Clinical practice recommendation

8.1.1.1.

Evidence-based psychological treatments (in particular, cognitive behavioural therapy) should be considered for the treatment of comorbid depression and anxiety disorders in line with existing NICE guidance for people who misuse cannabis or stimulants, and for those who have achieved abstinence or are stabilised on opioid maintenance treatment.

Current practice

Despite the recent increase in research on psychological treatments, current UK practice is not underpinned by a strong evidence base and there is wide variation in the uptake and implementation of psychological approaches to treatment across services. A number of factors may contribute to this situation. First, the emphasis in many community-based opioid treatment services is based on pharmacological management and supportive case coordination, with practice tending to be influenced more by the background and training of those delivering treatment within services than what research has shown to be effective. Second, a considerable amount of the evidence is extrapolated from other disorders (predominantly alcohol misuse) or other healthcare systems, for example the US or Australia, and inevitably this raises questions about the applicability of the evidence to UK drug misuse services. Third, there has been weak dissemination of the evidence base concerning psychological interventions until recently (Wanigaratne et al., 2005). Fourth, the limited availability of appropriately trained therapists also contributes significantly to variable access to such services in the UK (Lovell et al., 2003).

Standard care in the UK typically consists of keyworking (Knight, 2006) which, as a matter of good practice, involves the building of a therapeutic relationship with the service user and which includes:

• an initial care plan, if required, to address immediate needs (for example, providing information and advice on drug and alcohol misuse)
• harm-reduction interventions
• motivational interventions to enhance retention in treatment
• developing and agreeing the care plan with the client and implementation of the care plan – with interventions relevant to each stage of the treatment journey and regular care plan reviews.

While formal psychological interventions may be delivered by a keyworker, this activity is not part of the keyworking process per se. The keyworker may provide a level of ongoing face-to-face therapeutic support involving the use of some psychological techniques.

Most NHS drug services in the UK tend to focus on people who misuse opioids and to be dominated by substitute prescribing. People who misuse cannabis tend not to be seen as a priority and are rarely included in service contracts, although cannabis misuse is considered a more significant problem among young people and as such interventions for these service users are more readily available. Cocaine treatment services have been developed recently but tend to lack focus and use mostly education-based approaches, for which no evidence has yet been identified.

When evaluating the outcomes of the studies described below, it is important to consider that standard care in the US, where most of the research considered in this chapter has been conducted, may involve higher levels of care and regular counselling, which surpass that usually available in the UK. The American Society of Addiction Medicine (ASAM, 2001) has defined standard outpatient treatment in the US as organised, non-residential services with designated drug misuse professionals providing regular treatment sessions totalling fewer than 9 contact hours per week. Treatment might typically consist of weekly individual and/or group counselling, which would aim to address not only the drug misuse but also wider medical, psychological and social needs. ‘Treatment as usual’ in recent US-based multi-site clinical trials reflects this characterisation (for example, Peirce et al., 2006; Rawson et al., 2004). Timko and colleagues (2003) surveyed all 176 Veterans Affairs substance misuse treatment programmes across the US and found that nearly all (99%) provided some form of drug or alcohol counselling or psychotherapy as part of standard outpatient care, with correspondingly high (90%) utilisation by service users.

8.3.1. Introduction

This section reviews the evidence for psychological interventions alone for the treatment of drug misuse; that is, without pharmacological interventions. Most of this evidence is focused on studies of drugs for which there is, as yet, little or no evidence for effective pharmacological interventions or substitute prescribing, for example cannabis and cocaine.

While most of the literature is focused on adults over the age of 18 who misuse drugs, there also exists an evidence base around psychological interventions (in particular family and social-systems interventions) for adolescents and young people. For the purposes of this guideline, a young person is defined as an individual aged 16–18, and studies have been included for review only if they were judged to include a significant proportion of participants aged 16 or above. (In each given study, at least 50% of participants are aged 16 years or over; where such information is not provided, mean age is greater than or equal to 15.5 years.)

8.3.2. Definitions of interventions

8.3.3. Databases searched and inclusion/exclusion criteria

Information about the databases searched and the inclusion/exclusion criteria used for this section of the guideline are in Table 17.

Table 17

Databases searched and inclusion/exclusion criteria for clinical effectiveness of psychological interventions.

8.3.4. Studies considered¹⁰

The review team conducted a new systematic search for RCTs that assessed the efficacy of contingency management, CBT, interpersonal therapy (IPT), behavioural couples therapy (BCT), family-based interventions and short-term psychodynamic interventions.

In the review of standard CBT, two trials (CRITS-CHRISTOPH1999; MAUDE-GRIFFIN1998) met the eligibility criteria, providing data on 370 participants. Both trials were for cocaine dependence and were published in peer-reviewed journals.

In the review of relapse-prevention CBT, nine trials (BROWN2002; CARROLL1991; CARROLL1994; CARROLL1998; MONTI1997; MCKAY2004; STEPHENS1994; STEPHENS2000; STEPHENS2002) met the eligibility criteria, providing data on 1,314 participants. Of these trials, six were on cocaine dependence (BROWN2002; CARROLL1991; CARROLL1994; CARROLL1998; MONTI1997; MCKAY2004) and three were on cannabis dependence (STEPHENS1994; STEPHENS2000; STEPHENS2002). All trials were published in peer-reviewed journals.

For contingency management, 14 trials (BUDNEY2006; CARROLL2006B; HIGGINS1993; HIGGINS1994; JONES2004; KADDEN2006; PETRY2004; PETRY2005A; PETRY2005B; PETRY2006; RAWSON2006; ROLL2006; SHOPTAW2005; SHOPTAW2006) met the eligibility criteria, providing data on 1,498 participants. Of these trials, six were for cocaine dependence (HIGGINS1993; HIGGINS1994; PETRY2004; PETRY2005A; PETRY2006; RAWSON2006), one for cocaine and/or heroin dependence (PETRY2005B), three for methamphetamine dependence (ROLL2006; SHOPTAW2005; SHOPTAW2006) and three for cannabis dependence (BUDNEY2006; CARROLL2006B; KADDEN2006). All trials were published in peer-reviewed journals.

For couples-based interventions, three trials (FALS-STEWART1996; KELLEY2002; WINTERS2002) met the eligibility criteria, providing data on 123 participants. All trials were published in peer-reviewed journals and were for people who were cocaine dependent or heroin dependent (all participants in these trials underwent detoxification, if required, before receiving the intervention).

For family-based and social-systems interventions for young people, six trials (DENNIS2004 Study 1, DENNIS2004 Study 2, HENGGELER1999, LATIMER2003, LIDDLE2001, WALDRON2001) met the eligibility criteria, providing data on 708 participants. All trials were published in peer-reviewed publications.

For psychodynamic interventions, one trial (CRITS-CHRISTOPH1999) met the eligibility criteria, providing data on 247 participants. This trial was published in a peer-reviewed journal and was for cocaine dependence.

For interpersonal therapy, one trial (CARROLL1991) met the eligibility criteria, providing data on 42 participants. This trial was published in a peer-reviewed journal and was for cocaine dependence.

For cue exposure therapy, no trials met the eligibility criteria.

In addition, 37 studies were excluded from the analysis. The most common reason for exclusion was no drug-use outcomes (further information about both included and excluded studies can be found in Appendix 14). Forest plots and full evidence profiles can be found in Appendix 15 and 16 respectively.

8.3.5. Cognitive and behavioural interventions

Relapse-prevention CBT appeared to be effective for cannabis dependence, particularly compared with waitlist control. However, in one trial (Stephens et al., 1994), where the therapy was compared with a support group, no significant differences were found. This may be explained by the use of group therapy in this trial; individual therapy appears to be more effective (for example, Stephens et al., 2002).

Neither relapse-prevention nor standard CBT was effective for the treatment of cocaine dependence. No differences were found for abstinence and drug misuse outcomes compared with control groups (for summary study information and summary of the evidence see Table 18 and Table 19).

Table 18

Study information and summary evidence table for trials of CBT versus waitlist or standard care, for people who are cocaine or cannabis dependent.

Table 19

Study information and summary evidence table for trials of CBT versus waitlist or standard care, for people who are cocaine or cannabis dependent.

There is strong evidence that contingency management is associated with much longer continuous periods of abstinence for cocaine compared with control groups. People in contingency management groups were more likely to be abstinent from cocaine over 3, 6, 9 and 12 continuous weeks in both prize and voucher reinforcement studies. Only one study compared prize and voucher reinforcement, and this showed a trend favouring prizes (RR = 1.59; 95% CI: 0.94 to 2.69). Although less research has been conducted on its efficacy for methamphetamine and cannabis dependence, it also appears that during treatment contingency management is more effective than control or CBT for these groups of drug users. However, this difference was not sustained at follow-up (for study information and summary of the evidence see Table 20 and Table 21).

Table 20

Study information table for trials of contingency management for people who misuse drugs.

Table 21

Summary evidence table for trials of contingency management for people who misuse drugs.

8.3.6. Psychodynamic interventions

There was a lack of trials assessing psychodynamic interventions. The one included trial did not appear to be effective in terms of abstinence and illicit drug-use outcomes in comparison with a control group (for a summary of study information and evidence see Table 22).

Table 22

Study information and summary evidence table for trials of psychodynamic interventions for people who misuse drugs.

8.3.7. Family- and couples-based interventions

Couples-based interventions were consistently associated with abstinence both at end of treatment and at 6- and 12-month follow-up for people with primary stimulant or heroin dependence. In contrast, the evidence did not suggest family-based and social-systems interventions to be effective for young people (typically around 16 years of age) predominantly misusing cannabis, whether compared with CBT or a less active psychoeducational control (for further details see Table 23 and Table 24).

Table 23

Study information table for trials of family-based interventions for people who misuse drugs.

Table 24

Summary evidence table for trials of family-based interventions for people who misuse drugs.

8.3.8. Clinical summary

8.3.9. Health economics

Economic modelling for contingency management

A decision-analytic Markov model was developed to assess the cost effectiveness of contingency management versus standard care for people who misuse cocaine in the UK. Contingency management involved regular contacts with a case worker over 12 weeks, combined with reinforcement in the form of vouchers exchangeable for retail goods and services, awarded when weekly abstinence from cocaine was achieved. Standard care consisted of less regular contacts with a case worker over the 12-week period. The time horizon of the analysis was 52 weeks. Between 12 and 52 weeks people in both arms of the model were assumed to receive standard care.

Economic model structure

The economic model consisted of two health states, abstinence and non-abstinence. The model was run in weekly cycles. People receiving either intervention were assumed to move from the state of abstinence to that of non-abstinence and not vice versa. All people in the model were assumed to remain in treatment, as no data on retention rates were available from the systematic clinical-effectiveness review.

Costs and health benefits included in the analysis

The economic analysis adopted the perspective of the NHS. Costs consisted of intervention costs only. Additional healthcare costs, incurred by deaths from cocaine use, hospital admissions owing to poisoning and intoxication, as well as inpatient stays owing to mental and behavioural disorders caused by cocaine misuse (Godfrey et al., 2002) were not included in the analysis. Inclusion of such costs would most likely favour contingency management, since this has been shown to be more effective than standard care in achieving abstinence among cocaine users. The measure of health benefit used in the analysis was the QALY.

Effectiveness data utilised in the model

Effectiveness data used in the model were derived from meta-analyses of RCTs that compared the effectiveness of contingency management and standard care for people receiving treatment for cocaine dependence.

Abstinence rates for the 12-week intervention period were taken from studies that reported percentages of service users remaining abstinent from cocaine over a number of weeks within the intervention period. Follow-up data were based on one study that reported percentages of service users who were abstinent at 12 months’ follow-up. Table 25 presents the effectiveness data used in the economic analysis and the clinical studies from which these were derived. Details of the clinical studies are provided in Appendix 14.

Table 25

Data on abstinence rates utilised in the economic model.

For the 12-week intervention period the model utilised data on percentages of users remaining abstinent over a number of weeks during treatment. The percentages of users who remained abstinent over consecutive periods of weeks not reported in the trials (for example over 1 week, 2 weeks, 4 weeks, and so on), were calculated using the available data and assuming exponential fit. The weekly probability of moving from the abstinence to the non-abstinence state at follow-up (between 13 and 52 weeks) was calculated using the reported abstinence rates at 12 months (52 weeks) assuming exponential fit.

Cost data

Owing to lack of patient-level cost data, deterministic costing of relevant resources was undertaken. Relevant resource utilisation was estimated by the GDG and was subsequently combined with unit prices to provide the total intervention cost. For each intervention, the GDG estimated the frequency and duration of contacts with case workers and the frequency of urinalysis tests (dipsticks) undertaken for the detection of cocaine. Cocaine users in the contingency management arm were assumed to receive a £3 voucher for each week they remained abstinent from cocaine during the first 6 weeks in treatment, a £5 voucher for each week of abstinence during the next 6 weeks in treatment and a £10 voucher each time they were found to be abstinent in checks performed at 26, 39 and 52 weeks.

Case-worker unit costs (assumed to be equivalent to those of community nurses paid according to Band 6) were taken from Curtis and Netten (2006), and the price of urine dipsticks was ascertained by personal communication with a pharmacist. Resource utilisation estimates and unit costs associated with contingency management and standard care are presented in Table 26.

Table 26

Resource utilisation estimates and unit costs associated with contingency management and standard care.

Utility data

Utility values required for the estimation of QALYs were derived from Connock and colleagues (2007) and Adi and colleagues (2007). The utility values used in the analysis are presented in Table 27.

Table 27

Utility values used in the economic analysis.

Sensitivity analysis

A sensitivity analysis was undertaken to investigate the robustness of the results under the uncertainty characterising the model input parameters. Selected parameters were varied over a range of values and the impact of these variations on the results was explored. The following scenarios were tested:

• Change in the RRs of the percentage abstinence during treatment or at follow-up of cocaine users receiving contingency management versus standard care. The 95% CIs of RRs calculated in the guideline meta-analyses, as shown in Table 25, were used. Two scenarios examined the simultaneous use of the lower 95% CIs and the upper 95% CIs of all estimated RRs, respectively.
• Changes in the total value of vouchers received by abstinent service users undergoing contingency management. A 100% increase and a 50% decrease were examined.

Results

Base-case analysis

Contingency management was cost-effective over 52 weeks. The ICER of contingency management versus standard care was £11,222 per QALY from an NHS perspective. Full results of the analysis are provided in Table 28.

Table 28

Results of the economic analysis: total average costs and QALYs per user under contingency management or standard care, over 1 year of follow-up.

Sensitivity analysis

Results were sensitive to changes in the RRs of the percentage abstinence achieved by users receiving contingency management versus standard care. When the lower 95% CIs of all estimated RRs were used, the ICER of contingency management versus standard care became £82,631 per QALY. This result was caused by the uncertainty characterising the follow-up data. When only the RRs of abstinence rates relating to the 12-week intervention period were changed to the lower 95% CIs (and RRs of abstinence rates achieved at 52-week follow-up remained intact), then the estimated ICER was £13,093 per QALY. An additional threshold analysis was undertaken to identify the value the RR of the abstinent rate of CM versus standard care should reach in order for the ICER of CM versus standard care to remain below the NICE cost-effectiveness threshold. The results showed that the RR should equal 1.23 in order for the ICER to fall below the threshold of £20,000 per QALY. The ICER was robust when changes were made to the value of reinforcing vouchers. Full results of the one-way sensitivity analysis are provided in Table 29.

Table 29

Results of sensitivity analysis.

Limitations of the economic analysis and overall conclusions

The results of the analysis are subject to various limitations. In order to utilise the available efficacy data, a number of assumptions were required. Follow-up data on abstinence were available for 12 months only. Weekly abstinence rates between 12 and 52 weeks were estimated from these data. In order to construct the economic model it was assumed that once people were found not abstinent, they continued using cocaine and did not achieve abstinence thereafter. This assumption may not accurately reflect abstinence trends among users over time. Also, it was assumed that all users were retained in treatment, due to lack of evidence on drop-out rates.

Healthcare costs additional to intervention costs were not included in the analysis. These were costs incurred by deaths from cocaine use, and hospital admissions owing to poisoning and intoxication, as well as inpatient stays owing to mental and behavioural disorders caused by cocaine use (Godfrey et al., 2002). Voluntary sector costs, social services costs and productivity losses were also not captured in the analysis. If all these cost elements are expected to be lower when higher rates of abstinence are achieved, contingency management is likely to be more cost effective than the findings of the analysis suggest.

Despite the limitations of the analysis, the results indicate that contingency management may be a cost-effective option for people who misuse cocaine, especially when the wider economic consequences of cocaine misuse are considered. Further research is needed to explore more accurately the cost effectiveness of contingency management for cocaine users.

Economic modelling for CBT

A decision-analytic Markov model was developed to assess the cost effectiveness of CBT when compared with no intervention (a waitlist control) for people who are cannabis users. CBT involved regular contacts with clinical psychologists; the time horizon of the analysis was 17 weeks.

Economic model structure

The economic model consisted of two health states, abstinence and non-abstinence. The model was run in weekly cycles. People receiving either intervention were assumed to move from the state of abstinence to that of non-abstinence and not vice versa. All people in the model were assumed to remain in treatment, as no data on retention rates were available from the systematic clinical-effectiveness review.

Costs and benefits included in the analysis

The economic analysis adopted the perspective of the NHS. Only intervention costs were included in the analysis, as data on further potential healthcare costs incurred by cannabis users were not available in the literature. The measure of health benefit used in the analysis was the QALY.

Effectiveness data utilised in the model

Effectiveness data used in the model were derived from meta-analysis of RCTs that compared the effectiveness of CBT and waitlist for people who misuse cannabis. These RCTs were included in the systematic review of clinical studies undertaken for the guideline. The studies reported outcomes in the form of percentage of service users who were abstinent at 17 weeks. Table 30 presents the effectiveness data used in the economic analysis; details of the clinical studies are provided in Appendix 14.

Table 30

Data on abstinence rates utilised in the economic model.

Cost data

Resource use associated with the provision of CBT was estimated by the GDG and subsequently combined with UK unit prices to provide total intervention costs. It was estimated that people who misuse cannabis receive nine CBT sessions over a 17-week period. For the first 8 weeks, CBT sessions are received on a once-weekly basis, with the last session in week 12. Psychotherapists spend on average 45 minutes per session and it was assumed that after any session a dipstick is used to detect use of cannabis.

Unit costs of psychotherapy were taken from Curtis and Netten (2006) and the price of urine dipsticks was ascertained by personal communication with a pharmacist. Given that the comparator is waitlist control, its cost is zero. Resource utilisation estimates and unit costs associated with CBT are presented in Table 31.

Table 31

Resource utilisation estimates and unit costs associated with CBT.

Utility data

Utility values required for the estimation of QALYs were derived from Connock and colleagues (2007) and Adi and colleagues (2007). However, the above studies did not provide utility weights for service users not in treatment (that is, on a waiting list) becoming drug free. It was assumed that the difference in utilities between those in treatment who were abstinent and those in treatment who were not was equal to the difference in utilities between those not in treatment who were abstinent and those not in treatment who were not. The utility values that were used in the economic analysis are presented in Table 32.

Table 32

Utility values used in the economic analysis.

Sensitivity analysis

A sensitivity analysis was undertaken to investigate the robustness of the results under the uncertainty characterising the model input parameters. Selected parameters were varied over a range of values and the impact of these variations on the results was explored. The following scenarios were tested in the sensitivity analysis:

• Change in the RRs of the percentage abstinence over 17 weeks of service users receiving CBT versus waitlist. The lower and upper 95% CIs of RRs calculated in the guideline meta-analysis, as shown in Table 30, were used.
• Changes in the time the clinical psychotherapist spends on CBT sessions.

Results

Base-case analysis

The ICER of CBT versus waitlist control was £31,151/QALY, from an NHS perspective. This value is over the cost-effectiveness threshold of £20,000/QALY as set by NICE (NICE, 2006a). These results indicate that CBT compared with waitlist for cannabis users is not a cost-effective intervention. Full results of the analysis are provided in Table 33.

Table 33

Results of the economic analysis over 1 year of follow-up.

Sensitivity analysis

The ICER of CBT versus waitilist remained above £20,000/QALY under all scenarios examined. Full results of the one-way sensitivity analysis are provided in Table 34.

Table 34

Results of sensitivity analysis.

8.3.10. Clinical practice recommendations

8.3.10.1.

Drug services should introduce contingency management programmes – as part of the phased implementation programme led by the NTA – to reduce illicit drug use, promote abstinence and/or promote engagement with services for people who primarily misuse stimulants.

8.3.10.2.

Cognitive behavioural therapy and psychodynamic therapy focused on the treatment of drug misuse should not be offered routinely to people presenting for treatment of cannabis or stimulant misuse or those receiving opioid maintenance treatment.

8.4.1. Introduction

The use of psychological interventions in combination with drug maintenance treatment is by far the most common application of psychological interventions in UK statutory drug treatment services. The most widely used of the drug treatments is methadone, originally pioneered by Dole and Nyswander (1965) as a treatment for heroin dependence. Less commonly prescribed is buprenorphine, which is a partial opioid agonist but an accepted maintenance treatment for opioid misuse (NICE, 2006f). The rationale for maintenance treatment is that, by using a synthetic opioid, cravings are relieved and, by switching from heroin to a controlled drug, risks and harms associated with illicit drug use can be reduced (for example, injecting behaviour and illegal activities associated with obtaining drugs) and stability can be increased. This stability may create a platform from which to continue psychological work in order to cope with the risk of relapse, deal with associated problems and eventually aim to achieve abstinence and develop a drug-free lifestyle.

As previously discussed, current practice is very varied in the UK. The most common scenario is for people on a maintenance prescription to have regular contact with a worker where practical issues are discussed and reviewed. Furthermore, it is rare in UK services to deliver psychological interventions specifically focused on attempting to reduce illicit drug use within methadone or buprenorphine maintenance treatment programmes. Most commonly, a significant proportion of people in these programmes continue to experience a range of difficulties with other substances, including illicit drugs and alcohol.

8.4.2. Databases searched and inclusion/exclusion criteria

Information about the databases searched and the inclusion/exclusion criteria used for this section of the guideline are in Table 35.

Table 35

Databases searched and inclusion/exclusion criteria for clinical effectiveness of psychological interventions in combination with opioid agonist maintenance treatment.

8.4.3. Studies considered¹¹

The review team conducted a new systematic search for RCTs that assessed the efficacy and/or safety of contingency management, CBT, behavioural couples therapy, short-term psychodynamic therapy, family-based interventions and interpersonal therapy in combination with opioid agonist maintenance treatment.

For methadone maintenance treatment in combination with standard CBT, one trial (WOODY1983) met the eligibility criteria, providing data on 56 participants. This trial was published in a peer-reviewed journal.

In the review of methadone maintenance treatment in combination with relapse-prevention CBT, three trials (EPSTEIN2003; RAWSON2002; UKCBTMM2004) met the eligibility criteria, providing data on 146 participants. One trial (UKCBTMM2004) was unpublished (a full trial report was obtained from the authors) and the other two were published in peer-reviewed journals.

For methadone maintenance treatment in combination with contingency management, 12 trials (CHUTUAPE2001; EPSTEIN2003; MCLELLAN1993; PEIRCE2006; PETRY2002; PETRY2005C; PRESTON2000; RAWSON2002; SCHOTTENFELD2005; SILVERMAN1998; SILVERMAN2004; STITZER1992) met the eligibility criteria, providing data on 1,436 participants. All trials were published in peer-reviewed journals between 1992 and 2006.

For buprenorphine maintenance treatment in combination with contingency management, four trials (DOWNEY2000; GROSS2006; KOSTEN2003; SCHOTTENFELD2005) met the eligibility criteria, providing data on 243 participants. All trials were published in peer-reviewed journals.

For couples-based interventions, one trial (FALS-STEWART2001) met the eligibility criteria, providing data on 36 participants. This trial was published in a peer-reviewed journal.

For family-based interventions, one trial (CATALANO1999) met the eligibility criteria providing data on 132 participants. This trial was published in a peer-reviewed journal.

For psychodynamic interventions, two trials (WOODY1983; WOODY1995) met the eligibility criteria, providing data on 150 participants. All trials were published in peer-reviewed journals.

For programmes for treatment dropouts, no trials met the eligibility criteria.

In addition 24 studies were excluded. The most common reason for exclusion was not providing extractable data (further information about both included and excluded studies can be found in Appendix 14). Forest plots and full evidence profiles can be found in Appendix 15 and 16 respectively.

8.4.4. Cognitive and behavioural interventions

Consistent with the evidence reviewed above of primary stimulant or heroin misuse, behavioural couples therapy and family-based interventions for people undergoing methadone maintenance treatment were associated with reductions in illicit drug use. Psychodynamic interventions did not appear to be effective for reducing illicit opioid use for people undergoing methadone maintenance treatment but there was some evidence for benefit on the secondary outcome of stimulant use and for people who had higher levels of psychiatric comorbidities (for further details see Table 36 and Table 37).

Table 36

Study information table for trials of CBT and contingency management for people in opioid agonist maintenance treatment.

Table 37

Summary evidence table for trials of CBT and contingency management for people in opioid agonist maintenance treatment.

Relapse-prevention and standard CBT do not appear to be effective treatment options for people undergoing methadone maintenance treatment. The majority of trials found no benefit for either form of CBT in comparison with control groups for abstinence and reduction in illicit drug use. However, there was some evidence that standard CBT may be beneficial for a sub-sample who experienced high levels of psychiatric comorbidity.

Consistent with the evidence above of contingency management for cocaine misuse, there is good evidence that contingency management for people undergoing methadone maintenance treatment is strongly and consistently associated with longer, continuous periods of abstinence during treatment and point abstinence at 6- and 12-month follow-up. These findings were consistent for studies using vouchers, prizes and privileges as reinforcers.

However, there is no evidence in support of contingency management for people undergoing buprenorphine maintenance treatment. It appears that contingency management is not associated with improved abstinence and illicit drug-use outcomes for this population. Possible explanations for the lack of effectiveness include some studies reinforcing abstinence from more than one drug, and the low-value reinforcement utilised in one of the studies. However, while most contingency management studies for methadone maintenance focus on reinforcing abstinence from one particular drug and often have higher-value incentives, studies using contingencies similar to those for studies of people undergoing buprenorphine maintenance have also shown much greater effectiveness for methadone maintenance.

8.4.5. Couples-based, family-based, and psychodynamic interventions

There was consistent evidence of benefit for couples-based and family-based interventions. However, the evidence for psychodynamic interventions mostly suggested limited benefit (for further details see Table 38).

Table 38

Study information and summary evidence table for trials of family-based interventions and psychodynamic interventions for people in methadone maintenance treatment.

8.4.6. Biological testing during contingency management

An important component of contingency management is the role of biological testing for the monitoring of drug use, to ensure that incentives are provided for genuine periods of abstinence. The two main issues addressed in this section are the methods of testing that should be used in this intervention and frequency of testing.

Frequency of testing

The detection times for drugs such as opioids and cocaine as discussed above suggest that frequently biological testing is required in order to establish that service users are given incentives based on genuine abstinence from the target drugs. The detection times indicate that oral fluid tests would be required every day and urine tests probably three times per week, at least at the beginning of a contingency management intervention. No data assessing the use of oral fluid testing during contingency management has been found and therefore only data on urinalysis can be analysed in this section.

Most of the earlier studies (for example, HIGGINS1993, HIGGINS1994) used a fixed frequency of three urinalyses per week. This regimen, strictly based on the detection times of drugs in urine, was important in establishing the efficacy of contingency management and for ensuring that abstinence rates reported in such studies could be trusted. However, later studies that sought to implement contingency management in more naturalistic drug treatment settings were aware of the burden such frequent testing provided on services. Therefore a variety of frequencies have been researched in studies of contingency management that attempt to maximise robust drug testing but also seek to ensure that such programmes can be implemented in naturalistic settings.

Table 39 summarises the number of urinalyses used in contingency management studies for participants either with methadone maintenance treatment or without it. There were no studies that compared different frequencies of urinalyses directly; however, it is possible to draw some conclusions by comparing studies using different frequencies.

Table 39

Study information and number of urinalyses used for contingency management.

In relation to methadone maintenance treatment, the majority of studies requested that participants provide three samples per week throughout the time of the intervention. This is probably due to participants having more regular contact with services (in most studies participants had to visit the clinic daily to receive methadone), which made it less difficult to obtain urinalyses frequently. Including studies with one, two and three urine samples per week did not result in significant heterogeneity (I² = 26.7%). This suggests that the responses of the service users were generally consistent across the 12 RCTs and that using less frequent urinalyses did not significantly impact on these responses.

In relation to studies on contingency management for cocaine and/or opioid users without methadone maintenance treatment, there was more variety in how frequently urine samples were collected. Two studies used a tapering strategy where samples were collected three times per week in weeks 1–3, two times per week in weeks 4–6 and once per week in weeks 7–12. One study collected two samples per week, and two studies collected three samples per week. The frequency of urine samples collected had very little impact on the meta-analysis; the findings from studies using the tapering strategy, samples taken two times per week and samples taken three times per week were extremely consistent (I² = 0%).

In conclusion, where there is no clear evidence for the benefit of using a particular frequency of drug testing, there is a need to balance the drug detection time with the resource implications of collecting samples. The use of the tapering strategy discussed above for urinalysis appears to be the preferable option as it begins with three urine samples per week (as suggested by the expected drug detection time), but the frequency of testing is reduced as the service user progresses through the intervention. This approach requires a lesser burden on resources while providing frequent testing early on in the intervention when it is most needed.

8.4.7. Implementation studies of contingency management

Evidence for the efficacy of contingency management in the treatment of drug misuse has been available for over a decade (Petry et al., 2001) but it has not seen widespread implementation in the NHS or even in the US, where much of the efficacy research has been conducted. In this respect contingency management is not unlike many other non-pharmacological treatments where uptake of the intervention can be limited even after the publication of guidance specifically designed to promote its uptake (Sheldon et al., 2004; Grimshaw et al., 2004). Despite these similarities, contingency management appears to raise particular concerns about its implementation in routine care (Petry et al., 2001).

The concerns raised relate to a number of areas and include the attitudes of staff and senior managers, the particularities of the RCTs and the participants recruited to such studies, the costs associated with its implementation, the reluctance of service users to willingly participate in contingency management programmes and the cultural difference between the healthcare system of the US and other, particularly publicly funded, healthcare systems such as in the UK. All of these concerns are seen as potential barriers to effective implementation and will be discussed in light of evidence from implementation studies identified.

A number of studies (Willenbring et al., 2004; McGovern et al., 2004; Kellogg et al., 2005; Kirby et al., 2006; McQuaid et al., 2007; Ritter & Cameron, 2006) have looked at staff attitudes to contingency management and have reported a generally positive attitude by the majority surveyed. Four of the studies took place in the US, with one in Australia (Ritter & Cameron, 2006) and the majority of the participants were employed in publicly funded services such as the Veterans Administration substance misuse services. A number of studies used a questionnaire, the Provider Survey of Incentives (Kirby et al., 2006), to facilitate comparisons between services. In one such comparison, between the US and Australian services, the US showed more positive responses to contingency management, but a significant number of the Australian respondents were neutral rather than negative (Ritter & Cameron, 2006). More senior staff such as senior clinicians and programme managers tended to have more positive attitudes to contingency management, whereas other staff favoured the use of other psychosocial interventions such as CBT or motivational enhancement (McGovern et al., 2004). The specific objections raised by staff are well summarised by Kirby and colleagues (2006) and mirror findings from the other studies. They include the possibility that incentive programmes are viewed by treatment providers as being too costly and labour intensive, that they are too difficult to implement, and a poor fit with what clinicians are already doing, and that treatment providers are not adequately trained to administer contingency management. A recent report of implementing contingency management in the UK (the injectable-opioid clinic in Chelsea and Westminster Hospital) found broadly similar issues (McQuaid et al., 2007). Both staff and service users cautiously supported the incentive programme. In addition, staff perceived service users to be more stable and less likely to use illicit drugs during the intervention. However, consistent with other implementation studies, there were staff concerns about the ethical implications of using incentives in the treatment of drug misuse. McQuaid and colleagues (2007) noted the importance of discussing the theoretical basis of contingency management and its ethical implications in order to gain support from staff.

A number of studies have reported on the implementation of contingency management focusing on organisational responses and service-user outcomes. In the most comprehensive report, Kellogg and colleagues (2005) describe the introduction of contingency management into large publicly funded substance misuse services in New York. The services involved in the implementation programme included: eight methadone treatment programmes, 19 outpatient chemical dependency treatment programmes, eight inpatient detoxification units, two halfway houses, a residential programme run in partnership with a community-based provider, four hospital intervention and referral services, and an intensive case management programme. The programme sought to address the concerns commonly raised and provided important information on the necessary changes required from staff, the training and programmes required to support its implementation and the responses of service users. Unsurprisingly, key to successful implementation was the endorsement of the programme directors and the willingness of the directors and implementation team to engage with the concerns of staff. This also needed to be supported with a full educational and training programme that provided clear direction for staff, many of whom were unfamiliar with the basic principles of contingency management. A crucial element was that staff recognised contingency management as an intervention aimed at changing key behaviours and not simply rewarding people for being generally well behaved. While service-user-based quantitative outcomes in this study were positive, they were very limited and concerned only with increased participation, for example in vocational rehabilitation programmes. However, a series of interviews and discussions with staff and service users suggested that contingency management had: increased service-user motivation for treatment; facilitated therapeutic progress; improved the attitude and morale of staff; and promoted the development of more positive relationships not only between service users and staff, but also among staff members (Kellogg et al., 2005). In this study contingency management shifted from being an intervention that was viewed as being potentially problematic to integrate with other interventions to becoming the main focus of interventions with service users.

Three other studies report some service-user-based outcomes; the first, Petry and colleagues (2001), is a small case series that describes the successful use of contingency management in individuals with a range of substance misuse and psychiatric problems. The second, by Lawental and Eshkol (2006), describes the impact of the implementation of contingency management in a methadone maintenance programme in a drug treatment unit in Haifa, Israel. This study describes the outcomes for two groups before implementation of contingency management (n = 35) and after (n = 41) and reports an improvement of 36% in clean urine tests (chi sq. = 11.08, p < 0.01). No other adjustments were made to the delivery of the unit’s treatment programme other than the introduction of contingency management. The third, by Shoptaw and colleagues (2006), looks at the impact of contingency management on the reduced use of methamphetamine among gay and bisexual men in specialist HIV services in San Francisco. The intention of the programme was to reduce methamphetamine use and thereby also reduce risky sexual practices in a group with a high HIV prevalence. The group studied (n = 143) had a high rate of methamphetamine use, with 42.7% reporting daily use and a further 43.4% at least weekly use; 77.6% of the sample were HIV positive, with large numbers engaging in unprotected sex (for example, 70.6% reported unprotected anal sex in the previous month). The programme reported good recruitment rates, reduced drug use comparable with results in trials with similar populations (Shoptaw et al., 2005) and acceptability to service users. However, retention rates (30% at 12 weeks) were lower than in comparable programmes for non-HIV populations, which were possibly attributed to the lower reinforcement values offered. The costs were considered by the authors to be ‘modest’ and the implementation programme was continued following the completion of the evaluation.

8.4.8. Clinical summary

8.4.9. Health economics

Economic model structure

The economic model consisted of three health states:

• in treatment and abstinent
• in treatment and not abstinent
• not in treatment and not abstinent.

The model was run in weekly cycles. According to the model structure, hypothetical cohorts of the study population received the interventions under assessment and were followed for 52 weeks. People retained in treatment were either abstinent or not abstinent. People who dropped out or were lost at follow-up were assumed to have misused drugs and to have remained non-abstinent thereafter. During the 12-week intervention period, people in treatment were able to move between the abstinent and not abstinent health states. In contrast, at follow-up, people in treatment (that is, for both arms, in standard care and receiving methadone maintenance treatment) who were found not abstinent could not move back into the abstinent state. A schematic diagram of the Markov model is presented in Figure 4.

Figure 4

Schematic structure of the economic model.

Effectiveness data utilised in the model

Effectiveness data used in the model were derived from meta-analyses of RCTs that compared the effectiveness of contingency management and standard care in people who misuse cocaine and/or opioids undergoing methadone maintenance treatment.

Abstinence rates for the 12-week intervention period were taken from studies that reported percentages of service users receiving methadone maintenance treatment remaining abstinent from cocaine and opioids over a minimum number of consecutive weeks within the intervention period. Follow-up data were based on studies that reported percentages of service users who were abstinent at a specific point in time, that is, the end of the intervention period (12 weeks) and at 6-month follow-up. Table 40 presents the effectiveness data used in the economic analysis and the clinical studies from which these were derived. Details of the clinical studies are provided in Appendix 14.

Table 40

Data on abstinence rates utilised in the economic model.

From Table 40 it can be seen that the reported percentages of users remaining abstinent over the 12 weeks of the intervention are considerably lower than the respective percentages referring to the end of the intervention period (at 12 weeks). A possible explanation of this inconsistency could lie in the heterogeneity between the two sets of studies that provided the above results (abstinence over 12 weeks and abstinence at 12 weeks respectively). However, this difference remained when data from studies reporting percentages of users both over 12 weeks and at 12 weeks were examined (in SILVERMAN1998 and SILVERMAN2004 rates of abstinence for contingency management versus standard care over 12 weeks were 8.69% and 0% respectively, and at 12 weeks 58.69% and 15.56% respectively). This discrepancy in the rates of abstinence over 12 weeks and at 12 weeks is probably explained by the fact that although some users were found abstinent at the end of the 12-week intervention period, they were not abstinent over the whole 12 weeks; that is, they had been using cocaine and/or illicit opioids in the early weeks of the intervention but achieved abstinence at the end of the intervention period. In terms of the model structure, this means that during the 12-week intervention period users could move from the non-abstinent to the abstinent state (in addition to moving from the abstinent to the non-abstinent state).

It must be noted that for the 12-week intervention period the model utilises data on percentages of users remaining abstinent over a minimum number of consecutive weeks; data on total weeks of abstinence were not available in the clinical studies. Owing to lack of such data, the economic analysis conservatively assumed that during the intervention period each user had only one period of abstinence, between 1 and 12 consecutive weeks. The percentages of users who remained abstinent over consecutive periods of weeks not reported in the trials (for example over 4 weeks, 5 weeks, 7 weeks, and so on), were calculated using the available data and assuming exponential fit.

Users in treatment at follow-up (that is, under standard care and receiving methadone maintenance treatment, for both arms of the model) were assumed to move from the state of abstinence to that of non-abstinence and not vice versa. The weekly probability of moving from the abstinent to the non-abstinent state at follow-up was calculated using the reported abstinence rates at the end of the intervention (12 weeks) and at 6 months, and assuming exponential fit. This probability was also extrapolated to the period between 6 months and 52 weeks. In order to avoid high levels of heterogeneity between 12 weeks’ and 6 months’ data, only data from studies that reported percentages of abstinence both at 12 weeks and at 6 months were utilised.

Although evidence suggested that there were no significant differences in retention in treatment between contingency management and standard care, rates on retention in treatment reported in the clinical trials were utilised in the economic model. Such rates were primarily available at the end of the intervention (at 12 weeks) and at the 6-month follow-up. As with abstinence rates at follow-up, only data from studies that reported rates of retention in treatment both at 12 weeks and at 6 months were utilised. From the above data, a weekly drop-out rate for the first 12 weeks and a weekly drop-out rate between 12 weeks and 6 months were calculated assuming exponential fit. The latter was also applied to the period between 6 months and 52 weeks. Data on retention in treatment used in the economic analysis and the clinical studies from which these were derived are provided in Table 41.

Table 41

Data on retention in treatment utilised in the economic model.

Cost data

Owing to lack of patient-level cost data, deterministic costing of relevant resources was undertaken. For each intervention assessed, the GDG estimated the frequency and duration of contacts with case workers and the frequency of urinalysis tests (dipsticks) undertaken for the detection of cocaine and/or opioids. The GDG also estimated the average daily dose of methadone prescribed to the service users over the time horizon of the analysis. Estimated resource use was subsequently combined with UK unit costs to provide the total intervention costs. People in the contingency management arm were assumed to receive a £3 voucher for each week they remained abstinent from cocaine and opioids during the first 6 weeks in treatment, a £5 voucher for each week of abstinence during the next 6 weeks in treatment, and £5 vouchers each time they were found to be abstinent in checks performed at 26, 39 and 52 weeks.

Case-worker unit costs (assumed to be equivalent to those of community nurses paid according to Band 6) were taken from Curtis and Netten (2006). The price of urine dipsticks was determined by personal communication with a pharmacist. Methadone unit costs were taken from BNF 53 (British Medical Association, 2007). Resource utilisation estimates and unit costs associated with contingency management and standard care are presented in Table 42.

Table 42

Resource utilisation estimates and unit costs associated with contingency management and standard care.

Further healthcare costs, including costs associated with attendances at an emergency department, GP visits and inpatient care for physical health problems, as well as inpatient and outpatient mental healthcare, were based on resource use data derived from the NTORS study (Gossop et al., 1998). Using these data, Godfrey and colleagues (2002) estimated the annual healthcare costs incurred by people who misuse Class A drugs in England and Wales, excluding treatment for dependence. Costs were reported separately for people who misuse drugs not in treatment for dependence, for those in treatment for less than a year, and for those in treatment for more than a year. Costs relating to the first two categories of people who misuse drugs were utilised in the economic analysis. Table 43 provides healthcare resource use estimates and respective costs incurred by people who misuse drugs in England and Wales, as reported by Godfrey and colleagues (2002).

Table 43

Annual healthcare resource use and costs incurred by people who misuse Class A drugs in England and Wales.

From Table 43 it can be seen that healthcare costs are higher for people who misuse drugs in treatment than for those not in treatment. This finding suggests that increasing the number of those in treatment may result in an increase in healthcare costs in the short run. In addition, healthcare costs estimated by Godfrey and colleagues (2002) were not adjusted to take into account the impact of current drug misuse on future healthcare demands. As a consequence, potential future costs from infectious disease risks among people who misuse drugs have not been included in the above estimates of healthcare costs and, consequently, in the economic analysis undertaken for this guideline.

Godfrey and colleagues (2002) did not report data on PSS costs associated with drug misuse; for this reason, such costs have been assumed to be negligible in the economic analysis. Criminal justice system and crime victim costs, which were included in the non-reference case analysis, were available in Godfrey and colleagues (2002). Criminal justice system costs included costs associated with drug arrests, arrests for acquisitive crimes, stays in police custody, appearances in court and stays in prison. Crime victim costs referred to material or physical damage, crime victims’ loss, and expenditure in anticipation of crime. Table 44 provides estimates of crime-related costs for people who misuse drugs not in treatment and for those in treatment for less than a year, as reported in Godfrey and colleagues (2002).

Table 44

Annual criminal justice system and crime victim costs incurred by people who misuse Class A drugs in England and Wales.

It should be emphasised that the amount of healthcare costs and crime-related costs incurred by people who misuse drugs as reported in Godfrey and colleagues (2002) exclusively depended on whether people were engaged in treatment or not; the impact of effectiveness of treatment (in terms of achieving abstinence from drug misuse) on these costs was not discussed in the study. Therefore, the present economic analysis has not differentiated between abstinent users and non-abstinent users in treatment for estimation of costs.

Healthcare costs were adjusted to 2006 prices using the hospital and community health services pay and prices inflation rates (Curtis & Netten, 2005). The inflation rate for 2005/6 was estimated using the average value of the hospital and community health services pay and prices inflation rates of the previous 3 years. Crime-related costs were adjusted to 2006 prices using the Retail Prices Index (Office for National Statistics, 2007).

Utility data

Utility values required for the estimation of QALYs were derived from Connock and colleagues (2007) and Adi and colleagues (2007). The utility values in the economic analysis are presented in Table 45.

Table 45

Utility values used in the economic analysis of contingency management versus standard care for people misusing drugs receiving methadone maintenance treatment.

The final utility values for the health states ‘in treatment – reduction in drug use’ and ‘not in treatment – drug misuse’ were weighed according to the proportion of injectors in the population of people who misuse drugs receiving methadone maintenance treatment, reported in the NTORS study (Gossop et al., 2003). According to the study, the proportion of injectors in this population was 61% at initiation of treatment and 44% at 1 year of treatment. The economic model assumed that the proportion of injectors among people who misuse drugs who dropped out of methadone maintenance treatment was the same as that characterising the population at initiation of treatment in Gossop and colleagues (2003).

Results

Base-case analysis

Contingency management was cost effective over 52 weeks. The ICER of contingency management versus standard care was £15,219 per QALY from an NHS/PPS perspective and £74 per QALY from a wider perspective including criminal justice system and crime victim costs. Full results of the analysis are provided in Table 46.

Table 46

Results of the economic analysis: total average costs and QALYs per user under contingency management or standard care, over a year of follow-up.

Sensitivity analysis

From an NHS/PPS perspective, results were sensitive to changes in the RRs of the percentage abstinence achieved by users receiving contingency management versus standard care. When the lower 95% CIs of all estimated RRs were used, the ICER of contingency management versus standard care became £68,283 per QALY. It must be noted, however, that the base-case results were robust under changes in the RRs of abstinence rates referring to the 12-week intervention period only (that is, when RRs of abstinence rates achieved at follow-up remained intact). In this case, the ICER of contingency management versus standard care was £16,219 per QALY, which is below the NICE-set cost-effectiveness threshold of £20,000 per QALY (NICE, 2005b). It was therefore the uncertainty characterising the follow-up data used in the analysis that strongly affected the results.

The ICER was robust in changes in the value of reinforcing vouchers, in the use of lowest and highest estimates of healthcare costs reported in Godfrey and colleagues (2002), as well as in the exclusion of these costs.

When a wider perspective that included crime-related costs was considered (non-reference case analysis), contingency management was cost effective (that is, its ICER versus standard care was below £20,000 per QALY) under all scenarios explored.

Full results of the one-way sensitivity analysis are provided in Table 47.

Table 47

Results of sensitivity analysis.

8.4.10. Clinical practice recommendations

8.4.10.1.

Drug services should introduce contingency management programmes – as part of the phased implementation programme led by the NTA – to reduce illicit drug use and/or promote engagement with services for people receiving methadone maintenance treatment.

8.4.10.2.

Contingency management aimed at reducing illicit drug use for people receiving methadone maintenance treatment or who primarily misuse stimulants should be based on the following principles.

• The programme should offer incentives (usually vouchers that can be exchanged for goods or services of the service user’s choice, or privileges such as take-home methadone doses) contingent on each presentation of a drug-negative test (for example, free from cocaine or non-prescribed opioids).
• If vouchers are used, they should have monetary values that start in the region of £2 and increase with each additional, continuous period of abstinence.
• The frequency of screening should be set at three tests per week for the first 3 weeks, two tests per week for the next 3weeks, and one per week thereafter until stability is achieved.
• Urinalysis should be the preferred method of testing but oral fluid tests may be considered as an alternative.

8.4.10.3.

Staff delivering contingency management programmes should ensure that:

• the target is agreed in collaboration with the service user
• the incentives are provided in a timely and consistent manner
• the service user fully understands the relationship between the treatment goal and the incentive schedule
• the incentive is perceived to be reinforcing and supports a healthy/drug-free lifestyle.

8.4.10.4.

Drug services should ensure that as part of the introduction of contingency management, staff are trained and competent in appropriate near-patient testing methods and in the delivery of contingency management.

8.4.10.5.

Contingency management should be introduced to drug services in the phased implementation programme led by the NTA, in which staff training and the development of service delivery systems are carefully evaluated. The outcome of this evaluation should be used to inform the full-scale implementation of contingency management.

8.4.10.6.

Behavioural couples therapy should be considered for people who are in close contact with a non-drug-misusing partner and who present for treatment of stimulant or opioid misuse (including those who continue to use illicit drugs while receiving opioid maintenance treatment or after completing opioid detoxification). The intervention should:

• focus on the service user’s drug misuse
• consist of at least 12 weekly sessions.

8.4.10.7.

All interventions for people who misuse drugs should be delivered by staff who are competent in delivering the intervention and who receive appropriate supervision.

8.4.11. Research recommendations – contingency management

8.5.1. Introduction

Naltrexone is an opioid antagonist that blocks the euphoric and other effects of opioids, and therefore eliminates the positive rewards associated with opioid use. A recent health technology appraisal conducted by NICE (2006c) concluded that naltrexone may have some limited benefit in helping those who have been detoxified from opioids in remaining abstinent, although very limited evidence also suggests naltrexone to be more effective in individuals who are highly motivated. The health technology appraisal acknowledged that naltrexone loses its protective effect if the service user does not take the medication, and also recommended that people who are prescribed naltrexone engage in psychosocial interventions, such as counselling and self-help groups, that promote concordance with medication. However, the evidence presented suggests that only contingency programmes providing incentives for individuals to remain abstinent have a positive impact on naltrexone concordance and other outcomes. A central question is whether the wider evidence base for psychosocial interventions substantiates the recommendation in the health technology appraisal.

Naltrexone is not widely used in the UK, accounting for only 11,000–14,000 prescriptions per annum, not all of which would be for managing opioid dependence (NICE, 2006c). Where it is prescribed, it is not evident whether this is done as part of a comprehensive package of care that includes psychological intervention and general support.

8.5.2. Databases searched and inclusion/exclusion criteria

Information about the databases searched and the inclusion/exclusion criteria used for this section of the guideline is in Table 48.

Table 48

Databases searched and inclusion/exclusion criteria for clinical effectiveness of psychological interventions in combination with naltrexone maintenance treatment.

8.5.3. Studies considered¹²

The review team conducted a new systematic search for RCTs that assessed the efficacy of contingency management, interpersonal therapy, CBT, behavioural couples therapy (BCT), and psychodynamic and family-based interventions in combination with naltrexone maintenance treatment (see Table 49).

Table 49

Study information and summary evidence table for trials of psychological interventions in combination with naltrexone versus control.

In the review of naltrexone in combination with contingency management, three trials (CARROLL2001B; CARROLL2002; PRESTON1999) met the eligibility criteria, providing data on 171 participants. All trials were published in peer-reviewed journals.

For naltrexone in combination with relapse-prevention CBT, two trials (RAWSON2001; TUCKER2004B) met the guideline eligibility criteria, providing data on 253 participants. All trials were published in peer-reviewed journals.

For naltrexone in combination with family-based interventions, two trials (CARROLL2001B; FALS-STEWART2003) met the eligibility criteria, providing data on 216 participants. All trials were published in peer-reviewed journals.

In addition, two studies were excluded from the analysis. The most common reason for exclusion was poor study quality (further information about both included and excluded studies can be found in Appendix 14). Forest plots and full evidence profiles are provided in Appendix 15 and 16 respectively.

8.5.4. Psychosocial interventions in combination with naltrexone maintenance treatment

A recent NICE technology appraisal (NICE, 2007) concluded that psychosocial interventions should be provided for people undertaking naltrexone maintenance treatment in order to increase adherence to this medication. This section of the guideline assesses the efficacy of such psychosocial interventions in greater detail. The study information and summary evidence is in Table 49.

8.5.5. Clinical summary

Contingency management, behavioural couples therapy and family-based interventions were all associated with significantly improved outcomes during treatment, but there is very limited follow-up data in any of the six trials and no evidence of long-term benefit.

There were mixed results for CBT. The trial with a 52-week duration appeared to be effective, although a more recent 12-week trial did not appear to affect drug use.

Given the recommendation in the NICE technology appraisal for a specific psychosocial intervention to support the use of naltrexone (which currently has a very low rate of uptake in the NHS), current evidence would suggest that service user and clinician preference, and whether the service user is in close contact with a partner or family member, should direct the choice of contingency management, behavioural couples therapy and family-based interventions.

8.5.6. Clinical practice recommendation

8.5.6.1.

For people receiving naltrexone maintenance treatment to help prevent relapse to opioid dependence, staff should consider offering:

• contingency management to all service users (based on the principles described in recommendations 1.4.1.3 and 1.4.1.4)
• behavioural couples therapy or behavioural family interventions to service users in close contact with a non-drug-misusing family member, carer or partner (based on the principles described in recommendation 1.4.3.1 for behavioural couples therapy).

8.6.1. Introduction

There is a long tradition in North America and Europe of self-help groups for people who misuse drugs. Most of these offer a programme of recovery known as the 12-steps, which has its origins in AA. Self-help groups especially relevant to people who misuse drugs are Narcotics Anonymous (NA) and Cocaine Anonymous (CA). There are other self-help groups available that offer alternative philosophies and approaches, but these have not taken root in the UK to the same extent as 12-step groups. There is open access to groups; the only entry requirement is for individuals to acknowledge that they have a drug problem. In principle, individuals may attend simply with the desire to become abstinent. It is not a requirement to be drug free at first attendance nor to abstain from the prescribed use of medication (including methadone maintenance), although in practice disapproval of opioid maintenance treatment is not uncommon among some 12-step communities.

There have been few research studies into the acceptability of the 12-step programme among UK drug users; however, a series of studies conducted in London NHS inpatient detoxification services (for example, Best et al., 2001) suggested that people who were drug dependent reported more positive attitudes to NA/AA and to the 12-step programme than those who were alcohol dependent, and reported a greater intention to attend after detoxification.

8.6.2. Definitions of interventions

8.6.3. Databases searched and inclusion/exclusion criteria

Information about the databases searched and the inclusion/exclusion criteria used for this section of the guideline is in Table 50.

Table 50

Databases searched and inclusion/exclusion criteria for clinical effectiveness of self-help interventions.

The review team conducted a systematic search for RCTs and observational studies that assessed the efficacy of 12-step self-help groups. Seven studies met the inclusion eligibility criteria set by the GDG. Two were RCTs (McAuliffe, 1990; Timko et al., 2006), two were cohort studies (Moos et al., 1999; Ethridge et al., 1999), one was a prospective longitudinal study (Fiorentine & Hillhouse, 2000), one was a case series (Toumbourou et al., 2002) and one was a sub-analysis of self-help group participation in all groups of an RCT (Weiss et al., 2005). All studies were published in peer-reviewed journals.

In addition, 16 studies were excluded from the analyses. The most common reason for exclusion was diagnosis of comorbid psychosis.

8.6.4. Benefits of attendance at self-help groups

The majority of studies on self-help groups have looked at 12-step-based groups. Various studies show that 12-step involvement has a positive impact on outcomes. For example, Weiss and colleagues (2005) show that, while simple attendance did not predict drug use, active participation in self-help groups did predict lower cocaine use in the following month and increasing levels of participation produced a significant incremental benefit. Similar associations between NA attendance and improved drug-use outcomes are reported by Fiorentine and Hillhouse (2000). Four hundred and seventeen participants commencing outpatient substance misuse treatment had an intake interview and 8 months later a follow-up interview, in order to determine the relationship between drug treatment participation and 12-step involvement. Overall findings illustrate that individuals who regularly attended 12-step programmes prior to treatment had significantly higher rates of successful treatment completion. Fiorentine and Hillhouse (2000) also demonstrate an additive effect of engaging in treatment and a 12-step self-help group at the same time, as this results in significantly better treatment outcomes when compared with drug treatment or 12-step self-help group participation alone. In Australia, Toumbourou and colleagues (2002) conducted interviews with 91 new members entering NA self-help groups. At baseline, participants filled in questionnaires regarding sociodemographic status and attendance levels at 12-step self-help groups in the year prior to the first interview. At 12-month follow-up, they completed a second interview detailing levels of involvement, highest step completed and levels of weekly attendance at the self-help groups. Self-report measures indicated that higher and more stable levels of NA involvement were associated with less marijuana and hazardous alcohol use.

McAuliffe (1990) conducted an RCT comparing a recovery training and self-help programme with a control condition. The recovery training and self-help group received a combined programme of professionally led recovery skills workshops and weekly self-help group meetings (not 12-step). Improved drug-use outcomes were shown at 6 and 12 months in both a US and a Hong Kong sample. This may indicate that non-12-step self-help groups are also beneficial in reducing relapse.

There is consistent evidence that 12-step attendance mediates better substance misuse outcomes. However, it should be noted that in most studies reviewed above, attendance at self-help groups was assessed alongside other treatment programmes. Although there are clear associations between self-help group attendance and drug-use outcomes, the impact of self-help groups outside intensive treatment programmes has not been assessed in enough detail.

8.6.5. Facilitating self-help group affiliation

A variety of studies have assessed interventions that encourage self-help group affiliation. These interventions range from ‘intensive referral’, providing advice, information and a personal contact (Timko et al., 2006), to residential programmes with a strong 12-step focus.

A large-scale prospective cohort study (n = 3,018) conducted by Moos and colleagues (1999) revealed that people receiving 12-step-based treatment for drug and/or alcohol misuse had superior abstinence outcomes compared with those in CBT or eclectic (based on a combination of 12-step and CBT principles) treatment groups. Humphreys and colleagues (1999) sought to further investigate the relationship between post-treatment self-help group participation and abstinence. They suggest that the level of participation in self-help groups may mediate the relationship between self-help group involvement and abstinence; that is, those receiving 12-step-based treatment were more highly involved in the programme than those in either CBT or eclectic treatment programmes; thus, increased levels of participation may have facilitated positive outcomes.

Timko and colleagues (2006) investigated the effects of intensive versus standard referral to self-help groups (based on the 12-step model), in order to determine which method increased self-help group attendance over a 6-month period. Participants commencing substance misuse outpatient treatment were randomly assigned to either group; those in the standard referral group received a timetable of local meetings. Participants in the intensive referral group received the same material as those in the standard group, with the addition of an information pack detailing various aspects of 12-step meetings and a more intensive discussion of the benefits, and potential concerns, of attending 12-step meetings. They were required to keep a record of self-help group meetings they attended and give brief descriptions of their personal reactions to and thoughts regarding the meeting. Counsellors also arranged for the participants to meet a self-help group volunteer who would accompany them to their first meeting. At 6 months’ follow-up, the intensive referral group showed greater attendance of and participation in self-help groups compared with those in the standard referral group. Furthermore, those in the intensive referral group showed greater reduction in alcohol and drug use and were more likely to be abstinent compared with those in the standard referral group.

Ouimette and colleagues (1998), in a secondary analysis of the study by Moos and colleagues (1999), showed that there was a synergistic effect between outpatient aftercare provision and 12-step self-help group participation following treatment. Service users who participated in both did better than those who only participated in one or the other. Those who did neither had the poorest outcomes. Once again, this study showed that increased frequency of attendance and increased involvement in 12-step activities enhanced outcomes.

8.6.6. Clinical summary

There have been several studies assessing the use of self-help groups for people who misuse drugs. The majority of studies have been conducted on 12-step programmes. There is limited but consistent evidence from these studies that 12-step attendance is associated with abstinence from illicit drugs and alcohol, and fewer drug and alcohol problems. Furthermore, involvement in such programmes can be improved by interventions from healthcare professionals to encourage regular attendance and active participation in such groups.

8.6.7. Clinical practice recommendations

8.6.7.1.

Staff should routinely provide people who misuse drugs with information about self-help groups. These groups should normally be based on 12-step principles; for example, Narcotics Anonymous and Cocaine Anonymous.

8.6.7.2.

If a person who misuses drugs has expressed an interest in attending a 12-step self-help group, staff should consider facilitating the person’s initial contact with the group, for example by making the appointment, arranging transport, accompanying him or her to the first session and dealing with any concerns.

8.7.1. Introduction

This section focuses on the evidence for the use of psychological interventions as part of broader packages of care, in particular case management. Case management is a strategy to improve the coordination of care for people who misuse drugs. It was devised for people with complex and multiple needs. An individual worker, the case manager, is responsible for the coordination and, where necessary, provision of care for service users. Contact with the case manager is usually expected to be on a regular ongoing basis. Case management originated in the mental health field and since the early 1980s it has been used in substance misuse services, mostly in the US but also in some European countries (in particular the Netherlands and Belgium).

In UK practice, case management has not been applied systematically in the same way as it has in the US and other European countries. The closest to case management in the UK is the care planning and care coordination approach, which have recently been the focus of much attention from the NTA and the subject of the recent Health Commission and NTA review of services, establishing these as important areas for development in UK services (NTA, 2006a). One of the conclusions of this review is that there is wide variation in procedures across the country.

8.7.2. Definitions of interventions

8.7.3. Databases searched and inclusion/exclusion criteria

Information about the databases searched and the inclusion/exclusion criteria used for this section of the guideline is in Table 51.

Table 51

Databases searched and inclusion/exclusion criteria for clinical effectiveness of case management.

8.7.4. Studies considered¹³

The review team conducted a new systematic search for RCTs that assessed the efficacy of case management (see Table 52). For trials of intensive referral versus standard referral, two RCTs (STRATHDEE2006; ZANIS1996) met the eligibility criteria, providing data on 286 participants. For trials of case management with ongoing contact versus standard care, eight RCTs (COVIELLO2006; MARTIN1993; MEJTA1997; MORGENSTERN2006; NEEDELS2005: Study 1; NEEDELS2005: Study 2; SALEH2002; SORENSEN2005) met the eligibility criteria providing data on 2,623 participants. All trials were published in peer-reviewed journals.

Table 52

Study information table for trials of case management for people who misuse drugs.

In addition, five studies were excluded from the analysis. The most common reason for exclusion was not providing required outcomes (further information about both included and excluded studies can be found in Appendix 14).

8.7.5. Case management

A summary of study information and evidence from the included trials is provided in Table 52 and Table 53. For further details of forest plots and full evidence profiles see Appendix 15 and 16.

Table 53

Summary evidence table for trials of case management for people who misuse drugs.

8.7.6. Clinical summary

One of the difficulties when interpreting this evidence is the variation in the sample populations, as well as what constitutes ‘case management’ in different studies.

Bearing in mind these sources of variation, overall, the evidence available consistently suggests that both intensive referral and case management, whether limited to a ‘brief’ care planning session, or initial care planning with ongoing contact, is effective at engaging service users in treatment at different stages of the treatment process. In terms of effects on illicit drug use, however, the evidence is mixed, with the overall suggestion of the meta-analysis that there is no improvement in outcomes compared with standard care.

While all the studies reviewed are US-based and hence interpretation should consider the cultural and health system differences already outlined, it should be noted that a remarkably similar picture is presented in mainstream mental health contexts in the UK and US, in that case management tends to improve treatment engagement but does not itself necessarily make a difference to outcomes (for example, for schizophrenia; NICE, 2002). The current evidence implies that for people who misuse drugs, effective, structured psychological interventions must be delivered in addition to standard care planning in order to achieve improved outcomes.

8.7.7. Clinical practice recommendation

8.7.7.1.

In order to reduce loss of contact when people who misuse drugs transfer between services, staff should ensure that there are clear and agreed plans to facilitate effective transfer.

8.8.1. Introduction

Multi-modal care programmes for the purpose of this review are defined as including a combination of therapy activities delivered in intensive schedules of 10 hours per week or more. Content of these programmes varies but would usually include education, daily living skills and other psychologically based interventions (for example, CBT, relapse prevention and reinforcement-based approaches), mostly delivered in group format. Such programmes are not common in generic drug treatment services in the UK, although they are available in some areas. They are more commonly used within drug services linked to the criminal justice system as a way of providing more intensive programmes for those referred. The current use of these interventions in the UK is limited and their distribution is not well understood.

8.8.2. Definitions of interventions

8.8.3. Databases searched and inclusion/exclusion criteria

Information about the databases searched and the inclusion/exclusion criteria used for this section of the guideline are in Table 54.

Table 54

Databases searched and inclusion/exclusion criteria for clinical effectiveness of multi-modal care programmes.

8.8.4. Studies considered¹⁴

The review team conducted a new systematic search for RCTs that assessed the efficacy of multi-modal care programmes (see Table 55).

Table 55

Study information table for trials of intensive outpatient treatment, day treatment and reinforcement-based therapy.

In the review of intensive outpatient treatment, four trials (COVIELLO2001; MCLELLAN1993; VOLPICELLI2000; WEINSTEIN1997) met the eligibility criteria providing data on 717 participants. All trials were published in peer-reviewed journals.

In the review of day treatment, two trials (AVANTS1999, MARLOWE2003) met the guideline eligibility criteria providing data on 370 participants. All trials were published in peer-reviewed journals.

In the review of intensive outpatient treatment with reinforcement-based therapy, three trials (JONES2005, SILVERMAN2001, SILVERMAN in press) met the eligibility criteria providing data on 282 participants. Two trials were published in peer-reviewed journals and one was in press. For further details on included studies see Appendix 14.

8.8.5. Multi-modal treatment programmes

Table 55 and Table 56 summarise study information and evidence for multi-modal treatment programmes. For further details see appendix 15 and 16 for forest plots and full evidence profiles.

Table 56

Summary evidence table for trials of intensive outpatient treatment, day treatment and reinforcement-based therapy.

8.8.6. Clinical summary

The evidence related to intensive outpatient treatments and day treatments (defined respectively as at least 9 and 20 hours of group work per week) does not support the notion that ‘more is better’ when comparing more intensive treatments to standard outpatient treatment in relation to drug-use outcomes. There is some evidence that reinforcement-based treatment can improve drug-use outcomes, although real-world application of this type of intervention may be limited. It is important to note, however, that some of the standard practice in the US appears to be better structured and more intensive than routine outpatient UK practice.

8.9.1. Introduction

People who misuse drugs often experience high rates of unemployment (Crowther et al., 2001). Crowther and colleagues (2001) argue that there are a number of social and clinical reasons for helping people with serious mental illness to work that are also applicable to people who misuse drugs. From a social standpoint, high unemployment rates are an index of the social exclusion of people who misuse drugs. From a clinical standpoint, employment may lead to improvements in the outcomes of people who misuse drugs through increasing self-esteem, alleviating psychiatric symptoms, and reducing dependency and relapse.

8.9.2. Definition of interventions

Pre-vocational training – Any approach to vocational rehabilitation in which participants are expected to undergo a period of preparation before being encouraged to seek competitive employment. This preparation could involve either work in a sheltered environment (such as a workshop or work unit), or some form of pre-employment training or transitional employment (Crowther, et al., 2001).

Supported employment – Any approach to vocational rehabilitation that attempts to place service users immediately in competitive employment. It is acceptable for supported employment to begin with a short period of preparation, but this has to be of less than one month’s duration and not involve work placement in a sheltered setting, or training, or transitional employment (Crowther et al. 2001).

8.9.3. Databases searched and inclusion/exclusion criteria

Information about the databases searched and the inclusion/exclusion criteria used for this section of the guideline are in Table 57.

Table 57

Databases searched and inclusion/exclusion criteria for effectiveness of vocational interventions.

8.9.4. Studies considered¹⁵

The review team conducted a new systematic search for RCTs that assessed the efficacy and/or safety of pre-vocational training, supported employment and enhanced vocational interventions.

For pre-vocational interventions two trials (HALL1977; ZANIS2001) met the eligibility criteria, providing data on 150 participants. Both trials were published in peer-reviewed journals.

For supported employment, no trials met the eligibility criteria.

In addition, six trials were excluded from the analysis. The most common reason for exclusion was that the intervention was not likely to be applicable to UK drug treatment services (further information about both included and excluded studies can be found in Appendix 14).

8.9.5. Vocational interventions

Study information and a summary of the evidence for included trials is provided in Table 58. For further details on forest plots and full evidence profiles see appendix 15 and 16.

Table 58

Study information and evidence summary table for vocational interventions.

8.9.6. Clinical summary

There was a lack of studies assessing vocational interventions for people who misuse drugs. The two included trials found some positive data suggesting pre-vocational training may improve the likelihood of being placed in a job for at least 1 day. Further research is required to replicate these findings and would benefit from more long-term employment outcomes.

8.10.1. Introduction

There is an increasing recognition that drug misuse affects the entire family and the communities in which these families live. For example, the Home Office’s updated Drug Strategy (Home Office, 2002) includes targets on increasing access to help, advice and counselling for parents, carers and families of people who misuse drugs. Additionally, the NTA user satisfaction survey found that 25% of respondents felt that staff did not offer families and carers enough support (Best et al., 2006). Therefore there is a need to assess whether interventions for carers are effective.

8.10.2. Definitions of interventions

8.10.3. Databases searched and inclusion/exclusion criteria

Information about the databases searched and the inclusion/exclusion criteria used for this section of the guideline is in Table 59.

Table 59

Databases searched and inclusion/exclusion criteria for clinical effectiveness of psychological interventions for carers.

The review team conducted a new systematic search for RCTs that assessed the efficacy and/or safety of community reinforcement and family training and 5-step for families/carers of people who misuse drugs (see Table 59).

For community reinforcement and family training, two trials (Kirby et al., 1999; Meyers et al., 2002) met the eligibility criteria, providing data on 152 participants. Both trials were published in peer-reviewed journals.

For the 5-Step intervention, one trial (Copello et al., 2007) met the eligibility criteria, providing data on 114 participants. This trial is in press.

In addition, two trials were excluded from the analysis because they did not have control groups.

8.10.4. Community reinforcement and family training

In both trials (Kirby et al., 1999; Meyers et al., 2002), community reinforcement and family training was compared with 12-step-based self-help groups (including 12-step facilitation) for carers.

The primary outcomes of these studies were to encourage people who misuse drugs and who had refused treatment into treatment, to reduce carers’ reported problems (social/emotional, relationship and health-related) and improve their psychological functioning (mood and social adjustment). Neither study found statistically significant differences between community reinforcement and family training and 12-step-based self-help groups in relation to carer problems and psychological functioning. Kirby and colleagues (1999) found statistically significant changes from baseline for both groups in relation to carer problems and psychological functioning. However, Meyers and colleagues (2002) found no statistically significant differences (after Bonferroni corrections for multiple testing) in changes from baseline at 12-month follow-up.

8.10.5. 5-step intervention

Copello and colleagues (2007) conducted a cluster-randomised trial (number of clusters = 137, number of participants = 143) comparing two intensities of a 5-step intervention. Primary care professionals were trained how to offer the 5-step intervention and asked to recruit and deliver the intervention to family members of people who misuse drugs and/or alcohol. All family members had experienced significant distress and lived with the person who misuses drugs or alcohol in the last 6 months. The majority of the sample were relatives of people who misuse alcohol; only 41.2% were relatives of people who misuse drugs. The largest proportions of family members included in the study were wives (43.1%) and children (35.3%).

Each primary care professional was treated as a cluster and was randomised to either the full intervention or guided self-help condition. The ‘full intervention’ consisted of up to five sessions, while guided self-help comprised one session where the primary care professional introduced the self-help manual (based on the 5-step model used in the full intervention) to the family member and encouraged him or her to work through it in his or her own time.

The two primary outcomes related to physical and psychological health (symptom rating test) and coping (the coping questionnaire). No statistically significant differences were found between the full intervention and the guided self-help conditions for both physical and psychological health (WMD = 0.23; 95% CI, −4.11 to 3.65) and coping (WMD = 0.12; 95% CI, −5.42 to 5.19).

8.10.6. Clinical summary

For both community reinforcement and family training and 5-step intervention, there were no statistically significant differences found between these more intensive interventions and self-help (that is, 12-step self-help groups and guided self-help). It appears that self-help interventions are as effective as more intensive psychological interventions in reducing stress and improving psychological functioning for carers and families of people who misuse drugs.

8.10.7. Clinical practice recommendations

8.10.7.1.

Where the needs of families and carers of people who misuse drugs have been identified, staff should:

• offer guided self-help, typically consisting of a single session with the provision of written material
• provide information about, and facilitate contact with, support groups, such as self-help groups specifically focused on addressing families’ and carers’ needs.

8.10.7.2.

Where the families of people who misuse drugs have not benefited, or are not likely to benefit, from guided self-help and/or support groups and continue to have significant problems, staff should consider offering individual family meetings. These should:

• provide information and education about drug misuse
• help to identify sources of stress related to drug misuse
• explore and promote effective coping behaviours
• normally consist of at least five weekly sessions.