Reanalysis of a randomized controlled trial on promoting influenza vaccination in general practice waiting rooms. A Zelen Design

Christophe Berkhout, Jeroen De Man, Amy Willefert-Bouche, Suzanna Zgorska-Maynard-Moussa, Margot Badelon, Lieve Peremans, Paul Van Royen

Keywords: Influenza Vaccines / administration & dosage* ; Patient Education as Topic / methods* ; Routinely Collected Health Data ; randomized controlled trial

Background:
In 2018, we published a randomized controlled trial (RCT) assessing the effect of an advertising campaign for seasonal influenza vaccination using posters and pamphlets in general practitioners' (GPs) waiting rooms. No effect of the intervention could be demonstrated in 10,597 patients of 75 GPs. However, the immunization uptake increased in both arms of the study compared to the prior year, and we evoked an experimental artefact.

Research questions:
To deepen the investigations explaining the increased vaccination uptake, considering the risks of an experimental artefact. To check the external validity of our trial.

Method:
Registry based 4/2/1 Zelen designed cluster RCT with a two years follow-up of the study cohort. Study population included a cohort of 23,024 patients eligible to be vaccinated (i.e. ≥65 years of age or affected by a chronic condition) registered under 175 GPs and followed over four years from winter 2013-2014 until winter 2016-2017. The intervention ran during the 2014-2015 influenza vaccination campaign. A posteriori, we included a new control group (“Zelen group”) consisting of patients with the same characteristics of 100 GPs but with the GPs not aware of the trial. The main outcome remained the number of vaccination units delivered per study group (by community pharmacies). Data were extracted from the SNIIRAM warehouse claim database for the Lille-Douai district (Northern France). The intervention effect was assessed using generalized estimating equations.

Results:
The Zelen group did not show a difference with the population of the RCT in vaccination uptake, while taking into account a higher baseline vaccination in this group. Overall, the proportion of vaccinated patients increased from 51% to 70% over the four study years in the 3 groups. Vaccination uptake increased faster in the 60-65 and was higher among the >65. Being vaccinated the previous year was only a weak predictor of being vaccinated the subsequent year.

Conclusions:

Points for discussion:
What are the factors discriminating public health data from primary care data?

How can Zelen designs with routinely collected data be used for RCTs in primary care?

Challenging interpretation of routinely collected data from claim database warehouses