The approach taken in the economic evaluation was similar to previous studies. The objective was to measure service use over time (pre and post intervention) using established methods.63 Service use was measured for the 4 weeks before release and then in the periods up to 6 weeks and 6 and 12 months post release. Data were collected by researchers from case notes and records rather than directly from participants.
The first stage in the process was to clean the data and to identify any inconsistencies. This was followed by regrouping some of the service categories according to levels of use. Six categories of service use were collected: psychiatrist, psychologist, occupational therapist, mental health nurse, care co-ordinator and other. All other contacts were categorised into the following categories: other nurse, outpatient contacts, GP and social care. ‘Other nurse’ included contacts such as general, physical and mental health nurses, substance misuse workers and psychological therapy sessions. Outpatient included those other contacts that were based in hospitals (including the accident and emergency department). GP contacts were contacts in a primary care setting. Social care included social workers, housing officers, community support workers, counsellors and probation officers; it is acknowledged that the last two services are not strictly social care but the use of these was relatively low.
The second stage was to combine the service use data with appropriate unit cost information. These data were mainly derived from the annual compendium published by the Personal Social Services Research Unit at the University of Kent,70 with others coming from NHS Reference Costs 2012–2013.71 Unit costs were multiplied by the number of contacts to calculate a cost per person. These total costs were averaged across all people for the different follow-up periods.
Results are presented separately for the different follow-up periods. Given that each covers a different length of time, the comparability of the data between time periods is problematic. In the tables we report the number and percentage of participants who had a particular service contact, the mean number of contacts among those actually using a service and the mean of cost across all participants. A total number of contacts (excluding inpatient days) and a total cost are also reported. (It is assumed that it may be positive for a participant to have more contacts with health services; however, inpatient contacts may indicate a poor outcome, therefore inclusion in the measure for average contacts per person would be counterintuitive.)
Results
In the 4-week period before release, about two-thirds of both groups had contacts with care co-ordinators, and around one-third had contacts with psychiatrists and mental health nurses (). People having contact with ‘other nurses’ and ‘mental health nurses’ were slightly higher in the TAU group than in the CTI group, with frequency of ‘other nurse’ being particularly high in the TAU group. Overall, total contacts per person using services were similar between the groups, at around five per person.
Service use in 4 weeks before release
The highest mean cost in both groups, after the cost of CTI, was for psychiatrist contacts, followed by care co-ordinators and other nurses (). Together, these services made up 76% of costs for both the CTI and TAU groups. The total for the CTI group was 12% higher than for the TAU group.
Mean costs in 4 weeks before release (2013/14, £)
shows the service use in the 6 weeks after release. The CTI group made more use of care co-ordinators and psychiatrists than the TAU group. The overall average contact (excluding inpatient services) was higher for the CTI group. When looking at inpatient care it can be seen that few were admitted. However, the number of days in hospital differed. CTI participants had more days admitted to medical wards and TAU participants had more days admitted to psychiatric wards.
Service use from release to the 6-week follow-up
Even though very few participants were admitted to hospital, this care accounted for 77% of costs for the CTI group and 84% for the TAU group, excluding CTI costs (). Psychiatrist and care co-ordinator costs were around twice as much for the CTI group than for the TAU group.
Mean costs from release to the 6-week follow-up (2013/14, £)
Critical time intervention contacts account for majority of the costs from release to the 6-week follow-up (see ). This is because the majority of the intervention, in terms of contacts, is carried out in this time period. However, in reality, this cost would likely be more spread out across the follow-ups, as the work would be not be so concentrated in this period. Preparation work that does not involve actual patient contact would shift some of the cost before this period. Similarly, monitoring of patients in the final few months that does not typically result in regular face-to-face contact would also shift some of the costs to later in time. In addition, the single person with a 31-day stay in hospital is driving up the cost per person in the CTI arm as well as the intervention costs.
The period before the 6-month follow-up covers a period of 20 weeks. The number of people in the CTI arm having contacts with care co-ordinators was higher than at 6 weeks and the frequency of these contacts was also higher (). The TAU group had increased contact with mental health nurses compared with the 6-week follow-up point and also compared with the CTI group. The use of care co-ordinators was slightly higher in the CTI group at 6 months than in the TAU group. The average total service costs were again higher for the CTI group (). There is not a significant difference in the cost of mental health nurse contacts between the two groups because of the lower frequency of contacts.
Service use from the 6-week follow-up to the 6-month follow-up
Mean costs from the 6-week follow-up to the 6-month follow-up (2013/14, £)
At the 12-month follow-up, the CTI group’s use of care co-ordinators remained similar to the previous follow-up and noticeably higher than for the TAU group (). The mean number of contacts with care co-ordinators was also far higher in the CTI group. Contacts with psychiatrists are slightly higher and more frequent in the CTI group. A higher number of the TAU group have contacts with mental health nurses; however, the frequency of these contacts is lower. Overall, the number of service contacts in the CTI group was about three times higher than in the TAU group.
Service use from the 6-month follow-up to the 12-month follow-up
shows that psychiatrists and care co-ordinators are the drivers of costs in the CTI arm. For the TAU group, inpatient care is the main cost driver. Overall, the costs are almost identical including mental health nurse contacts.
Mean costs from the 6-month follow-up to the 12-month follow-up (2013/14, £)
At each time point, excluding the 12-month follow-up, the mean costs are higher for the CTI group (). The cost per week is helpful when comparing between time periods. It can be seen that costs increase initially for both groups and then fall after the 6-week follow-up.
Average costs per person across all the follow-up periods (2013/14, £)
Without the imputed CTI intervention costs, the average service use costs in the intervention arm for the 4 weeks before release and the period from release to the 6-week follow-up are £195.35 (down from £717.19) and £440.21 (down from £2592.80), respectively. Both the service costs at 4 weeks before release and at the 6-week follow-up are not significantly different once CTI contacts are taken out, possibly because of the small sample size and high number of dropouts.
Service costs over the whole follow-up period in relation to age show large differences. For those participants aged 20–29 years the mean cost was £2432.94, for those aged 30–39 years it was £4521.58, for those aged 40–49 years it was £1647.20, and for those aged ≥ 50 years it was £2087.87. Participants with schizophrenia had mean costs of £3217.07, whereas for those with depression the cost was £1867.48. Participants who were diagnosed with schizophrenia complicated by a PD had average service costs of £3528.85. The remaining had an average cost of £2445.49. Those with schizophrenia complicated by alcohol or drug misuse had average costs of £3382.34 compared with £2735.07 for those without. Participants with an Axis II diagnosis had average costs of £3071.89 compared with £2546.88 for those without.
White participants’ costs were higher (£3523.04) than black (£2272.75) or Asian (£1953.12) participants’ costs. Perhaps not surprisingly, costs were substantially higher for participants who were unemployed (£3539.55) than for those in work (£1385.40).
Regression analysis performed on the cost data showed that service costs for unemployed participants were 1.5 times higher than for those who were in employment, and that costs for those not at work due to sickness (and those absent from work due to illness) were, similarly, 1.5 times higher than for those in employment. However, this relationship becomes insignificant when you exclude inpatient costs, suggesting that this relationship is driven by the inpatient costs, a service that relatively few people used. The cost breakdowns previously are for only the patients who had full service use information for the year, which totalled 84 participants. These costs also include the imputed cost of the intervention.
Combining the clinical effectiveness from the primary outcome measure, a difference of proportion of people engaged of 0.161 at 1-year follow-up and, costs, a difference of £2485.10 in favour of the CTI arm, gives a cost per additional person engaged of £15,426. To determine if this is cost-effective depends on society’s willingness to pay for an additional person to be engaged with mental health services at 1 year following release.
