Statistics from Altmetric.com
- Cardiovascular disease
- randomised controlled trial
- cardiac rehabilitation
- coronary artery disease
The RAMIT trial was funded in 1997 to randomise 8000 patients to cardiac rehabilitation (CR) or routine care and would have provided sufficient power to demonstrate the effect of CR on mortality in a single trial. Despite the efforts of the investigators, the study sponsors (NHS Research and Development Programme), requested early closure due to low recruitment. Given the level of all cause mortality in the control group (84/910, 9.2%) the trial would have needed around 3100 patients in each arm, to detect a 20% reduction at 80% power. In fact the study had <1000 in each arm of the trial. Despite that, over the last 10 years, the ‘negative result’ from RAMIT has been discussed widely. This is the first published account and we welcome the discussion it is provoking (see page 637).
What RAMIT tells us about the efficacy of CR
RAMIT was not designed as a trial of ‘efficacy’, that is, to demonstrate whether or not CR ‘works’ but as a pragmatic trial of its effectiveness as provided ‘in real life’. The results showed no benefit for survival, psychosocial status or HRQOL. In addition the CR group was less likely to be physically active at 12 months than the control group. These results are opposite to those produced by the Cochrane Collaboration Systematic Review and meta-analyses of CR,1 2 all of the reviews of CR by authoritative bodies around the world, from the American Heart Association3 to the WHO4 and is contrary to established clinical guidance5 along with insurance remuneration agreements, in most countries of the world.
Since RAMIT was published the Cochrane CR group has included its data alongside the existing 47 RCTs. This resulted in a slight fall in the estimated, relative, all cause mortality benefit for CR from 13% to 11%. Critics of CR often claim that much of the evidence was produced before the era of modern cardiological treatments and therefore may not be relevant. However a separate analysis, by the Cochrane CR Group, clearly shows no strong evidence of a reduction (slope: 0.008, 95% CI −0.002 to 0.0186, p=0.101) in relative mortality effect over time.
Although it has often been said that the RAMIT trial ‘showed that CR doesn't work’ this is not correct or indeed what the paper's authors claim which is, that CR as ‘typically delivered in the UK’ may not be effective.
What RAMIT tells us about the effectiveness of CR
The suggestion that some CR programmes in the UK may not be providing all of the benefits shown in RCTs is not new. Others have pointed out that some programmes bear little resemblance to the treatment protocols used in some of the RCTs and have questioned the assumption that these programmes are delivering effective treatment.6 This was one of the reasons for the British Heart Foundation providing funding to establish the National Audit of Cardiac Rehabilitation (NACR).
For 5 years the NACR Annual Report has documented (for the 348 programmes in England, Wales and Northern Ireland) the huge variation in: staffing levels and multi-disciplinary involvement (eg, dietetics, physiotherapy, psychology, occupational therapy), the duration and frequency (eg, 4 to 20 weeks, once or twice weekly), the intensity of exercise prescribed, the methods used to change health behaviour (eg, lectures, cognitive-behavioural methods, written materials) and method of delivery (eg, individual, group based, group based with ‘home exercise’, outpatient, self-management at home, home-based and menu based). The uptake rates, refusal rates, dropout rates and time on the waiting list, all proxy indicators of quality also vary widely.7 For example, in one Trust a part time single-handed nurse or physiotherapist may be attempting to provide a ‘multi-disciplinary’ rehabilitation programme: in another the same patient caseload and case mix may be served by a multi-disciplinary team of 5 full-time staff. We believe that it would be surprising if there were not variation in outcomes. The NACR also collects electronic audit data at the patient level which when pooled across all programmes, has shown that the key outcomes targets for CR defined in the National Service Framework for CHD are being met or exceeded, with the exception of referral, which on average is less than half that advocated. These improvements, are shown both immediately post CR and at 12 months post discharge. One of the most striking longer term outcomes for patients is the reduction in those who were sedentary from 49% before CR to 29% 12 months after CR and of those meeting the UK recommended guidelines for physical activity increased by 15% at 12 months.7 The RAMIT finding that CR patients were less likely to be physically active suggests that we may not be looking at the same population.
Recent observational studies of CR conclude that the extent of benefit is dependent on the dose and mode of intervention.8–11 A study of Medicare beneficiaries (over 600 000 patients) carried out a robust analysis of over 70 000 CR patients against matched non-CR controls. Relative mortality risk reduction of 58% and 34% occurred following CR at 12 months and 5 years respectively. The authors concluded that the dose of CR influenced outcome and that one 5-year death would be averted for every 12 patients receiving CR.11 A recent review of CR trials and observational studies over the last 40 years confirms its contemporary effectiveness.12
The NACR will soon have amassed nearly half a million case records and we are seeking to model; staffing, case mix and rehabilitation processes to determine if some configurations of provision are more effective than others and if so, to attempt to define which are the most cost-effective. Additionally, by linking our data to other cardiac datasets, including survival data, we hope to be able to compare matched cases that did or did not take part in CR and examine the effect on survival in England, Wales and Northern Ireland. It is inconceivable that ethical permission would be gained for any further randomised trials in CR and this is the only realistic way to demonstrate effectiveness in reducing premature mortality.
What could explain the RAMIT result?
For a pragmatic trial it is essential to capture a representative sample of clinical practices and patients and to accurately describe them when publishing findings.13 Unfortunately some conventions of trial reporting were not observed in the RAMIT paper.13 14 Most importantly the usual ‘Consort diagram’ showing recruitment and losses is missing.14 Requests, from professional groups (eg, BACPR), have been made to the main author for further information, so far without success. The investigators did set some criteria for inclusion, ‘more than one discipline’, ‘multifactorial (including exercise, education and counselling)’, ‘ten hours minimum duration’ and the willingness to randomise patients out of treatment. The authors anticipated that this latter requirement might hamper recruitment of programmes and we believe this may have also biased the sample. Even in 1997 CR was regarded, by most of its practitioners, as a treatment that reduced mortality. Those that did agree may have represented a small, atypical, sample. The great majority of patients are keen on CR, the NACR shows that <20% of those invited say they ‘can't’ attend or are ‘not interested’. In RAMIT patients were asked to take part in a trial that would impose a 50% risk that they would miss out on CR or, presumably, if they refused be certain to receive it. Could it be that more of those happy to take the risk of exclusion were those who either had the least to gain or those who weren't keen to take part? We don't claim that self-selection was the only explanation; merely that it is one of the possibilities that can't be explored due to a lack of information.
The response to RAMIT
Online publication of RAMIT has already led to a number of CR programmes being challenged by Doctors, managers and commissioners, to provide counter evidence. It is beyond the scope of this editorial to do more than list some of the benefits demonstrated in the literature, they include: biological (eg, improved cardiac perfusion, less ischaemia, remodelling in CHF, improved endothelial function); medical (eg, reduced mortality, reduced cardiac risk scores, improved adherence to medication); economic (including reduced readmissions, something RAMIT data also supports); social and psychological (eg, improved HQROL, reductions in anxiety and depression). If CR were a medication, it seems likely that every cardiac patient would be offered the treatment: yet, almost everywhere in the world, CR is dramatically under prescribed.
Less than 5% of CR programmes in the UK say that they have a doctor as part of the multi-disciplinary team. It seems likely that the lack of this medical patronage is one of the reasons why, in the last 10 years, CR in England has remained a Cinderella service which is in contrast to the immense improvement in cardiological and cardiac surgery services. The NACR Annual Reports and the accompanying BHF Campaign for CR have, prompted a number of health policy initiatives from the Department of Health. These include a Department of Health Commissioning Pack for CR for England,15 which defines the core elements of a CR pathway and new payment by results tariffs for CR. Together with emerging guidelines, in each of the four nations in the UK, CR is ‘en route’ to reducing the variability in the delivery of rehabilitation and provide, at least, a minimum and uniform standard of service for all patients.
The RAMIT trial was not designed to and does not increase our knowledge of the efficacy of CR. The evidence that CR significantly reduces relative all cause and cardiac mortality in a highly cost-effective manner remains little changed.
There is unjustifiable variation in funding, staffing, content of programmes and referral across CR programmes in England, Wales and Northern Ireland. The RAMIT trial reminds us that CR, like any other treatment, can be delivered ineffectually: but its results should not be generalised to ‘CR as practised in the UK’. Clinical audit of all CR programmes is an essential safeguard to ensure that services are being delivered effectively. Rigorous observational analysis of CR services, using the NACR dataset, will help us identify how best to deliver CR and provide evidence for future clinical guidelines.
Professor Rod Taylor of the Cochrane CR Group Peninsula Medical School.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.