Third national biobank for population-based seroprevalence studies in the Netherlands, including the Caribbean Netherlands

The seroprevalence of National Immunization Programme (NIP)-targeted diseases is periodically monitored in the Netherlands (NL) by national seroepidemiological (PIENTER) studies, executed by the National Institute for Public Health and the Environment (RIVM) in collaboration with Public Health Services and municipalities. The first serosurvey was performed in 1995/1996 [1] and the second in 2006/2007 [2].

Gathering seroepidemiological data forms an important tool for the evaluation and optimization of the NIP, and gives insight into the protection against infectious diseases in different (sub)groups in the population. The results of previous Dutch serosurveys have contributed to vaccine policy, e.g., during the measles epidemic in 2013/2014, data on seroprevalence (particular the maternal antibody levels in infants) from the second Dutch serosurvey in 2006/2007 were used to advise on measles-mumps-rubella (MMR)-vaccination at an earlier age [3]; tetanus seroprevalence data led to an advice regarding a more restricted use of tetanus prophylaxis [4]; and the decision to revaccinate against meningococcal C disease at an adolescent age with a tetravalent vaccine were partly based on these data [5].

Since the last serosurvey in 2006/2007 several adaptations in de NIP were applied (Table 1) and some (small) outbreaks occurred, e.g., measles outbreak in the Dutch Bible Belt in 2013/2014 [6], mumps among vaccinated young adults [7] and increased incidence of meningococcal serogroup W disease since 2015 [8]. Events like these will have impact on the immune status in the population and justify close investigation. In addition, monitoring the seroprotection of the population is required at regular time intervals as vaccination can affect the dynamics of infectious diseases on the long term, for instance leading to an increasing age of infection or waning antibody levels, e.g., against diphtheria and measles [3, 9]. For these reasons, a third seroepidemiological study (PIENTER-3) was performed [10] to identify (new) population groups at risk for infectious diseases and to evaluate the adaptations made in the vaccination scheme in order to improve the overall quality of the programme.

This third study has been extended with the collection of saliva and faecal samples, as well as oro- and nasopharyngeal swabs, creating a more comprehensive biobank for the Dutch population. With the collection of these diverse human materials, accompanied by extensive individual information from questionnaires and the linkage with other data sources (medication histories and environmental exposures), this biobank harbours a wealth of information.

Importantly, this serosurvey has been expanded to include the Caribbean Netherlands (CN) for the first time (Health Study Caribbean Netherlands, HSCN). The Caribbean islands Bonaire, St. Eustatius and Saba (together CN) are officially part of NL and considered public entities under Dutch law since October 10, 2010. Hence, public health falls under the direct responsibility of the Dutch government, e.g., the supply, execution and monitoring of the NIP. Strikingly, no data on protection against infectious diseases and associated risk factors are available. The need for knowledge is underlined by outbreaks of measles and diphtheria in neighboring countries in Latin America [11, 12] and epidemics of vector-borne diseases in the Caribbean region (e.g., zika, dengue and chikungunya) [13‐15].

This paper outlines the design of population-based cross-sectional serosurveys in NL and CN, its study population and response rates. Subsequently, future research possibilities of this extensive data collection and experience with conducting large population-based public health research in CN will be described.

A third national biobank among the general population of NL has been generated and will be an important tool to evaluate infectious disease control in NL and contribute to public health policy. The seroprevalence data will provide insight into the effectiveness of the Dutch NIP and direction for improvement. These results will also be of value for future outbreak management. In addition, data collection has been extended to CN for the first time, which resulted in an extensive amount of information that will enable us to support future public health policy on these Caribbean islands, e.g., regarding tropical pathogens. Moreover, besides serum collection, a large number of additional materials have been collected, which allows us to look into relevant emerging topics, such as antibiotic resistance and the microbiome. Hence, this biobank offers unique opportunities to investigate infectious diseases in a much broader sense.

A high number of persons participated in this study, which enables us to perform most seroepidemiological (sub)analyses with sufficient power as calculated beforehand. Response rate (blood and questionnaire) in the NS sample was 14.4% and in CN 23.4% and, in line with other international studies [20‐22], females and elderly as well as natives were highest responders among the invitees. Overall, the NS sample (including the oversampling of non-Western migrants) is rather comparable to the Dutch population, especially regarding urbanisation degree and religion [23, 24]. Religion is an important variable with regard to seroprevalence as vaccination coverage among several denominations in NL is much lower compared to the rest of NL [16]. Females as well as indigenous Dutch, people from SAN and other non-Western countries – other than Morocco and Turkey – were slightly overrepresented (e.g., 54.2% females and 78.1% indigenous Dutch in our sample vs. 50.4% and 73.2%, respectively, in the Dutch population) [25]. Overrepresentation of SAN and non-Western migrants was due to the oversampling. Further, participants had a lower educational level as compared to the general Dutch population (31.4% is low-educated in the Dutch population vs. 38.6% in our sample) [26]. Generally, for each age strata in the NS sample, sufficient participants were included, although overall we reached marginally lower numbers per age strata than in the 2006/2007 study, except for 0-year-olds. For the LVC sample, sufficient ORI-participants were included in each age strata, except for the oldest age group (60–89 years).

The CN sample in this study is generally a good reflection of the total population on the islands concerning religion and educational level [27, 28]. People born in non-Western countries (which consists of 93% Latin Americans in our CN sample) and males are slightly less represented in our study population though (namely, 14.4% Latin Americans and 44.8% males in our sample vs. 19.3% and 51.5%, respectively, in the CN population) [29].

Socio-demographic dissimilarities compared with the general populations of NL and CN, due to selectivity in response and the sample design, will be taken into account by weighting the participants on a set of variables (age, sex, ethnicity and degree of urbanisation). An in-depth non-response analysis for NL will be carried-out and published in the near future. Moreover, having applied an identical robust design for the third time ensures maximum comparison with previous studies and opens opportunities for changes over time and modelling analyses. A less costly and less extensive design of sample collection, for instance via residual sera, is more prone to selective response and lacks the opportunity to collect additional materials as well as data on various characteristics of participants to perform risk factor analyses.

The response rate in the NS sample (14.4%) was lower as compared to the two previous serosurveys performed in NL (50% in 1995/1996 and 32% in 2006/2007). Nonetheless, low(er) response rates have also been reported in other recently conducted large population-based studies in NL (e.g., ‘NL de Maat genomen’, phase 1: < 20% [30] and ‘Lifelines’: 24.5% [31]) as well as abroad, for instance in the United Kingdom (‘UK biobank’ 5.5% [32]). Likewise, participation in health examination studies in other European countries (e.g., ‘FINRISK’, Finland, ‘HSE’, England, ‘DEGS’, Germany, etc.) all show a decrease in responses over the past decades [21] and this declining trend is also observed in a large serosurveillance study in America (‘NHANES’) [20]. Although the underlying reasons for this worldwide decreasing trend in study participation is not exactly known and may differ per study and country, our non-response questionnaire indicated that most persons did not have time to come to the clinic or were dreading blood collection. It has been previously reported that participation in a population-based survey including collection of blood, especially in children, is likely to be low [33, 34]. Although we tried to organize several consultation hours at centralized locations, travel time might have also been a conflicting factor for some people, especially in larger municipalities where response rates were lowest. Other sampling options, such as self-sampling finger pricks or house visits, could therefore be considered in the future to overcome some of the hurdles of reluctance to participate. This implies, however, that laboratory techniques should be suitable for analyses in low volumes of blood or other sources of material, like DBS or saliva. Last, as participants often do not receive any personal results when participating in health studies, we suspect that, in nowadays more individualized societies, there might be less willingness to contribute without direct personal benefit; hence, future studies should consider such incentives if feasible.

The response rate in CN (23.4%) was higher compared to NL. Possible explanations might be a high awareness of the study as a relatively high proportion of island residents were invited to participate, the extensive local media attention, and the presumably less individualistic island culture. Previous studies on the islands reported higher participation rates (e.g., ‘Omnibusenquête’ (2013) [35]: 40–62%; ‘Kon Salu ta…’ (‘How healthy is…’) (2002): 80–86%) [36‐38], however a completely different approach was used in these studies: house visits with multiple contact efforts, the design (solely questionnaire), and longer duration of the study, i.e., our efficient time planning limited adjustment during the study in order to increase response. Further, the population registry was not fully up-to-date and the delivery of the invitations by mail was challenging for the local postal department, which both could have negatively affected our net response. This reaffirms that logistical matters are unavoidable for these small islands and thus future studies should consider building in large(r) time margins. Moreover, population-based studies including sample collection are unfamiliar among the CN population. In order to increase awareness and response, an extensive communication plan was made tailored per island using all communication tools available. We experienced that personal appeal, repetition and promotion via key figures was most beneficial in term of response. Nevertheless, initially we might have missed some participants due to taboos towards collection of faecal material. In line with our study in NL, future studies in CN could consider introducing additional collection with ditto incentive regarding such samples to increase overall response.

In conclusion, this third Dutch biobank offers unique future research possibilities. Over 9000 blood samples and questionnaires as well as additional materials, such as saliva, oro- and nasopharyngeal swabs and faeces, have been collected. This offers the opportunity to perform thorough seroprevalence studies assessing immunity against and risk factors for (candidate) NIP-targeted diseases, taking into account different infection dynamics in the Caribbean region. Furthermore, for the first time, these data will inform us on the occurrence of and risk factors for tropical pathogens in the Caribbean region, such as zika, dengue, chikungunya, West Nile virus and yellow fever. Moreover, we will be able to connect the serological data to results from additional materials collected (via molecular typing and bacterial cultures) and environmental- and pharmacological data. Hence, we are able to gain relevant new insights into the emerging fields of the microbiome, antibiotic resistance and carriage of pathogens in relation to vaccination responses, allergies, environmental- and lifestyle factors [39]. Besides the extensive data collection, the vast majority of participants gave consent for participation in a potential follow-up study enabling a nested-case cohort study in the future. Summarized, this large biobank forms a great base for research in the field of infectious disease epidemiology and its changing dynamics and consequently, this knowledge can guide future public health policy in NL and CN.