Methadone‐Assisted Opiate Withdrawal and Subsequent Heroin Abstinence: The Importance of Psychological Preparedness

Steve Jones, PhD , Barbara Jack, PhD, Julie Kirby, PhD, Thomas L. Wilson, PhD , Philip N. Murphy, PhD 5 Department of Psychology, Faculty of Health, Social Care, and Medicine, Edge Hill University, Ormskirk, UK Evidence‐Based Practice Research Centre, Faculty of Health, Social Care, and Medicine, Edge Hill University, Ormskirk, UK Faculty of Health, Social Care, and Medicine, Edge Hill University, Ormskirk, UK School of Applied Social Sciences, De Montford University, Leicester, UK Department of Psychology, Edge Hill University, Ormskirk, UK


INTRODUCTION
Interventions targeting opiate abstinence pose challenges to patients arising from withdrawal symptoms, 1 followed by longer-term challenges around establishing an opiate-free life. The latter may include craving to use opiates again, 2 pressure to relapse from opiate misusing acquaintances, 3 and the need to replace opiate misuse as a coping mechanism for problems in living. 4 Opiate abstinence treatments occur in a variety of settings including hospital wards, outpatient clinics, and prisons. These may differ regarding the extent of psychosocial support available, with potential consequences for patients' affective states and commitment to treatment objectives. UK treatment guidelines emphasize the importance of patients' preparedness for opiate withdrawal, and their active engagement in treatment decisions, particularly regarding the discontinuation of opiate substitution treatment (OST) in favor of a withdrawal regime. 5,6 However, pressure from governmental sources encourages abstinence-orientated treatments wherever possible. Concerns have consequently arisen regarding adverse outcomes due to patients being channeled toward abstinence without sufficient preparation and commitment to this step, particularly where the withdrawal stage will occur in prison. 7 In other governmental jurisdictions, including the United States, OST is not generally available in prisons despite evidence of its effectiveness. Opiate withdrawal is, therefore, enforced regardless of patient preparedness, with a consequent decreased likelihood of future treatment engagement and an increased likelihood of further dangerous opiate misuse. 8,9 Self-efficacy is an element of psychological preparedness for addiction treatment, which has been examined in relation to a variety of substances across treatment contexts. [10][11][12] Self-efficacy is generally understood to refer to a person's belief in their ability to achieve specific goals. 13 For addiction treatment, this implies attaining treatment objectives including, where appropriate, abstinence from a drug of addiction. Where OST has taken the form of methadone maintenance there is evidence that self-efficacy is positively related to achieving treatment goals such as enhanced life stability and avoiding illegal drugs. [14][15][16][17] For abstinence-orientated treatment, there are few studies examining the predictive value of self-efficacy at treatment admission for subsequent outcomes. Psychometric measures of admission self-efficacy for outpatient methadone withdrawal have been shown to remain stable during the stabilization stage of medication, but to diminish progressively with tapering methadone doses. 18 Illicit opiate use in this sample increased during the methadone taper. Another study reported that higher admission self-efficacy for a brief inpatient methadone detoxification predicted a greater likelihood of refusing post-detoxification medication to maintain abstinence. 19 However, the success rate for maintaining abstinence was not reported, and this study measured self-efficacy using an ad hoc scale comprising one question. Confidence in one's ability to achieve treatment goals has sometimes been reported as being synonymous with self-efficacy. Visual analogue scale admission measures of confidence for successful treatment completion have positively predicted heroin abstinence 3 months later, but not inpatient methadone detoxification completion. 20,21 Overall, the lack of consistent use of psychometrically validated measures of self-efficacy in studies of opiate withdrawal and abstinence weakens the conclusions possible regarding its role as an outcome predictor for such treatment.
The present multisite study compared patients' selfefficacy scores from a validated psychometric instrument at admission for methadone withdrawal (T1) in three treatment settings. The predictive ability of these scores was subsequently tested at 6-month follow-up (T2) for treatment outcomes, and also for maintenance of contact with T1 treatment services. Predictive ability for the latter outcome was tested due to the risk of contact being lost with patients following OST termination, and the negative consequences which can follow. 8,9 The treatment settings were a prison, a hospital inpatient unit, and a communitybased clinic. Self-efficacy does not exist in a psychological vacuum, and two other aspects of preparedness for treatment examined were mood state, and attributions for treatment outcomes. It may be noted that depression has been associated with on-going opiate misuse, 19,22 while attributions emphasizing taking responsibility for one's treatment progress have been related to favorable outcomes. 23,24 However, such attributions are frequently measured retrospectively, rather than being potential predictors as in this present study. Finally, as variables concerning criminal activity and judicial consequences can impact upon heroin users differently, [25][26][27] the relationship of such variables to measures of psychological preparedness was also examined.

METHOD Design and Participants
Treatment setting at T1 was a between-participant independent variable comprising three patient/participant groups recruited from facilities in the north-west of England. The prison group (PG) was recruited from HMP Liverpool, the inpatient group (IG) from a National Health Service (NHS) facility, and the community-based treatment group (CG) from an NHS drug dependency clinic. As access to incarcerated patients was only possible in a male prison, only male participants were recruited in the other treatment settings to avoid gender becoming a confounding variable across settings. All patients were within the first week of a methadone programme orientated toward withdrawal at T1. Recruitment to the study was based upon successive admissions, subject to willingness to participate, and interviewer availability. The total sample for the study at T1 was N = 72, with n = 24 for each group. The GPower software programme 28 indicated that n = 24 would provide power at 0.85 for large effects (ie, f = 0.40), for intergroup comparisons comprising three groups with a conventional alpha level of P ≤ .05. Additionally, the total sample of N = 72 would be sufficient for power at 0.85 for two-tailed correlations with an effect size of r = .35 and a conventional α level of P ≤ .05. Decisions concerning regression analyses using T2 variables as outcomes were made on a post hoc basis, subject to the sample size available.

Treatment Protocols
All three treatment settings conformed to UK guidelines for treating drug misuse 5,6 regarding OST, and the need for patient readiness for opiate withdrawal and long-term abstinence. Methadone was prescribed in response to the needs of the individual in all treatment settings, with dosages being dependent upon the severity of withdrawal symptoms. Starting doses were typically in the range of 30 to 50 mg of oral methadone daily. There was a target reduction rate of 2 mg per week for the PG, while the target completion time for the IG was between 3 and 4 weeks. The variability in the reduction rate for the CG, in response to patient need and distress prevents the reporting of an average or notional rate.

Psychological Preparedness
Psychological preparedness for treatment at T1 was assessed by the following instruments.
Stages of Change Readiness and Treatment Eagerness Scale (SOCRATES 8D). This self-completion instrument yields scores on three scales representing, respectively, recognition of a drug problem, ambivalence regarding drug use, and taking steps regarding drug use. 29 Higher scores indicated higher levels of these properties.
Drug-Taking Confidence Questionnaire (DTCQ-8). The DTCQ-8 30 is a self-completion measure of confidence for avoiding drug use across eight situations. Total scores pooled across all situations were used, with higher scores indicating higher self-efficacy.
Confidence for Treatment Scale (CTS). The CTS 21 gave patients the following question and instruction: "How confident are you that you will be drug-free following your drug treatment. Place a cross on the line below." The scale comprised a horizontal line 100 mm in length, with the left pole labeled "No confidence," and the right pole labeled "extremely confident." Scores were recorded in millimeters measured from the left pole. Higher scores indicated higher confidence.

Attribution of Treatment Responsibility Scales (ATRS).
Based upon the format used by Eiser et al, 31 patients were presented with the question "Many heroin users fail when they try to give up the drug because:" to which they responded on each of five scales. Each scale represented one pole of two bipolar dimensions, these being "internal attribution-external attribution" and "stable attribution-unstable attribution." The five scales presented to participants were: (A) "It is just too difficult for them"; (B) "Because they do not try hard enough"; (C) "Because they do not know the best way to set about it"; (D) "Because of the kind of people they are"; and (E) "Because of bad luck" (external and unstable). The response format for each scale was "1" ("strongly disagree"), "2" ("disagree"), "3" ("neither agree nor disagree"), "4" ("agree"), and "5" ("strongly agree"). Internality scores were calculated by summing the ratings of (B) and (D), and subtracting from this sum, the sum of the ratings of (A) and (C) (ie, Internality = (B + D) − (A + C)). For stability scores the calculation was as follows: Depression Anxiety Stress Scales (DASS-21). This instrument comprises 21 items, generating subscale measures for depression, anxiety, and stress. Higher scores indicated higher levels of these scale properties. 32

Background Measures
The Addiction Severity Index (ASI) 33 recorded patients' demographic details, family situations, histories of illegal drug consumption, treatment for drug-related problems, and engagement in criminal behavior.

Procedure
T1 data were collected through structured interviews with patients conducted by a member of the research team (SJ) in a confidential location within the respective treatment settings. Written informed consent was obtained from all patients. Where possible, T2 data were also obtained through interviews conducted in the same facility as T1 interviews. However, T2 interviews were difficult to arrange due to constantly changing residential addresses for patients in the community, which included those discharged from hospital or prison since T1. Consequently, the presentation of T2 data are limited to self-reports of the use or not of heroin since T1, and patients contact status with the service where the T1 withdrawal had started.

Ethical Scrutiny
The data collection procedure was approved by the relevant NHS Research Ethics Committees, prison authorities, and the research ethics scrutiny procedures of Edge Hill University. The rights of all patients regarding confidentiality, withdrawal from the study, anonymity, and protection from harm while participating were respected at all times. Ethical approval did not include access to records of criminal offenses or to patients' treatment notes. However, the Participant Information Sheet emphasized that the researcher had no role in making treatment decisions, and would respect the confidentiality of their data in relation to both their clinicians and legal authorities. It was emphasized, therefore, that there were no incentives to misinform the researcher. Data were stored in accordance with the UK Data Protection Act 1998.

Analytic Strategy
Intergroup comparisons were primarily conducted using between-participants analysis of variances (ANOVAs), with post hoc pairwise comparisons being evaluated as two-tailed against a Bonferroni adjusted α level of P ≤ .017. Where the assumptions of parametric ANOVA were violated, nonparametric Kruskal-Wallis ANOVAs were conducted, with post hoc comparisons utilizing the Mann-Whitney U test being evaluated in the same way. Relevant correlational analyses were conducted across the whole sample. Logistic regression analyses were conducted on two dependent variables (DV) measured at T2, these being "heroin use or not since T1," and "contact maintained maintenance or not with T1 treatment services." The power and validity of these analyses were protected by minimizing the number of independent variables included as predictors using hierarchical variable entry, so that the ratio of participants per predictor remained as far below the accepted maximum criterion of 10:1 as possible. 34,35 Psychological Preparedness for Opiate Withdrawal Month 2020

Background and Drug Misuse-Related Variables at T1
The mean period of current incarceration for PG patients at T1 was 6.63 months (SD = 8.58 months). Table 1 (Sections A to C) summarizes results for background and drug misuse variables. The post hoc comparisons show that the PG had committed more drug offenses than the IG. PG patients had also committed more burglaries and had longer total imprisonment durations than the other groups. Regarding drug consumption, the PG reported less heroin consumption than the other groups in the 30 days prior to T1, but reported the highest prevalence of cocaine consumption within this timeframe. CG patients were receiving significantly higher daily methadone doses at T1 than the other groups.

Treatment Preparedness Variables at T1
Table 1 (Section D) summarizes results for the treatment preparedness variables at T1. The IG scored significantly higher than the PG on both the DTCQ-8 and CTS, while also scoring significantly higher on the SOCRATES problem recognition scale than the CG. There were no significant intergroup differences on the other SOCRATES scales, or for any of the mood and attribution scales. Table 2 summarizes the significant correlations between treatment preparedness variables and demographic, crime-related, and drug misuse variables, across the whole sample. Age was negatively correlated with both the SOCRATES ambivalence and taking steps scales, while selfefficacy was negatively correlated with lifetime duration in prison, and the number of days in the previous 30 days on which heroin and methadone, respectively, had been used.

Heroin Use Between T1 and T2
Sample attrition meant that T2 data were only available from 48 patients comprising 13 PG, 18 IG, and 17 CG patients. Three of the 24 patients lost to the study at T2 were contactable but declined to participate, with contact being lost with the remainder (ie, n = 21; 29.2% of the T1 sample) by both the researcher and the treatment services supervising the withdrawal at T1. Only 15 of the 48 patients retained at T2 reported having remained heroin-free since T1, with 10 of these being either in prison or receiving continuing medication from drug services. Of the 33 patients who had used heroin since T1, 29 were either in prison or receiving continuing support from drug services. Where data were available, use of heroin or not since T1 was not associated with treatment setting ( χ 2 [2, N = 48] = 4.778, ns).
Hierarchical logistic regression analyses for the prediction of heroin use or not since T1 were conducted with a limit of two independent variables at one time to preserve statistical power. Of the five variables in Table 3 (Section A) showing significant (or near to significant) T2 differences in measures taken at admission, CTS ratings were excluded from the regression analyses due to high correlations with the other variables. The DTCQ-8 scores were entered in model 1, and yielded a significant improvement in outcome prediction (ie, compared with random prediction (model χ 2 [1, N = 48] = 4.831, P = .028, −2 log likelihood = 54.794, Cox and Snell R 2 = .096). Model 2 comprised, successively, the other variables in Table 3 (Section A), one at a time. None of these variables significantly improved the prediction of relapse compared to Model 1. Further details of model 1 are shown in Table 4 (Section A) where the positive B coefficient for DTCQ-8 scores indicates that higher scores were related to an increased likelihood of no further heroin use since T1.

Maintenance of Contact With T1 Treatment Services at T2
Patients' maintenance of contact with T1 treatment services was not associated with treatment setting ( χ 2 [2, N = 72] = 2.625, ns). Table 3 (Section B) shows the three variables with significant T2 differences in measures taken at admission between patients maintaining contact or not. As there were only three variables showing significant differences, they were entered as a block of independent variables in a logistic regression analysis using maintenance of contact or not as the dichotomous DV. This yielded a significant improvement in the prediction of contact maintenance T2 (model χ 2 [1, N = 68] = 11.990, P = .007, −2 log likelihood = 70.399, Cox and Snell R 2 = .162). Further details of this model are  shown in Table 4 (Section B). None of these independent variables had significant Wald χ 2 values.

DISCUSSION
The present findings show that self-efficacy measured at T1 by the DTCQ-8 30 was an effective predictor of subsequent self-reported heroin use or not at a 6-month follow-up. T1 intergroup comparisons showed that the IG scored significantly more highly than the PG on this measure, and also on the CTS. This result adds to the existing literature concerning the predictive relationship of self-efficacy and confidence measures for successful treatment completion, to outcomes concerning methadone withdrawal and the maintenance of abstinence. [18][19][20][21] Of particular note is that these findings came from a robust psychometric instrument for measuring self-efficacy rather than an ad hoc scale. 19,30 These results suggest that self-efficacy is a measurable variable potentially usable by clinicians as an indicator of patients' preparedness for completing a methadone withdrawal programme and remaining heroin-free. Its use as such would, however, need to be as part of a battery of indicators informing clinical judgment regarding patients' preparedness for opiate abstinence and the gradual removal of OST, in accordance with clinical guidelines. 5,6 As the present data do not indicate whether patients high in selfefficacy sought an inpatient setting, whether the inpatient setting boosted patients' self-efficacy, or whether a synergistic interaction of these two explanations occurred, further research is required to examine the implications for clinical practice regarding the effective development of patients' self-efficacy across different treatment settings, in addition to the use in practice of self-efficacy measures as indicators of treatment preparedness.
Psychological preparedness for opiate withdrawal and abstinence is a complex construct where self-efficacy does not function alone. DTCQ-8 scores were negatively correlated with methadone treatment duration in the 30 days prior to T1. This may reflect a need to obtain sufficient personal and lifestyle stability through OST before withdrawal and abstinence become conscionable objectives to patients, thus reflecting the preparedness required for ceasing OST highlighted in clinical guidelines. 5,6 The strong positive correlation between T1 methadone doses and attributions to luck for treatment success may similarly reflect a lack of preparedness, and a risk of the negative outcomes associated with OST termination without adequate preparation. 8,9 Psychological preparedness is also related to the broader social context of patients' lives, as shown by the negative correlation (approaching statistical significance) between lifetime duration in prison and both DTCQ-8 and CTS scores. The PG scored worse on both of these preparedness variables than the IG. This suggests that limitations to drug availability through being in prison were not associated with enhancing treatment preparedness, which would be consistent with the negative consequences of premature OST termination in prison reported elsewhere. Prison incarceration has been shown to weaken relationships with people and activities not associated with crime, and to reduce prospects for legitimate employment. 27 Such consequences may be related negatively to treatment preparedness, which, in turn, highlights the need for an integrated network of clinical and social care provision for patients with addiction problems. 5,6 In summary, low DTCQ-8 scores may indicate the appropriateness of deferring OST termination and the initiation of opiate withdrawal, in favor of interventions to raise self-efficacy for a subsequent withdrawal.
This study provided no evidence of emotional states predicting outcomes for methadone withdrawal and abstinence, despite evidence for emotions playing an important role in addictive behaviors. 19,22 This role may not, therefore, include mediation of outcomes for specific withdrawal episodes. The lack of a difference in depression between the PG and the other groups at T1 is noteworthy, given the constraints of a prison environment. Additionally, no relationship between T1 attributions for treatment outcomes, and subsequent actual outcomes were found. Although this suggests that such attributions at admission do not have predictive value, this does not limit the indicative value of attributions for progress and outcomes recorded later in the treatment process. 23,24 OST termination can lead to loss of contact with treatment services and negative outcomes arising from continued opiate use. 8,9 Although no variables concerning methadone treatment were related to maintaining contact or not with T1 services in the present sample, higher SOCRATES ambivalence scores and younger age were related to this outcome. These relationships could indicate a lack of preparedness for the opiate abstinence treatment available. For patients who remained in contact with T1 services, these two variables, which were negatively correlated with each other, were not related to reported heroin use between T1 and T2. Their role as predictor variables may therefore be limited to maintaining contact or not with services.
This study has some important limitations. Sample attrition was a challenge to T2 data collection, with some nonsignificant results possibly being due to low statistical power. However, attention to maintaining statistical power in the logistic regression analyses did lead to meaningful findings. As only male participants were studied, further research is required concerning preparedness indicators for comparable interventions for female patients. There was also reliance on self-reported data regarding drug use and crime. However, this is not uncommon in the literature. Toxicological verification of abstinence is not always possible, and may not be ethically desirable following treatment termination with regard to maintaining mutual respect between former patients and professionals.
In conclusion, this study highlighted self-efficacy beliefs for treatment outcomes as potential predictors for opiate withdrawal modified by methadone and subsequent abstinence. Patients' age and ambivalence regarding change emerged as potential predictors for maintaining contact with treatment services.
This study was part of a successful partially self-funded PhD project undertaken by the lead author (SJ). Additional financial support was provided by the Faculty of Health and Social Care at Edge Hill University.