Overall, 4,845 articles were identified with our search strategy; of these, 4,827 were excluded from the review (Figure 1) because of the above mentioned reasons. A total of 292,278 subjects were involved at baseline (268,885 subjects at follow-up). Four publications, involving 17,329 subjects, studied the effect of physical activity on weight gain and obesity [16–19]. Six publications, involving 134,188 subjects, investigated the effect of physical activity on CHD [20–26]. Five publications, involving 84,647 subjects, studied the effect of physical activity on type 2 diabetes mellitus [27–31]. Six publications, involving 15,006 subjects, investigated the effect of physical activity on Alzheimer’s disease and dementia [32–37]. Some studies included more than one disease accounting for the discrepancy in the overall number of subjects and included studies. The maximum follow-up time ranged from 6 to 60 years.
Effect of physical activity on weight gain and obesity
Overall, the studies included in this review showed a negative relationship between physical activity and weight gain or obesity over time. Additional file 1: Table S1 summarises the examination data and the used survey sizes for the included studies on the long-term relationship between physical activity and weight gain and obesity.
An important study analysing the development of obesity depending on physical activity is the Aerobics Center Longitudinal Study (ACLS) conducted by the Cooper Clinic, Texas . Between 1970 and 1998, DiPietro et al.  examined 2,501 healthy men aged between 22 and 55 years at baseline and five years later. The daily physical activity level was negatively related to the weight gain during the follow-up time. Those people who reduced their daily physical activity level gained a considerable amount of weight, while those people who maintained the same level of activity during the study did not gain weight. Further, those people who increased their physical activity level during the study experienced weight loss. DiPietro et al.  reported that a daily physical activity level with a metabolic rate of at least 60% above the resting metabolic rate is necessary for losing weight. Hence 45 to 60 minutes of brisk walking, gardening or cycling should be included in the daily routine to maintain weight in middle-aged men.
Gordon-Larsen et al.  investigated the relationship between walking and weight gain. In the Coronary Artery Risk Development in Young Adults (CARDIA) Study, they examined 4,995 women and men aged between 18 and 30 years at baseline (1985/1986) who were re-examined 2, 5, 7, 10 and 15 years later. After 15 years, there was a negative association between 30 minutes walking per day and weight gain depending on the percentile of baseline weight. Data for people in the 25th percentile of baseline weight showed no significant relation between walking duration and weight gain. In contrast, data for people in the 50th percentile of baseline weight revealed that for every 30 minutes of daily walking the weight gain was 0.15 kg per year less for men and 0.29 kg per year less for women. Finally, data for people in the 75th percentile of baseline weight showed the smallest weight gain: for every 30 minutes of walking per day, men reduced their weight gain by about 0.25 kg per year and women by about 0.53 kg per year without making any other changes to their habitual lifestyle. Hence, the results of this study indicate that participants with a higher baseline weight benefit more from being physically active (for instance, for women: the total weight gain in 15 years was 13 kg for inactive women compared to only 5 kg for active women).
Hankinson et al.  used the same study population (CARDIA) to investigate the physical activity level in relation to a 20-year weight gain. Of 1,561 men and women, those with high habitual activity at the 20-year follow-up had a smaller increase in mean BMI, waist circumference and weight per year compared than those with low habitual activity. Men and women maintaining higher activity gained 2.6 and 6.1 kg less weight over the 20-year period than men and women with low activity, respectively. In addition, the results of that study indicated that women benefit more from maintaining a higher physical activity level than men and that maintaining higher activity levels during adulthood may lessen weight gain during the course of their life.
The Copenhagen City Heart Study by Petersen, Schnohr and Sorensen  linked cross-sectional and 10 year long-term analyses, to determine the development of weight gain. They examined 3,653 women and 2,626 men at three measurement points at 5-year intervals. The participants were aged between 20 and 78 years at baseline. Results of the three cross-sectional examinations (1st at baseline, 2nd after five years, 3rd after 10 years) also showed a negative relationship between physical activity and weight. The preventing effects of medium leisure time physical activity (LTPA) on obesity were lower than those of high LTPA for both genders. The longitudinal analysis revealed a significant direct correlation between the level of LTPA and the risk of becoming obese for men but not for women. In contrast to the results of the cross-sectional analysis, the more active participants had a higher risk of becoming obese. Moreover, the results of that study indicate that obesity may lead to physical inactivity.
Therefore, the results of the first three studies [16–18] suggest a negative correlation between physical activity and weight gain after several years of follow-up (greater physical activity leads to less weight gain). In contrast, the fourth study  provided evidence that being more physically active leads to a greater risk of becoming obese. They suggest that obesity influences the development of physical inactivity; however they did not discuss possible causes and effect relations. These results raise the question of the causality of the relationship between physical activity and weight gain. Detailed information, results and limitations of each study are presented in Additional file 1: Table S1.
Effect of physical activity on coronary heart disease (CHD)
Of all modern diseases, coronary heart disease (CHD) has received the most scientific scrutiny. Overall, most studies reported a negative relationship between physical activity and the occurrence of CHD for physical activity levels above the minimum energy expenditure. Additional file 2: Table S2 summarises the examination data and the used survey sizes of the included studies addressing the longitudinal relationship between physical activity and coronary heart diseases.
In 1948, the National Heart, Lung and Blood Institute founded by Kannel et al. established the Framingham Heart Study. This research group investigated the general causes and the development of coronary heart disease in 5,209 men and women, aged 30 to 62 years at baseline . The results revealed a negative association between the physical activity level and the emergence of CHD events and overall cardiovascular mortality [38–40]. Lee and Paffenbarger  compared the results of the Framingham Heart Study with data for 18,835 men who graduated from Harvard University between 1916 and 1950 and established the Harvard Alumni Health Study. In five mail-back surveys, researchers investigated the association between physical activity and stroke  and other CHD .
The relationship between energy expenditure and the incidence of stroke showed a u-shape pattern . Specifically, spending at least 2,000 to 3,000 kcal additional energy per week on physical activity was necessary for reducing the risk of stroke. These results were reassessed for all CHD  in 12,516 Harvard Alumni over the course of 16 years (from 1977 through 1996). For CHD in general, the relationship between energy expenditure and the incidence of CHD showed the same u-shape pattern but the curve was shifted towards lower additional energy expenditure: spending at least 1,000 kcal additional energy per week on physical activity was necessary to reduce the risk of CHD. Hence, moderate to vigorous additional physical activity of about 2,000 to 3,000 kcal (min. 1,000 to 2,000 kcal) per week appear to reduce the overall risk for CHD, stroke and other diseases (e.g. hypertension).
Comparable results were also reported by the Honolulu Heart Program [22, 23] including 8,006 men of Japanese ancestry aged 45 to 68 years at baseline who lived in Oahu, Hawaii. After 16 years, the physical activity reported at baseline was negatively related to CHD events and mortality. However, it is important to note that these results were partially mediated through the effects of hypertension, diabetes mellitus, cholesterol and BMI.
The studies cited in the next section had similar results but also featured the following additional findings.
The Alameda County Health Study by Kaplan et al.  reported the dependency of CHD mortality on several health factors and behaviour by quantifying the relative risks of various covariates (age, sex, perceived health, mobility impairment, heart problems, high blood pressure, diabetes mellitus, shortness of breath, current smoking, low BMI and social isolation) in 6928 men and women. After including all covariates, a protective effect of LTPA is still noticeable.
Gillum et al.  investigated the relationship between physical activity and stroke incidence in The National Health and Nutrition Examination Study I Epidemic Follow-Up Study on 5,852 persons aged 24 to 74 years at baseline and reported comparable results as above studies [20–23]. However, while the u-shaped relationship between physical activity and the incidence of stroke was confirmed for men, for women greater physical activity was negatively linearly associated with the incidence of stroke. In addition, recreational physical activity was not associated with the incidence of stroke in African American subjects, yet a significant interaction between heart rate and the incidence of stroke was observed only for African American subjects. The authors provided limited discussion of these differing results between Caucasian and African Americans.
To investigate the link between obesity and associated diseases, Li et al.  quantified the relative risk of developing CHD dependent on obesity and physical activity. They followed 88,393 Nurses aged 34 to 59 in their Nurses’ Health Study from 1980 to 2000. Being overweight and obese was significantly associated with increased risk of CHD. In addition, increased levels of physical activity were related to a graded reduction in CHD risk. Further, greater absolute mass (in kg) gained during adulthood predicted a higher CHD risk. The study concluded that obesity and physical inactivity contribute independently to the development of CHD in women.
Overall, all studies included in this review section showed a predicted negative relation between physical activity and the risk of CHD over time. Two studies [20, 21] showed that a minimum additional energy expenditure of 1,000 to 2,000 kcal per week is necessary to achieve health related results. Limitations of these studies comprise the inclusion of very specific and selected participants (e.g. Harvard Alumni in the Harvard Alumni Heart Study and Nurses in the Nurses’ Health Study). In addition, these results cannot be generalized for the general public because of the selected social and ethnic backgrounds of participants and unbalanced gender distributions. In addition, most studies used Caucasian subjects alone. Hence, additional research on other ethnicities is necessary to obtain generalizable results. Moreover, the summarized studies were not designed to clarify the causality of the relationship between physical activity and CHD events. Additional research on the impact of other lifestyle factors as mediators or moderators of the relationship between physical activity and CHD is necessary. Detailed information, results and limitations of each study are presented in Additional file 2: Table S2.
Effect of physical activity on type 2 diabetes mellitus
While the incidence of type 2 diabetes mellitus in older people has increased rapidly , all studies reported a negative relation between physical activity and the risk of type 2 diabetes mellitus. Additional file 3: Table S3 summarizes the results of the included studies that investigated the long-term relationship between physical activity and type 2 diabetes mellitus.
In their Nurses’ Health Study involving 70,120 nurses aged 40 to 64, which has been on-going since 1976, Hu et al.  investigated the relationship between participants’ physical activity level and the development of the relative risks for type 2 diabetes mellitus. Physical activity was negatively related to the incidence of type 2 diabetes mellitus even after adjusting for BMI where participants with higher physical activity levels had a lower relative risk of acquiring type 2 diabetes mellitus than those who with a lower physical activity level.
Berenzen et al.  and Demakakos et al.  reported generally comparable results in 653 men and women in the Copenhagen City Heart Study and in the English Longitudinal Study of Ageing covering different age groups, respectively. In addition to the negative relation between physical activity and the incidence of type 2 diabetes mellitus, Demakakos et al.  showed that moderate to vigorous physical activity (performed at least once per week) is necessary to achieve a positive effect on health and to reduce risk of type 2 diabetes mellitus. Stratifying their results by age revealed that with increasing age a higher intensity per training session or even several sessions per week are required to achieve the same risk reduction.
A high body weight or obesity, often described by the relation between body weight and body height (body mass index—BMI), and socioeconomic status are strong covariates for the relationship between physical activity and the incidence of type 2 diabetes mellitus. For instance, Katzmarzyk et al.  analysed the association between obesity, physical activity, cardiorespiratory fitness and the incidence of type 2 diabetes mellitus in their Physical Activity Longitudinal Study involving 1,543 men and women. Obesity and physical fitness, but not physical activity, were significant predictors of the incidence of type 2 diabetes mellitus. Mozaffarin et al.  added lifestyle factors in their analysis of the risk of type 2 diabetes mellitus in 4,883 participants of the Cardiovascular Health Study. Low-risk lifestyle factors included physical activity above the median level, dietary score in the upper two quintiles, having never smoked, no alcohol, a body mass index below 25 kg/m2 and a waist circumference below 88 cm for women or below 92 cm for men. With every healthy lifestyle factor the incidence for type 2 diabetes mellitus decreased by 35%. For people scoring lowest (that is, were the healthiest) in every lifestyle factor, an 82% lower risk for type 2 diabetes mellitus was predicted compared to all other patients. In addition, it was predicted that if these associations were causal, 8 of 10 cases of type 2 diabetes mellitus could be prevented.
All studies [28–31] reported a negative relationship between physical activity and the incident risk of type 2 diabetes mellitus. However, there are other factors than physical activity that are important in the development of type 2 diabetes mellitus. For instance, the results of the Physical Activity Longitudinal Study by Katzmarzyk et al.  suggest that not only the presence or absence of physical activity is a determining health factor but that the level of obesity and physical fitness also has an influence on the relationship between physical activity and the state of health. However, it is difficult to confirm these conclusions because of the small number of longitudinal studies that consider physical fitness and other lifestyle factors. In addition, the precise mechanism of how physical activity acts to reduce the risk of type 2 diabetes mellitus, such as through altered insulin sensitivity or altered insulin production, is still unknown.
Effect of physical activity on Alzheimer’s disease and dementia
The relationship between physical activity and dementia, particularly Alzheimer’s disease, is important for the general public because the incidence of dementia increases with increasing age . Additional file 4: Table S4 summarizes the results of the included longitudinal studies on the relationship between physical activity and Alzheimer’s disease and dementia.
The few existing studies [32–37] found that physical activity is negatively related to the incidence of Alzheimer’s disease and dementia in healthy men and women. Physically active people are at a lower risk of developing cognitive impairment and have a higher cognitive ability score. Interestingly, activities with low intensity, such as walking, are negatively related to the incidence of dementia and Alzheimer’s disease . These results indicate that regular physical activity may be an important and potent factor preventing cognitive decline and dementia in healthy older people. Most studies on Alzheimer’s disease and dementia originate in the field of Psychology. The link between physical activity and Alzheimer’s disease and dementia in healthy participants at baseline has only been reported in very few studies [32–37], further emphasizing the overall lack of studies and specifically the lack of long-term studies that include people without dementia or Alzheimer’s disease. Most studies included people who had already been diagnosed with dementia or Alzheimer’s disease to research the development of the diseases. Detailed information, results and limitations of all included studies on physical activity and Alzheimer’s disease and dementia are presented in additional file 4: Table S4.