Health Survey as of 2014

What does the survey comprise?

Purpose

To provide the most complete picture possible of developments in the health, medical contacts, lifestyle and preventive behaviour of the Dutch population.

Survey population

Persons aged 0 years and over living in private households. For some topics, other age limits apply.

Statistical unit

Persons.

Year survey started

Statistics Netherlands has been conducting an annual Health Survey since 1981. In the period 1997-2009, the Health Survey was part of the Continuing Survey on Living Conditions (POLS). As of 2010, the Health Survey is being conducted as an independent survey again.

Information on previous editions of the Health Survey (2010-2013) can be found here:

Health Survey, 2010-2013

As of 2014, there has been a partnership regarding the lifestyle aspects within the Health Survey, the Lifestyle Monitor. Various parties, focusing on lifestyle, are working together. In addition to Statistics Netherlands, these are the National Institute for Public Health and the Environment (RIVM), Trimbos-instituut, Rutgers, Sao Aids Nederland, VeiligheidNL and Voedingscentrum.In years that the European Health Survey, EHIS, is conducted, the EHIS-questions are included in het Health Survey. Statistics Netherlands provides the relevant data to Eurostat. This was the case in 2014 and 2019. The next edition of EHIS will take place in 2025

Frequency

Annual.

Publication strategy

The figures are definitive.

How is the survey conducted?

Sampling technique

A representative sample of persons is selected from the Personal Records Database (BRP). Sampling is distributed evenly over all months of the year.

Observation method

As of 2014, the Health Survey is conducted in the so-called ‘mixed mode’ setup. Sample units from the population are asked to participate via the internet (CAWI – Computer-Assisted Web Interviewing). Non-respondents are reapproached to participate in a face-to-face interview by way of Computer-Assisted Personal Interviewing (CAPI).

As of 2018, a new target group strategy is being adopted. This means that not all reapproachable CAWI non-respondents are actually reapproached for a CAPI interview. In cases where a particular target population has yielded a relatively high response rate in CAWI, a smaller sample of this target population is being reapproached for CAPI. In this way, the CAPI mode is used more selectively with special focus on the groups with low CAWI response rates. The final total response will thus better reflect the total population. The target groups have been determined based on the background characteristics on which information was available during sampling. Examples of background characteristics include age, income and origin country.

With effect from 2021, the target group strategy has been further optimized, so that CAPI was used even more selectively, attempting to get the response chances per target group as equal as possible.

Since 2018 so called incentives are being used. This means that respondents have a chance to win a prize. This is described in the invitation letter. The purpose of this is to increase the response.
The questionnaire contains questions of a personal nature, CAPI respondents are given the opportunity to complete those parts of the questionnaire by themselves (CASI – computer-assisted self-interviewing).

In 2020 and 2021, the intended observation method could not be fully implemented. As a result of the corona epidemic and the associated measures, face-to-face interviewing was disrupted. This form of data collection was nog possible in a few months and was only possible in a limited extend in a few other months. More about this can be found in the section ‘weighting’
Respondents

Persons living in private households. For respondents under the age of 12, the survey questions are answered by a parent or caretaker.

Sample size

Each year, the aim is to have a net sample (ie the ultimate number of respondents) of at least 9,500 people.

The response rate in the years 2014 to 2017 was 60 to 65 percent. The gross sample (the number of people approached) was therefore approximately 15 thousand in those years. The response percentage has decreased by introducing the target group approach in 2018, due to the selectively reapproaching this new approach included. After all, the aim was not to achieve the highest possible response rate, but to get the response rates per target group as equal as possible. In 2023, 19,373 people were contacted and the net response was 9,531 (response rate 49.2).

In the years 2020 and 2021, the net response was lower than the target (just over 8,700 and just over 8,400 respectively). This was a result of corona measures, which disrupted the observation process.

Checking and correction methods

Every year this sample survey is checked for plausible results, internal consistency and completeness.

Weighting

To correct for differences between the composition of the net sample and the total population, a weighting factor is applied based on the following characteristics: gender, age, migration background, marital status, urban environment, province, province, household size, income, wealth, and survey season, and as of 2018, target group.In 2020 and 2021, the observation of the Health Survey was disrupted by corona(measures). An attempt has been to correct for the partial loss of face-to-face observations. More about this can be found in the Toelichting berekening kwartaal- en jaarcijfers Gezondheidsenquête 2020 (in Dutch only).

What is the quality of the results?

Accuracy

The health survey entails a sample, so the figures are subject to chance fluctuations. Standard margins can be used to determine reliability margins around the figures. The lower and upper limits of the 95% confidence interval are shown in the StatLine tables.

If the number of respondents for a given classification category is less than 100, no results are presented due to oversized margins.

Comparibility over the years

Some caution should be taken into account when comparing results as from 2014 with results over the period up to and including 2013. Between 2013 and 2014, several significant changes were introduced in the Health Survey.

Up to and including 2013, the survey consisted of two research components: the first part of the questionnaire allowed respondents to complete the survey via the internet, telephone or personal interview. The second part of the questionnaire always had to be done via the internet or on paper.

However, as of 2014 the survey has consisted of only one part. Respondents can only complete the entire questionnaire via the internet or personal interview. In addition, changes were made to the structure and composition of the questionnaire from 2014 onwards, and in some cases there have been changes to the question semantics. Furthermore, the weighting of the Health Survey has been modified.

The changes between 2013 and 2014 led to a number of method contraventions, as shown by a method breakdown analysis. More information on this analysis and the established method breaks can be found here in Dutch: Methodebreuk Gezondheidsenquête 2014.

As a result of corona measures, the size and composition of the response file on 2020 was different than in other years. This has been corrected for using time series models, see paragraph ‘weighting’. Nevertheless, an impact on the results cannot be completely ruled out.

The introduction and optimization (in 2021) of the target group approach (see: Observation method) has influenced the figures for some outcome variables. Additional research has been conducted on this by CBS, RIVM and Trimbos Institute, in which the observation strategy of 2021 was simulated on the data from 2014 to 2019. In this way it was possible to estimate what the outcomes would have been for some core variables for those years if the 2021 observation strategy had already been applied in those years.

For the core variables on non-prescription drug use, smoking, daily smoking, excessive alcohol consumption and compliance with the physical activity guidelines, differences were found between the published and simulated outcomes in some years. These differences were mainly found in the years 2014 to 2017. A memorandum has been written about the additional analyses, in which it is advised to continue to use the published figures form previous years.

More about this analysis and the results can be found in the memorandum (in Dutch): Dataverzamelingsproces Gezondheidsenquête/Leefstijlmonitor 2014-2021 

Downloads

Related items