< Previous | Contents | Next >

EFFECTIVE AND COST-EFFECTIVE MEASURES TO REDUCE ALCOHOL MISUSE IN SCOTLAND: A LITERATURE REVIEW

CHAPTER TEN RELAPSE PREVENTION

Introduction

10.1 Relapse prevention programmes consist of a combination of psychosocial and pharmacological interventions aimed at maintaining abstinence or problem free drinking following detoxification. The studies reviewed are summarised briefly in tables 10.9 (effectiveness) and 10.10 (cost-effectiveness) at the end of the chapter.

Effectiveness of relapse prevention

Types of intervention

10.2 There are three main forms of psychosocial intervention (Fuller and Hiller-Sturmhofel, 1999):

• cognitive-behavioural therapy (CBT) is designed to help patients to identify high risk situations for relapse and to develop coping strategies;
• motivational enhancement therapy (MET) aims to motivate patients to change behaviour; and
• 12-Step Facilitation Therapy (TSF) is based on the AA approach and provides 12 consecutive activities that should be achieved during the recovery process.

One review has considered the use of transcendental meditation (Alexander et al 1994).

10.3 Pharmacotherapy can be based on aversion therapy or anticraving therapy. Aversion therapy has been available for more than 50 years and relies on an unpleasant reaction between the medication and alcohol. The most common aversive medication is disulfiram. Anticraving medications are a more recent form of therapy and fall into two categories. Opioid or opiate antagonists (such as naltrexone) act by blocking the pleasant effects of alcohol, thus reducing the desire to drink. Acamprosate is another drug aimed at reducing craving although the mechanism by which it achieves this effect is less clear.

10.4 Other drug therapies have been evaluated for effectiveness in relapse prevention. These are mainly drug treatments that affect mood states such as anxiety and depression and include lithium, benzodiazepines and SSRIs (selective serotonin reuptake inhibitors).

Effectiveness of relapse prevention

Psychosocial interventions

10.5 Psychosocial interventions are considered to be effective. Although interventions have not been subject to randomised trials, largely because of the problems of withholding interventions from controls, it is generally accepted that spontaneous remission (recovery without intervention) occurs in about 1/3 of cases (Babor, 1995 cited in Raistrick et al 1999). This provides a benchmark for judging the effectiveness of interventions.

10.6 The best evidence on effectiveness is drawn from a large US study, Project MATCH. 1,726 patients were randomly assigned to receive CBT, MET or TSF. 952 patients had only received outpatient treatment; 774 patients were receiving outpatient aftercare following inpatient treatment. Table 10.1 summarises the results across all interventions for these groups. Interventions were more effective in the group that had previously received inpatient care but there had been no matching between the sample groups. The total percentage achieving abstinence or controlled drinking was 56% to 60% compared with an estimated spontaneous remission rate of 33%.

Table 10.1 12 month follow up results Project MATCH

10.7 Project MATCH found little difference in the overall effectiveness of the 3 interventions. Short-term effectiveness was lower for MET (at 3 months) but longer-term results were similar at 3 years. The main objective of Project MATCH was to investigate whether matching patients to therapies could increase effectiveness but results in this area were rather limited.

10.8 One review of family therapy found this to be effective (Edwards and Steinglass, 1995). The impact of family therapy depended on the gender of the alcoholic, the investment in the relationship and family support. Transcendental meditation has also been demonstrated to be effective in one review (Alexander et al 1994).

Pharmacological interventions

10.9 There is some evidence that Disulfiram may reduce drinking frequency and amount drunk but no evidence of an increase in abstinence rates (Garbutt et al 1999; Hughes and Cook 1997). A recent review of supervised Disulfiram suggests that this may be more effective for selected patients (Brewer et al, 2000).

10.10 Both Naltrexone and Acamprosate have been shown to delay return to drinking. Short-term results for Naltrexone from small trials suggest that relapse rates may reduce by half (Table 10.2).

Table 10.2 Comparison of relapse rates for Naltrexone and placebo

10.11 Most studies of Acamprosate have shown an increase in the number of alcohol free days and higher abstinence rates at 6 months but this effect may diminish over time. Studies with a longer intervention period (48-52 weeks) have shown more sustained effects.

Table10.3 Examples of outcomes for Acamprosate versus placebo

10.12 There is little evidence that other drug therapies are generally effective for reducing or preventing alcohol use. SSRIs have a mild and transient effect in moderate drinkers but no effect on alcohol dependent patients (Lejoyeux 1996; NIAAA 2000). However, it is important that co-existing psychiatric problems are appropriately treated. Treating depression, for example, improves outcomes for drinking regardless of the type of antidepressant.

Population groups

10.13 The populations studied have tended to be determined by alcohol use characteristics rather than any demographic characteristics. Results for specific groups have not been reported. Drug interventions tend to exclude pregnant women.

Settings

10.14 Comparison of settings has not been carried out in any of the studies reviewed. Where reported, interventions have mainly been in outpatient settings.

Providers

10.15 Comparison of providers has not been carried out in any of the studies reviewed.

Intensity of treatment

10.16 None of the literature reviewed considered this issue in detail. Evidence from Project Match suggests that more intensive treatments (CBT and TSF) have better short-term outcomes than less intensive methods (MET) but longer-term outcomes are similar. Two trials of Acamprosate found a significant difference in abstinence rates at 52 weeks following 48-52 weeks treatment, and a significant difference remained at 104 weeks. This contrasts with studies of shorter-term interventions (6 months) where effects were eroded during follow up. A statistical analysis of treatment studies found that more intensive treatments had higher abstinence rates (Monahan and Finney 1996).

Quality and relevance of the evidence

10.17 The main problem with the evidence on psychosocial interventions is the lack of standardisation in the interventions. The effectiveness depends on what is delivered and how it is delivered and this needs to be reproducible if the same results are to be achieved.

10.18 Trials of Naltrexone are mostly small but of good quality. The reported follow up periods are short. There is poor reporting of the nature of the psychosocial interventions used alongside most of the evaluations of drug treatments. Compliance with drug therapy is a problem. Most studies report results on an intention to treat basis and thus the estimates of effectiveness are not compromised. However, drop out rates are fairly high in these research studies and may be higher in routine care settings. This has implications for the overall impact of intervention programmes.

Cost-effectiveness

Psychosocial interventions

10.19 Holder et al (2000) investigated the medical care costs prior to and following the initiation of alcohol treatment in Project MATCH. In Project MATCH, alcoholics were volunteer subjects in a study designed to administer three major forms of outpatient alcoholism treatment (TSF, CBT, and four sessions of MET) in a randomised clinical trial. The treatments were examined in relation to ten primary matching variables. Subjects attended on average two-thirds of the scheduled sessions. Nonetheless, they significantly reduced their drinking amounts and frequency of drinking from 25 days to 6 days per month at one year following treatment. One third of the purely outpatient treatment group drank without bingeing (>33%), compared with one fifth (<20%) who abstained throughout the one-year follow-up. Thus, fairly low levels of high-quality outpatient treatment succeeded in greatly reduced drinking among alcohol-dependent subjects for a substantial follow-up period. Only two of the primary matching variables were found to have significant effects, psychiatric severity and network support for drinking.

10.20 Holder et al investigated the medical and health care impacts of the MATCH treatments. The authors presented the findings of a longitudinal study of before and after treatment costs for 279 patients from 430 project MATCH patients at two of the nine MATCH research units. The medical care costs used in this study were inpatient costs, outpatient costs and total medical costs. Medical care cost data were compiled from hospitals, insurance companies and health care providers, with service providers located by the use of self complete forms issued to trial participants. Holder et al's findings showed treatment costs to the health service were reduced, and that matching patient characteristics to alcohol treatments at intake can reduce health care costs. Table 10.4, taken from Holder et al, shows the findings in terms of client characteristics and healthcare cost savings.

Table 10.4 Clinical effectiveness and cost savings of MATCH treatments

From Holder et al (2000): Parentheses show the treatment which is most likely to produce medical cost savings)

10.21 The results show health care cost savings to be dependent upon the key patient characteristics of alcohol dependence, psychiatric severity and the level of network support for drinking. The MET treatment appears most likely to generate cost savings in patients with low psychiatric severity and low network support for drinking, whilst for the most severe patients in these two groups, CBT generates greater savings. For patients with high levels of alcohol dependence, the TSF treatment produces greater health care cost savings.

Pharmacological interventions Acamprosate

10.22 Annemans et al (2000) investigated the cost-effectiveness of acamprosate in maintaining abstinence amongst weaned alcoholic patients. Costs of treating patients with acamprosate were compared to no pharmaceutical treatment over 24 months. Average costs of relapse were calculated from insurance data based on a Belgian survey among GPs. A Markov model was used to model the movement of patients over time through monthly stages. Possible states were ambulatory follow-up, ambulatory detoxification, institutionalised detoxification and, after detoxification, institutionalised follow up, lost to follow up and death. If there was no relapse, patients would stay in same state; if they relapsed they could either lapse or binge. Probabilities were put into the model from literature based evidence. Costs were applied to each of the states and probabilities computed for patients moving between states. The results were taken from a randomised-controlled trial of 448 weaned alcoholics.

10.23 Simulations were calculated from a 24-month run of the model. Patient outcome was not included in the article, although good cost information provides a breakdown of the main constituent cost components. Total expected cost for the acamprosate strategy is 211,986BEF compared to 233,287BEF for the 'no treatment' strategy. The authors concluded that acamprosate was cost saving to the health care provider, yielding average net savings of 22,000 BEF (£3,370) per patient over 24 months. However, the model had only six states, for simplicity, although there could be many more complications. The results are sensitive to the probability of relapse and under different conditions may not be cost saving. The authors most notably omit wider costs (e.g. productivity etc) and also health benefits.

10.24 Shadlich and Brecht (1998) investigated the incremental cost per additional abstinent alcoholic for adjuvant acamprosate compared to a standard care baseline. Outcome data were taken from the PRAMA study, a randomised-controlled trial, and epidemiological data. Expert opinion and official and administrative statistics were used for cost data. The modelling exercise used a decision tree analysis of a simulation scenario with 500,000 alcoholics. The definition of 'alcoholic' was not made clear. RCT data were based on patients satisfying five of the DSM-III-R criteria for alcohol dependence, with a mean duration of alcoholism of ten years.

10.25 The treatment regimen evaluated was adjunct acamprosate for 48 weeks in addition to standard care. Both treatments took place after an alcohol detoxification and the patients had to be completely abstinent for a minimum of 14 and maximum of 28 days before admission into the study. Some detoxification was inpatient based. The comparator was a treatment of counselling or psychotherapy according to the routine practices of the 12 participating psychiatric out-patient centres with placebo. Health outcomes were measured by abstinence in the medication free follow-up period, but exactly how this was measured was not defined. In addition, adverse health effects of alcohol dependence syndrome were considered (alcohol psychoses, alcoholic fatty liver, acute alcoholic hepatitis and alcoholic liver cirrhosis) but these were only included in the analysis of health costs avoided.

10.26 The treatment costs showed additional medication costs of acamprosate for 48 weeks valued at DM6.78 per day. This was the only active treatment cost valued. It was assumed that costs of standard care were the same for both groups and therefore no information was given. Health care costs were projected for four major health outcomes, psychosis, dependence syndrome, alcoholic hepatitis and alcoholic liver cancer, and measured by the probability of events for the non-abstinent group from expert opinion and standard hospital costs. Total programme costs were estimated at DM 2,169,000 for 1000 patients, with overall cost savings to society of —DM 2602 (£822) per additional abstinent patient.

Quality and relevance of the evidence

10.27 The authors provide an interesting study that could be replicated using data from Scotland. There are issues about standard care and how acamprosate can be administered at the ideal period when people have become first abstinent. However, the study fails to account for a number of potential benefits from treatment and main outcome is limited. Acamprosate is a therapy currently used in the UK. The Edinburgh clinic was one of the centres in the original trials, although the results from the Scottish site did not suggest that there were additional benefits from the treatment.

Inpatient versus outpatient aftercare

10.28 McCrady et al (1986) examined the cost-effectiveness of partial hospitalisation versus inpatient settings after brief inpatient alcohol treatment. A population of alcoholics in need of detoxification or intensive inpatient care, and diagnosed as an alcohol abuser or alcohol dependent, were randomly assigned to partial hospitalisation (PHT) or inpatient treatment (EIP) after inpatient evaluation and / or detoxification. Partial hospitalisation was a hospital detoxification, followed by a period where the patient commuted to hospital from home for 6.5 hours a day. For inpatient treatment, participants continued as inpatients but went into the same programme as the partial patients.

10.29 The costs of treatment were taken from hospital bills, with attendances at hospital costed at an average visit cost, and hospitalisations costed at an average daily rate. Treatment costs included room and board and ancillary charges, whilst daily treatment program costs included work time missed and child care costs. The health care costs of hospitalisation were $183 per day, and of outpatient treatment were $35 per visit. Total treatment costs are shown below (table 10.5). Wider costs measured included legal problems, frequency and problem consequences, income from missing work and child care during hospitalisation. Child care costs were $1.50 per hour, based on 8 hours per day for PHT and 9.5 hours for EIP.

Table 10.5 Per Patient costs (1980 dollars) for PHT and EIP

10.30 Total treatment costs were lower for the PHT group compared to the EIP group. Initial treatment costs were $1700 lower for PHT and outpatient and rehospitalisation costs were comparable. The PHT group was most likely to be hospitalised at other facilities. Unit costs of improvement were defined by the difference between baseline and follow-up abstinent days divided by total treatment costs, to give abstinent days per $100 invested. In the PHT programme this was 5.4 days per $100 and for EIP, 4.2 days per 100. The treatment cost per abstinent subject was $18,935 (PHT) and $21,637 (EIP). The total treatment cost per abstinent or moderate drinking subject was $9,966 (£6,788) (PHT) and $13,222 (£9,007) (EIP).

10.31 The authors concluded that partial hospitalisation offers a more cost-effective alternative to the treatment of alcoholic patients when compared to inpatient treatment. However, 32% of clients did not respond to the extended follow up. There was also a heavy reliance on self-report data, with relatives, or other individuals nominated by the client, reporting more frequent drinking than clients. In terms of study generalisability, wider application will depend on local effectiveness data. If programmes are similarly effective, a less resource intensive intervention is likely to be more cost-effective.

10.32 Pettinati et al (1999) investigated whether patients had better outcomes with inpatient rather than outpatient treatment. A population of 93 inpatients and 80 outpatients with DSM-III-R diagnosis of alcohol dependence was evaluated at treatment entry to a private healthcare setting. Both patient groups followed the same clinical programme based on a 12-step programme of AA, individual, marital, family and group counselling in the intensive treatment period of 4 weeks of inpatient and 6 weeks of outpatient care. Programme differences centred on amount of treatment hours and attendance at support groups. Inpatients attended therapy during the day and AA at evenings. Outpatients attended sessions approximately 1-2 evenings a week, AA meetings on the evenings that they did not attend therapy sessions, and a family education programme during weekends.

10.33 Treatment costs were calculated from service billing and adjusted for geographic and institution specific charges. Wage losses and transportation costs to outpatient sessions were also added. Effectiveness was defined by the probability of returning to drinking given psychiatric severity and/or number of drinking consequences at treatment entry. The average cost per successfully completing inpatient was $9,014 (£6,140) and for an outpatient, $1,420 (£967). This indicates inpatient treatment costs were approximately 6.5 times that of outpatient. The authors calculate a 'cost-effectiveness' ratio at three, six and 12 months follow up as a ratio of inpatient:outpatient costs and find the ratios to be 4.5:1, 5.3:1 and 5.6:1 respectively. The authors conclude that this is a modest 'cost-offset' effect.

Quality and relevance of the evidence

10.34 Much more meaningful cost-effectiveness results could have been illustrated by combining the costs and outcomes in the more traditional way. The study is also confined to a group of patients in the higher socio-economic class brackets, located in a private health care setting, and also uses a non-random study technique. Furthermore, cost information is not well presented and the value of the study is very limited.

Behavioural marital therapy

10.35 O'Farrell et al (1996a) published a cost-benefit and cost-effectiveness analysis of behavioural marital therapy (BMT) with and without relapse prevention (RP) sessions for alcoholics and their spouses, conducted using a population of male alcoholics at the Veterans Affairs Medical Center, USA. The authors addressed whether alcohol related health and legal costs decrease in the 12 months after, compared with 12 months before, an outpatient BMT programme, and whether decreased legal and health costs exceed programme costs. The additional cost savings of the health and legal systems were calculated when RP is added to BMT.

10.36 After participating in behavioural marital therapy, 59 couples with an alcoholic husband were randomly assigned to receive 15 relapse prevention sessions over the following 12 months. Costs were analysed retrospectively to calculate the health and legal costs at intake and follow up and differences between BMT and BMT plus RP. The study did not provide details of how costs were derived but estimated $2,279 per abstinent subject in BMT, and $3,280 in the BMT plus RP group. O'Farrell et al estimate savings in health care and legal system costs as a result of adding the RP component to BMT using a hospitalisation per diem rate of $260.73, $43.54 per day for halfway house stays, $63.01 per day for jail stays.

10.37 In terms of value for money, BMT appears more cost-effective than BMT plus RP in terms of percentage of days abstinent per $100 spent (7% improvement versus 4% improvement per $100 treatment costs). For BMT, the monetary costs of health and legal system utilisation are reduced by an average $4200 per patient, compared with $1,259 average costs of treatment. Comparing BMT and BMT plus RP, the monetary value of benefits minus treatment costs were $4,189 (£2,853) and $1,725 (£1,175) respectively.

10.38 A similar study by Farrell et al (1996b) randomly assigned 36 married male alcoholics who had started individual therapy to one of three treatments: counselling alone, counselling plus BMT, or counselling with an interactional couples group. The interactional therapy encouraged the sharing of feelings and problem solving through discussion and verbal insight into each couple's relationship. The BMT couples group included weekly homework and behavioural rehearsal to promote sobriety through an Antabuse Contact and to increase positive couple and family activities and teach communication skills.

10.39 The resources used to provide the interventions consisted of outpatient treatment sessions, with per session costs taken from VA cost accounting information. The cost of an outpatient mental health visit for alcohol dependence was taken from the Cost Distribution Report at the VAMC (cost $54.55). No further details are presented. Table 10.6 summarises the results of the cost-benefit analysis of the treatments.

Table 10.6 Cost-Benefit results for single versus couples counselling

10.40 The costs of hospital and alcohol treatment were taken from the Veterans Affairs cost schedules and the number of visits was recorded by timeline follow back by patients. The Massachusetts Department of Corrections provided the cost of average stays in jail. Costs of delivering treatments were deducted from the dollar reductions in utilisation to estimate the cost savings. The benefit to cost ratio computed then shows positive cost offsets where ratios exceed unity.

10.41 By adding BMT or interactional couples therapy to the counselling, the number of therapy sessions roughly doubled, as did the cost from about $450 for eight sessions to about sixteen at a cost of $900. The results showed that there were additional costs involved in terms of health and legal costs when interactional couples therapy was added to individual counselling, with a negative benefit-to-cost ratio. However, costs did fall for the BMT and individual counselling only treatments. The authors point to the greatest reduction in costs as a result of the counselling only intervention, and close examination of the data shows that the incremental impact of adding BMT to individual counselling could have a very small, even negative, marginal impact.

10.42 In terms of cost-effectiveness analysis, results are presented in units of outcome per $100 of programme expenditure. In terms of units of improvement in percent days abstinent, from pre-treatment to one year follow up, individual counselling plus BMT yields 5.4 days improvement per $100 invested, individual counselling plus interactional therapy yields 4.3 days and individual counselling only generates 13.6. In terms of costs to produce one continually abstinent participant, for one year, counselling alone cost $1,350 (£920), counselling plus interactional therapy cost $3,580 (£2,440) and counselling plus BMT, $2,143 (£1,460). The results suggest counselling alone to be the most cost-effective treatment.

Quality and relevance of the evidence

10.43 O'Farrell et al's results clearly show the consequences of treatment, but lack details of the costs of the interventions. The study results are also limited by the small sample sizes in the treatments (n=12, n=12 and n=10). In addition, the study results should not be generalised beyond the population in question. The participants were elderly Americans and it is questionable whether the results would be replicated in a UK population with different characteristics, in a very different health care system which operates with different costs.

Modelling of results for Scotland

10.44 The following simulation draws on the work of Schadlich and Brecht (1998) and combines data from the paper with local UK costs to estimate the potential cost-effectiveness of delivering a relapse prevention programme. However, since there are no health outcome data, the simulation is limited to a cost-offset approach, whereby resource costs of avoided adverse health outcomes are compared with the cost of treatment.

10.45 Schadlich and Brecht investigated the cost-effectiveness of acamprosate therapy in Germany. Acamprosate is provided for 48 weeks for 1000 patients as an adjunct to standard care and compared with a placebo (1000 patients). The study examined trial evidence on adverse events as a consequence of treatment with acamprosate or standard care. In terms of an incremental cost-effectiveness analysis the costs of the standard treatments should cancel out leaving the cost of the acamprosate treatment over and above standard care. In addition, the cost of an average GP consultation has been included for the prescribing of acamprosate (cost £18 for a 9.36 minute average consultation).

Costs

10.46 Campral EC has a cost of £24.95 (BNF, September 2000) for an 84 tablet pack. Based on a dose of 3 tablets a week, for 48 weeks, 12 packs would be required per patient at a total cost of £299.40. The total cost of administering the treatment to 1000 patients is shown in table 10.7. The total additional cost of treating 1000 patients with adjunct acamprosate is £317,400.

Table 10.7 Costs for Acamprosate treatment in Scotland

Consequences

10.47 Consequences of the two arms of the trial are listed in chart 10.1 in terms of alcoholic psychoses (ICD291), alcohol-dependence syndrome (ICD303), acute alcoholic hepatitis (ICD571.1) and alcoholic liver cirrhosis (ICD571.2).

Chart 10.1 Adverse health consequences - Acamprosate trial

10.48 Using the data provided on mean length of stay in hospital and in rehabilitation units, the table 10.8 shows the estimated health care consequences as a result of standard care and adjuvant acamprosate treatment. Hospital costs are discounted at 6% per annum, based on the assumption of 10 years before the onset of alcoholic psychoses and alcoholic liver cirrhosis, 4 years before alcohol dependence syndrome, and 5 years before acute alcoholic hepatitis. The adverse consequences in the acamprosate treatment arm total £2,612,742 compared with £3,596,960 for the standard care arm.

Table 10.8 Resource consequences for Acamprosate and standard care

Costs and Consequences

10.49 The incremental costs of acamprosate therapy and the incremental benefits of treatment have been combined to estimate potential cost-offset effects. Treating 1000 patients with acamprosate for 48 weeks, together with a single GP visit, costs approximately £317,400. The savings in terms of hospital care for adverse events avoided in the Acamprosate group are £984,218. This is a very conservative estimate as residential rehabilitation savings are excluded from the calculations. However, some care should be exercised in interpreting this result. The saving represents the value of resources that are released and may be used to treat other patients; it is unlikely that financial savings of the same magnitude could be realised. It should be noted that this analysis excludes any health benefits as a result of treatment.

Limitations of the modelling

10.50 The simulation is based on the cost of the "additional" treatment to standard care, i.e. that prescribing acamprosate as an adjunct requires an additional 2 GP consultations. The simulation suggests that the addition of this drug may result in net resource savings. However, there is a much smaller base of evidence for developing the type of model illustrated in this section compared to that for brief interventions. The model is sensitive to the assumptions made about the type of adverse events related to the relapse of treatment, i.e. patients return to heavy drinking. Fewer adverse events or lower Scottish costs per adverse event may suggest a lower net benefit. However, it should be noted that there has been no "value" calculated for the health benefit (for example in terms of QALYs) that arise from this treatment. The results therefore indicate that there is considerable scope for alcohol treatments to have cost-effectiveness ratios well below current UK benchmarks and indeed some may be resource saving

Further research

10.51 Studies examining the separate and combined effects of Naltrexone and Acamprosate are reported to be underway. Optimal treatment length also needs to be examined and longer-term follow up is required. Better studies of the separate and combined effects of psychosocial and pharmacological interventions are required as existing studies have not described clearly the content of the psychosocial intervention. There is scope to develop economic model using clinical data from the UK for wider simulations than the outline model used in this chapter. Such models combining clinical effectiveness data, data on adverse events of continued patterns of harmful drinking and local cost data could be used to compare different alcohol treatments, the impact of expanding expenditure on alcohol treatment or the potential for new drug treatments such as Naltrexone and Acamprosate combinations. Collecting more routine data from existing programmes could also be used to monitor existing treatments against research expectations.

< Previous | Contents | Next >