Presentation on theme: "«Les enjeux méthodologiques, opérationnels et éthiques» «Methodological, operationnal and ethical challenges» JC Desenclos, Institut de Veille Sanitaire."— Presentation transcript:
«Les enjeux méthodologiques, opérationnels et éthiques» «Methodological, operationnal and ethical challenges» JC Desenclos, Institut de Veille Sanitaire & ANRS AC25
Why this seminar ? ANRS « Actions coordonnées » : to promote the emergence of research projects from the research community on priorities or insuficiently covered needs AC25, AC18 & AC 12 : hard to promote community intervention trials ! Why not a research seminar ?
Health interventions applied on a community-wide basis have been increasingly used in public health research over the past several decades Community-based interventions efficient way to increase the generalizability of public health program benefits Needs to demonstrate to health policy makers the positive impact of health intervention beyond that based on individual-based clinical trials The emergence of community health intervention trials is a shift in health research from studies that focus primarily on the individual to those that focus on community groups. The emphasis on interventions focusing on communities has created new methodological, operational and ethical challenges Why doing community health intervention trials ? D’après Atienza & King, Epidemiol Rev, 2002
Innovative actions are needed to reduce hepatitis and HIV infection and their health & social impact to individuals and communities Biological, clinical & epidemiological and social sciences research : –improve our understanding of the processes that contributes to the spread and the health & social impact of these infections –new tools, strategies & policies for prevention & intervention –new opportunities to reduce further their health and social impact Innovative tools or health social strategies must be evaluated at community level throught community-based intervention trials However, community-based intervention trials under utilized by the French public health research community Why this seminar for ANRS ?
Our aim ! Foster the development of the experimental evaluation of innovative prevention & health, behavioral and social strategies among the French researchers for research promoted by ANRS in France and internationally En donner le goût !
EXPERIMENTAL Exposure manipulated by Investigator DescriptiveAnalytic Exposure NOT manipulated by Investigator OBSERVATIONAL Cohort Case-control Case-series Cross-sectional Ecological Quasi-experimental and Randomized trials Study Designs (internal validity) Individuals Clusters Source: Mertens T
Methodological - scientific issues (1) Two components –definition of the experimental social intervention or program to bring desired changes and health improvement –research design to evaluate / assess / quantify interventions outcome and impact Although relatively similar to clinical trials for basic principles, quite a few methodological, operational and ethical issues
Methodological - scientific issues (2) Definition and conceptualization of the intervention Is a community control trial needed ? Enough evidence to get into a trial ? Definition of the community of interest Randomization ? –if yes how ? –Individual vs cluster ? Statistical power ? Monitoring the delivery of the intervention and adherence Outcome measurements (subjective vs objective…) Cohort vs cross-sectional analysis Accounting for secular trends Statistical issues for the analysis of community trials …
Understanding how community-based interventions improve health is as critical (if not more) than evaluating the outcome It requires the specification of conceptual or theoretical models Need for a strong interaction between social sciences and quantitative epidemiology Methodological - scientific issues (3) Atienza & King, Epidemiol Rev, 2002
Operational challenges Balancing scientific methodology and feasibility Trial organization & planning Identify and recruit appropriate communities Resources and funding Information & communication Political, institutional and social issues… Community involvement… Recruitment & training of study personnel… Mono-centre vs. multi-centre trials (multi-country) Quality control and quality assurance Data collection Interim analysis Scientific monitoring (steering, scientific boards…) …
Ethical issues (1) Follow principles for clinical trial, however, specific issues –the behavioral & social nature of the intervention –community vs individual approach in intervention delivery Ethical consideration (1, 2) –Autonomy/informed consent –beneficience –non-maleficience –assessment of the risk –risk – benefit ratio –justice –access to care –confidentiality –community invlovment & partnership 1.Peter Cleaton-Jones: the first randomization trial of male circumcision: what were the ethical issues ? Plos Med; 2005:2:e287 2. MIMH Collaborative HIV/STD prevention trial group. Ethical issues. AIDS 2007;21:S69-S80.
Ethical issues (2): Challenges specific to international collaborative research Sponsor-funding of trial in poor countries by rich countries Potential to exploit vulnerable international population Should be independent of any moral or ideological value from sponsoring country (ies) Should address needs of host population and nation Acceptable level of care Secure access to intervention if conclusive after trial is over Contribute to capacity building of host country Moving research into policy practice MIMH Collaborative HIV/STD prevention trial group. Ethical issues. AIDS 2007;21:S69-S80.
Seminar organization Conferences : 3 introductory & 1 methodological Five 2 hours thematic sessions –sexual issues (2) –screening and care strategies (2) –prevention challenges among drug users –based on experience in the North and South –A « table ronde » on role of institutions For each session: –a coordinator –three invited speakers –a discussant that focuses on methodological and ethical issues and the impact on policy decision –a moderator to keep the time and write a synthesis of the session, including issues discussed
Pas mal d’échecs ! Ce n’est pas que cela ne marche pas ! –mais c’est plus compliqué qu’à première vue –méthode de recherche en santé public toujours en évolution Echecs: –pas de financements –échec dans la mise en œuvre –pas de mise en évidence de l’effet Importance de l’analyse de ces échecs Design alternatifs
Randomised controlled trials of interventions to prevent HIV InterventionNumber of trialsImpact on HIV incidence CompleteOngoing Behaviour change73Little overall impact Bacterial STD tx50Impact in one trial Herpes suppressive tx12No overall impact Male circumcision3050-60% reduction in risk in all 3 trials Female diaphragm & gel10No overall impact Oral pre-exposure prophylaxis 13No overall impact Vaginal microbicides85No overall impact HIV vaccine41No overall impact Source: H Weiss
No true effect of the intervention Conceptualisation of the intervention Study too small to detect significant effect Poor adherence reduced efficacy in trial Possible reasons for lack of impact: Adapted from : H Weiss
Some key statistical issues in HIV prevention trials Sample size & power –Choice of study population –Choice of endpoint Methods of analysis Factors influencing measured efficacy –Adherence –Withdrawal Some issues in studies among HIV+ individuals Alternative designs – alternative analyses Source: H Weiss
MIRA Study Design All women received risk reduction counseling, free male condoms and diagnosis and treatment of curable STIs Women were followed quarterly for 12-24 months + Source: N Padian
MIRA Results from ITT (n=4948) HIV Incidence rate Relative Hazard (95% CI) All Sites4.01.05 (0.84 – 1.32) Harare2.71.20 (0.83 – 1.74) Durban6.80.95 (0.69 – 1.31) Johannesburg3.41.05 (0.60 – 1.87) Source: N Padian
Evaluating multi-component technologies vs. New technology + Effective prevention Effective prevention New technology Effective prevention Effective prevention Intervention Control Attributable fraction Source: N Padian
Adherence to multi-component interventions vs. New technology ± Adherence + Effective prevention ± Adherence Intervention Effective prevention ± Adherence Control Source: N Padian
Summary RCT is gold standard Trials must be powered adequately Tempting to be over-optimistic in sample size assumptions (to get funding!) Allow for reduced HIV incidence within trial Adherence major issue in most HIV prevention strategies Need to improve methods of estimating adherence ITT is least biased – but often informative to also allow for as-treated analyses Causal models allow for non-adherence