To the best of our knowledge, this is the first study which, by applying a questionnaire in parallel to both participants and nonparticipants, aimed to identify reasons for (non)participation in health research among participants and nonparticipants of a population-based health-related study among the elderly. Previous research examined the reasons either in qualitative studies with small groups of individuals [12,13,14] or were of a descriptive nature by recording the reasons either only among participants [8] or nonparticipants [7, 15]. In contrast to those studies, we used an analytical approach by applying multivariate techniques with automatic selection to identify the most important reasons for (non)participation. In contrast to other studies, the participants and nonparticipants of our influenza vaccination study did not differ much in terms of basic sociodemographic and health-related characteristics such as age, Body Mass Index, self-reported myocardial infarction and cancer. However, multivariable analyses showed that the risk of nonparticipation increased significantly among females, those with diabetes, and a poorer self-perceived health status, indicating a possible selection bias.
The identified reasons for (non)participation in health research were the combination of various reasons and can be summarized into three groups: a) reasons associated with personal benefit (e.g. receiving information about personal health status), b) reasons associated with invasiveness of the study instruments (e.g. blood draws), and c) reasons associated with time demand by the study. In addition, altruistic motives and data security issues were significantly associated with (non)participation, but only in the sensitivity analysis. The latter analysis excluded the nonparticipants whose main reason for not participating in the influenza vaccination study was that they had already received the vaccine for that season. The remaining nonparticipants may be considered “true” nonparticipants and the two items identified in the sensitivity analysis (altruistic motives and data security issues) should therefore also be taken into account.
Personal benefit in form of monetary and nonmonetary advantage has been observed in many studies and is used as an incentive to increase response rates [16]. Receiving information on personal health was found to be a driving reason for participation in genomic studies among elderly individuals in Switzerland [17]. Benefits from participation in a health study should be clearly communicated to potential study participants to increase the response rates, e.g. through addressing them adequately in the consent forms. In line with our finding on time demand by the study, Gaertner et al. observed that “no time” was among the top five reasons for nonparticipation reported in a health survey among elderly nonparticipants in Berlin, Germany [15]. Alternative methods of data collection that reduce the time demands by a study should be considered, such as home visits, self-collection of biosamples, or web-based data collection. However, web-based data collection might not be feasible in the current elderly population. This might change considerably in the future, as a result of aging of computer/internet users and increasing acceptance of electronic media among senior citizens.
There was considerable overlap between the subgroups who considered low time demand and low number of medical interventions important (see Fig. 2). This observation makes sense from the point of view of human behavior, as “impatient” personality traits or perceived or real high demands on an individual’s time may likely affect the responses to these two questions similarly. On the other hand, there was only a very small percentage of nonparticipants who, in addition, did not find receiving health information a motivating factor. Thus, we did not identify a specific subgroup of individuals with particularly adverse attitudes toward participating in health research.
A monetary incentive turned out to not be a determinant of (non)participation. This is in line with results from our other population-based studies in urban Northern Germany, i.e. a resource-rich setting. In a pilot study in Pretest 2 of the German National Cohort, a monetary incentive was not associated with participants’ willingness to participate in future studies [18]. Likewise, in a study on serial home-collection of anterior nasal swabs for S. aureus surveillance, a monetary incentive did not increase participants’ compliance with the study protocol [19]. These findings are important for cost-efficient planning of future studies in populations such as ours, but may not be transferable to research settings in resource-poor countries.
The response rate to the influenza vaccination study was very low. Therefore, new strategies are required to recruit adequate numbers of study participants for studies where representativeness of the general population is important. Sampling based on the residents’ registry alone turned out to be an inefficient way to recruit a study population representative of the general aging population. Although we oversampled older age groups to achieve representativeness in terms of age, it is difficult to control for other parameters. For example, we missed frail individuals living in nursing homes. Thus, one needs to consider other sampling approaches in addition to the traditional register-based sampling, such as recruitment in nursing or residential care homes for elderly individuals, family physicians’ offices, or other places with a high accumulation of senior citizens, such as senior citizens’ associations or community support structures for the elderly. A recent systematic review showed that response rates were higher among population-based studies which apply face-to-face interviews, provide home visits for examinations, or had a less intensive study protocol [20]. We did not address the first two aspects in our present study, but it appears plausible that, in addition to offering a stream-lined study protocol with few interventions, providing home visits could further increase participation rates in health-related studies among elderly individuals.
Limitations of the study
This study has two potential limitations. First, only one third of the nonparticipants returned the nonresponder questionnaire. This subsample may be biased in terms of selected sociodemographic characteristics and may thus not be representative of the general population. For instance, we failed to obtain information about nonparticipants who could not complete and/or mail the questionnaire due to frailty or dementia. Second, the 10-item questionnaire was administered to the participants of the vaccination study at the last follow-up visit. Experiences made while participating in the study may have affected their attitudes toward health research and participation in health-related studies.