Table 4.

Characteristics of studies by design, follow-up, measurement, limitations and confounding factors.

Author/Year	Study design/Follow-up period	Measurement	Limitations and confounding factors (according to the authors)
Brownson RC et al., 1996	R I C-S survey, two samples/60 months	No comprehensive information on the accuracy of ‘Behavioral Risk Factor Surveillance System’ data during the study period	Potential response bias due to a lack of phone coverage of certain sociodemographic groups. Effects of national programs: National High Blood Pressure Education Program and National Cholesterol Education Program.
Tudor-Smith C et al., 1998	R I C-S survey, two samples/60 months	Brief interview (BI) Self-completed questionnaire (CI) Response rate: I – in 1985: BI=88%, CI=67% in 1990: BI=79%, CI=61% C – in 1985: BI=84%, CI=64% in 1990: BI=77%, CI=61%	Control group sample size at baseline was too small to provide sufficient statistical power for analysis. Diffusion of Heartbeat Wales projects and programs to control group was faster and to a far great extent than initially expected. Introduction of a national program. Favorable secular trends in smoking and dietary choices.
Puska P et al., 1983	R I C-S survey, three samples/120 months	In 1982, the cuff sphygmomanometer was longer than the one used in 1972/1977.	Finnish health service system and cultural factors.
Nafziger NA et al., 2001	R I C-S survey, two samples, cohort/60 months.		Cross-contamination and testing effects. Insufficient sample size.
Carleton RA et al., 1995	R I C-S survey, six samples/102 months	Response rate: 68%	National education programs, commercial marketing programs and secular trends. Time of economic difficulty – high unemployment and low incomes.
Huot I et al., 2004	R I C-S survey, two samples/48 months	Food frequency questionnaires-restricted food list, choice of frequency and difficulty remembering foods eaten in the past as well as their quantity.	Study population not representative of the adult population of the participating communities. Secular trends. Study design. Insufficient participation rates. In urban site, activities were directed mainly toward children, and parental participation was low. Cross-contamination.
Winkleby MA et al., 1996.	R I C-S survey, three samples/108 months		Positive and negative secular trends. Health promotion activities through popular press and voluntary agencies. Advent of broad-based federal programs. Antismoking legislation and education.
Nguyen QN et al., 2012	R I C-S survey, two samples/36 months	Self-reported behavioral questionnaire – recall bias. Arterial pressure measure in one visit and weight measure influenced by the harvest cycle.	Blood tests conducted in only part of the sample due to budget constraints. Epidemiological transitional status – rural populations adopting urban lifestyles. Negative effects of globalization. Hawthorne effect.
Schuit AJ et al., 2006	Cohort/60 months	Data influenced by seasonalityResponse rate: 80%	Study individuals were involved in previous monitoring studies. Age-associated changes in cohort design. Study not blinded.
Wanda Wendel-Vos GC et al., 2009	Cohort/60 months	Response rate: 80%	Study individuals were involved in previous monitoring studies. Measurements in C group were conducted over a longer time span than those of I group. Study not blinded.
Kottke TE et al., 2006	R I C-S survey, four samples/48 months	Self-reported data. Biological data from Mayo Clinic records.	Reliability of biological data differed from that of data collected in the context of high-quality research protocols. No control group. Community physicians with practices that focus on healthy lifestyles. Antismoking legislation. Aggressive marketing of low-carbohydrate/high-saturated fat diets.
Lupton BS et al., 2003	Cohort/36 months	Response rate – second invitation: I-61% and C-70%	Lack of randomization of communities, differences in lifestyle factors at baseline, secular trends, countywide intervention programs and crossover contamination. Inhabitants’ worries about high morbidity and mortality from coronary disease. Improvements in social conditions during the study period.
Lupton BS et al., 2002	Cohort/36 months	Response rate – second invitation:I and C-70%	Lack of randomization of communities, baseline differences, secular trends, countywide intervention programs and crossover contamination.
Record NB et al., 2000	Ecologic, observational, retrospective/240 months		Undetected secular trends. Concurrent initiatives.Death certificate data reliability – including possible influence of program awareness by local physicians. Program intensity measure. Economic fluctuations.
Hoffmeister H et al., 1996	R I C-S survey, three samples/84 months	Response rate: I-74.5/73.0/71.6% C-66.7/71.4/69.0%	Nationwide programs, regional preventive activities, self-help initiatives.
Luepker RV et al., 1994	R I C-S survey, seven samples Cohort/72-84 months	Response rate: Cross-sectional-78.7 Cohort-67.1%	Lack of randomization of intervention communities. Favorable secular trends in exposure to coronary heart disease risk reduction messages and activities. Cross-contamination.
Weinehall L et al., 2001	R I C-S survey, ten samples, Control – three samples/120 months	C – Participation rate: 76.7-81.3%	Intervention area with high cardiovascular disease incidence. Data collection: I – October and November C – January to April
Wood DA et al., 2008	Matched, cluster-randomized, controlled trial/41 months		Statistically underpowered – number of patients and partners was much smaller than expected, heterogeneity between patients, characteristics and pairs of centers. Knowledge of possible audit among randomized usual care centers. Characteristics of non-responders.
Mortality rates MRFIT, 1990	Cohort/120 months		Each death certificate was independently coded by two nosologists, and disagreements were adjudicated by a third nosologist.

R I C-S=Repeated Independent Cross-Sectional, I=Intervention group, C=Control group.