Effects of case management after severe head injuryBMJ 1994; 308 doi: https://doi.org/10.1136/bmj.308.6938.1199 (Published 07 May 1994) Cite this as: BMJ 1994;308:1199
- R J Greenwood,
- T M McMillan,
- D N Brooks,
- G Dunn,
- D Brock,
- S Dinsdale,
- L D Murphy,
- J R Price
- Department of Neurological Sciences, St Bartholomew's Hospital, London EC1A 7BE
- Atkinson Morleys Hospital, Wolfson Rehabilitation Centre, and St George's Hospital Medical School, London SW20 0NR
- Wellcome Neuroscience Group, Glasgow G12 0AA
- Department of Biostatistics and Computing, Institute of Psychiatry, London SE5 8AF
- Correspondence to:Dr Greenwood.
Objectives: To examine the effects of early case management for patients with severe head injury on outcome, family function,and provision of rehabilitation services.
Design: Prospective controlled unmatched non-randomised studyfor up to two years after injury.
Setting: Four district general hospitals and two university teaching hospitals, each with neurosurgical units, in east central, north, and north east London and its environs.
Subjects: 126 patients aged 16-60 recruited acutely and sequentially after severe head injury. All received standard rehabilitation services in each of20the six hospitals and districts: case management was also provided for the 56 patients admitted to three of the hospitals.
Main outcome measures: Standard measures of patients' physical and cognitive impairment; disability and handicap; and affective, behavioural, and social functioning and of relatives' affective and social functioning. Relatives' perception of burden; changes in patients' and relatives' housing, financial, vocational, recreational, and medical needs; and ongoing requirements for care and support; and the amount and type of paramedical input provided were assessed with structured questionnaires.
Results: For a given severity of injury, case management increased the chance and range of contact with impatient and outpatient rehabilitation services. However, duration of contact was not increased by case management, and there was no demonstrable improvement in outcome in the case managed group. Any trends were in favour of the control group and could be accounted for by group differences in initial severity of injury.
Conclusions: Widespread introduction of early case management of patients after severe head injury20is not supported, and early case management is not a substitute for improvement in provision of skilled and specialist rehabilitation for patients.
Despite evidence of rehabilitation ameliorating residual problems after severe head injury there is considerable shortfall in provision of such rehabilitation
Case management has been suggested as a way of improving delivery of services that are available
In this study early case management was provided for up to two years after severe head injury
In this study early case management was provided for up to two years after severe head injury
Case management increased patients' contact with rehabilitation services but did not increse the duration of this contact or improve outcome for patients or relatives
Case managment is not a substitute for improved provision of rehibilitation services
There is increasing evidence that the effect of residual problems after severe head injury can be ameliorated by rehabilitation.1 Despite this, there is a considerable shortfall in the provision of rehabilitation for victims of head injury in the United Kingdom.2,3 The services that do exist provide early rehabilitation in the acute sector,4,5 when physical deficits (which usually recover rapidly and completely) are most obvious.6 On later re-entry to the community, cognitive, behavioural, and emotional disturbances cause psychological and vocational failure, social isolation, and a burden on carers, and these often increase over time.6 Despite this, little or no rehabilitation is provided more than six months after injury.4,5
Newcombe et al proposed that fragmentation of responsibility obstructed the successful rehabilitation of such patients.7 We proposed that service shortfalls and a tendency for patients to fall through the net of existing services may also result from inertia in the system, patients' inertia, and carers' and professionals' confusion about the nature of the problems requiring treatment after head injury.8 Coordination of services that are potentially available, particularly in the community, by a case manager has proved effective in other vulnerable populations such as psychiatric patients in the United States*RF 9-12* and the frail elderly in the United Kingdom.13 In Britain case management has also been advocated, though not evaluated, for people with mental handicap14 and is expected to underpin the coordination of community care for several client groups in this country.15
In principle, therefore, there is good reason to expect that case management would provide flexible and cost effective improvement in service provision in the United Kingdom for patients after head injury. However, there is no empirical evidence to show this, and our paper describes the first prospective controlled study to evaluate the effectiveness of one model of case management in improving service input and patient outcome after severe head injury.
Patients and methods
Between March 1987 and March 1988 we established working practices to be adopted by a case manager.8,16 Briefly, after an assessment the case manager formulated a proactive rehabilitation plan, for which detailed clinical knowledge of problem areas after severe head injury was crucial, and facilitated cooperation and involvement of patients, relatives, and professionals. Other than giving general information about head injury and acting as an informal counsellor and a support for the patient and family, the case manager did not provide any formal treatment, retraining, or education but recruited such services from other agencies as required.
At the same time we evaluated service provision after severe head injury in 11 hospitals and districts that fed into two neurosurgical units in north London.5 We used these results to select two groups of three hospitals with roughly similar services and populations for the present study. The hospitals and districts were matched on their type, their location, the number and severity of admissions and discharges after head injury, and the hospital and community facilities. Each group contained one teaching hospital with a neurosurgical unit and two district general hospitals, one peripheral urban and one suburban.
We allocated case management to one group of hospitals by the toss of a coin. Thus, patients in the case managed hospitals received normal services plus case management while those in control hospitals received normal services alone. Between March 1988 and August 1991 we prospectively compared the outcome of case management of severely head injured patients from the hospitals and districts that received case management with outcome in the control hospitals and districts. Ethical approval was obtained from the hospitals involved at each stage of the study.
We chose a design in which hospitals rather than patients were randomised to case management or control services because the advantages of prospectively randomising patients were confounded by two factors. Firstly, there would be talk between staff and patients in the two groups in the same hospital - possibly in the same ward and even in adjacent beds - so that any change in referral practice or facilities generated by the case manager would affect both treated and control groups. Secondly, it would be impossible to blind an assessor to an individual patient's allocation.
We recruited patients who had sustained a closed head injury, were aged 16-60, had been in coma for six hours or had a post-traumatic amnesia of more than 48 hours, who were or whose main carer was resident in the health district of one of the hospitals in the study, and whose family had given informed consent. We excluded patients who had received hospital treatment for drug or alcohol misuse, psychiatric disturbance, or a disorder of the central nervous system during the previous year. We also excluded patients who were of no fixed abode or if follow up was unlikely. The patients were recruited acutely - within seven days of injury - whether or not they had had neurosurgery.
Between March 1988 and November 1990 eligible patients were admitted sequentially to the study. An assessor collected data at entry and at six, 12, and 24 months after injury to determine service provision and patient and family outcome.
On recruitment, each patient's biographical and medical details were recorded on an entry form previously developed in Glasgow. Severity of injury was rated by the duration of coma, the minimum Glasgow coma score on admission,17 and the duration of post-traumatic amnesia with the Galveston orientation and amnesia test.18 Physical impairment and a rating of disability and handicap were recorded with the Bond neurophysical scale19 and the disability rating scale20 respectively. Neurosurgical input, the days and hours of paramedical input, and the duration of stay in intensive care, neurosurgery, and other wards were recorded with forms used in previous studies in Glasgow.
At six, 12, and 24 months of follow up the assessor recorded physical impairment with the Bond scale19 and cognitive impairment with standard tests of verbal and non-verbal intelligence and memory*RF 21- 25* and of language comprehension and fluency26,27 compared with an estimate of premorbid verbal intelligence.28 Changes in the patients' affective, behavioural, and social functioning; changes of the patient's personality; and changes of affective and social function and burden in relatives were assessed with standard instruments*RF 29-32* and questionnaires developed in Glasgow.33 The Barthel index,34 an extended version of the Glasgow outcome scale,35 the Glasgow assessment schedule,36 and the disability rating scale provided global ratings of patients' impairment, disability, and handicap and of the time spent in contact with paramedics, clinical psychology, and other agencies after discharge. Changes in the patients' and relatives' housing, financial, vocational, leisure, and medical needs and ongoing requirement for hours of care, support, and supervision were recorded with questionnaires used previously in Glasgow. Consumer satisfaction with services and information given to patients and relatives and satisfaction with case management was recorded by telephone interview near the end of the study with analogue self rating scales.
Five hypotheses were tested in our study: firstly, that control variables, including those relating to severity of injury, would not differ between the two groups of patients; secondly, that case management would decrease the duration of the initial hospital stay but increase the number of patients in contract with rehabilitation and the duration of this contact; thirdly, that the degree of cognitive impairment in the patients would not be altered by case management; fourthly, that patients' potential for employment and quality of life would be improved by case management; and fifthly, that subjective and objective burden, affective disturbance, and medical input would be reduced in relatives of case managed patients.
We performed routine data description and analysis with SPSS.37 We calculated odds ratios for differences in referral rates, with and without correction for differences at intake and possibly hospital identity, by fitting a linear logistic model38 using the statistical program EGRET.39 Differences in time trends for the outcome variables were analysed with variance components models with the program REML.40 In the latter models the random effects were subject, treating hospital (when it was non-negligible), and the usual residual term. The fixed terms were time, input covariates measuring severity (log score of post-traumatic amnesia, log duration of coma, and coma score), and group (case managed v control). The effect of severity of injury on rate of improvement was modelled by a time by log post- traumatic amnesia interaction, and possible differences between groups in rates of recovery were tested by fitting a group by time interaction. As this was almost always non-significant and more often than not in the wrong direction (case managed patients improving more slowly than the controls), this was usually dropped for the model and treatment effects estimated by the main effect of group. In all analyses 95% confidence intervals were calculated if appropriate.
A main reason for choosing a random effects model to analyse trends in outcome measures - rather than the more familiar methods of trend analysis - was the patterns of missing values in the data files. Traditional methods such as the use of multivariate analysis of variance38 cannot cope effectively with missing data. There were inevitably a few drop outs and missing outcome interviews, but the main source of missing data arose from the design of the trial. Patients were recruited up until six months before the end of the study. Early recruits therefore had follow up data for six, 12, and 24 months. Many patients had data for six and 12 months but not for 24 months, and the last patients to be recruited had data for six months only. The systematic paterns of missing data thus arose from the design of the study and the random date of patients' injury. Observations can therefore be regarded as missing completely at random in the analyses. Random effects models have no difficulty in coping with data of this type and were therefore selected as appropriate means of analysis.
In total, 56 case managed and 70 control patients were admitted to the study. Eight case managed patients and six controls died and were excluded from the study, as were four patients who were untraceable and two who were out of the country at follow up. Thirty one case managed patients and 29 controls were followed up for two years. Table I shows the numbers of patients at each stage of assessment.
Table II shows the baseline characteristics of the patients. The two groups were similar for age, sex ratio, cause of injury, alcohol intake at injury, premorbid intelligence quotient (IQ), and number of patients unemployed or with financial or housing problems before injury. The case managed patients, however, were more severely injured on entry to the study: they were unconscious for significantly longer, had more respiratory complications and tracheostomies and less conservative management, had lower minimum Glasgow coma scores on admission to hospital, and had longer post-traumatic amnesia. Subsequent analysis of service provision or outcome was adjusted for these differences and the effect of hospital membership by stratification or matching.
Table III shows the number of patients referred to rehabilitation services, and table IV shows the time spent in rehabilitation. Case management increased the number of patients in contact with inpatient rehabilitation and hospital outpatient services, but the duration of contact did not differ between groups. There was a tendency also for case management to increase numbers of patients in contact with hospital and outpatient or community services, although in the last case the numbers were small.
After head injury of a given severity case management increased the chance but not the length of admission to a rehabilitation unit. Case management increased the number of referrals to physiotherapy and occupational therapy and especially to clinical psychology, social work, and speech therapy. However, when contact occurred, the hours of contact were not increased by case management. In addition, inpatients with shorter post-traumatic amnesia were more likely to be referred for rehabilitation if they were case managed (table V).
Overall, therefore, case management increased the chance of a given patient coming into contact with rehabilitation facilities as an inpatient or an outpatient. However, the duration of that contact, estimated either in days admitted or hours of treatment, was not increased except for hours of treatment in a neurosurgical unit. This was probably related to an increased length of stay in a neurosurgical unit in the case managed group rather than treatment time per day, which was slightly less in this group.
Measures of patients' residual physical and cognitive impairments, personality, and affective and social functioning at six, 12, and 24 months after injury were not significantly different in the two groups: table VI shows the results for 24 months. Table VII shows the patients' residual disability and handicap, their potential for competitive employment, and their actual employment. There was no significant difference in these measures between the two groups even when severity of injury was allowed for by examining outcome separately in patients with more than two weeks of post-traumatic amnesia.
Family and relatives
Table VIII shows the results at 24 months' follow up for distress reported by patients' families and relatives (estimated by standard questionnaires and direct questions about medical contact by the relatives), changes in the use of their leisure time, and their perception of stress within the family. Even when allowance was made for severity of injury, signficantly more relatives in the case managed group reported that the accident had had a “major” rather than “no” or “some” effect on the family. Other measures showed no significant difference at six, 12, and 24 months except that at 12 months more relatives in the case managed group reported a change in leisure activities since the injury (P=0.01).
Table IX shows relatives' responses to questions about the financial consequences of the injury and the effects it has had on the household, including the hours of care or supervision required by the patient, at six, 12, and 24 months' follow up. Significantly more families in the case managed group thought that their household routine had been upset at 12 months, and this was confirmed by similar though not significant results at 24 months (P=0.7). In addition, more patients in the case managed group required someone to stay in the house to look after them at 12 and 24 months (P=0.04 and P=0.01 respectively).
Summary of outcome
None of the variables measured showed even a trend in favour of benefit in the case managed group let alone significant benefit. Any trends were in favour of the control group: for example, slightly less burden and significantly less effect on the family unit (table VIII) or routine (table IX) and a tendency toward better standard scale scores of disability and handicap (table VII). However, these differences can largely be accounted for by the difference between the groups in initial severity of injury.
Over the past 10 years the concept of case management has emerged as an important element of health care systems for elderly and especially mentally ill and mentally handicapped patients.41 External case management is also advocated in the United States for “catastrophic” illness42 and after severe head injury,*RF 43-45* although it has never been subjected to proper evaluation in these patients. Our study is the first to do this after severe head injury.
All models of case management aim to link the right people with the right problems46 by networking and responsible scheming.47 Our case manager acted as an independent case manager without a budget48 and adopted an enabling rather than therapeutic role,49 not providing any formal treatment but recruiting from other agencies. The work involved outreach and home based rather than office or outpatient based contact with patients, their relatives, and the appropriate services; it was felt that only in this way could rehabilitation programmes be individualised and supervised proactively. The work was time consuming, and each of the three people who acted as case manager during the study found it difficult to manage more than about 20 patients at any one time. The model thus incorporated elements of the “assertive”12 or “clinical”50 case management process advocated by Holloway for the mentally ill.51 The model appeared to be accepted by other professional groups in hospitals and the community and by the patients' families, 19 of 20 families stating at two year follow up that the case manager was “very” or “extremely” helpful.
We recruited patients acutely within seven days of injury rather than months after injury to explore whether case management influenced the provision of rehabilitation services in hospital as well as in the community and the duration of hospital stay. We also thought that early contact by the case manager with the family would provide the best opportunity to ameliorate family distress. Our results show that this model of case management for patients with severe head injury significantly increased the number of patients in contact with formal rehabilitation in hospital and in the community. This effect was greatest for clinical psychology, social work, and speech therapy - services to which referrals are usually made relatively rarely.5
Despite case management increasing contact with formal rehabilitation, no difference in outcome was found between the two groups of patients during two years of follow up. We had expected to find no significant change in measures of physical and cognitive impairment after rehabilitation since the aim of such training is to reduce dependence despite impairment rather than necessarily reducing impairment. Unexpectedly, however, we found that increased referral to rehabilitation did not improve functional ability in or outside the home, increase return to work, reduce family distress, or reduce levels of supervision and care. Similarly, early contact with and provision of information to relatives by the case manager did not have the predicted effect of reducing reported distress despite case management being regarded as helpful.
A possible explanation for these apparent anomalies is that, while case management increased referral to rehabilitation services, it did not increase hours of treatment. The importance of the length of time spent in rehabilitation training is shown by studies of rehabilitation programmes in the United States: programmes which reported a positive effect on outcome in patients with severe head injury involved hundreds of hours of rehabilitation training.*RF 52-54* Our model of case management provided advice, support, advocacy, goal planning, and referral but not interventional training itself. A similar model has failed to show benefit in patients with chronic mental illness.55 We therefore suggest that the failure of our model of case management to benefit severely head injured patients is because the model is unable to modify practice sufficiently to provide patients with significantly more rehabilitation training and because of the shortfall of available skilled and specialist rehabilitation for head injured patients in the United Kingdom.*RF 2-5*
It is unlikely that insensitive outcome measures or the small numbers of patients in each group might have obscured the benefit of case management. Although the standard outcome measures available for patients at the start of the study in 1986 were relatively insensitive, we also used the battery of scales and questionnaires that have been developed in Glasgow over the past 20 years. These allow for detailed inquiry into psychosocial functioning of patients and family members after severe head injury. They have previously been shown to be sufficiently sensitive to detect changes in burden in families and deteriorating psychosocial function in patients6 or patients' level of return to work56 and could reasonably be expected to reflect an effect of case management. Furthermore, any trends in favour of better outcome were found in the control group. This is probably accounted for by the group difference in initial severity of injury, but it makes it unlikely that there was any obscured benefit in the case managed group despite the small size of the groups.
We examined only one model of case management after severe head injury, and the patients were case managed for a maximum of two years after injury (after which problems may well increase rather than decrease6). Thus, our results throw no light on the efficacy or otherwise of other models of case management. For example, case management may benefit patients with complex problems resulting from multiple handicap or at a long time after injury, when rehabilitation input is minimal4,5 and potentially mutable problem areas can be more readily identified. In addition, case management might be made effective if it emphasised modification of professionals' work practices, something we clearly failed to do in this study, or if it operated in the context of an improved range and quality of rehabilitation resources for patients.
This study was funded by the King Edward's Hospital Fund for London, the Joint Research Board of St Bartholomew's Hospital, and the Department of Health.