Should we Review Single Interventions or All Comparators in a Decision-Making Context?

Jefferson T., Demicheli V., UK Cochrane Centre, Oxford, UK

Introduction: Currently the majority of Cochrane reviews are carried out on single interventions. Topics are mainly chosen on me basis of reviewers' suggestions (see abstract, Jefferson and Milan). This practice is in keeping with the Collaboration's bottom-up approach but may be in conflict with requirements for decision-making. Decision-makers assess a range of alternative interventions that will compete for scarce resources to tackle health care problems. We report on how the use of Cochrane reviews illustrated the "best buy" strategy for prevention and early treatment of influenza in healthy adults.
Objectives:
Methods: Work consisted in systematically reviewing available knowledge on the topic upon which to base a decision-making model for policy in me setting of the British Army. We began by carrying out three Cochrane reviews: one on vaccines, one on amantadine and rimantadine and one on the yet unlicensed neurammidase-inhibitor antiviral compounds.
Results: The reviews simplified our decision tree. Before the Cochrane reviews our tree had prevention/treatment combinations leading to 11 decision nodes and 165 terminal nodes. After the reviews we were able to eliminate many interventions on the grounds of safety and effectiveness and we were left only with preventive alternatives (1 decision node and 12 terminal nodes). The pruning of our tree (with the reasons) is summarised in the Table.

Items	Pre-Cochrane Reviews	After Cochrane Reviews
Choice of the best single alternative for PREVENTION	Oral vaccines; Aerosol vaccines Parenteral vaccines; Oral Amantadine; Oral Rimantadine; Inhaled Zanamivir; Oral Ro 64- 0796 (GS 4104)	Parenteral vaccines; Oral Amantadine Oral Rimantadine; Oral Ro 64-0796 (GS 4104) (Aerosol and oral vaccines are less effective or there are minimal differences and do not represent currently a real alternative. Zanamivir trials only apparently included laboratory confirmed outcomes)
Choice of the best single alternative for TREATMENT	Oral Amantadine, Oral Rimantadine; Inhaled Zanamivir; Oral Ro 64-0796 (GS 4104)	None (all compounds shortened duration of illness by 0.5 days)
Table: Possible alternatives to prevent and treat influenza, before and after reviews of the evidence.

Discussion: Our systematic review of me topic eliminated spurious alternatives, avoiding a waste of resources in further evaluating them and the formation of misleading conclusions.

Rome 1999 O60