Study design and participants
The “German epidemiological study on ankle brachial index” (getABI) is a large-scale epidemiological trial with a cross-sectional part and a longitudinal part. The methods and design of the study have been described elsewhere in greater detail [14, 19]. Briefly, 34 vascular physicians across Germany trained and supervised 344 representative general practitioners in their vicinity. An assessment of primary care attendees, irrespective of their reason for seeing the doctor, was conducted in a prespecified week in October 2001. An average of 20 (maximum 25) eligible patients per practice fulfilling the inclusion criteria (age ≥ 65 years, patient being legally competent and able to cooperate appropriately, and providing written informed consent) were recruited over this week. The only exclusion criterion was life expectancy ≤6 months. A total of 6,880 patients were included in the study. Results reported in this paper mostly refer to the cross-sectional part of the 5-year follow-up. As part of the 5-year follow-up telephone interview, a sample of 1,376 participants was interviewed on sport participation. The protocol of getABI was approved by the University of Heidelberg (Germany) Ethics Committee. All patients gave written informed consent. The study was conducted according to the “Good Epidemiological Practice” recommendations issued by the “German Working Group Epidemiology” [2].
Outcome
Sport participation
Participation in sporting activities was assessed as part of the 5-year follow-up telephone interview using a modified version of the “Freiburger” questionnaire [17]. The activities asked and the summarized total time devoted to these activities are listed in Tables 2 and 3. The total time was given in three categories depending on the stated activity (either about 15 min, about 30 min, about 60 min or longer during the last week; or about 1 h, about 2–3 h, about 4 h or longer during the last month).
The main outcome for the logistic regression analysis was the binary parameter sport participation (“sporty”): yes or no. Participants who performed at least one sporting activity during the week (cycling) or the month (other sports) prior to the interview were defined as sporty. The recall of different time periods for cycling and other sports is a limitation of the “Freiburger” questionnaire. Cycling was defined as a sporting activity independent of the purpose (to run errands, for leisure, on bicycle tours, or on a stationary bicycle). This definition was based on the consideration that, for elderly people, cycling is a demanding task that requires not only a certain endurance capacity but also coordination, balance, and reaction speed. Looking at the metabolic requirements, Ainsworth et al. assign 4.0 metabolic equivalents even to “bicycling, <10 mph, to work or for pleasure” [1]. In contrast, walking either for leisure or to run errands was considered an everyday activity (not a sporting activity) because even for elderly people it does not require special motor skills and is usually of less than moderate intensity [1].
Covariables
Sociodemographic variables
The participants’ sex and date of birth were documented by the general practitioner at baseline. For the definition of the variable “education level,” the participants’ highest graduation (no graduation/comprehensive school (Hauptschulabschluss)/high school (Realschulabschluss or Mittlere Reife)/university entrance qualifications (Fachabitur or Abitur)) was assessed by the general practitioner at baseline. The country of birth was evaluated by telephone interview (5-year follow-up).
Cardiovascular risk factors
Waist circumference was measured in the general practitioner’s practice by standard protocol at the 3- and the 5-year follow-up of getABI. Waist circumference at 5-year follow-up was used for analysis. If this value was missing, waist circumference at 3-year follow-up was used. The current smoking status was documented at baseline (nonsmoker/former smoker/current smoker).
Health status
Both subjective and objective measures of health status were used: self-reported health status, report of the general practitioner, and blood test results.
Self-reported general health was assessed by telephone interview (5-year follow-up) by the use of the Short-Form 8 Health Survey (SF-8™ 4-week recall version). The instrument yields an eight-part profile of functional health and well-being. Two SF-8 composites can be calculated: the SF-8 physical component score (PCS) and the SF-8 mental component score (MCS). These scores are linear combinations of the eight items based on their respective importance for physical functioning and mental functioning. Higher scores represent better health status. The survey has demonstrated good reliability and validity [38].
The general practitioners documented the following vascular events at the beginning and throughout the trial (until follow-up at 5 years): cardiovascular events (myocardial infarction, cardiac revascularisation), cerebrovascular events (stroke, revascularization at carotids), and peripheral arterial events (amputation, peripheral revascularization, intermittent claudication).
Subjects were defined to have diabetes mellitus if they had been assigned the clinical diagnosis by their physician, if they were receiving any antidiabetic medication (insulin or oral), or if their HbA1c (glycosylated hemoglobin) was ≥6.5%. Subjects were defined to have a lipometabolic disorder if they had been assigned the clinical diagnosis by their physician, if they were receiving cholesterol-synthesizing enzyme (CSE) inhibitors or fibrates, if their serum total cholesterol was ≥200 mg/dl, or if their serum triglycerides were ≥150 mg/dl. Subjects were defined to have arterial hypertension if they had been assigned the clinical diagnosis by their physician or if they were receiving an angiotensin-converting enzyme (ACE) inhibitor or an angiotensin type 1 receptor (AT1) antagonist or diuretics. Diabetes mellitus, lipometabolic disorder, and arterial hypertension classifications were done at baseline.
Statistical analysis
Statistical analyses were performed with SAS version 9.1 (SAS Institute Inc., Cary, NC, USA). A logistic regression analysis (both univariate and adjusted for covariables) was performed to assess the association between the main outcome sport participation (sporty yes/no) and the following binary sociodemographic parameters, cardiovascular risk factors, and parameters of health status: age (≤/> median age (76 years)), sex (female/male), birthplace (inside/outside Germany), graduation (at least/less than university entrance qualifications (Fachabitur)), waist circumference (men </≥ 102 cm; women </≥ 88 cm) [20], smoking status (smoking/not smoking), SF-8 PCS (>/≤ median PCS (48.3)), SF-8 MCS (>/≤ median MCS (55.4)), history of (at least one) cardiovascular event (yes/no or unknown), history of (at least one) cerebrovascular event (yes/no or unknown), history of (at least one) peripheral arterial event (yes/no or unknown), diabetes mellitus (yes/no or unknown), lipometabolic disorder (yes/no or unknown), and arterial hypertension (yes/no or unknown).
Persons with missing values in one of the following variables were excluded from all analyses: sporting activities, age, sex, birthplace, graduation, waist circumference, SF-8 PCS, SF-8 MCS, walking (leisure/errands). In case of contradictory answers, data were analyzed as obtained. The level of significance was set at p < 0.05 for all analyses.