Summary
This study examined the impact of a complex intervention — vertical integration, involving an acute hospital and 10 GP practices — on unplanned care. Across the 10 practices involved, pooled rates of ED attendances did not change significantly after vertical integration. However, there were significant reductions in the rates of unplanned hospital admissions (−0.11, 95% CI = −0.18 to −0.045, P = 0.0012) and unplanned hospital readmissions (−0.021, 95% CI = −0.037 to −0.0049, P = 0.012), per 100 patients per month. These effect sizes represent 888 avoided unplanned hospital admissions and 168 readmissions for a population of 67 402 patients per annum. The estimated savings from the reductions in unplanned care are in excess of £1.5 million.
Given the general lack of progress in reducing the demand for unplanned care (a few notable exceptions aside),21 these findings suggest that vertical integration appears to remove barriers to more effective coordinated patient care in a diverse set of GP practices, although the detailed mechanisms remain unclear.
Strengths and limitations
This is an observational study and, as such, does not provide proof of causation. Nonetheless, sophisticated statistical methods were used to estimate a ‘causal’ effect. The background of this study provided appropriate context for the implementation of the SC study design. The nature of the SC design also meant that the explicit control of confounders was unnecessary because synthetic controls incorporate both observed and unobserved time-varying confounders.22
The use of an SC study assumed that no other major events differentially impacted potential control practices, or that the intervention did not surreptitiously impact on geographically adjacent, non-treatment practices. Other (non-vertical integration) practices in Wolverhampton were included in the donor pool in an attempt to mitigate this. Nonetheless, it is possible that the observed effect was due in part to a set of wider changes in the Wolverhampton area along with the vertical integration programme (see Supplementary Figure S1 for details). It is also possible that the findings are compromised by not fully capturing the unplanned activity of practices in the donor pool that were outside the catchment area of the target hospital, as this might have led to unaccounted differences in the SC outcome estimates, owing to varying care quality.
The significant size of the study population across multiple diverse GP practices and their synthetic controls strengthens the reliability of the findings and reduces the risk of bias through exclusion of subgroups. Nonetheless, the study was based in one health economy and further work would be required to assess the generalisability of the findings.
Although the causal mechanisms involved were not examined, there are several potential explanations for the findings.
RWT sought to increase the accessibility of primary care and made appointments available for patients to book. This has been delivered, in part, by adjusting the skill mix in practices, and ensuring that appointments are handled by the most appropriate staff. If this increase in appointments has occurred at a greater rate than control practices, and these additional appointments have been taken up by patients, and these appointments have been used to improve the control of patients’ long-term conditions, then this may have led to a reduction in GP-referred unplanned hospital admissions. Given that no decreases in the rate of ED attendances were observed, it seems unlikely that these additional appointments have been used to manage patients’ urgent care needs.
RWT has developed a management information system for GP practices joining the vertical integration scheme, which provides data on patients’ use of primary, community, and hospital services. If this information system provides superior insight and access to GPs, then it may enhance GPs’ surveillance of their at-risk population and support the creation of novel interventions to reduce their risk of admission.
As a result of the vertical integration programme, GPs are represented on a number of RWT’s management committees. This, along with the process of employing practice staff directly, may increase the extent of joint working and coordination between primary and secondary care. If GPs can access the advice of secondary care clinicians more readily, this might lead to a greater confidence in managing patients without the need for admission through sharing the clinical risks between primary and secondary care, which may lead to a reduction in unplanned admissions.
Finally, RWT has committed to the principles of population health management and now employs several public health specialists. This increased emphasis on systems thinking may reduce the reliance on hospital care.
Comparison with existing literature
Unplanned hospital admissions in the UK rose by 47% between 1998 and 2013, from 3.6 million to 5.3 million, with only a 10% increase in population over this period. These admissions are expensive: in 2012 they cost the NHS £12.5 billion.23
Although reviews suggest that risk prediction models may have a role to play in identifying those people who are at risk of unplanned admission to hospital, their impact on unplanned care remains limited, with suggestions that they may increase unplanned care.24,25 The authors of the PRISMATIC study suggest that risk stratification tools may increase emergency hospital admissions by focusing GPs’ attention on patients at the highest risk of admission at the expense of patients with lower but more modifiable risks.26 More promising approaches based on greater integration between primary care and community services have been advocated, with preliminary evidence of a 14% reduction in unplanned emergency admissions against a 28.5% increase in the background. (The 14% reduction represents the change in Frome, a small town in Somerset — where the integration described by Abel et al happened — and the 28.5% increase represents a change in the rest of Somerset, excluding Frome.)21
The study findings with respect to unplanned hospital readmissions corresponds with the findings of a vertical integration study in Portugal, which also found that vertical integration has the potential to reduce hospital readmissions.27 However, a study of vertical integration in the US noted that the process of organisational change led to increases in emergency hospital use, with limited evidence of enhanced quality of care.28
Implications for research
This study contributes to the evidence base on the validity of vertical integration as a means of improving service coordination and reducing unplanned hospital use, although many questions remain unanswered and require further study. Although teasing out cause and effect relationships in a complex adaptive system is challenging, perhaps the most crucial question is to determine how and why vertical integration has been effective. The mechanisms that led to lower unplanned care need to be studied. The wider intended or unintended consequences of vertical integration also need attention, as do the costs of vertical integration. Moreover, further studies are required to assess the generalisability of the findings.
Notwithstanding the limitations, the current study offers some evidence that vertical integration may be useful in addressing the year-on-year increase in unplanned hospital admissions in the UK.
