|
|
|
No recommendation before 1995
There will not be a recommendation about the feasibility of a trial to make heroin available in a controlled manner to dependent users this year. This will allow time for detailed discussion of a proposed evaluation design and will allow a number of projects still underway to be completed. In addition, with both the ACT and NSW governments facing elections early in 1995, it is unlikely that there would be a suitable climate for rational debate on the issues, regardless of whether the recommendation was for or against a trial.
One of the two new working papers presents a proposal for how a trial should be evaluated if it eventuates. The proposal has both advantages and disadvantages. Given that evaluation is central if a trial goes ahead, we are planning a thorough discussion of evaluation issues over the next few months.
Work in progress includes a detailed examination of the history of prescribing of heroin and morphine to dependent users in the UK, USA, Sweden and the Netherlands, as well as current prescribing in the UK and Switzerland. We are also preparing working papers on the outcome measures for a trial, the day-to-day running of a trial, the effects of heroin on driving, as well as detailed costings for a trial. In addition, we are planning a national community survey of attitudes to a trial.
New Working Papers
Civil liability issues associated with a "heroin trial"
Natasha Cica
Working Paper Number 11
In terms of civil liability, the legal duties owed by researchers in a heroin trial may be classified under three headings: liability in battery; liability in negligence; and liability for disclosing confidential information.
Each of these areas of liability is discussed in this paper. Particular attention is paid to situations where the researchers' legal duties are unclear or may be conflicting.
Liability in battery arises out of the touching of another person without that person's legally valid consent. Researchers therefore must obtain a valid consent to trial procedures from each trial participant. For this to occur, the following requirements must be satisfied:
These four requirements are examined in this paper.
Researchers will be liable in negligence to a trial participant or other person affected by the conduct of the trial if the following elements are all present:
Each of these elements of liability in negligence is discussed. Issues raised include the scope of the duty owed by researchers towards children born damaged as the result of their parent's participation in the trial.
The discussion of liability for disclosing confidential information centres on the Epidemiological Studies (Confidentiality) Act 1992 (ACT), which specifically applies to the proposed heroin trial.
The legal rules outlined in this paper should be viewed as minimum standards only. The researchers should focus less on avoiding liability than on ensuring that each individual trial participant, and each other individual to whom a duty of care is owed, is treated with maximum respect.
Statistical Issues in Planning a Randomised Controlled 'Heroin Trial' Robyn G. Attewell and Susan R. Wilson with a Foreword on an evaluation strategy (Gabriele Bammer) and an Appendix on possible outcome measures (Gabriele Bammer, Nova Inkpen and David McDonald) Working Paper Number 12
This paper is concerned with the major statistical issues in planning a randomised controlled trial to compare two management regimes for the treatment of dependent heroin users. The background to the evaluation strategy is presented in the foreword. The primary purpose of the trial is to address the question (for people who are or have been on a methadone program):
"Does providing heroin as an optional addition to, or replacement for, methadone improve outcomes over and above improvements obtained with oral methadone alone? Outcomes include health, HIV risk behaviours, criminal behaviour, social functioning and licit and illicit drug use."
A trial would therefore have two arms: an expanded availability arm in which injectable heroin and oral methadone were both available and a control arm where only oral methadone was available.
This paper discusses, in this context, power and sample size considerations, the requirements for a good trial and issues associated with attracting and retaining volunteers from the study population.
A table is provided with a comprehensive range of values for determining whether the available study population size will be adequate for the level of effect in which one is interested. This table is used in conjunction with an appendix which lists, from surveys conducted with ACT methadone clients, likely control group values for various outcome measures. Suppose the sample size were to be 100 in each group (total 200). Then 80% of the time one would detect improvements as small as about 8% for outcome measures having frequencies around 10% (or 90%) in the control group, and this increases (decreases) to about 18% for outcome measures having frequencies around 50% in the control group.
There are four main requirements for a good trial, assuming the treatment groups, the participants, and the content of the outcome questions have been decided. They are absence of systematic error, precision, range of validity, and calculation of uncertainty.
Randomisation reduces the likelihood of systematic error, while lack of 'blinding', use of self-report data, and intermingling of study participants may contribute to such error. There is also discussion of randomisation of participants who are not independent, for example couples, and analysis of their data.
Precision can be increased by increasing the number of participants, by repeated observations of participants and by making concomitant observations of participants. Inclusion in the analysis of concomitant variables that strongly affect the 'outcome' variables, for example length of time on the methadone program, may (under certain conditions) increase the precision with which the effect of the treatment regime can be estimated, and increase the statistical significance of this effect.
The wider the range of conditions investigated in the trial, the greater is the confidence in the extrapolation of the conclusions. This is limited by the size of the study population. There is also subtle subjective bias which affects interpretation of the estimates of group differences. Extrapolation of the result to other situations and participants depends on one's point of view and this bias is neither fixed nor numerically measurable.
For this trial, there should not be any confounding factors at the initial design stage, but they may result as a consequence of differential retention of participants in the two treatment groups.
Attracting and retaining volunteers may well be the most challenging part of the trial. It can be feared that individuals who volunteer in the hope of access to heroin and who are then assigned to the 'methadone only' arm will drop out. 'Drop out' may be also related to concomitant variables. For example, one can imagine that those who have only been on the methadone program for a short length of time may be more likely to drop out. Dropping out would lead to a loss of power and dropping out related to concomitant variables would lead to confounding. A table is presented giving the maximum 'detectable as different' percentage for the expanded availability group, assuming a total original sample size of 200, no drop-outs in this group, and a 'reasonable' drop-out rate of 50% from 100 to 50 in the control group, over a range of control group percentages. Slightly larger improvements would need to be seen to be 'detectable as different' at the same power. The table can also be used to show the effect of changing the ratio of assignment to the expanded availability group and the control group from 1:1 to, for example, 2:1 in order to increase a participant's odds of obtaining the expanded availability arm. The additional degree of improvement which would need to be seen to be 'detected as different' is relatively small. However, if such a recruitment strategy was used, retention rates, particularly in the control arm would have to be high.
Police Workshop
In late May a workshop about policing issues was held. It was chaired by Chief Inspector Frank Hansen of the NSW police, who co-organised it with Detective-Superintendent Ted Foster OIC Drug Operations in the ACT. Local and interstate police attended as did Dr Grant Wardlaw, Acting Director, AIC and study co-ordinator, Dr Gabriele Bammer. The issues discussed included:
There was general agreement that the proposed trial design would minimise the potential of a trial to attract users to the ACT. The two most important issues needing further work were seen to be trial security and possible problems from people driving under the influence of heroin.
The workshop provided a valuable opportunity for the project to have input from police about concerns with aspects of the trial and to discuss strategies for dealing with practical difficulties.
'ATTITUDE'
The ABC television series Attitude screened a program about the process of the Feasibility Study on June 1. We decided to cooperate with the program because we believed it was an opportunity to provide information to the general public which will aid rational discussion about illicit drug policy and proposals for policy reform. The response to the program was overwhelmingly positive and most people saw it as an even-handed presentation of the difficult issues involved.
Briefings
In early August we were invited to brief the South Australian Legislative Council Select Committee on the Control and Illegal Use of Drugs of Dependence about the Feasibility Study. We have also been briefing politicians, businesspeople, policy makers and others about the research results and process, as well as the need for debate when the recommendation is made.
Changes to the advisory committee
Dr Grant Wardlaw is now Acting Director of the Australian Institute of Criminology and thus replaces Professor Duncan Chappel as Co-chair of the Advisory Committee.
Mr Richard McCreadie, Deputy Commissioner Tasmanian Police, has replaced Dr Mike MacAvoy. He is the Chair of the National Drug Strategy Committee.
Changes to the research team
Projects completed: Nova Inkpen Research Assistant, AIC (outcome measures for a trial)
New projects: Dr Andy Butlin has joined NCEPH as a Visiting Fellow to work on a national survey of community attitudes to a trial.
Further Information
For further information about the feasibility research contact: Dr Gabriele Bammer, National Centre for Epidemiology and Population Health, Australian National University, Canberra ACT 0200, Phone: (06) 249 0716 Fax: (06) 249 0740
New Publications
Working Papers
*Cica, N. 1994. Civil liability issues associated with a "heroin trial". Feasibility Research into the Controlled Availability of Opiods Stage 2 Working Paper Number 11.
* Attewell, R.G. and Wilson , S. R. 1994. Statistical Issues in Planning a Randomised
Controlled 'Heroin Trial' with a Foreword on an evaluation strategy (Gabriele
Bammer) and an Appendix on possible outcome measures (Gabriele Bammer, Nova
Inkpen and David McDonald). Feasibility Research into the Controlled Availability
of Opiods Stage 2 Working Paper Number 12.