residents and potentially improved the methods used to identify care home residents. The MDT is one of the elements most likely to have had an impact on hospital use and not knowing what proportion of the residents in our study were seen by the MDT makes it more difficult to know how different the care that the vanguard residents received was compared to the comparison residents. We were also unable to evaluate the effect of being seen by the MDT.

The matched control residents had broadly similar age, gender, health conditions and previous hospital use to the vanguard residents. They lived in care homes that were broadly similar in terms of provision of nursing or residential care. Furthermore, our statistical evaluation used risk-adjusted analyses, which lends further rigour and validity to the study. However, there might be other unobserved differences between the vanguard residents and matched control residents that were not (or could not be) measured, such as informal care they received from relatives, staffing levels in care homes, or quality of care received in hospital, and therefore could not be adjusted for. But by choosing matched control residents living in Wakefield, we ensured that the two groups were more likely to be similar in ways that could not be observed or measured and also that both groups had access to the same health services. There was no statistically significant difference in death rates between the two groups, which lends support to the notion that their baseline health profiles were comparable.

The pool of potential control residents was small so the same control person was reused multiple times. This introduces correlation, which means that the width of the confidence intervals is underestimated and therefore the results should be interpreted with caution.

Previous Next