German health interview and examination survey for adults (DEGS) - design, objectives and implementation of the first data collection wave

DEGS is primarily designed as a periodically repeated national health interview and examination survey of adults in Germany. The target population comprises adults 18–79 years of age with permanent residence in Germany according to local population registries. In order to lay the ground for longitudinal studies, persons who had participated in the 1998 national health interview and examination survey (GNHIES98) were invited to take part in DEGS1, provided they had agreed to be recontacted and were still contactable [6]. Representativeness of DEGS1 as a cross-sectional survey is maintained by extending the population sample based on a two-stage sampling procedure according to power analyses and sample size estimations as described below.

The implementation of DEGS1 conforms to the principles outlined in the Declaration of Helsinki [7] and to the German Federal Data Protection Act. The DEGS1 study protocol was consented with the Federal and State Commissioners for Data Protection and approved by the Charité-Universitätsmedizin Berlin ethics committee in September 2008 (No. EA2/047/08). Participants provided written informed consent prior to the interview and examination.

Overall 17410 persons, of whom N = 6402 or 37% were GNHIES98 participants, were invited to participate in DEGS1. After excluding non-deliverable survey contacts, persons unable to provide written informed consent and persons with insufficient German language skills, the overall net sample consisted of 16305 adults, of whom N = 6358 or 39% were GNHIES98 participants. Overall, 8152 adults (4283 women; 3869 men) participated in DEGS1. Total numbers of the study population and response rates are summarized by age group and GNHIES98 participation status in Table1. Response rates were consistently higher among GNHIES98 participants compared to newly sampled individuals.

Both interview and examination data are available for 89% of the study population (N = 7238; 3765 women, 3473 men). The remainder of the study population (N = 914; 518 women, 396 men) have interview data only, because they were unable or unwilling to come to one of the local study centers. Cross-sectional and trend analyses will be restricted to 7988 DEGS1 participants 18 to 79 years of age, with examination as well as interview data available for 7116 of these. Longitudinal studies will include a total of 3959 persons 28–91 years of age who participated in DEGS1 as well as in the GNHIES98.

The DEGS1 sampling protocol was developed in co-operation with the Leibniz Institute for the Social Sciences (GESIS), Mannheim, Germany [8]. Like previous German health interview and examination surveys [1, 2, 4] DEGS1 is based on two-stage stratified cluster sampling. Primary sampling units (PSUs) are the communities. PSUs were sampled from a list of German communities stratified according to districts and the BIK classification system, which takes into account the grade of urbanization, regional population density, and administrative borders [9]. Sampling was done with probability proportional to community size using the Cox procedure for controlled rounding [10]. Selected metropolitan areas were generally represented by several PSUs. Selected communities with less than 1000 inhabitants were combined with neighboring small communities to form a single PSU. Within PSUs, random samples of individuals, stratified by 10-year age groups, were drawn from local population registers.

Since DEGS1 included GNHIES98 participants, the DEGS1 PSU sample had to build upon a total of 120 PSUs previously sampled for the 1998 survey. In order to remain representative at the population level, additional PSUs (N = 60) were sampled. In newly added PSUs, all participants were newly sampled. Within GNHIES98 PSUs, only newly recruited individuals are included in the age group 18–28 years due to the aging of the GNHIES98 cohort by 10–14 years. Persons 30 years and older were sampled as necessary, in order to replace the number of participants expected to decline participation in DEGS1 or to be lost to follow-up. The number of newly sampled PSUs and individuals per PSU was determined by statistical power and sample size considerations as described below.

Given the cluster design, simple variance estimates of prevalence estimates will underestimate total variability and hence estimation error, because individuals within PSUs can be assumed to show greater similarities than individuals randomly selected from the entire population of Germany. The increase in estimation error is quantified by the design effect (Deff). Total sample size needs to be increased in proportion to Deff to attain the same level of precision as in simple random sampling [11]. Deff depends on the average number of individuals within PSUs (m) and the degree of similarity within clusters which is estimated by the intra class correlation coefficient (ICC): Deff = 1 + ICC(m-1) [12]. It is therefore more efficient to increase the total number of PSUs rather than m, in order to keep the total sample size as small as possible. Reasonably precise estimation of a sex and 10-year-age group–specific prevalence of 1% (95% CI to be attained: 0.5–2.1%) would require a total sample size of about 7500. In this calculation, Deff is ignored as the number of persons in a PSU within any sex-age-group can be expected to be very small. Analyses of GNHIES98 data showed that ICC can be assumed to take on values between 0.005 and 0.036. Based on the experience that survey logistics permit the examination of 40–50 persons per PSU, design effects would be 4–15% higher with 50 instead of 40 persons seen per PSU. The required precision could most efficiently be reached based on a total number of 180 PSUs and an average of m = 42 persons seen per PSU, yielding a total sample size of 180 × 42 = 7560 [8]. This sample size also proved to provide 90% power to detect a 3% change in overall obesity prevalence in analyses for time trends at the 5% significance level based on two-sided tests [13].

Eligible persons were invited to participate in the survey by letter. An information booklet illustrated the study objectives, procedures and logistics. A response card was enclosed to fill in telephone numbers and preferred contact times. Participants were offered a small monetary incentive.

Invitation letters were sent approximately five weeks prior to the survey visit. In order to optimize response rates, survey activities were announced in local newspapers, and if possible, on local television and radio stations. In addition, small survey reports were published at regular intervals in national medical and public health journals. Persons willing to participate were contacted by telephone for scheduling and further information. A confirmation letter including the time of the appointment, location plan, and preparatory instructions was sent thereafter. Persons with appointments in the morning were asked to keep an overnight fast and not eat or drink anything but water before blood testing, unless they were diabetic. Persons with afternoon appointments were asked to fast for at least four hours, unless they were diabetic. Participants were asked to bring along immunization records as well as the original containers of all medications (prescription and over-the-counter) used within 7 days prior to the survey appointment.

Personal contact was sought to persons who had not responded to the invitation letter within four weeks. Exclusion criteria (such as insufficient language proficiency or inability to provide informed consent) were recorded. Persons willing to participate but facing organizational problems (e.g. lack of transportation, time constraints) were offered transportation and flexible scheduling, such as appointments early in the morning or in the evening.

Fieldwork in DEGS1 extended over three years. Data collection started on November 25, 2008 and ended on November 26, 2011. PSUs were successively visited by two mobile study teams according to a random touring schedule, in order to avoid a systematic bias of study results by seasonal or time trends. Duration of stay at a particular PSU was limited to one week. In order to include a sufficiently high number of study participants per PSU, eligible persons had to be scheduled for appointments ranging from 7:30 a.m. to 6:00 p.m. during the week and from 7:30 a.m. to 10:30 a.m. on Saturdays.

Study teams consisted of specifically trained health professionals, including a study physician and three technicians. Interviews and examinations took place in community-owned facilities that were rented for the purpose of the survey.

Data collection in DEGS1 was based on: (1) a semi-quantitative self-administered food frequency questionnaire (FFQ); (2) automated assessment of currently used medications (AmEDa); (3) collection of urine and blood samples; (4) standardized measurements and physical performance tests (M/T); (5) a standardized physician-administered computer-assisted personal interview (CAPI); (6) a standardized self-filled questionnaire (SFQ).

The FFQ questionnaire was mailed to study respondents along with the confirmation letter. It included instructions and examples for completion. Participants were asked to take the completed questionnaire to the survey appointment.

AmEDa as well as standardized measurements and tests were conducted by study technicians. Unique product identifiers (Pharmazentralnummer, PZN) on original drug containers brought to the survey site were scanned and automatically coded according to the latest version of the WHO Anatomical Therapeutic Chemical (ATC) Classification System [14]. Information on medication source (doctor’s prescription, over-the–counter, family medicine cabinet), indication, dose, frequency and duration of use, and intake within the past 24 hours was documented. Missing information due to incomplete matching (about 27% of scanned drugs) or missing original containers (about 1.5% of participants) was completed by telephone follow-up contacts and hand searching.

Body weight and body height were measured in underwear with shoes removed. Waist and hip circumferences were measured according to a standardized protocol. Three consecutive automated blood pressure measurements were taken on the right arm three minutes apart with participants sitting and having been at rest for five minutes.

Spot urine samples were collected and dipstick tested on site. Venous blood samples were collected using Vacutainer EDTA and gel tubes. The time of blood draw and number of hours since last meal were recorded. Participants who came in fasting were served a breakfast or snack thereafter. Blood was centrifuged and separated. Serum specimens as well as urine specimens were aliquoted and stored at −40°C within one hour. EDTA whole blood tubes were shaken and kept in the original collection tubes at 4°C. A 100 μl whole blood aliquot intended for folate analysis was hemolyzed under light protection and stored at −40°C. A full blood count was performed on site. Additional whole blood, serum and urine specimens were transported by car at −40°C at the end of each survey week for further analysis and storage at the central epidemiology laboratory unit at the Robert Koch Institute, Berlin. Clinical chemistry and allergic sensitization analyses were performed within 6–8 weeks. Specific analyses and confirmatory tests relating to the diagnosis of infectious diseases were commissioned to reference laboratories. Extra serum, urine and whole blood aliquots were stored at −40°C.

Applying the CAPI, study physicians obtained a detailed medical history including family history. The interview was administered in German language, hence participants had to be sufficiently proficient in speaking and understanding German. Self-administered questionnaires filled in at the survey site were available in different languages (German, Turkish, Serbo-Croatian, Russian, English). Foreign language versions were offered to participants with German as a foreign language who were able to communicate in German, but had limited capacities for reading and writing in German. Different versions of the questionnaire were used for the population of working age (up to 64 years of age) and for persons 65 years and above. A short version of the self-administered questionnaire restricted to core questions was offered to those who lacked the strength or time to answer the full-length questionnaire.

Completing the entire survey program required an average of three hours. GNHIES98 participants who had moved outside their PSU and were unable to come to the survey site were asked to complete a computer-assisted telephone interview (CATI) covering CAPI questions and to answer a self-administered postal questionnaire.

DEGS1 aims to analyze health status, health risks and resources, functional capacity levels, and disability in the German adult population. Prevalence estimates will be obtained and time trends will be analyzed in comparison with data from the GNHIES98. Changes in health status and risk factors over time at the individual level will be studied among DEGS1 participants who already took part in the GNHIES98. Major study objectives include:

As summarized in Table2, constructs and instruments applied in DEGS1 were kept compatible to those used in the GNHIES98 as far as possible. The CAPI covered a total of 32 health conditions including hepatitis (A, B, C, E), prevalent chronic diseases and cardiovascular events (stroke, myocardial infarction). Participants were asked whether a particular health condition was ever diagnosed by a physician, and if so, at what age or in what year the diagnosis was first made, and whether the condition required any current medical treatment or medication use. Persons with a lifetime history of a particular disease were asked, whether the disease was still present during the 12 months preceding the interview. Disease specific information was also obtained.

In addition to self-reported morbidity, detailed information on current medication surveyed by AmEDa, and various objective health measures will contribute to analyze health status and health risks in the adult population. Infectious diseases and immunization status will be evaluated based on self-reported childhood infections and vaccinations, immunization records, and laboratory test results. These include serum concentrations of antibody titers against several viruses and bacteria as well as nucleic acid amplification testing for confirmatory diagnosis of hepatitis B and C in serum and diagnosis of gonorrhea and chlamydia infections in urine. Medication use and objective health measures will help to validate self-reported chronic health conditions, to identify previously unknown disease and hidden health risks, and to analyze patterns of co- and multimorbidity as well as aspects of treatment effectiveness and quality of care. A list of chronic health conditions covered in DEGS1 and information available from interview and examination is given in Table3.

Information on other facets of health, such as somatic symptoms and complaints, various aspects of mental health, and subjective health was collected via self-administered questionnaires based on standardized instruments (Table2). Self-administered questionnaires also served to collect detailed information on reproductive history among women, a history of benign prostate disease among men, injuries, and among persons 65 years and older, a history of fractures and falls as well as fear of falls.

Functional capacities were assessed by standardized questionnaires (vision and hearing impairment) and tests. Aerobic endurance fitness tests using submaximal endurance bicycle ergometry were restricted to the study population 18–64 years. Exclusion criteria were defined using the translated version of the Physical Activity Readiness Questionnaire (PARQ) [15]. Among persons 65 years of age and older, functional capacities relevant to daily life activities were assessed (Table2). Isometric handgrip strength in both hands was measured while participants were standing upright using the Smedley S dynamometer. A simple test sequence was used to assess balance (Romberg test, semi-tandem, tandem, one-leg stand), lower leg muscle strength and coordination (five-chair rise test), and mobility (Timed “Up & Go”). Cognitive functioning was assessed using the digit symbol substitution test (DSST) from the authorized German version of the Wechsler Adult Intelligence Scale, 3rd revision (WAIS-III).

Health risks and resources covered in DEGS1 include potentially modifiable behavioral risk factors (e.g. dietary habits, tobacco use, and risky alcohol consumption), environmental factors (e.g. living and working conditions, social support, lay caregiving), socio-demographic context variables, and health care services utilization. The FFQ was self-developed and designed to estimate the frequency and amounts of 53 food groups consumed during the past 4 weeks. It was validated against two non-consecutive 24 hour recalls and showed an overall reasonable validity [16]. Compared to the FFQ applied in the GNHIES98, the DEGS1 FFQ has more food items and includes estimation of portion sizes, which permits a more quantitative estimation of food intake. The instruments also cover a different time frame. Alcohol consumption and tobacco use were assessed by self-developed questions already applied in the GNHIES98. Validated German versions of established instruments were additionally applied in DEGS1 to identify persons with risky drinking behavior or nicotine dependence. Socio-demographic variables were assessed according to standards recommended by the German Epidemiology and Preventive Medicine Associations [17]. Education questions meet the International Standard Classification of Education (ISCED) [18].

A separate mental health study module (DEGS1-MH) was implemented for in depth assessment of mental health disorders, help-seeking behavior and cognitive function. The data were collected using computer-assisted face-to-face interviews administered by specifically trained psychologists. DEGS1 participants were asked for their consent to be recontacted for a study center or at home appointment within four weeks after the basic DEGS1 clinic visit.

The core instrument of DEGS1–MH is a computer-assisted German version of the Munich Composite International Diagnostic Interview (DIA-X/M-CIDI) [19], a modified and translated version of the World Health Organization Composite International Diagnostic Interview (CIDI) [20]. The CIDI is a fully structured interview for the assessment of mental disorders according to the definitions and criteria of ICD-10 and DSM-IV. It consists of several sections tailored to assess mood disorders (major depression, dysthymic disorder, bipolar disorder); anxiety disorders (panic disorder, specific phobias, agoraphobia, generalized anxiety disorder, social phobia); post-traumatic stress disorder; obsessive-compulsive disorder; substance use disorders (alcohol abuse and dependence, abuse and dependence of pharmaceuticals, nicotine dependence); eating disorders; somatoform disorders (pain disorders, undifferentiated somatoform disorder); psychotic symptoms (screening without further differential diagnosis); treatment of mental disorders. The CIDI is a diagnostic instrument intended for use in clinical practice as well as in clinical and epidemiological research. It is suitable for use in population-based studies to estimate the prevalence of mental disorders, to assess severity of mental disorders and associated burden of disease, and to analyze treatment patterns, treatment barriers, and aspects of quality of care [20, 21].

Cognitive function was assessed with a neuropsychological test battery that covers important cognitive domains such as episodic memory, working memory, executive function, and cognitive speed. The battery consists of established cognitive tests that were used in epidemiological studies before and were mostly derived from widely-used test batteries such as the German version of the CERAD battery [22, 23].

The DEGS1-MH is carried out in close cooperation between the Robert Koch Institute and the Institute of Clinical Psychology and Psychotherapy, Center of Clinical Epidemiology and Longitudinal Studies (CELOS) at the Technische Universität Dresden. CELOS was already commissioned to carry out the Mental Health Module of the GNHIES98 [24, 25]. This opens the perspective for analyses of changes in the prevalence of mental health problems as well as for longitudinal analyses based on individuals participating in both surveys.

Persons sampled and eligible for DEGS1 who declined participation were asked to provide the reason for non-participation. In addition, they were asked to fill in a short questionnaire covering essential health-related characteristics, such as self-rated health, history of long-standing chronic illness, smoking habits, educational background, and doctor visits. This will permit comparison of participants and non-participants with respect to age, sex, education, health status, health-related behavior, and health care services utilization.

The DEGS1 data set will comprise approximately 2000 items per study participant. Data management has been carried out in parallel with data collection based on highly standardized and partly automated procedures for data processing and plausibility checking. This will assist in providing high quality and timely data. A final set of key data was available for main pre-planned analyses 6 months after the completion of fieldwork. Data from the DEGS1 survey wave will be made available to the scientific community as a public use file in 2014.

In order to assure that estimates derived from DEGS1 are representative at the national level, two sets of sample weights were computed :one set for all participants aged 18–79 years, one set for those with examination data available. Sample weights adjust for different sampling probabilities within the design strata and correct for deviations between the design-weighted net sample and German population statistics (December 31, 2010) based on an iterative raking procedure using age, sex, federal state of residence, community BIK classification category, nationality (German yes/no), and education level according to ISCED [18] using population statistics dating from 2009. For GNHIES98 participants included in DEGS1, the weighting procedure additionally includes adjustments for the re-participation rate.

The DEGS1 study operations manual contains standard operating procedures (SOPs) for all DEGS1 interview and examination components as well as internal and external quality control measures [8]. Members of the study teams underwent an initial two-week training session leading to accreditation. They were continuously supervised and reassessed at 6-month intervals. Internal quality control was carried out by senior RKI supervising staff. Measures of internal quality assurance encompassed: regular field visits to the study teams, bimonthly reviews of logbook entries for meeting requirements of daily calibration procedures and transportation of blood and urine specimens, continuous data checks for completeness and plausibility. External quality assurance was carried out by specifically experienced scientists of the Bremen Institute for Epidemiology and Prevention Research (BIPS) based on field observation and data audits.

Participants received immediate feed-back on clinically relevant test results based on written summary reports prepared by study physicians. Within six to eight weeks, study participants received a standardized letter summarizing clinically relevant study results including clinical chemistry, test results for hepatitis, chlamydia and gonorrhea infections, specific allergy testing, body mass index, blood pressure, thyroid volume, and physical fitness. Study participants will also be informed about immunity status against poliomyelitis, measles, mumps, and rubella as test results become available.

The DEGS1 steering committee at the Robert Koch Institute welcomes collaborations with external researchers. Collaborations may relate to input of specific know-how and professional expertise to analyses that were given a particularly high level of priority and therefore pre-planned to be completed within two years after finishing fieldwork. Additional analyses will require a research proposal, which will be reviewed by the DEGS1 steering committee and the Robert Koch Institute Research Council in order to ensure data protection and privacy regulations, public health relevance, data efficiency, and scientific quality. For further information please visit the DEGS1 study homepage at http://​www.​degs-studie.​de.