eTable 2. Overview of measurement and data acquisition methods and quantitative variables used, by period of data acquisition.

	Time periods of studies
	1990–1992 (OW91)	1997–1999 (BGS98)	2008–2011 (DEGS1)
Measurement of biomarkers
Blood pressure
Sphygmomanometer	Hawksley random-zero sphygmomanometer	mercury sphygmomanometer (Erkameter 3000)	Datascope Accutorr Plus
Cuffs	12 × 23 cm (arm circumference<20 cm); 12 × 28 cm (arm circumference 20–40 cm); 14 × 40 cm (arm circumference >40 cm)	8 × 20 cm (arm circumference<20 cm); 12 × 28 cm (arm circumference 20–40 cm; 14 × 40 cm (arm circumference >40 cm)	10.5 × 23.9 cm (arm circ. 21.0–27.9 cm); 13.5 × 30.7 cm (arm circ. 28.0–35.9 cm); 17.0 × 38.6 cm (arm circ. 36.0–46.0 cm)
Number of measurements	2	3	3
Measurement used to compare 1990–1992, 1997–1999 and 2008–2011	systolic blood pressure (mm Hg), second measurement	systolic blood pressure (mm Hg), second measurement	systolic blood pressure (mm Hg), second measure‧ment
Measurement used in publications to date	mean of the first and second measurements	mean of the second and third measurements	mean of the second and third measurements
Spyhgmomanometer- and cuff-related bias	underestimation of the systolic blood pressure by up to 3 mmHg with the random-zero method, but overestimation of the blood pressure through use of the medium cuff with arm circumfrences up to 40 cm, of the order of 3 mmHg (mean) according to sensitivity analyses of the population mean	obsolete cuff rule for large arm circumferences; calibration of values for comparison with Datascope in order to take sphygmomanometer- and cuff-related differences into account	the cuffs meet the criteria of current guidelines; the sphygmomanometer meets the validation ‧criteria of international specialty societies
Serum cholesterol, total
Measuring device	SMAC (Technicon Corporation, Tarrytown, NY, USA)	MEGA (Merck, Germany)	Architect ci2800 (Abbott, Germany)
Method of analysis (parameter)	cholesterinoxidase-peroxidase-4-aminophenazone-phenol	cholesterinoxidase-peroxidase-4- aminophenazone-phenol	cholesterinoxidase-peroxidase-4-aminophenazone-phenol
Parameter used to compare 1990–1992, 1997–1999 and 2008–2011	mean serum total cholesterol level in mmol/L	mean serum total cholesterol level in mmol/L	mean serum total cholesterol level in mmol/L
Biases due to device and method of ‧analysis	same method of analysis; the two changes of measuring instrument probably had little effect on the measured values (11) the varying percentage of fasting subjects across surveys probably had a negligible effect (conclusion from adjusted and stratified analyses)
Serum glucose
Measuring device	SMAC (Technicon Corporation, Tarrytown, NY, USA)	MEGA (Merck, Germany)	Architect ci2800 (Abbott, Germany)
Method of analysis (parameter)	glucose-oxidase-peroxidase-4-aminophenazone-phenol	glucose-oxidase-peroxidase-4-aminophenazone-‧phenol	hexokinase
Parameter used to compare 1990–1992, 1997–1999 and 2008–2011	mean serum glucose in mmol/L	mean serum glucose in mmol/L	mean serum glucose in mmol/L
Biases due to device and method of ‧analysis	The percentage of fasting subjects (≥ 8 hours before blood drawing) rose from each survey to the next, from 9.5% in the OW91 to 26.6% in the BGS98 and 49.0% in the DEGS1. This may have led to an overestimation of the mean serum glucose concentration in the OW91 compared to the DEGS1. Sensitivity analysis: consideration of the fasting time as an independent categorical variable (<4 hr, ≥ 4 to<8 hr, ≥ 8 to <10 hr, ≥ 10 hr) in the models for estimating differences in trends yielded similar trends to those obtained when fasting time was not considered. this was true of all serum glucose trends presented here, except among men in western germany: in this subgroup, consideration of the fasting time altered the p-value for differences in trends between the two periods 1990–1992 and 2008–2011 from significant (p<0.001; unadjusted mean glucose values of 5.56 mmol/l with 95% ci [5.48; 5.63] in the ow91 vs. 5.30 mmol/l [5.21; 5.39] in the degs1) to insignificant (p = 0.188; adjusted mean glucose values of 5.38 mmol/l [5.29; 5.46] in the ow91 vs. 5.29 mmol/l [5.18; 5.39] in the degs1).
Weight and height
Measuring device	weight without shoes, lightly dressed: measurement accuracy 0.1 kg height: measurement accuracy 0.5 cm	weight without shoes, lightly dressed: SECA electronic scale, measurement accuracy 0.1 kg height: yardstick built into SECA scale, measurement accuracy 0.1 cm	weight without shoes, in underwear: SECA electronic column scale 930, measurement accuracy 0.1 kg height: portable stadiometer (Holtain Ltd./UK), measurement accuracy 0.1 cm
Method of analysis (parameter)	BMI: weight/height² (in kg and m, respectively)	BMI: weight/height² (in kg and m, respectively)	BMI: weight/height² (in kg and m, respectively)
Parameter used to compare 1990–1992, 1997–1999 and 2008–2011	BMI ≥ 30 kg/m² (WHO obesity threshold)	BMI ≥ 30 kg/m² (WHO obesity threshold)	BMI ≥ 30 kg/m² (WHO obesity threshold)
Biases due to device and method of analysis			change in instruction to subjects (from lightly clothed to underwear only)
Questioning methods
Self-reported high blood pressure
Questions	self-administered questionnaire of the NUST2: Do you now have, or have you ever had, any of the following ‧diseases? (I have it now / I don�t have it any more / I don�t know whether I still have it / I have never had it) high blood pressure, hypertension self-administered questionnaire of the East-West Survey: Which of the following diseases have you ever had? (if yes, then: only in the past 12 months / in the past 12 months and previously / treated with drugs) high blood pressure, hypertension	(1) self-administered questionnaire: Which of the following diseases have you ever had? (yes / no / I don�t know) high blood pressure, hypertension (2) CAPI Was the subject ever given a diagnosis of any of the following diseases or health disturbances by a physician? (yes/no) high blood pressure, hypertension	CAPI Was your blood pressure ever found to be elevated or too high? (yes / no / I don�t know) Subjects were told only if they asked that a sys tolic blood pressure above 140 mmHg or a diastolic blood pressure above 90 mmHg is considered to be elevated.
Variables used to compare 1990–1992, 1997–1999, and 2008–2011	high blood pressure (ever): yes/no	high blood pressure (ever): yes/no (CAPI)	high blood pressure (ever): yes/no (CAPI)
Reporting bias	The mode of questioning changed from a self-administered questionnaire (OW91, BGS98) to a CAPI (BGS98, DEGS). Any ensuing reporting bias was probably small, as there was no significant difference between the self-reported and CAPI-derived prevalences in the BGS98. Sensitivity analysis: comparison of self-reported and CAPI-derived prevalences (weighted and age-standardized) in the BGS98 (complete-case sample, 5385 subjects): (1) CAPI: overall 22.7 [21.2; 24.2]; men 22.5 [20.6; 24.6]; women 22.8 [20.9; 24.8] (2) Self-reported: overall 22.6 [21.3; 24.0]; men 22.0 [20.2; 24.0]; women 23.1 [21.4; 25.0]
Self-reported hyperlipidemia
Questions	self-administered questionnaire of the NUST2: Do you now have, or have you ever had, any of the following diseases? (I have it now / I don�t have it any more / I don�t know whether I still have it / I have never had it) high cholesterol, high lipids self-administered questionnaire of the East/West Health Survey:   Which of the following diseases have you ever had? (if yes, then: only in the past 12 months / in the past 12 months and previously / treated with drugs) high cholesterol, high lipids	(1) self-administered questionnaire: Which of the following diseases have you ever had? (yes / no / I don�t know) high cholesterol, high lipids (2) CAPI Was the subject ever given a diagnosis of any of the following diseases or health disturbances by a physician? (yes/no) high cholesterol, high lipids	CAPI Did a doctor ever tell you that you had a disorder of lipid metabolism? This term refers to high levels of fatty substances such as cholesterol or triglycerides. (yes / no / I don�t know)
Variables used to compare 1990–1992, 1997–1999, and 2008–2011	hyperlipidemia (ever): yes/no	hyperlipidemia (ever): yes/no (CAPI)	hyperlipidemia (ever): yes/no (CAPI)
Reporting bias	The mode of questioning changed from a self-administered questionnaire (OW91, BGS98) to a CAPI (BGS98, DEGS). Any ensuing reporting bias was probably small, as the self-reported prevalence in the BGS98 was not significantly higher than the CAPI-derived prevalence. If a correction were to be made for the observed difference, the prevalence for 1990–1992 would have to be adjusted downward, and this would only reinforce the observed trend. Sensitivity analysis: comparison of self-reported and CAPI-derived prevalences (weighted and age-standardized) in the BGS98 (complete-case sample, 4775 subjects): (1) CAPI: overall 27.6 [25.8; 29.4]; men 30.1 [27.8; 32.5]; women 25.1 [22.8; 27.6] (2) Self-reported: overall 28.6 [26.9; 30.3]; men 31.4 [29.0; 33.9]; women 25.8 [23.6; 28.1]
Self-reported diabetes mellitus
Questions	self-administered questionnaire of the NUST2: Do you now have, or have you ever had, any of the following diseases? (I have it now / I don�t have it any more / I don�t know whether I still have it / I have never had it) high blood sugar, diabetes self-administered questionnaire of the East/West Health Survey: Which of the following diseases have you ever had? (if yes, then: only in the past 12 months / in the past 12 months and previously / treated with drugs) high blood sugar, diabetes	(1) self-administered questionnaire: Which of the following diseases have you ever had? (yes / no / I don�t know) high blood sugar (diabetes mellitus) treated with insulin high blood sugar (diabetes mellitus) not treated with insulin (2) CAPI Was the subject ever given a diagnosis of any of the following diseases or health disturbances by a physician? (yes/no) high blood sugar (diabetes mellitus) treated with insulin high blood sugar (diabetes mellitus) not treated with insulin	CAPI Did a doctor ever tell you that you had high blood sugar or diabetes?. (yes / no / I don�t know)
Variables used to compare 1990–1992, 1997–1999, and 2008–2011	diabetes (ever): yes/no	diabetes (ever): yes/no (CAPI)	diabetes (ever): yes/no (CAPI)
Reporting bias	The mode of questioning changed from a self-administered questionnaire (OW91, BGS98) to a CAPI (BGS98, DEGS). Any ensuing reporting bias was probably small, as the self-reported prevalence in the BGS98 was not significantly higher than the CAPI-derived prevalence. If a correction were to be made for the observed difference, the prevalence for 1990–1992 would have to be adjusted downward, and this would only reinforce the observed trend. Sensitivity analysis: comparison of self-reported and CAPI-derived prevalences (weighted and age-standardized) in the BGS98 (complete-case sample, 5638 subjects): (1) CAPI: overall, 4.58 [3.94; 5.33]; men 4.86 [4.03; 5.84]; women 4.31 [3.50; 5.29] (2) Self-reported: overall 4.93 [4.24; 5.75]; men 5.39 [4.46; 6.50]. women 4.48 [3.68; 5.45]
Lack of exercise
Questions	How often do you engage in physical exercise? regularly, more than 2 hours per week regularly, 1 to 2 hours per week less than one hour per week not at all	How often do you engage in physical exercise? regularly, more than 4 hours per week regularly, 2 to 4 hours per week regularly, 1 to 2 hours per week less than one hour per week not at all	How often do you engage in physical exercise? not at all less than one hour per week regularly, 1 to 2 hours per week regularly, 2 to 4 hours per week regularly, more than 4 hours per week
Variables used to compare 1990–1992, 1997–1999, and 2008–2011	no exercise: yes/no	no exercise: yes/no	no exercise: yes/no
Reporting bias	The social desirability of exercise may have led to an underestimation of the prevalence of lack of exercise. Analyses have shown that self-reported information about exercise overestimates the amount of exercise taken in comparison to objective measures (accelerometry or the doubly-labeled water technique) (12, 13). It is hard to estimate the effect, if any, of the change in the number and order of answer categories on the comparability of data across surveys.
Smoking status
Questions	Have you ever smoked, or do you smoke now? I used to smoke but don�t any more. I smoke now. I have never smoked.	Have you ever smoked, or do you smoke now? I have never smoked. I smoke now. Yes, every day. Yes, occasionally. I used to smoke. I have quit smoking in the past 12 months (…).	Do you smoke at all at present, even in small amounts? Yes, every day. Yes, occasionally. No, not any more. I have never smoked.
Variables used to compare 1990–1992, 1997–1999, and 2008–2011	current smoker: yes/no	current smoker: yes/no	current smoker: yes/no
Reporting bias	The social undesirability of smoking may have led to an underestimation of its prevalence. Analyses have shown that self-reported information about smoking underestimates the amount of smoking in comparison to objective measures (urine cotinine concentration). Current smokers sometimes say they are ex-smokers and are misclassified as such (14). It is hard to estimate the effect, if any, of the changes in the questions and answer categories on the comparability of data across surveys.
Determination of drug use
Antihypertensive drugs
Data acquisition	standardized interview to determine drug use in the past 7 days	standardized interview to determine drug use in the past 7 days	standardized interview to determine drug use in the past 7 days
Drug coding ATC code ATC code of the WHO for BGS98 ATC code of the WIdO for DEGS1 Recoding of the EPhMRA code in the ATC code for the OW91	antihypertensive drugs (ATC codes C02) or diuretics (ATC code C03) or β-adrenoreceptor antagonists (ATC code C07) or calcium-channel blockers (ATC code C08) or drugs affecting the renin-angiotensin system (ATC code C09)	antihypertensive drugs (ATC codes C02) or diuretics (ATC code C03) or β-adrenoreceptor antagonists (ATC code C07) or calcium-channel blockers (ATC code C08) or drugs affecting the renin-angiotensin system (ATC code C09)	antihypertensive drugs (ATC codes C02) or diuretics (ATC code C03) or β-adrenoreceptor antagonists (ATC code C07) or calcium-channel blockers (ATC code C08) or drugs affecting the renin-angiotensin system (ATC code C09)
Indicator used to compare 1990–1992, 1997–1999 and 2008–2011	use of antihypertensive drugs: yes/no	use of antihypertensive drugs: yes/no	use of antihypertensive drugs: yes/no
Recall bias	Underestimation of the prevalence of drug use because of recall bias cannot be ruled out.
Cholesterol-lowering drugs
Data acquisition	standardized interview to determine drug use in the past 7 days	standardized interview to determine drug use in the past 7 days	standardized interview to determine drug use in the past 7 days
Drug coding ATC code ATC code of the WHO for BGS98 ATC code of the WIdO for DEGS1 Recoding of the EPhMRA code in the ATC code for the OW91	drugs affecting lipid metabolism (ATC code C10)	drugs affecting lipid metabolism (ATC code C10)	drugs affecting lipid metabolism (ATC code C10)
Indicator used to compare 1990–1992, 1997–1999 and 2008–2011	use of lipid-lowering drugs: yes/no	use of lipid-lowering drugs: yes/no	use of lipid-lowering drugs: yes/no
Recall bias	Underestimation of the prevalence of drug use because of recall bias cannot be ruled out.
Antidiabetic drugs
Data acquisition	standardized interview to determine drug use in the past 7 days	standardized interview to determine drug use in the past 7 days	standardized interview to determine drug use in the past 7 days
Drug coding ATC code ATC code of the WHO for BGS98 ATC code of the WIdO for DEGS1 Recoding of the EPhMRA code in the ATC code for the OW91	antidiabetic drugs (ATC code A10)	antidiabetic drugs (ATC code A10)	antidiabetic drugs (ATC code A10)
Indicator used to compare 1990–1992, 1997–1999 and 2008–2011	use of antidiabetic drugs: yes/no	use of antidiabetic drugs: yes/no	use of antidiabetic drugs: yes/no
Recall bias	Underestimation of the prevalence of drug use because of recall bias cannot be ruled out.

ATC code, anatomic-therapeutic-chemical code; BMI, body-mass index; BGS98, German National Health Interview and Examination 1998; CAPI, computer-assisted telephone interview; CI, confidence interval; DEGS1, German Health Interview and Examination ?Survey for Adults 2008–2011; EPhMRA, European Pharmaceutical Market Research Association; NUST2, German National Examination Survey 1990; OW91, East/West (German) Health Survey 1991; WHO, World Health Organization; WIdO, Research Institute of the Local Health Insurers in Germany.