The need for ongoing, prompt, cost-efficient and comprehensive monitoring and evaluation of drug treatment is well recognised. The scientific literature supports the use of randomised controlled trials (RCTs) as the ‘gold standard’ research design for evaluating interventions and treatment outcomes. However, in community substance-use treatment settings, the use of RCTs is not always possible or appropriate. There are ethical issues and questions about relevance, feasibility and costs.

Given the limitations of RCTs and the demand for ongoing, timely and effective monitoring of treatment outcomes, Comiskey and Snel sought to test the feasibility of linking laboratory data and client intake data and its usefulness for modelling retrospectively five-year longitudinal drug treatment outcomes in an Irish opiate treatment setting.1

Methodology

A multi-site, retrospective, longitudinal cohort study was implemented to evaluate outcomes for opiate users across Dublin from January 2006 to December 2010.

Longitudinal urinalyses were extracted from two national laboratories which provide substance use screening tests to 17 drug treatment sites and to two buses providing mobile needle exchange services. During the five -year study period a total of 1,734,283 test results were identified for 330,802 urine samples presented by a total of 4,518 unique individuals across Dublin.

In addition to the urinalysis database, data were extracted at a number of ad hoc time points from all the treatment sites which use a client electronic record system, which is known as the Drugs and AIDS Information System (DAIS). This system records data about drug-users seeking treatment, including their demographic characteristics, assessment prior to treatment, prescriptions, hepatitis C status, treatment programmes, needle exchange programmes and rehabilitation integration services. During the study period, 5,430 records relating to 2,832 unique individuals were recorded in DAIS; of these unique individuals, 97 per cent  were Irish, 72 per cent were male, 59.7 per cent had no children and seventeen known deaths were recorded. These data were linked, via a unique client identity number, at the individual client level to the longitudinal urinalysis data.

Once the full database of unique clients was created, with each repeat episode of treatment linked, captured and recorded for that client, the longitudinal outcome variables in the form of drug positive urinalyses results were derived. Outcomes were tracked sequentially, with the first urinalyses denoted as time point 1 (the intake/baseline measurement) and each subsequent urinalysis result was then recorded as test 1, 2, 3 … up to a maximum of 260.

Results

Across the five-year period 62 per cent of the urine samples tested positive for benzodiazepines, 43 per cent for cannabis, 40 per cent for opiates and 11 per cent for cocaine. Analysis of substances used at treatment intake, at six months and at one- to five-year follow-ups, revealed:

Differences in urinalysis protocols: Extraction of urinalysis data from the two national laboratories revealed that treatment sites varied considerably regarding the number and frequency of urinalysis tests conducted; for example two of the seven DAIS treatment clinics requested over 83 per cent of analyses, but accounted for only 68 per cent of all DAIS clients. The researchers attribute these variations to a difference in treatment-site philosophy and practice rather than variations in client characteristics.

A further inconsistency occurred between the stated tests conducted as part of a routine urine test and what was found in the analysis: while 99.9 per cent of urine samples were tested for opiates and cocaine as required by policy, only 72 per cent were tested for benzodiazepines.

Age of first drug use: Clients who did and did not test positive for opiate use at year five were compared on a number of variables. No significant difference was found apart from age at first drug use (mean age of 15.53 years vs. mean of 14.63 years, p=0.008)

Opiate use: The proportion of clients who tested opiate positive fell from 61.8 per cent at initial treatment intake to 12.5 per cent at the end of the five-year period. The researchers’ time series analysis of weekly proportions opiate positive had predicted 16 per cent (95% confidence interval: 7%–25%) of clients would be opiate positive at the end of the five years.

Other drug use: Significant increases were found in benzodiazepine use, and significant increasing effects of concurrent cocaine and benzodiazepine use on the likelihood of using opiates. It was also possible to link and describe the changes in the methadone doses prescribed: analysis of the DAIS system revealed there were only minor changes in the doses of methadone prescribe over the five years.

Conclusion

With this research the authors have demonstrated that data from existing multi-sited, cross-sectional sources can be linked, matched, mined and modelled to develop prompt, retrospective, sequential outcome results that are useful for policy makers, service providers and service users.

1 Comiskey CM and Snel A (2016) Using clients’ routine urinalysis records from multiple treatment systems to model five-year opioid substitution treatment outcomes. Substance Use & Misuse (51/4): 498–507. https://www.drugsandalcohol.ie/25363/