Health services around the world are attempting to improve care for people with long-term conditions, as currently it is often fragmented and expensive. Many interventions have been tried and tested. To the long list of evaluations another can now be added – that of Birmingham OwnHealth (published today in the BMJ).
Birmingham OwnHealth was England’s largest example of telephone health coaching, established in 2006. Operating as part of routine clinical practice, it initially enrolled patients with diabetes, chronic obstructive pulmonary disease or cardiac problems.
Our evaluation (done in partnership with Ernst & Young) found that the service was associated with an increase in unplanned (“emergency”) hospital admissions, compared with patterns seen for a matched control group that we selected retrospectively from other parts of the country.
It’s possible that, by improving people’s understanding of when to seek help but not providing a responsive service, patients instead sought help elsewhere – for example at emergency departments
The intervention looked promising as a way to help patients manage their health problems. On enrolment into the service, patients received their own care manager – specially-trained nurses employed by NHS Direct who often worked with patients over several years.
During monthly phone calls, patients and nurses worked through eight modules, aiming to improve patients’ understanding of their conditions and to help them make lifestyle changes if needed.
Birmingham OwnHealth care managers could refer patients onto existing services, such as mental health care and social care. GPs referred patients into the service and were also offered monthly phone calls to discuss their patients.
Previous evaluations of Birmingham OwnHealth had found high levels of patient satisfaction and improvements in clinical metrics (such as glycated haemoglobin and body mass index) ,the latter among patients with poorly controlled diabetes. So why did the intervention not reduce hospital admissions?
One possibility is that the theory behind the intervention was not quite right – in other words, perhaps the activities associated with the intervention were unlikely to reduce admissions (even if they might have improved care in other important aspects).
For example, telephone calls were almost exclusively made outbound from nurses to patients. It’s possible that, by improving people’s understanding of when to seek help but not providing a responsive service, patients instead sought help elsewhere – for example at emergency departments.
Another possibility is that the theory underlying Birmingham OwnHealth was right, but the execution needed some refinement. The cohort of patients examined in the study (2,700 patients recruited between 2006 and 2008) represented fairly early experience, when it can take time for interventions to begin to operate at full throttle.
For example, it can take time to integrate new services into the care pathway. It’s also possible that the context in which the intervention operated (e.g. the financial incentives) were not supportive of the goals of the programme. Regardless of the explanation, Birmingham OwnHealth has been closed down, so it is too late to go back and refine the intervention.
Should health coaching be introduced into other parts of the NHS? Other large studies have shown similar impacts to ours. For example, a retrospective study of health coaching in the US Medicaid found 20 per cent more emergency department visits among intervention patients than controls, summed over the two years following enrolment. Yet, there are some more positive signs.
A large randomised controlled trial, also in the US, found that health coaching was associated with reductions in overall admissions for people with long-term conditions, even though no change was detected in the number of admissions that occurred through the emergency room.
This intervention had some distinguishing characteristics, including the use of a predictive model to identify high-risk patients for intervention, and an element of shared decision making for ‘preference-sensitive conditions’. It also operated in the context of employer-based health insurance.
Although reducing the number of avoidable hospital admissions is not the only aim of new interventions, it is often seen as a potential way to release resources for investment elsewhere in the system. Unfortunately, the elements that make up a successful hospital avoidance scheme are not yet fully understood, and might differ according to local context.
One approach for future interventions would be to apply the evaluation method used for Birmingham OwnHealth in a more ‘real time’ fashion, rather than retrospectively. This might allow potential problems with service delivery to be identified sooner, and improvements to be made.