Impact of centralising acute stroke services in English metropolitan areas on mortality and length of hospital stay: difference-in-differences analysis
BMJ 2014; 349 doi: https://doi.org/10.1136/bmj.g4757 (Published 05 August 2014) Cite this as: BMJ 2014;349:g4757All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
A fascinating piece. It is very heartening to see high quality systems research, something we need more of. However, I would like to propose an alternative population health perspective on this topic given.
1. Peter Elton's rapid response it maybe hard to generalise from one place to another
2. one would need to treat (in a centralised service compared to a non centralised service) 90 odd incident strokes (c1% ARR) to prevent a death
3. with all the inherent cost of the reconfiguration (one may never actually realise the benefit of this in cold hard cash unless you ACTUALLY take the capacity OUT of the system etc etc) and politics of closing units
4. whilst the evidence of benefit of alteplase from some perspectives is contested, rapid thrombolysis (arguably the point of centralisation - moving to a better system etc etc) and prevention of death & disability is perhaps a little bit less contested. However there is a wealth of published data and maybe audits but all the audits Suggesting that that the vast majority DONT get thrombolysed rapidly enough....
5. and the work done on Isle of Wight about the population benefit of different methods of acheiving a reduction in morbidity associated with stroke (an earlier version of STAR tool) giving us a methodology to compare the health gain of different methods of reducing stroke morbidity - ie comparing hyper acute stroke unit with population BP control
then, arguably it follows that:
if the objective is the reduction of morbidity and mortality from stroke.....
and there is limited time, attention (policy makers, clinicians and commissioners) and cash to effect improvement
and thus from a population health perspective
it may be considerably more beneficial in pop health terms to focus your energy on primary prevention of stroke
salt (HTN prevention)
pop management of known BP
Improving anticoagulation rates in AF
and arguably continued policy attention to centralisation of acute stroke care is essentially for a marginal gain compared to main goal from pop h perspective
I accept others have different views, and many make a valid case to improve stroke care now. A population health perspective may give a different angle
a thought experiment
References
LSE / Health Foundation
STAR - socio-technical allocation of resources
http://www.health.org.uk/learning/star/
LSE working paper no 5
Estimating the health gains and cost impact of selected interventions to reduce stroke mortality and morbidity in England
2011
M Airoldi, G Bevan, A Morton, M Oliveira and J Smith
http://www.grammatikhilfe.com/management/documents/WP5_-_Estimating_the_...
Competing interests: No competing interests
Improving acute care for stroke saves lives and reduces disability. It is also cost-effective. This study has shown that the reconfiguration of stroke services in London resulted in a significant decline in risk adjusted mortality, with approximately 168 lives saved at 90 days after admission, and a 7% reduction in length of stay in the first 21 months. This is a tribute to the courage and vision of all involved, as well as an example to other large urban areas worldwide. An important aspect of the London achievement was the increased use of community services, especially Early Supported Discharge (ESD).
While this evidence of the success of the model is encouraging it is of significant concern to note that four years on from the acute reconfiguration there has been limited improvement in the services commissioned to support life after stroke. Despite the emphasis on the importance of services aimed at maximising independence and achieving a good quality of life there are still significant gaps in the provision of ESD, community rehabilitation teams with stroke specific skills, psychological and emotional support and six month reviews in the capital. Following Andrew Marr’s stroke in 2013 his wife, Jackie Ashley, commented on the difficulty of accessing rehabilitation in the community. In an article in The Telegraph in August 2013 she reported that she had been contacted by a large number of stroke survivors who expressed the view that hospital care was excellent, but that support on discharge back into the community was very limited.
ESD is a particularly important element of post-acute care that is vulnerable in the current economic climate. It is relatively cheap (a typical case-load of 100 stroke survivors needs 1.0 WTE Physiotherapist, 1.0 WTE Occupational Therapist, 0.4 WTE Speech and Language Therapist, 1.0 WTE nurse and around 1 PA of consultant time) (CLARCH, 2013), but has proven benefits in patient outcomes, length of stay and readmissions. Up to 50% of stroke survivors can benefit from access to this type of multi-disciplinary stroke specialist team (National Stroke Strategy, 2007).
It is worrying that cracks are already beginning to show in the system as community services are beginning to be reduced (Nigel Hawkes, 2014). During the summer months half of the hyper acute stroke units have been over occupied for significant periods of time. Data from the London Ambulance Service does not demonstrate a significant increase in the number of potential strokes admitted. However, there is difficulty moving the increased number of stroke survivors through the pathway. Data from the Stroke Sentinel National Audit Programme indicates that over the last financial year 15 of the 24 acute stroke units have increased lengths of stay.
Unless appropriate services are commissioned in the community to allow stroke survivors to access the support they need outside of the hospital environment there is a risk to continuing performance of this World-leading service model.
References
Collaboration for Leadership in Applied Health Research and Care (2013) Leading change in Early Supported Discharge. www.clarch-ndl.nihr.ac.uk
National Stroke Strategy (2007) Department of Health
Nigel Hawes bmj2014;349:g5067
Competing interests: No competing interests
The authors are appropriately cautious in their discussion when they state: “we cannot rule out the possibility that the differences in outcomes could be caused by variations in severity of stroke between Greater Manchester and London and the rest of England.” However the press release, which led to considerable national media coverage showed no such caution: the reconfiguration in London has reduced mortality, saving an additional 96 lives a year compared with the rest of England…The less radical centralisation in Greater Manchester had no effect on mortality over and above the decline seen in the rest of England during the study period”.
The meta-analysis of intravenous thrombolysis with alteplase for acute ischaemic stroke showed a substantial reduction of disability at the possible expense of a small increase in mortality (1). This indicates that the optimum configuration of services should not only depend on reviewing mortality but also on other measures of quality of care. Even so, it is important to consider the effect of reconfiguration on mortality.
The population stroke mortality data for persons aged under 85 (diagnosis at death starts to become unreliable over the age of 85) shows that mortality in Greater Manchester declined faster than the national average in recent years whereas Greater London followed the national mortality rate closely. The most comparable area to Greater Manchester, West Yorkshire also showed a decline (Tom Hennell, personal communication):
FIGURE
One of the authors of the BMJ paper commented that: “I think it would be difficult to attribute the decline in stroke mortality observed in the population (no. deaths/population size) over time to reconfigurations in acute stroke pathways without first accounting for the decline in stroke incidence, which we know has been substantial during the last 10-20 years.” (Morris, personal communication). This is true for the overall decline in stroke mortality but less true of the difference between areas although still possible due to differential changes occurring more sharply at the time of reconfiguration such as changes in ethnic minority composition. It is also possible that the larger population decline is due to other factors such as more vigorous public health measures in Greater Manchester than in London but there is little evidence to support that.
As the greatest relative decline in population stroke mortality occurred since partial centralisation in Greater Manchester (which does need to be confirmed by data from subsequent years), it is important to consider whether there could have been reasons for the greater apparent reduction in stroke mortality following hospitalisation in Greater London than elsewhere. One possibility is that in a highly centralised service, patients who would previously have been diagnosed as having a transient ischaemic attack are diagnosed as a stroke. Since the 2014 ICD-10-CM definition of a transient ischaemic attack is “A disorder characterized by a brief attack (less than 24 hours) of cerebral dysfunction of vascular origin, with no persistent neurological deficit” if clinicians at a highly specialist centre are more likely to detect minor persistent neurological deficit after 24 hours, the denominator for strokes will be increased by the inclusion of patients with a low risk of mortality. Whether this is so, the very nature of biases is that others may exist which have not been considered.
It is dangerous to be categorical that the there has been a much better outcome in London than in Greater Manchester based on the adjusted mortality following hospitalisation when some of the unadjusted mortality shows no significant difference, the length of stay (which may reflect disability) shows a greater decline in Greater Manchester as does the population stroke mortality. We should not rule out that there has been an improvement in London not experienced to the same extent elsewhere but any judgement should be made after analysing all the data.
1 Emberson J, Lees KR, Lyden P, Blackwell L, Albers, Bluhmki E at al Effect of treatment delay, age, and stroke severity on the effects of intravenous thrombolysis with alteplase for acute ischaemic stroke: a meta-analysis of individual patient data from randomised trials The Lancet - 6 August 2014 DOI: 10.1016/S0140-6736(14)60584-5
Competing interests: No competing interests
The improvement in London and Manchester stroke services is clearly significant and has offered real clinical benefits to patients. There are however a number of issues regarding this paper and more broadly that significantly limit the extent to which it can be concluded that increased centralisation of stroke (and other) services will automatically result in clinical outcome improvement.
The decision to compare London and Manchester services to the rest of England rather than a more suitably matched control significantly weakens any conclusions about the effect of centralisation. A comparison between London (and Manchester) stroke services and other well performing but differently configured services (as measured by Stroke Sentinel National Audit Programme performance) would seem to potentially offer more information about optimal service configuration, although heavily confounded by funding as previously suggested in Dr Hill’s rapid response. The current comparison shows that well performing services do better than the average as would be expected, but the effects of resource and underlying clinical service quality are major confounders which limit the extent to which conclusions may be drawn regarding the role a centralised service configuration plays in this.
As acknowledged by the authors, the functional status of patients (and new institutionalisation rates) are important metrics in evaluating change in stroke outcomes. Whilst any improvement in mortality is at face value extremely welcome, it would be useful to know if this translates to more survivors with a good (or at least not extremely poor) functional outcome. Future studies may require mortality and functional status data to be collected beyond three months to fully address this.
The authors correctly highlight that centralisation of stroke services cannot easily be a universal solution geographically, even if robust evidence of its’ superiority was to emerge. Additionally, it seems illogical to suggest that their findings could inform the centralisation of other services, whilst citing highly disease specific examples in support of this point. Centralisation may be the answer for a number of services, but this needs robust disease or procedure specific evaluation before we commit significant amounts of ever-scarcer resources to major changes in the way the NHS provides care.
Competing interests: No competing interests
The study authors fail to discuss the impact of significant financial investment in services in the results achieved. The London Stroke Strategy was supported by an additional total £23 million in capital funding as reported by Healthcare for London (1). The cost improvement programme highlighted the need for an additional £13 million investment in HASUs to provide the imaging and nursing needs to improve the service, and a £7 million staffing investment in SUs. My understanding is that similar investment in services was not made for Manchester, nor is there intent for similar investment elsewhere in the United Kingdom despite heavy promotion of the perceived merits of greater centralisation of stroke services.
The study does not therefore go on to discuss whether this largescale investment in stroke services received by London Trusts confounded the mortality and length of stay through resource availability to provide medical care, therapy and social care planning. Improvements in resource in all these areas could reasonably account for the improvements seen in mortality, length of stay and adherence to best practices outlined in the Stroke Sentinel National Audit Program (SSNAP, formerly SINAP). No attempt appears to be made to address this significant potential confounder; or discussion whether equivalent resources in non-centralised settings would be able to offer similar outcomes.
A health economics analysis of the service funding per patient relative to health outcomes UK-wide would be more valuable in assessing whether investment in stroke services would provide better long term cost-efficacy to the wider healthcare economy.
1. http://www.londonhp.nhs.uk/wp-content/uploads/2011/03/London-Stroke-Stra...
Competing interests: No competing interests
As the authors make clear, in the penultimate paragraph of Implications, this study doesn't help much with system redesign in non-metropolitan areas, such as Sussex (where I live). Telemedicine will help in those places, as the authors say, but that requires rotas of stroke physicians as well as robust electronic communications, neither of which seem to be readily achievable across the whole patch. And, although there are bound to be minimum and maximum volumes for optimum outcomes, there is still no conclusive evidence as to what these are. In Sussex, therefore, the search for ways of reducing mortality and making other improvements continues.
Competing interests: No competing interests
Re: Impact of centralising acute stroke services in English metropolitan areas on mortality and length of hospital stay: difference-in-differences analysis
Authors’ reply
We agree with Gooderham that the findings of our study are unlikely to be relevant in rural areas.
We agree with Hill, and we stated in our paper, that a health economic analysis of the reconfigurations in London and Greater Manchester would be valuable. The analysis should consider whether or not the changes in London and Greater Manchester are cost-effective. As Hill points out this ought to account for the up-front investments in London and Greater Manchester. Among other things it also ought to account for the relatively high costs per day of hyper acute stroke care, the impact on mortality and functional outcome, and the lifetime costs incurred by the NHS (primary, secondary and community care), social services and families caring for stroke survivors at different levels of functional outcome. We are planning to undertake such an analysis as part of our study.[1]
Matthew Rudd suggests that rather than use the rest of England as a comparator we ought to have used controls consisting of other well performing but differently configured services. Our comparator was acute stroke services treating patients living in urban areas of the rest of England, which we think is appropriate. The difference-in-differences design that we used was necessary so that we could compare changes over time between areas. Because there is no clear definition of what a “suitably matched” control might be, had we selectively chosen the comparator we were concerned we would have been accused of cherry picking. We agree with Rudd (Elton makes a similar point) that functional outcomes are important; further research to investigate the impact of centralising acute stroke services on these outcomes would be beneficial. We said in the paper that our findings could inform the centralisation of other specialist health care services, meaning that other such services in metropolitan areas might consider whether or not a hub-and-spoke model is appropriate. We were not suggesting that the London acute stroke model ought to be unthinkingly applied to other specialist services, and agree with Rudd that changes only ought to be made following suitably robust evaluation.
Elton made several comments. We were unable to control for stroke severity in our analysis and noted this as a weakness, but as we stated it was not possible to detect any trend in differences in stroke severity between London, Greater Manchester and the rest of England from supplementary audit data. The press release that Elton referred to presents the main findings of the paper, including annualised figures of the number of lives saved in London. Elton provides a graph showing data on declining population stroke mortality for four areas (number of stroke deaths among persons aged under 85 years in each area divided by number of people aged under 85 years living in each area) during the period 2001 to 2012. We maintain the view expressed in the email correspondence reported by Elton that these data are not appropriate for evaluating the impact of changes to acute stroke services. The reason is that the trends shown may have nothing to do with acute stroke care, but could reflect other factors throughout the stroke pathway, including declining incidence of stroke over time.[2] Elton suggests the decline in stroke incidence may have been the same in every area, but provides no evidence to justify this. We believe that to evaluate acute stroke care it is more appropriate to measure outcomes among people who have had a stroke. Based on his data, Elton states mortality in Greater Manchester declined faster than the national average in recent years whereas London followed the national mortality rate closely. There is no statistical analysis to support this, and if it is true, it is unclear if these trends had anything to do with the centralisation of acute stroke services in London and Greater Manchester. Given that centralisation in both areas took place in 2010, that as Elton says there was a similar decline in West Yorkshire (where no equivalent reconfiguration took place) to the one seen in Greater Manchester, and that the trends in the graph are linear from 2006, there is little evidence in his data to suggest that centralisation had any effect on population stroke mortality. But, as stated, we think the metric Elton uses to evaluate the impact of changes to acute stroke services in 2010 is inappropriate. Elton goes on to suggest the decline in mortality in London found in our analysis may be because patients with transient ischaemic attack (TIA) in London after the reconfiguration were more likely to be incorrectly diagnosed as having a stroke. One of the cornerstones of hyper acute stroke care is speedy access to imaging, and comprehensive assessment and diagnosis to determine appropriate treatment. Consequently, if anything it is less likely that TIA patients receiving hyper acute care in London and Greater Manchester are misclassified as stroke patients. Elton also comments on the differences between our unadjusted and adjusted analyses. Our unadjusted figures also show a small decline in mortality in London compared with the rest of England, but there was some evidence of difference-in-differences with respect to age, sex, type of stroke, and deprivation in Greater Manchester and age, ethnic group, and type of stroke in London, indicating the adjusted analysis was necessary. We note the data in Elton’s graph have not been adjusted for potential confounders.
Cutting and Gompertz make an important point that services received by stroke patients after being discharged from the hospital ought to be appropriately supported in order to maintain the benefits seen in the acute stroke setting. We wholeheartedly agree with this view.
We agree with Fell that primary prevention of stroke is also important and ought not to be neglected.
Stephen Morris
Rachael M Hunter
Angus I G Ramsay
Ruth Boaden
Christopher McKevitt
Catherine Perry
Nanik Pursani
Anthony G Rudd
Simon J Turner
Pippa J Tyrrell
Charles D A Wolfe
Naomi J Fulop
References
1 Fulop N, Boaden R, Hunter R, McKevitt C, Morris S, Pursani N, et al. Innovations in major system reconfiguration in England: a study of the effectiveness, acceptability and processes of implementation of two models of stroke care. Imp Sci 2013;8:5.
2 Lee S, Shafe ACE, Cowie MR. UK stroke incidence, mortality and cardiovascular risk management 1999–2008: time-trend analysis from the General Practice Research Database. BMJ Open 2011;1:e000269. doi:10.1136/bmjopen-2011-000269
Competing interests: See article for statement of potential competing interests.