A pilot randomised controlled trial of brief versus twice weekly versus standard supervised...

9
A pilot randomised controlled trial of brief versus twice weekly versus standard supervised consumption in patients on opiate maintenance treatmentRICHARD HOLLAND 1 , CATRIONA MATHESON 2 , GERALDINE ANTHONY 2 , KENNEDY ROBERTS 3 , SAKET PRIYARDARSHI 3 , ANN MACRAE 4 , EDDIE WHITELAW 3 , SIVARAM APPAVOO 5 & CHRISTINE BOND 2 1 Norwich Medical School, University of East Anglia, Norwich, UK, 2 Centre of Academic Primary Care, Aberdeen University, Foresterhill Health Centre, Aberdeen, UK, 3 Glasgow Addiction Service, Glasgow, UK, 4 NHS Highland Substance Misuse Service, Royal Northern Infirmary, Inverness, UK, and 5 Substance Misuse Service, Royal Cornhill Hospital, Aberdeen, UK Abstract Introduction and Aims. Methadone maintenance remains the mainstay of treatment for opiate dependence in Scotland. Guidelines recommend supervised self-administration for at least 3 months, yet this is often interpreted as long-term supervision. However, there is no evidence base for deciding the optimal period of supervision.We tested the feasibility of conducting a randomised controlled trial (RCT) of different supervision models. Design and Methods. Three armed pilot RCT, set in three Scottish treatment areas, recruited opiate-dependent patients who had received methadone treatment for 3 months. Participants were randomised to: (i) no supervision; (ii) twice weekly supervision; or (iii) daily supervision for further 3 months.As a pilot, key process measures were: recruitment rates, follow-up rates and treatment fidelity.We also wanted to estimate effect sizes of two co-primary outcomes for a full RCT: treatment retention and illicit heroin use.The recruitment target was 60 participants. Results. One hundred and two eligible patients were identified, 60 (59%) participated, and 46 followed up (77%). Study fidelity was good with two participants moving group.Those randomised to no supervision were significantly happier with their group allocation. No significant differences were found in primary outcomes, although retention decreased with increased supervision, while illicit heroin use was least in those most supervised. Discussion and Conclusions. It is possible to recruit and randomise participants to an RCT comparing different forms of supervised consumption. Pilot data suggest increased supervision may reduce illicit heroin use, but may decrease retention.This should now be tested in a large-scale multicentre RCT. [Holland R, Matheson C, Anthony G, Roberts K, Priyardarshi S, MacRae A, Whitelaw E, Appavoo S, Bond C. A pilot randomised controlled trial of brief versus twice weekly versus standard supervised consumption in patients on opiate maintenance treatment. Drug Alcohol Rev 2012;31:483–491] Key words: adherence, opiate maintenance treatment, RCT, supervised consumption. Introduction In Scotland over 22 000 patients are receiving metha- done as a substitute treatment for opiate dependence [1]. Almost 60% have their consumption supervised in a pharmacy [2]. Since 1999, guidance has promoted supervised consumption for 3 [3] to 6 months [4] to ensure patients prescribed these medications take them, and to prevent drug diversion [5]. Diversion may mean patients are under-treated and continue illicit heroin use; and diverted drugs put others at risk [6]. Equally, regular contact through supervision may enhance treatment, but implies a lack of trust and imposes a need to attend daily. Richard Holland BM, BCh, BA, DPH, DA, FFPH, PhD, Professor of Public Health Medicine, Catriona Matheson BSc, MSc, PhD, Senior Research Fellow, Geraldine Anthony RN, RM, BSc, MSc, Research Fellow, Kennedy Roberts BSc, MBChB, FRCGP, Retired Clinician, Saket Priyardarshi Bsc Hons, MBChB, MRCGP, MSc, Clinical Lead in Addictions, Ann MacRae MBChB, MRCPsych, Specialty Doctor, Eddie Whitelaw DMHN (Diploma Mental Health Nursing), Research Nurse, Sivaram Appavoo MBBS, MRCPsych, Specialty Doctor, Christine Bond PhD, MEd, BPharm, FRPharmS, FFPH, FHEA, Professor of Primary Care. Correspondence to Dr Richard Holland, Norwich Medical School, University of East Anglia, Norwich Research Park, Norwich, Norfolk, NR4 7TJ, UK. Tel: +44-1603-593574; Fax: +44-1603-593572; E-mail: [email protected] Received 2 December 2010; accepted for publication 19 October 2011. REVIEW Drug and Alcohol Review (June 2012), 31, 483–491 DOI: 10.1111/j.1465-3362.2011.00394.x © 2011 Australasian Professional Society on Alcohol and other Drugs

Transcript of A pilot randomised controlled trial of brief versus twice weekly versus standard supervised...

A pilot randomised controlled trial of brief versus twice weeklyversus standard supervised consumption in patients on opiatemaintenance treatmentdar_394 483..491

RICHARD HOLLAND1, CATRIONA MATHESON2, GERALDINE ANTHONY2,KENNEDY ROBERTS3, SAKET PRIYARDARSHI3, ANN MACRAE4, EDDIE WHITELAW3,SIVARAM APPAVOO5 & CHRISTINE BOND2

1Norwich Medical School, University of East Anglia, Norwich, UK, 2Centre of Academic Primary Care, AberdeenUniversity, Foresterhill Health Centre, Aberdeen, UK, 3Glasgow Addiction Service, Glasgow, UK, 4NHS HighlandSubstance Misuse Service, Royal Northern Infirmary, Inverness, UK, and 5Substance Misuse Service, Royal CornhillHospital, Aberdeen, UK

AbstractIntroduction and Aims. Methadone maintenance remains the mainstay of treatment for opiate dependence in Scotland.Guidelines recommend supervised self-administration for at least 3 months, yet this is often interpreted as long-term supervision.However, there is no evidence base for deciding the optimal period of supervision.We tested the feasibility of conducting arandomised controlled trial (RCT) of different supervision models. Design and Methods. Three armed pilot RCT, set inthree Scottish treatment areas, recruited opiate-dependent patients who had received methadone treatment for 3 months.Participants were randomised to: (i) no supervision; (ii) twice weekly supervision; or (iii) daily supervision for further3 months. As a pilot, key process measures were: recruitment rates, follow-up rates and treatment fidelity.We also wanted toestimate effect sizes of two co-primary outcomes for a full RCT: treatment retention and illicit heroin use.The recruitment targetwas 60 participants. Results. One hundred and two eligible patients were identified, 60 (59%) participated, and 46 followedup (77%). Study fidelity was good with two participants moving group.Those randomised to no supervision were significantlyhappier with their group allocation. No significant differences were found in primary outcomes, although retention decreasedwith increased supervision, while illicit heroin use was least in those most supervised. Discussion and Conclusions. It ispossible to recruit and randomise participants to an RCT comparing different forms of supervised consumption. Pilot datasuggest increased supervision may reduce illicit heroin use, but may decrease retention.This should now be tested in a large-scalemulticentre RCT. [Holland R, Matheson C, Anthony G, Roberts K, Priyardarshi S, MacRae A,Whitelaw E, AppavooS, Bond C. A pilot randomised controlled trial of brief versus twice weekly versus standard supervised consumptionin patients on opiate maintenance treatment. Drug Alcohol Rev 2012;31:483–491]

Key words: adherence, opiate maintenance treatment, RCT, supervised consumption.

Introduction

In Scotland over 22 000 patients are receiving metha-done as a substitute treatment for opiate dependence[1]. Almost 60% have their consumption supervised ina pharmacy [2]. Since 1999, guidance has promotedsupervised consumption for 3 [3] to 6 months [4] to

ensure patients prescribed these medications takethem, and to prevent drug diversion [5]. Diversion maymean patients are under-treated and continue illicitheroin use; and diverted drugs put others at risk [6].Equally, regular contact through supervision mayenhance treatment, but implies a lack of trust andimposes a need to attend daily.

Richard Holland BM, BCh, BA, DPH, DA, FFPH, PhD, Professor of Public Health Medicine, Catriona Matheson BSc, MSc, PhD, SeniorResearch Fellow, Geraldine Anthony RN, RM, BSc, MSc, Research Fellow, Kennedy Roberts BSc, MBChB, FRCGP, Retired Clinician, SaketPriyardarshi Bsc Hons, MBChB, MRCGP, MSc, Clinical Lead in Addictions, Ann MacRae MBChB, MRCPsych, Specialty Doctor, EddieWhitelaw DMHN (Diploma Mental Health Nursing), Research Nurse, Sivaram Appavoo MBBS, MRCPsych, Specialty Doctor, Christine BondPhD, MEd, BPharm, FRPharmS, FFPH, FHEA, Professor of Primary Care. Correspondence to Dr Richard Holland, Norwich Medical School,University of East Anglia, Norwich Research Park, Norwich, Norfolk, NR4 7TJ, UK. Tel: +44-1603-593574; Fax: +44-1603-593572; E-mail:[email protected]

Received 2 December 2010; accepted for publication 19 October 2011.

bs_bs_banner

R E V I E W

Drug and Alcohol Review (June 2012), 31, 483–491DOI: 10.1111/j.1465-3362.2011.00394.x

© 2011 Australasian Professional Society on Alcohol and other Drugs

Only two previous randomised controlled trials(RCTs) have investigated different models of supervi-sion in substance misuse. The first was a US factorialdesigned RCT (n = 150) which compared supervisedconsumption twice weekly versus five times weekly,and 80 mg versus 50 mg of methadone. The authorsfound significantly higher dropout rates in those visitingfive times weekly [7]. More recently, an AustralianRCT compared supervised and unsupervised con-sumption of buprenorphine-naloxone and found littledifference in retention at 3 months (57% vs. 61%,respectively) [8].

In the UK no RCTs have been conducted. Observa-tional studies have suggested drug-related death ratesare lower in areas with supervision [9], and that drug-related deaths [10] or deaths due to methadoneoverdose have fallen after its introduction [11].

UK practice varies, with many English units super-vising for approximately 3 months, while Scottish prac-tice generally involves prolonged supervision [12].Thisuses resources which could be used to provide othertreatments. Equally, prolonging supervision may disad-vantage patients from ‘normalising’ their lives. Thus,briefer supervision may not only be cheaper, but mayencourage clients to continue treatment and stabilisetheir lives.

This study sought to establish whether an RCTcould be conducted to test the effectiveness and effi-ciency of three different models of supervision after aninitial 3 months of daily supervision: no supervision butdaily pick-up; reduced frequency of supervision (2 daysper week) with daily pick-up on other days; and con-tinued daily supervision.

Methods

Subjects, setting and recruitment

The setting was community pharmacies in the ScottishHealth Board areas of Greater Glasgow and Highland.This was subsequently extended to Grampian becauseof slow recruitment. Participants were eligible if theyhad confirmed opioid dependence and had beenmanaged with supervised methadone for 3 months. A2-week ‘window’ before or after the 3-month point wasallowed as not all clients could be seen exactly at3 months.

Exclusion criteria

Participants were not eligible if they were: under16 years of age; pregnant or breastfeeding; incapable ofproviding informed consent; on methadone doses>120 mg; or had too severe a medical or psychiatriccondition in the opinion of the treating physician.

Randomisation

Eligible participants who agreed to participate wererandomised using an automated telephone randomisa-tion system, stratified by site, to one of three arms: nosupervision but daily pick-up; reduced frequency ofsupervision (2 days per week) with daily pick-up onother days; and continued daily supervision.

Interventions

The intervention period was for 3 months. All clientswere dispensed methadone by a community pharmacy.Those randomised to ‘no supervision’ discontinuedsupervision on entry to the study (i.e. after completing3 months of supervised consumption), but collectedtheir medication daily. Those randomised to ‘twiceweekly supervision’ were dispensed daily, but super-vised on only 2 days per week. Those allocated toreceive daily supervised consumption attended theirpharmacy where their pharmacist dispensed theirmedication to be swallowed under their supervision.Daily meant 6 days per week (occasional clientsattended 5 or 7 days per week) depending on usualpharmacy opening hours.

Participants and researchers were not ‘blinded’ to thegroup allocation.

Outcome measures and length of follow up

As a pilot study we were particularly interested inrecruitment rates, follow-up rates, and fidelity to groupallocation after randomisation, due to either personaldissatisfaction with their treatment, or clinician prefer-ence. This ‘treatment fidelity’ was measured by theproportion randomised remaining in their allocatedtreatment arms at 6 months.

In order to assess potential effect sizes for our iden-tified outcomes in a future definitive study, data werecollected on participants at baseline (just prior to ran-domisation) and at 3 months post randomisation (i.e.until participants had been treated for 6 months). Wewere interested in two co-primary outcome measures:retention in treatment and illicit heroin use (self-reported, using the Maudsley Addiction Profile (MAP)[13]) at follow up. A participant was considered as ‘notretained’ where a gap in prescribing of over 5 daysoccurred.

Secondary outcomes included: urine drug results(collected from patient notes), other illicit drug use(self-report using MAP); alcohol use measured by thebrief Alcohol Use Disorders Identification Test (briefAUDIT); addiction severity and social functioningmeasured by the MAP; changes in psychological func-tioning (MAP); changes in quality of life (Short Form-

484 R. Holland et al.

© 2011 Australasian Professional Society on Alcohol and other Drugs

12, SF-12) [14]; changes in criminal behaviour (MAP);treatment satisfaction (Treatment Satisfaction Ques-tionnaire); and adverse events (e.g. drug overdose/death). Additionally, we asked participants, at followup, ‘are you happy with the group you were allocatedto?’ and explored issues around drug diversion. Thelatter involved participants completing two questions:accessibility of street methadone (4-point Likert scalefrom impossible to very easy); and their view of theprice change of methadone in the last month (3-pointscale: gone up, no change or gone down). The MAP,brief AUDIT and SF-12 have all been previously vali-dated, and extensively used in previous studies.

Analysis and sample size

Data were entered into spss Version 15.0 (SPSS Inc.,Chicago, IL, USA) and a 10% random check conductedfor accuracy. Associations between group and categori-cal variables were assessed using the c2-test (or Fisher’sexact test for small expected frequencies). Descriptivestatistics were calculated to inform the main processmeasures of recruitment, follow-up rates, group fidelityand acceptability of randomisation. Changes betweenbaseline and follow up across groups were assessed usingthe Kruskal–Wallis test as the distributions of mostoutcomes were skewed. Unless specified, a per protocolanalysis was performed as one of the main focuses was toexamine rates of study completion.An intention-to-treat(ITT) analysis was also undertaken for the two primaryoutcome measures (treatment retention and illicitheroin use). Responses to open questions were analysedusing a thematic analysis where comments were groupedinto key emerging themes.

This study was not powered to detect a clinicallymeaningful difference between groups. Instead, weaimed to recruit 60 participants in order to assessrecruitment strategies, study fidelity and potentialeffect sizes to inform a future sample size calculation fora definitive study.

Results

A total of 102 potential participants were approached,of whom 60 (59%) entered our study. Reasons for notparticipating included: 18 failed to attend their baselineinterview (18.6%); 11 declined (10.8%); seven (6.8%)were clinically unsuitable (severe alcohol use or othermental health issues); and six (5.9%) were unsuitablebecause of work commitments, moving area or enteringrehabilitation. Recruitment proved slow to begin with,but improved once extended to two additional siteswithin Glasgow, and to Grampian.

Table 1 describes participants’ baseline characteris-tics collected on all participants, and demonstrates that

the groups were reasonably equivalent, although thesmall numbers randomised (n = 60) did lead to somebetween-group differences. Those in the twice weeklygroup were more likely to be female and to be youngerin age. Those supervised were more likely to be unem-ployed. Baseline drug use appeared worst in those ran-domised to be unsupervised according to self-report,although urine results indicated no difference. Physicalhealth measured by MAP appeared worst in the super-vised group, although physical functioning (SF-12) didnot differ between groups. Psychological health mea-sured by MAP did not appear to differ between groups,but SF-12 data suggested the supervised group hadpoorer psychological health at baseline.

Study fidelity and follow up

Only two participants were moved from their allocatedgroup (3.3%, 95% confidence interval 0–11.5%). Bothhad been randomised to ‘no supervision’ and were sub-sequently supervised at the request of their clinicianbecause of concerns about child safety (n = 1) andcontinued chaotic behaviour (n = 1). Both had thesame prescriber. None of the other eight prescribersrequested changes to participants’ group allocation.

In total, while 10 participants were considered notretained in treatment at 6 months, 14 participants(23%) failed to complete follow-up questionnaires: twowere moved group because of concerns about childsafety; four entered prison; and eight did not completefollow-up questionnaire (see Figure 1). The remaining46 participants completed full follow-up questionnaires.

Primary outcomes

Illicit heroin use. Table 2 shows that while heroin useappeared to decrease in both the fully and twice weeklysupervised groups, from a median of 1 day to no use,those unsupervised did not appear to decrease use,although their baseline use was higher (a median of2 days use/month). These differences were not statisti-cally significant whether analysed using a per protocolapproach, or using an ITT approach.

Retention in treatment. Retention in treatmentappeared best in those who were unsupervised (17/19,89%), and worst in those supervised (15/20, 75%), withtwice weekly supervision appearing to yield an interme-diate effect (18/21, 86%) (Table 3).This was not statis-tically significant, although there was some suggestion ofa dose–response effect with more supervision worseningretention (P = 0.47 ITT, Fisher’s exact test).This trendwas stronger by per protocol analysis where retention inthose unsupervised was 94% (16/17), and again lowest

Supervised consumption in opiate treatment 485

© 2011 Australasian Professional Society on Alcohol and other Drugs

in those supervised (73%, 16/22), but remained notstatistically significant (P = 0.21).

Secondary outcomes

Results demonstrated that there were no significantdifferences between groups in the number of days anyillicit substance had been used (MAP), physical health,social functioning (days of contact, and conflicts withpartners, family and friends), crime involvement,quality of life (SF-12), or treatment satisfaction. Urinedrug results also demonstrated no between-group dif-ferences at follow up, although all groups showedreductions in use of illicit drugs, in particular opiates,with a statistically significant (P = 0.046) within-groupdecrease observed for the supervised group (data avail-able from the authors).

Alcohol use appeared low in this sample, with 20non-drinkers recruited. The proportion identified withproblem drinking (brief AUDIT score >4) reduced inthe supervised group (from 47% to 33%), while itincreased or remained the same in the other two groups(between-group P = 0.05). In contrast, psychological

health appeared to improve most in the twice weeklysupervised group (reducing from a score of 14 to ascore of 6), while staying the same or worsening in theother two groups (P = 0.03). No adverse events wererecorded in any group.

Participants responded to questions about illicitmethadone availability without problems. Resultsimplied that there was no change in methadone avail-ability or price between baseline and follow up. At bothtimes methadone was considered quite or very easy toaccess by over 50% of participants. Prices differedbetween Glasgow and the other sites. In Glasgow allparticipants stated a price of £10 per 100 mL; else-where prices were £15–20 per 100 mL.

Participants were asked about their views of groupallocation. Marked differences between groups wereevident with two-thirds of those in the unsupervisedgroup happy about their group allocation, but only 30%of those in the twice weekly group and 14% of those inthe supervised group happy (P = 0.004). When askedfor any comments about their allocated group, a strongtheme from unsupervised participants was of reducedstigma when not needing to take their medication in a

Table 1. Baseline comparison of groups as allocated

GroupUnsupervised

(n = 19) 2¥ weekly supervised (n = 21)Supervised

(n = 20)

Male, n (%) 14 (73.7) 12 (57.1) 16 (80)Median age (IQR) 35 (29, 40) 31 (29, 43) 35 (30, 40)Unemployed, n (%) 10 (52.6) 14 (66.7) 17 (85.0)Location

Glasgow, n (%) 11 (57.9) 10 (47.6) 12 (60)Grampian, n (%) 7 (36.87) 11 (52.4) 6 (30)Highland, n (%) 1 (5.3) — 2 (10)

Urine toxicology Negative urines = 7 Negative urines = 13 Negative urines = 9Positive urines = 12a Positive urines = 8a Positive urines = 11a

• Opiates = 9 • Opiates = 8 • Opiates = 8• Codeine = 1 • Codeine = 0 • Codeine = 0• Benzosb = 7 • Benzos = 3 • Benzos = 5• Cannabis = 3 • Cannabis = 0 • Cannabis = 1• Cocaine = 1 • Cocaine = 0 • Cocaine = 0

MAPc, median (IQR)Median no. days anyc substance used 12.5 (2.8, 31.8) 6 (2, 22) 8 (4, 30)Psychological health scored 11.5 (7, 17.3) 14 (7.5, 17) 13 (9, 21)Physical health scored 10.5 (3.8, 15) 8 (2.5, 27) 20 (16, 25)% time in conflict with partner 50 (0, 50) 10 (5, 45.8) 6.7 (0, 26.7)% time in conflict with friends 0 (0, 0) 0 (0, 13.3) 0 (0, 0)

Brief AUDITScore (IQR) 4 (1, 9.3) 4 (0, 6) 3 (0, 8)% problem drinkinge 42.9 47.1 46.7

SF-12, median (IQR)Physical functioningf 48.5 (39.5, 55.1) 53.2 (46.4, 55.9) 45.8 (32.8, 56.7)Psychological functioningf 47.7 (35.9, 55.3) 48.4 (30.8, 51.4) 33.4 (29.5, 40.7)

Treatment satisfaction score, median (IQR) 12 (9.8, 15.3) 10 (6.5, 13) 13 (7.8, 16)

aParticipants tested positive to one or more substance in urinalysis. bBenzos = benzodiazepines. cMAP = Maudsley AddictionProfile. dHigh scores = worse health state for MAP. eAUDIT >4. fHigh scores = better health state for SF-12.

486 R. Holland et al.

© 2011 Australasian Professional Society on Alcohol and other Drugs

pharmacy. The reverse was true for those supervised(Table 4).

Discussion

This study demonstrated that it is possible to recruitto a study comparing three different approaches to

supervision. Only 10% of potential participantsrefused to enter the study. Equally, clinicians used totight supervision requirements proved willing to relaxtheir practice. Indeed, only 7% of clients were deemedunsuitable for the study by clinicians. Furthermore,only one of nine clinicians changed clients (n = 2,3.3%) from unsupervised to supervised (our measure

Registered/eligible participants(n = 102)

Refused/unsuitable to participate (n = 42)Declined = 11Failed to attend baseline interview = 14Failed to engage in treatment = 4Clinician deemed unsuitable = 7Other reason = 6

Randomisation(n = 60)

(59% eligible participants)

Received intervention as allocated (n = 17)Subsequently moved group to be supervised (n = 2)

Lost to follow up (n = 3)Prison (n =1)Unknown (n = 2)*

Lost to follow up(n = 4)Prison (n = 1)Unknown (n = 3)

Analysed (n = 14) Analysed (n = 17)

Unsupervised (n = 19) Supervised

(n = 20)

Lost to follow up (n = 5)Prison (n = 2)Unknown (n = 3)

Received intervention as allocated (n = 21)Withdrawn from study(n = 0)

Received intervention as allocated(n = 20)Withdrawn from study(n = 0)

Analysed (n = 15)

Supervised twice weekly (n = 21)

Figure 1. CONSORT flow diagram. Flow chart describing progress of participants through the pilot methadone supervision trial.

Supervised consumption in opiate treatment 487

© 2011 Australasian Professional Society on Alcohol and other Drugs

of treatment fidelity). In both cases, the reasons wereappropriate, but potentially meant they should nothave entered the study. However, recruitment wasslower than we had anticipated. Thus, a future, defini-tive RCT would need to plan to recruit across a widerange of drug treatment sites and to ensure researchstaff were able to recruit participants throughout theweek, in a way that was impossible in this small pilotstudy.

This small study was not statistically powered toestimate the effectiveness of different forms of supervi-sion. Nevertheless, our findings are of interest as resultssuggest that increasing levels of supervision decreasedretention (a negative finding), while reducing illicitheroin and alcohol use (a positive finding).This was notstatistically significant (with the exception of thealcohol result) and should be treated with caution.However, it tentatively suggests that there is a dilemmain drug treatment between a harm minimisationapproach favouring retention, which may be improvedby relaxing supervision, and a recovery approach,where eliminating heroin use is the key objective, whichneeds tighter supervision. A further complexity isadded when risk to others of ‘leaked’ methadone isconsidered, particularly in light of strong observationalevidence demonstrating reduced methadone overdose

deaths after introduction of supervised consumption[9–11].

In many drug treatment centres treatment protocolsmay be ‘fluid’ allowing clinicians to take responsibilityfor the decision as to whether or not to maintain aclient on supervised treatment. Nevertheless, ourrecent national survey within Scotland has demon-strated marked variety in supervision practice [15].This seems to vary more between centres than withincentres as ‘standard practice’ will occur. Thus, it isimportant to determine underlying principles of goodsupervision practice, which are indeed evidence-based. Such principles require data from a larger,better powered RCT. Such an RCT could test longerversus shorter fixed period supervision, as well as amore flexible client-centred approach, where only aminimum period is defined.

Our findings were supported by participants’ views oftheir randomisation groups.While all clients either werehappy or did not mind being unsupervised, over 40% ofthose supervised did not like this. It is not surprising thatclients would rather take their methadone in private[16,17]; indeed it was clear that stigmatisation in phar-macies is still a major problem. Despite considerabletraining of pharmacy staff and resources in the form ofprivate areas/consulting rooms, this issue needs further

Table 2. Illicit heroin use (last 30 days)

Group

Unsupervised 2¥ weekly supervised Supervised

Between-groupP-value

Median days heroinused in last 30 days

(IQR)

Median days heroinused in last 30 days

(IQR)

Median days heroinused in last 30 days

(IQR)

Per protocol analysis Baseline 2 (0, 13.5) 1 (0, 2) 1 (0, 3) 0.563Follow up 2.5 (0, 13.8) 0 (0, 2) 0 (0, 7) 0.945

Intention-to-treatanalysis

Baseline 1 (0, 13) 1 (0, 2) 0 (0, 2) 0.288Follow up 2 (0, 13) 0 (0, 2) 0 (0, 2) 0.933

Table 3. Retention in treatment

Group (intention-to-treat analysis)Unsupervised 2¥ weekly supervised Supervised Between-group

P-value(n = 19) (n = 21) (n = 20)

Retained in treatment at end of follow up, n (%) 17 (89.5) 18 (85.7) 15 (75.0) 0.449

Group (per protocol analysis)Unsupervised 2¥ weekly supervised Supervised Between-group

P-value(n = 17) (n = 21) (n = 22)

Retained in treatment at end of follow up, n (%) 16 (94.1) 18 (85.7) 16 (72.7) 0.210a

aFisher’s exact test.

488 R. Holland et al.

© 2011 Australasian Professional Society on Alcohol and other Drugs

review. However, some degree of feeling different toother customers is inevitable, given the nature of havingto take their medication on the premises. To deal withclients who feel stigmatised within their pharmacy to the

extent that it affects their treatment, it may be appropri-ate for some to be dispensed long-term from their drugtreatment centre, or to be encouraged to use a pharmacymore distant from their home.

Table 4. Positive and negative comments from participants

What were the positive (good) things about the group you were randomised to?

Unsupervised Supervised twice weekly Supervised daily

It was just great, got take home & thisis the first time I think I have madegood progress in cutting down, I wascareful about storing it as well, abreak from being eyeballed to deathat chemist

I liked not having to be watchedswallowing my meth at thepharmacy everyday. Two days ismore than enough.

Helped me to cope better. it makes mecome to clinic and dont mind thedaily pick up.

Relaxed time to go to chemist, nothaving to queue. other customersdid not know why I was there.

Days on which unsupervised felt lessstigmatised

Keep stability in dealing withpharmacy daily

Not having to wait in the chemistevery day, felt like a normalcustomer

No time restraints about attending.Other customers did not knowattending for methadone

No strict time to take methadone,helped as working full-time. Othercustomers did not know why he wasin chemist

Was good but times got changed as myurine was dirty & I missed someappts

Took a while to get used to but it wasgreat. Taking it at night was great.Liked the flexibility. Made Sure Ilocked it in the cupboard away fromthe children (aged 2, 5, and 7 yearsold).

Less trips to the pharmacy

What were the negative (bad) things about the group you were randomised to?

Unsupervised Supervised twice weekly Supervised daily

Because too relaxed & felt lack ofstructure led to more heroin use.

being stared at by people in thepharmacy and its a pain now as Igot a job in a garage and have tokeep asking to come in later orsneak away at lunch time to get mymeth.

Too close- near neighbours house.Difficult to be discreet

Risk of getting dropped/smashed whatwould you do? or loose lid?

Still having to go to a chemist everyday

Bumping into people at chemist

Still tied to the chemist. Bottle openedeasily and often spilled in mypocket. The chemist would notreplace it even when I showed themin my coat. need to use better screwcaps bottles.

Stigma & feels addiction teams toblame, as only a prescription ‘writingservice’

Having to stand in a chemist everyday, people (normal customers)know you were in for methadone

Temptation of selling methadone if Imskint. The just supervision al thetime has no trust but you cant reallyfail. it says a lot if you do want tostop it you get it unsupervised. if iwas complete

Hates sitting in the pharmacy-sweating often 20 mins waiting.Limited with daily pick up. Feelslabeled as a drug user- and veryaware people watching and staringjust hate it as trying to get better.

Having to attend chemist every day Having to stand in chemist every day& being easily identified as ‘a junkie’

Supervised consumption in opiate treatment 489

© 2011 Australasian Professional Society on Alcohol and other Drugs

Strengths and weaknesses

This study had a small sample size and was notdesigned to detect a difference in primary outcomesbetween the groups. However, randomisation did notyield exactly equivalent groups with unemploymentbeing most common in the supervised group, and moresubstance use occurring in the unsupervised group atbaseline. Equally, given the form of the interventionthere was no possibility of blinding participants.However, a future larger study could blind researchersto the participants’ allocation when collecting follow-updata. In this study, limited resources meant this was notpossible, although bias was unlikely for our measure ofretention (one of our two primary outcomes), or forurine data.

Study retention was reasonable for a substancemisuse study with 77% followed up over the 3-monthtimescale. A larger RCT should have a longer follow-uptime of at least 6 to 12 months. Such a study shouldplan for the retention rate to drop further, but havesufficient resources to include more intensive follow-upmethods, such as obtaining telephone numbers of sig-nificant others of study participants.

Methodological aspects

This pilot has demonstrated the importance of under-taking such a pilot before carrying out a full RCT. Notonly did the pilot assist with ensuring the quantitativemethods were appropriate (e.g. the outcome measuresand recruitment techniques), but by including a morequalitative component in this study we are better able totailor a future study accordingly. The qualitative com-ments demonstrated that relaxing supervision did havethe effect intended with many positive comments fromthe participants. Equally, shortly after our studystarted, clinicians commented that twice weekly super-vision with daily pick-up did not appear a realistic alter-native treatment. Interestingly, this did not appear to bethe case for the clients in that group who appreciatedany reduction in the number of times that their doseswere supervised (see Table 4). Thus, we wouldcommend to others a need to include formal, or infor-mal qualitative data collection, within pilot and fullstudies, in an effort to better determine the meaning ofinterventions for those participating.

However, the difference in the supervision modelsthat we tested was reasonably modest as all interven-tions included daily pick-up of methadone. Inevitably,this may have had implications in terms of the effectsize seen. Future studies should consider both issues ofsupervision intensity, and dispensing frequency, as bothof these interventions are likely to be important toclients and may have synergistic effects.

Methadone leakage

It is important to note that supervision of methadone hastwo potential levels of effect. This study explored theeffects at an individual level. However, imposition ofsupervision was initiated in the late 1990s also for itscommunity level effect; reducing drug diversion andwider access to methadone.This community level effectis much harder to capture by an RCT such as this. In thispilot, we attempted to gather data on the availability andestimated price of ‘street methadone’ as a crude measureof methadone leakage. Inevitably, this is a difficult aspectto study accurately. Our small study, with only 40 par-ticipants on reduced, or no supervision, across a widearea was unlikely to change illicit methadone availability.Nevertheless, it was considered useful to assess if ourquestions could be used in future. Completion of thesequestions was good, suggesting a large trial could usethis as a future crude measure of leakage.

Clearly, any future large-scale reduction in supervi-sion could potentially increase the availability of illicitmethadone. Interestingly, our results indicate that theprice of methadone in Glasgow is approximately halfthat in Grampian or Highland, despite Glasgow’s highsupervision rates. Whether this reflects the largernumbers in treatment in Glasgow and therefore greateravailability is unclear.We would suggest that any futurestudy of supervision frequency collect data in all treat-ment centres included, asking all new clients (irrespec-tive of entry into the study) their experience of access toillicit methadone. These data should be collected pro-spectively in advance of the trial starting, during thetrial, and after the trial has ended, in an effort tomeasure the potential community-level impact of achange in supervision practice.

Conclusions

This study demonstrated that it is possible to recruitand retain a high proportion of suitable participants toa study investigating different approaches to supervisedmethadone consumption. Treatment fidelity was good.In addition, the outcome measures appeared sensitiveenough to pick up significant differences betweengroups.

Our study gave an interesting indication of a poten-tial conflict between client retention and clinicalprogress.Those randomised to the unsupervised groupwere significantly more likely to be happy with theirrandomisation group, and appeared more likely to beretained in treatment. In contrast, those retained in thesupervised group appeared to be using less illicitopiates. This suggests that a larger, multicentre RCTshould be funded to clarify the relationship betweenduration of supervision, retention in treatment and use

490 R. Holland et al.

© 2011 Australasian Professional Society on Alcohol and other Drugs

of illicit heroin in substitution treatment. This wouldprovide much needed evidence as to the most appro-priate supervision policy to promote and fund.

Acknowledgements

Thanks to the Chief Scientist Office for funding thisproject and all the patients who took part.

Ethical approval

Ethical approval for the study was provided by theNorth East Scotland Research Ethics Committee.

References

[1] Scottish Government. Review of methadone in drug treat-ment: prescribing information and practice. Edinburgh:The Scottish Government, 2007.

[2] Matheson C, Bond CM, Tinelli M. Community pharmacyharm reduction services for drug misusers: national servicedelivery and professional attitude development over adecade in Scotland. J Public Health (Oxf) 2007;29:350–7.

[3] Independent Expert Working Group. Clinical guidelines ondrug misuse and dependence update 2007: consultationdraft June 2007. London: National Treatment Agency,2007.

[4] Advisory Council on the Misuse of Drugs. Reducing drugrelated deaths: a report of the Advisory Council on theMisuse of Drugs. London: The Stationery Office, 2000.

[5] Departments of Health. Drug misuse and dependence:guidelines on clinical management. London: The Statio-nery Office, 1999.

[6] Green H, James RA, Gilbert JD, Harpas P, Byard RW.Methadone maintenance programs—a two-edged sword?[Review] [20 refs]. Am J Forensic Med Pathol 2000;21:359–61.

[7] Rhoades HM, Creson D, Elk R, Schmitz J, Grabowski J.Retention, HIV risk, and illicit drug use during treatment:methadone dose and visit frequency. Am J Public Health1998;88:34–9.

[8] Bell J, Shanahan M, Mutch C, et al. A randomized trial ofeffectiveness and cost-effectiveness of observed versusunobserved administration of buprenorphine-naloxone forheroin dependence. Addiction 2007;102:1899–907.

[9] Weinrich M, Stuart M. Provision of methadone treatmentin primary care medical practices: review of the Scottishexperience and implications for US policy. [see comment].JAMA 2000;283:1343–8.

[10] Seymour A, Black M, Jay J, Cooper G, Weir C, Oliver J. Therole of methadone in drug-related deaths in the west ofScotland. Addiction 2003;98:995–1002.

[11] Strang J, Hall W, Hickman M, Bird SM. Impact of super-vision of methadone consumption on deaths related tomethadone overdose (1993–2008): analyses using OD4index in England and Scotland. BMJ 2010;341:c4851.

[12] Cameron I, Naji S, Lind C, McGeganey N. A survey ofNHS Services for opiate dependents in Scotland. Report tothe Effective Interventions Unit. Edinburgh: ScottishExecutive, 2001.

[13] Marsden J, Gossop M, Stewart D, et al. The MaudsleyAddiction Profile (MAP): a brief instrument for assessingtreatment outcome. Addiction 1998;93:1857–67.

[14] Ware J Jr, Kosinski M, Keller SD. A 12-Item Short-FormHealth Survey: construction of scales and preliminary testsof reliability and validity. Med Care 1996;34:220–33.

[15] Anthony G, Matheson C, Holland R, et al. Methadone pre-scribing under supervised consumption on premises: AScottish clinician’s perspective on prescribing practice.Drug Alcohol Rev 2011; Online ahead of print. DOI:10.1111/j.1465-3362.2011.00316.x.

[16] Matheson C. Views of illicit drug users on their treatmentand behaviour in Scottish community pharmacies: implica-tions for the harm-reduction strategy. Health Educ J1998;57:31–41.

[17] Matheson C. Privacy and stigma in the pharmacy: illicitdrug users’ perspectives and implications for pharmacypractice. PJ 1998;260:639–41.

Supervised consumption in opiate treatment 491

© 2011 Australasian Professional Society on Alcohol and other Drugs