Results

Three hundred and fourteen patients were randomised: 184 to receive the additional intervention (the “trial” group) and 130 to receive only existing standard services (the “control” group). The figure shows the flow of patients in the trial. Seven patients were excluded because they were notified to the head injury register more than six weeks after injury. At six month follow up 218 (69%) randomised patients were interviewed, 132 (71%) from the trial group, 86 (66%) from the control group. There was no evidence of selective follow up of either group (χ²=0.23). The groups were comparable for age and sex at randomisation (table 2).

Statistical tests on data collected at six month follow up relating to the accident itself and the patient’s premorbid social situation indicated no significant differences (p<0.05) between the trial and control groups for age, sex, marital status, employment situation, cause of injury, or associated injuries sustained or severity of head injury as measured by post-traumatic amnesia (table 3).

At six months, the trial group’s totalled ratings on the RHFUQ32 were significantly lower than those of the control group (Mann-Whitney U test:z=−2.54; p=0.01), indicating that the trial group patients experienced less disruption of social activities. Table 4 shows the frequencies of specific disabilities reported and the range of ratings.

The trial group also had significantly lower ratings than the control group on the RPQ27 (Mann-Whitney Utest: z=−2.27; p=0.02), indicating that the trial group patients experienced fewer or less severe post-concussion symptoms. Table 5 shows the frequency of specific reported symptoms and range of totalled ratings.

Table 6 shows subgroup analysis of the totalled ratings on the RHFUQ and RPQ for the control and trial groups, each divided according to severity of head injury (as measured by post-traumatic amnesia). When patients were classified using traditional criteria into mild, moderate, severe, or very severe groups28 the main benefit appeared in the mild and moderate groups. When different cut off points for post-traumatic amnesia were used, patients with post-traumatic amnesia<7 days seemed to benefit most from the early intervention service, at least for better social outcomes.

Six of the patients interviewed at 6 months had been randomised to the control group but had received the early intervention. All these patients had significant scores on the RHFUQ (median=12, range 8–21) or RPQ (median=20, range 2–33), and all but one had post-traumatic amnesia⩽24 hours. They were almost certainly a self selected group of patients with appreciable problems who had asked to receive a service. When they were excluded from analyses, Mann-Whitney test results remained significant for both RHFUQ scores (z=−2.04, p=0.04) and RPQ scores (z=−1.92, p=0.05).

The intervention given to all trial group patients included written information and advice about coping with head injury. After attempts to contact them by telephone or letter, 44 (24%) did not respond and proved impossible to contact. Nineteen (10%) were contacted and given advice over the telephone, 35 (19%) were assessed face to face, 62 (34%) were assessed face to face and required additional telephone support, and 21 (12%) were assessed face to face and required further outpatient contact. Of those 52 trial group patients who were not interviewed at six month follow up, all had been given or sent written information: 29 (56%) had not responded to the invitation to contact the service, three (6%) were contacted and given advice over the telephone, six (11%) had had face to face contact, 10 (19%) were assessed face to face and required additional telephone support, and four (8%) were assessed face to face and required additional out patient contact. The six control group patients who had contacted the service for help were given written information and two were seen once, two needed additional telephone contact, and two additional face to face contact. Three patients randomised to the trial group died before the service could contact them and two patients in the control group died before the six month follow up assessment.

The assessing clinicians guessed whether or not patients had received a service and were 61% successful (X²=23.92) which is at greater than chance levels. This was mainly because a large proportion of control patients told the clinicians that they had not received any intervention before their assessments were completed.

Discussion

The results of this large randomised controlled trial indicate that patients who received specialist intervention after head injury had significantly less social disability and significantly less severe post-concussion symptoms six months after injury (p<0.05), compared with those who did not receive the service.

The problems of defining and registering patients with head injury have been discussed previously in relation to the earlier randomised controlled trial.24 There is no agreed single diagnostic criterion: not all patients even attend hospital and the diagnosis may be missed. It is inevitable that not all patients will have been registered and some clinicians might not agree on the diagnosis in every case. Moreover, it is possible that patients who had had severe injuries and were considered likely to die were not notified. However, all patients surviving a severe head injury would have been registered because team members visited the ward every week. The number of patients in the age range studied was that anticipated and we think that the patients randomised were representative of patients admitted to and discharged alive from most trauma wards with head injury.

Baseline information collected at the time of admission was minimal as there was no routine assessment at the time of notification to randomisation. Practical considerations such as staff resources available and the difficulty in tracing and contacting such a large number of people, many of whom were well, made it impossible to undertake a formal assessment within the first few days. Furthermore, the process of contacting and interviewing all randomised patients would probably have been therapeutic, as providing information and reassurance would have been unavoidable, affecting data from the control group. Data in hospital records are often unreliable and incomplete. The main prognostic factor, post-traumatic amnesia, was assessed at six months after injury in all patients interviewed and this has been shown to be valid and reliable.28 The randomisation process was successful (tables 2 and 3) in that there were no significant epidemiological differences between trial and control groups (p<0.05). We have no reason to doubt that the groups were comparable.

The percentage of patients followed up (69%) was better in this study than in the previous randomised control trial,3 and was good for this particular population of patients. A more complete follow up would be extremely difficult to achieve in a group of patients who often may have recovered fully, most of whom who work, and for whom there is no obvious incentive to cooperate with researchers. The information needed to get in touch with younger patients is often unavailable. For some patients, their head injuries may have arisen in traumatic circumstances which they do not wish to discuss. We have no reason to think that follow up was biased towards either group. It is likely that most of those patients not interviewed at 6 months had few symptoms. The clinicians undertaking the 6 month follow up had not been involved in giving treatment and we think that the follow up data collected were not biased.

The outcome measures used were those already tested for reliability and validity in this type of research.3 We did not use any specific tests of cognitive function. This would have reduced the follow up rate significantly (they cannot be done quickly or by telephone). Moreover, from the perspective of the patient it is their symptomatology and disability that concerns them. The lack of data on cognitive skills does make it more difficult to understand how the intervention helped patients, but does not reduce the validity of the findings.

Two possible reasons why an overall benefit to the trial group patients was shown in the present study but not the previous randomised control trial3 suggest themselves. The earlier study included a large proportion of patients with no post-traumatic amnesia (48%), or, with post-traumatic amnesia<1 hour ( 72%), who had no social disability and no post-concussion symptoms at six month follow up (57% and 43% respectively). In the present study, only 40% of patients had post-traumatic amnesia<1 hour, only 39% had no disability at follow up, and only 28% had no post-concussion symptoms at follow up (tables4 and 5). The present study was influenced far less, therefore, by the “flooring” effects of large numbers of trivial head injuries. Secondly, more patients in the present study received additional intervention rather than information alone: 33% had extra telephone support and 12% outpatient intervention, compared with 13% and 6% respectively in the first randomised control trial

Tables 4 and 5 can be used to estimate the practical benefits arising from the service. Table 4 suggests that the most common social disabilities to be benefited by intervention included relationship with partner, coping with family demands, ability to participate in social activities, and, ability to enjoy leisure activities. Table 5 suggests that the most common post-concussion symptoms to be reduced by intervention included concentration difficulties, headaches, fatigue, sleep disturbance, irritability; and, frustration. The mean totalled ratings for the control and trial groups were 8 and 5 respectively on the RHFUQ, and 14 and 10 respectively on the RPQ.

This suggests that the social disabilities which a representative control group patient might experience at six months could include moderate problems in their relationship with their partner, moderate problems in enjoying leisure activities, and mild problems coping with family demands. A comparable trial group patient might only experience moderate problems in enjoying leisure activities. Similarly, the post-concussion symptoms of a representative control group patient could include severe problems with concentration and fatigue, and moderate problems with irritability and headache. A comparable trial group patient might experience only mild problems with fatigue and concentration, and moderate problems with irritability and headache. The results suggest that the intervention helped patients significantly in coping with social disability, probably by encouraging the use of better compensatory strategies. They also show that the intervention reduced the severity of post-concussion symptoms, probably by minimising the effects of a vicious cycle of anxiety exacerbating post-concussion symptoms and vice versa.

Tables 4 and 5, however, also show that at 6 months after injury a significant number of patients continued to experience some social disability (61%) and post-concussion symptoms (72%). This highlights the danger of giving bland reassurance which can cause anger and frustration in patients if their symptoms and disabilities do in fact persist.19

Table 6 shows the subgroup analysis suggesting that the patients who benefited the most from the intervention were those whose post-traumatic amnesia was <7 days. Those with post-traumatic amnesia ⩾7 days appeared not to benefit. The importance of these findings are difficult to assess because the number of very severely injured patients was small, although explanations do suggest themselves. Firstly, it is possible that a ceiling effect arose from the last group as the outcome measures were not designed for patients with very severe injuries. Such patients often have quantitatively and qualitatively different problems from those whose injuries were less severe. Secondly, the more severely injured patients (who are likely to have remained in hospital many months, and often transferred to a specialist centre and received rehabilitation) may not have benefited from the service the input of which was more appropriate for patients well enough to be discharged home and no longer in post-traumatic amnesia. This finding is consistent with the lack of influence seen in the study of case management.21 Thirdly, it is possible that benefits for this group of patients might not be evident as early as six months after injury.

In summary, the results suggest that all patients with head injury with post-traumatic amnesia<7 days would benefit from routine early intervention by a specialist service if their head injuries or other injuries are sufficiently serious for them to be admitted to hospital. A previous study suggested that patients experiencing significant post-concussion symptoms, or increased anxiety, or significant depression, or evidence of post-traumatic stress, early after injury, were those most likely to experience persisting post-concussion symptoms later on.33 They would probably also benefit from routine early intervention, as would all patients with post-traumatic amnesia⩾1 hour who attend accident and emergency departments only and are not admitted to hospital.3

The Medical Disability Society’s recommendation that all patients with head injury should be followed up routinely requires modification. On the evidence of these results we recommend that all patients admitted to trauma wards with a head injury should be offered early specialist follow up services routinely, as should any patient with post-traumatic amnesia⩾1 hour who attends an accident and emergency department. Specialist services should also be available to patients attending an accident and emergency departments or consulting their general practitioner who report appreciable post-concussion symptoms symptoms, depression, anxiety or post-traumatic stress within a few weeks of their injury, regardless of their post-traumatic amnesia.