"Ageing is the process that converts fit adults into frailer adults with increased risk of illness, injury, and death" (Miller, 1999). Ageing is neither a developmental process nor a disease, although a great proportion of older people have some overt or subclinical disease. It is useful but difficult to distinguish whether the age-related changes observed in the human beings are due to ageing itself or consequences of life- style factors, disease processes or genetic factors. These statements offer a basic definition of ageing. However, ageing is a process which individuals encounter in various ways and at different speeds. Ageing also influences different organ systems in varying ways and magnitudes. It is apparent that age-related changes in various physiological characteristics, for example, in cardiovascular fitness vary between the time periods (age cohorts). Studies on these age-related changes encounter problems of cohort and time-of measurement effects, and selection bias due to health (Ingram, 1999).

Physical exercise has effects counter those of age on many body structures and functions (Shephard, 1997a). Cardiovascular fitness, for example, has been found to be markedly higher among older endurance- trained men and women than among older less active subjects (Fitzgerald et al., 1997; Wilson and Tanaka, 2000). Cross-sectional comparisons in aerobic capacity between age groups usually tend to find minimal age- related changes in aerobic capacity because of enhanced selection by health and level of fitness in surviving older age groups. Older people who are fitter and healthier are more likely to survive than their frailer counterparts. Katzel et al. (2001) reported that longitudinal reductions in the absolute VO_2max (in ml·kg^-1·min^-1) among older endurance athletes were two to three times as large as those predicted by cross-sectional analyses or those found longitudinally in their sedentary counterparts. However, Saltin (1986) have suggested that the decline in VO_2max in endurance athletes is comparable regardless of the athletes are studied cross-sectionally or longitudinally.

Since the publication of the origin of the species by Charles Darwin in 1859 it has been claimed that the fittest individuals of different species survive the longest (Balady, 2002; Darwin, 1859). This survival advantage among physically active and fit individuals has been thought to be true in older people (Goraya et al., 2000). Considerable beneficial effects of physical exercise on physical performance have been found in numerous longitudinal surveys and training studies. Recent training studies suggest that adaptation to endurance training is also preserved among the oldest old people (Binder et al., 2002; Malbut et al., 2002; Vaitkevicius et al., 2002). It has been suggested that the magnitude of the gain in aerobic capacity by endurance training is related to age of subject, duration of exercise bouts, length of training programme, and pretraining VO_2max (Green and Crouse, 1995). Most of these cross-sectional and longitudinal studies on cardiovascular fitness have, however, been performed among selected physically active and healthy older people. More data are needed to confirm the beneficial effects of training among the oldest old people aged 75 and over, especially among women, and among those having disabilities.

Strenuous exercise can also have deleterious effects. Few researchers have reported any adverse effects among elderly people both either during acute exhaustive exercise or prolonged physical training. Most of the adverse effects reported have been minor musculoskeletal problems. A large proportion of these studies, again, have been conducted among healthy people below age 75. Subjects with health problems have usually been excluded.

The standard guidelines issued by American College of Sports Medicine (ACSM, 2000) in exercise testing and prescription differ in a few details between elderly and younger people. Exercise testing under the supervision of a physician is recommended for most elderly people during maximal tests and prior to participation in vigorous exercise (ACSM,2000; Fletcher et al., 1995). However, the value of exercise testing both for health screening and predictive purposes among elderly people has been questioned (Gill et al., 2000). It has been suggested that the safety margin in the case of physical exercise is narrower among older than younger people (Kallinen and Alen, 1995). Among older individuals the adverse effects of physical exercise may occur with a lower dose of exercise than among younger individuals. It may also be that older people benefit proportionally more from lower doses of exercise. A relatively few randomised controlled trials have been performed that address the issue of dose-response relationships among elderly people.

To address the issue of the overall beneficial or harmful responses of physical exercise, the effects of physical exercise on cardiovascular functions, fitness and health were examined among elderly people in cross-sectional and longitudinal population-based studies. The role of exercise testing was also clarified among elderly people.

The primary research objectives were the following:

1. To study the effect of physical activity level on cardiovascular fitness and effort tolerance in elderly women in a cross-sectional study.

2. To study whether exercise test status and exercise-test results, including exercise capacity, are predictive for risk of death in community-dwelling elderly men and women.

3. To study the beneficial and possible adverse effects of an 18-week strength and endurance training programme on cardiovascular functions, fitness and health among elderly women in a population- based randomised controlled trial.

4. To examine the acute and prolonged effect of acute exercise in the form of a progressive cycle ergometer test on cardiac arrhythmias, ST segment depressions and other cardiovascular adverse effects in community-dwelling elderly people.

Effort tolerance in elderly people with different physical activity backgrounds (Study 1)

Study design
Sixty physically active women (PA) aged 66-85 years were selected from a sample of 600 members of Finnish sport organisations on the basis of a physical activity questionnaire. Most of them had a life-long training background and were still active in various sports (running, cross-country skiing, track and field, gymnastics). As a control group (CO) a random sample of 71 sedentary women aged 70-81 years was taken from the population register of the rural municipality of Jyväskylä. The control women performed light habitual physical activities (housekeeping, shopping, walking). None of them, however, reported regular endurance type training or other vigorous physical exercise.

Fifty-two (87%) of the PA and 42 (59%) of the CO took part in the laboratory examinations (Sipilä and Suominen, 1993). Those women who failed to attend the laboratory examinations were interviewed by phone about the reasons to their non-participation in the further examinations. Almost all of the control groups explained their non-participation by reference to health problems, and the physically active women failed to take part mostly because of travel arrangements. The ethical committee of the Central Finland Central Hospital approved the study protocol and all of the subjects gave their written informed consent.

Laboratory examinations
Before the laboratory investigations questionnaires were sent to the subjects to elicit background data concerning their health, medication, and physical activity. The laboratory examinations consisted of various anthropometric and physiological measurements. A detailed physical examination was done to establish he contraindications for the exercise test (Table 10).

Blood pressure, body height and weight, body fat, and 12-lead ECG were taken before the exercise. The bioelectrical impedance measurements were performed at 10.00-10.30 h. The machine was calibrated daily with standard resistor. Before the measurements the subjects had fasted 3-4 h and not exercised for at least 12 hours. In our laboratory, the coefficient of variation between two consecutive bioelectrical impedance measurements has been in the order of 2-3%. In addition erythrocyte sedimentation rate and blood hemoglobin concentration were measured before the exercise. Serum gamma glutamyltransferace, alanine aminotransferase activities, and creatinine concentration were analysed afterwards to exclude hepatic or renal disturbances.

Cycle ergospirometry
Subjects were to perform an exhaustive cycle ergometer exercise to their volitional maximum. The subjects were asked to keep the pedalling frequency within the limits of 50-60 rpm as far as possible and a weight of 200-300 g was added after every 2 minutes to the cycle's basket connected to a mechanical braking system (Monark 814 E, Varberg, Sweden). The mean incremental loading after every 2 minutes was 18 W, depending on pedalling frequency.

The subjects were encouraged to continue pedalling the ergometer to their personal maximum unless they experienced any exceptional symptoms (chest pain, dizziness, severe breathlessness, musculoskeletal pain). ECG leads II, V1, V5 and the well-being of the subject were monitored continuously; 12-lead ECG, and brachial arterial cuff pressure were recorded at minimum intervals of 2 minutes. Oxygen uptake, carbon dioxide production, ventilation, and respiratory exchange ratio (RER) were measured after each 30 seconds with a gas analyser (Minjhardt Oxycon-4, Odijk, Holland). The indications for exercise cessation were based on the guidelines of ACSM 1995 (Table 11).

The point of maximal effort was evaluated by the laboratory personnel on the basis of objective signs of subject's exhaustion i.e. breathlessness, or, if, for example, the pedalling frequency and work power consistently decreased in spite of the subject's effort. On the other hand if the exercise was terminated for medical criteria (Table 11) or if the subject suddenly stopped without having reached exhaustion, the test was considered submaximal. According to their ability to perform the exercise submaximally or maximally, the physically active subjects and controls were divided into submaximally exercised (submax.) and maximally exercised subjects (max.).

Analysis of plasma C-ANP and N-ANP
To evaluate the cardiac load a 10 ml blood sample from subject's (sitting) cubital vein was taken by venapuncture in EDTA tubes immediately before and within one minute after the ergometer test for the plasma C-ANP and N-ANP analysis (Thibault et al., 1987). The subjects were highly co-operative and thus good quality samples were obtained in all cases except in three cases in which we failed to get a blood sample within a minute after the exercise.

The blood samples were placed immediately in ice and centrifuged within 60 minutes at +4 ^oC. Plasma was separated and stored at -80 ^oC for further analysis. Plasma C-ANP and N-ANP radioimmunoassays were carried out by using the methods described previously (Vuolteenaho, 1985a; Vuolteenaho et al., 1985b; Vuolteenaho et al., 1992). For the C-ANP assay plasma samples were extracted with Sep-Pak C18 cartridges (Waters, Milford, MA) with a recovery of 82 ± 10%. N-ANP was measured directly from 25 l plasma samples. The sensitivities of the C-ANP and N-ANP assays were 2 pmol·l^-1 plasma and 30 pmol·l^-1 plasma, respectively. The intra- and interassay coefficients of variation were < 10% and < 15%, respectively.

The two blood samples for C-ANP and N-ANP analysis were also taken at 15-20 minute intervals from those subjects (n= 13) excluded from the exercise to determine the normal variability of C-ANP and N-ANP over time. N-ANP was measured in all of the 42 exercising PA and in 19 out of the 22 exercising CO. The sample size of N-ANP was reduced because of problems in the venapuncture of three subjects. C-ANP was measured in 30 exercising PA and 16 exercising CO. The marked reduction of 15 in the sample size of C-ANP was due to technical problems (12 because of technical failure in the centrifuge thermostat and 3 because of problems in performing the venapuncture).

Statistical analysis
All data were expressed as means ± SD. The difference between the groups at the baseline was analyzed with one-way ANOVA. The effects of physical loading on the response of C-ANP and N-ANP were assessed using ANOVA with repeated measures. Pearson's product moment correlation test was done to determine connections between C-ANP and N-ANP with the physiological variables. Multiple linear regression analysis was used to determine connections between aerobic capacity and the other variables. Subjects with β- blocker were excluded for all statistical analyses concerning heart rate and rate pressure product (RPP) data. An alpha level of 0.05 was set to mark statistical significance. The statistical software package used was SPSS for windows version 6.0 (SPSS Inc., USA).

The predictive value of exercise testing for survival among 75-year-old men and women (Study 2).

Study design
The study population and the laboratory examination protocol are presented in Figure 2. Complete background, exercise test and mortality data were finally obtained from 282 persons. The ethical committee of the University of Jyväskylä approved the study protocol and all of the subjects signed a written informed consent. About two weeks before attending the research centre all the subjects were interviewed individually at home on the subject of their health, current medication and living habits, e.g. current smoking and physical activity.

Physical activity was assessed by the question "Which of the following descriptions best corresponds to your current level of physical activity?" The subject selected one of six alternatives ranging from 1 (mostly light activities sitting in one place) to 6 (participating in competitive sports) (Grimby, 1986). Subjects were divided into three categories according to their physical activity level: category 1 (mostly light physical activities or activities sitting in one place), category 2 (moderate physical activities about 3 hours per week), category 3 (moderate physical activities or sports at least 3 hours per week or heavy physical activities at most 4 hours per week). Participation in competitive sports was excluded from the categories as it did not apply to any of the subjects. Physically inactive subjects were considered to belong to category 1, intermediately physically active to category 2, and physically very active to category 3. Current smoking status was categorized as non-smoking, smoking occasionally, or smoking daily.

At the research centre a standard physical examination was carried out by a physician to check general health and medication prior to the functional capacity tests. The number of chronic diseases was calculated according to physicians' diagnoses of chronic conditions lasted of over 3 months of duration. Blood pressure, body height and weight, and 12-lead ECG were recorded before attempting the cycle ergometer exercise. Body height and body mass were measured using standard procedures. Body mass index (BMI) was calculated as the relation of body mass (kg) to squared body height (m). In addition, blood hemoglobin, and glucose concentration were measured before the test. Hemoglobin A1c, total cholesterol and HDL-cholesterol concentrations were analyzed from serum samples afterwards. The contraindications for the cycle ergometer exercise were based on the guidelines issued by the American College of Sports Medicine 1986(a non-exercise test group).

A progressive cycle ergometer exercise was administered to 74.0% of the men and 66.0% of the women (77 men, 126 women). The indications for terminating the cycle ergometer exercise were assessed by the same physician (a specialist in sports medicine) on the basis of the guidelines issued by the American College of Sports Medicine 1986 (Tables 10 and 11).

The Ethical Committee of the University of Jyväskylä approved the study protocol and all of the subjects gave their written informed consent to participation in the study.

Cycle ergometer exercise
Subjects were to perform an exhaustive cycle ergometer test to their volitional maximum pedalling at a frequency of 50 to 60 rpm for as long as possible, and a weight of 300 or 600 g was added after every 2 minutes to the a load basket which was connected to the mechanical braking system of the ergometer (Monark 814E, Varberg, Sweden). The amount of work achieved during each minute was then calculated on the basis of the load and the actual number of revolutions, the latter registered by a counter. The average initial load was 29 watts, and the average incremental load 18 or 36 W. Oxygen uptake was not measured directly in this study.

The point of maximal effort was evaluated by the test personnel on the basis of objective signs of the subject's exhaustion, e.g., breathlessness, or when pedalling frequency and work power fell consistently in spite of the subject's efforts to maintain them (exercise test completion group). On the other hand the test was considered to have been prematurely terminated if the exercise was stopped for medical reasons or if the subject suddenly stopped without having reached exhaustion (exercise test termination group). All the exercise tests were performed with the subjects on their current medication. Thirty-seven percent of the subjects were taking nitrates and 12% β- blockers. Digitalis and diuretics were being used by 21% and 27%, respectively.

The number of the subjects in the non-exercise test group (NEX), exercise test termination group (TER), and exercise test completion group (CMP) are presented in Table 12. The proportion of subjects whose exercise test was prematurely terminated was higher among men than among women (Table 12).

Analysis of mortality
Time of death for all participants who died during the nine-year follow-up was obtained from the official register of the province of Central Finland and from hospital records. The principal causes of death were classified and registered up until the year 1994 in accordance with the 1977 International Classification of Diseases (WHO, 1977) and thereafter in accordance with the 1992 International Classification of Diseases (WHO, 1992). Mortality was followed up from the baseline studies (1989 and 1990) until the end of 1998. Complete mortality data was obtained for all the deceased subjects. The most frequent causes of death were cardiac-related and ranged between 28 and 49% of cases in the different groups. The figures of deaths from cancer ranged between 12 and 36%.

Statistical analysis
Standard methods were used to calculate the means and standard deviations of the variables. Students t- test (two-tailed) or analysis of variance (ANOVA, Least Significance Difference test, LSD- test) were used in comparisons of means for continuous variables. Contingence tables and χ²-test were used to compare categorical data.

Survival distributions were analyzed by using statistics D (Lee and Desu, 1972) for the non- exercise test group, exercise test termination group, and exercise test completion group. The Cox proportional hazards model (Cox, 1972) was used to estimate the relative risk for mortality in terms of exercise-test status, sex, various physical characteristics, health status, physical activity, and physical performance. Subjects who did not die during the follow-up were assigned a survival time of 9 years. For those who died during the follow-up, survival time was calculated as the time from the beginning of the study to the day they died.

Three main models were created to calculate the hazard ratios for death: Model 1 with unadjusted hazards ratios for all three study groups, Model 2 with adjusted hazards ratios for all three study groups, and Model 3 with adjusted hazards ratios for the exercise test completion group. The adjusted models were controlled by sex, number of chronic diseases, physical activity level, and, additionally in the exercise test completion group, cycling power. The adjustment was done step by step by adding one factor after another into the new model. The covariates included in the models were selected on the basis of their known or suspected association with mortality. Covariates not having a significant association with mortality were excluded from the main three final models presented in the results. The excluded covariates were BMI, smoking, blood hemoglobin, blood glucose, hemoglobin A1c concentrations, total cholesterol and HDL-cholesterol ratio. Cumulative models were created to calculate the hazard ratios between three groups: two sub-groups of the exercise test termination group, i.e., termination for cardiovascular reasons, and termination for other reasons, and an exercise test non-termination group.

The computations were done with SPSS 10.0 software (SPSS Inc., Chicago, IL).

Improving cardiovascular fitness by endurance and strength training (Study 3)

Study design
A postal questionnaire about health, medication, physical activity level, and functional status was send to a random sample of 76- to 78-year-old women (n= 240) drawn from the population register of the city of Jyväskylä, Finland. Of the 157 women (65.4%) who responded to the questionnaire, 65 women reported no severe diseases or functional impairments which could exclude their performing the cycle ergometer test and the subsequent strength or endurance exercise programme on the basis of a physician's evaluation and the guidelines for exercise testing and prescription issued by ACSM (Table 10).

Fifty-four women participated in the laboratory examinations. The remaining 11 who did not attend the laboratory examinations explained their non-participation by lack of time, travelling, and poor health. Forty-two women with no contraindications for physical exercise were randomly assigned to strength (n= 16), endurance (n= 15), and control (n= 11) groups. The randomisation was performed manually by drawing lots after deciding the number of subjects in each group.

The number of subjects in the two exercise groups was intentionally larger than that in the control group to overcome the possible higher drop-out in the exercise groups. Some of the subjects had chronic conditions (Table 13) which, however, did not exclude them from physical training. Nine subjects had no prescribed medications, 13 subjects had a nitrate as medication, 9 subjects had a ß- blocker, and 6 subjects used calcium channel- blockers respectively. Only one person used digitalis. The level of physical activity in all of the groups studied averaged 3 in six class scale (Table 13), i. e. around 3 hours moderate physical activity per week, involving domestic tasks such as cooking, cleaning, straightening up the room, making beds, and ordinary gardening, walking longer distances, and cycling. None of the subjects reported regular vigorous physical exercise.

Twelve subjects in the exercise groups and eleven in the control group completed the intervention. Of the seven women who withdrew from the study, six stopped because of illness and one was unwilling to continue. The study was approved by the ethical committees of the Central Finland Central Hospital and of the University of Jyväskylä. A written informed consent was received from all the subjects.

Laboratory examinations
Body height and body mass were measured using standard procedures. Lean body mass (LBM), and body fat were assessed by bioelectrical impedance (Spectrum II, RJL Systems, Detroit, MI). Maximal isometric force was measured in a sitting position on a custom-made dynamometer (Sipilä et al., 1996).

A clinical examination, blood tests (glucose, haemoglobin, erythrocyte sedimentation rate), right brachial arterial cuff pressure after 10 minutes' rest, resting and exercise ECG preceded the exercise capacity tests to set the contraindications for exercise (Table 10).

Cycle ergospirometry
The subjects performed a symptom-limited cycle ergometer exercise to their volitional maximum according to the protocol described in the methods section of study 1. Oxygen consumption, carbon dioxide production, ventilation and respiratory exchange ratio were assessed by a gas analyser (Minjhardt Oxycon-4, Odijk, The Netherlands). The cycle ergometer and the exercise test were done at the baseline and after the 18-week intervention in all subjects. The strength measurements were additionally performed after 9 weeks of the intervention.

Physical training
Both exercise groups participated in an 18-week progressive physical training program which consisted of a 2-week orientation phase and a 16-week training phase. The subjects had supervised 1-hour training sessions twice a week during the orientation phase and three times a week during the training phase. At the beginning of each training session, both groups had a warm-up period of 10 minutes, including brisk walking and stretching. At the end of the session, stretching exercises for the major muscle groups were performed. The subjects in strength and endurance groups participated in 85-95% of the exercise sessions.

Strength group trained the major muscle groups (muscles of the thigh, calf and trunk) on equipment using compressed air as a resistance (HUR, Kokkola, Finland). The intensity of the training was gradually increased from 60% (first 3 weeks) of the 1 RM (repetition maximum) to 75% of the 1 RM (last 4 weeks). 1 RM was defined as the heaviest load the subject could move in acceptable way throughout the complete range of motion. The resistance was individually adjusted according to the one-repetition maximum test (1 RM) measured at 2-wk intervals. To obtain the 1 RM, the initial resistance was set close to the previous 1 RM result. The resistance increment was 0.25 bar, which corresponded to 2.5 kg in knee flexion, 3 kg in the knee extension and 5kg in the combined knee-hip extension.

After the 18-week training period, the 1RM values increased on the average 19 - 60 % depending on the muscle group tested. The individual variation between the two final 1 RM measurements averaged 2%. During the strength training sessions the subjects performed three to four sets of 8-10 repetitions with a 30-s pause between sets. Each repetition lasted for 4-6 s, with no more than 2 s rest between the repetitions.

The endurance group walked on an indoor track twice a week and had step aerobics once a week. The training heart rate was individually adjusted on the basis of the initial cycle ergometer test. The training intensity was gradually increased from 50% (first 5 weeks) to 80% (last 4 weeks) of the initial heart rate reserve. The training heart rates were controlled by heart rate monitors (Sport Tester, Polar Electro, Kempele, Finland) during the training sessions. The walking session lasted 20 min in the first 2 weeks and 30 min in the subsequent 16 weeks. The walking distance during indoor walking increased from the average value of 1500 m (range 1200-2200 m) to the 2700m (range 2400-3300 m) in the end of the training period. The step-aerobics session lasted for 40 min, during stepping on a bench 0.10 m or 0.15 m in height.

Controls were instructed to continue their habitual physical activity levels. All subjects, including the subjects in the control group kept a diary concerning their physical activities during the intervention to detect possible changes in physical activity level. A routine clinical examination was repeated after 9 and 18 weeks of the intervention to all subjects. Any new symptom was examined by the supervising physician and the training stopped if necessary.

Brachial arterial cuff pressure was measured before exercise sessions. Ambulatory ECG was registered in 3 subjects in the training groups because of chest pains in order to exclude possible cardiac disease or dysfunction behind the symptoms.

Statistical analysis
Standard statistical procedures were performed to calculate means and standard errors (SE). The differences between the groups at the baseline were analysed using either the one-way ANOVA or chi-square test. The effects of training were assessed using two-way ANOVA with repeated measures. An alpha level of 0.05 marked statistical significance in the group comparisons for independent samples. If the significance of the interaction of group by time in ANOVA with repeated measures was p < 0.10, the training effect was localised utilising simple contrasts with a p < 0.05. The statistical power of detecting a significant interaction was 0.614 for peak power (W·kg^-1) and 0.238 for peak oxygen uptake (ml·kg^-1·kg^-1). The statistical software package used was Sigma Stat program version 2.0 (Jandel Scientific Corp, USA) or SPSS for windows (SPSS Inc., Cary, NC).

Cardiac adverse effects and acute exercise (Study 4)

Study design
The general design of the study is presented in Figure 3. The study population was the same as in study 2. The target population consisted of all 75-year-old residents of the city of Jyväskylä, Central Finland (n= 388). Of them 295 (76%) entered the laboratory for the assessment of their health and functional capacity (Heikkinen, 1997). Before entering the laboratory all of them were interviewed individually at home about their background data of health, functional capacity and living conditions. A questionnaire of health and physical activity was given to t he subjects to fulfil before entering to the laboratory examinations. Physical activity was assessed with the question "Which of the following descriptions best corresponds to your current level of physical activity?" The subject selected one of 6 multiple-choice alternatives (Grimby, 1986).

A detailed physical examination was done in the laboratory by a physician to check general health and medication prior to the functional capacity tests. Blood pressure, body height and weight, body fat (by bioelectrical impedance, Spectrum II, RJL Systems, Detroit, MI), and 12-lead ECG were recorded before cycle ergometer exercise. In addition, the erythrocyte sedimentation rate, blood haemoglobin and glucose concentration were measured before the exercise. Total cholesterol and HDL-cholesterol concentrations were analysed from serum samples afterwards. The contraindications for cycle ergometer exercise were based on the guidelines issued by ACSM 1986 (Table 10).

A progressive cycle ergometer exercise was then administered to 74.0% of the men and 66.0% of the women (77 men, 126 women) (population group, PG). The indications for terminating the cycle ergometer exercise were assessed by the same physician (a specialist in sports and exercise medicine) on the basis of the guidelines issued by ACSM 1986 (Table 11). Maximal isometric strength measurements, 10-meter walking test, and stair-mounting test were performed by the subjects on the same day as the cycle ergometer exercise. Additionally, some psychological tests were administered in the intervals between the physical capacity tests. The subjects stayed in the laboratory for the assessments for an average of 6 hours on the same day.

Any possible contact by the subjects with the central hospital, to which all acute cardiac cases are referred, during the 24-hour period after the laboratory examinations was to be obtained from the hospital's patient records. Thirty-six subjects out of the PG were invited to the ambulatory ECG recordings and the second cycle ergometer exercise on the basis of the prior clinical examination, laboratory tests (blood hemoglobin and glucose, ECG) and initial exercise test on the cycle ergometer. Of this number 28 subjects (16 women and 12 men) with no contraindications to heavy physical exercise participated in the second cycle ergometer exercise and ambulatory ECG recordings (holter group, HG).

The primary inclusion criterion for the subjects in the HG group was that they were able to attain their peak physical effort in the initial cycle ergometer test. The second criterion was that the subjects should not have had any regular medications influencing on the amount of heart arrhythmia and ST segment changes. Thus, the subjects of the HG were free of any observable limitations on physical loading and were apparently healthy and asymptomatic of cardiovascular diseases. One person, however, had previously suffered myocardial infarction and one had mild bronchial asthma and medicated hypothyreosis.

The ethical committee of the University of Jyväskylä approved the study protocol and all of the subjects signed a written informed consent.

Cycle ergometer exercise
The subjects (both in PG and HG) were to perform an exhaustive cycle ergometer exercise to their volitional maximum, as described in the methods section in study 2. All the exercise tests were performed with the subjects under their current medications. Thirty-seven per cent of the subjects were taking nitrates and 12% β-blockers. Digitalis and diuretics were being used by 21% and 27%, respectively.

Ambulatory ECG
Ambulatory ECG recordings were taken 24 hours before and after the cycle ergometry (HG). Portable two-channel tape recorders (Marquette Electronics Inc., USA) were used with electrode placement to obtain modified V₁ and V₅ leads. Before final electrode placement the height of the calibration signal and R-signal were checked (minimum 10 mm in the modified V₅) and the electrodes were replaced if necessary.

The recorders were applied between the hours 8-11 a.m. and were removed at the same time the following day. The subjects went home wearing the recorder on a waist belt and continued with their normal daily activities. They were asked to maintain the daily activities performed over the two consecutive 24-hour periods comparable. The subjects kept a diary of their activities and symptoms. They were also asked to push the 'event' button of the holter device if they had any exceptional symptoms, i.e. severe breathlessness, dizziness, feelings of arrhythmia or chest pains.

The ECG tapes were analyzed in the routine scanning mode with a Marquette Laser Holter (Marquette Electronics Inc., USA) electrocardioscanner by the same technician. The number of ventricular premature beats (isolated, couplets, runs), supraventricular beats (isolated, runs) and ST segment depressions were compared between the two consecutive time-matched recordings before and after the maximal exercise. Rhythm strips were obtained to document the complexity of arrhythmia and the nature of each ST segment depression.

The criteria for calculating ST segment depressions were as follows: every ST segment depression was checked and registered if it was > 1.0 mm from the baseline at 60 milliseconds after the J-point and lasted at least 1.0 minute. There should also be an interval of at least 2 minutes between two separate ST segment depressions. In relative ST segment depressions the basal time constant was 2 hours. Slowly ascending, horizontal or downsloping ST segment depressions were taken into account for comparison. The recordings of 5 subjects were excluded from the data analysis, two because of poor quality of the ECG signal in the ambulatory ECG recordings, one because of incomplete exercise data, one because of left bundle branch block, and one because of β-blocker medication. Altogether the ambulatory recordings of 12 men and 11 women were available for the statistical analysis.

Statistical analysis
Standard statistical procedures were performed to calculate means and standard errors (SE). The differences between the men and women in the PG and HG were analysed doing T-test, Mann- Whitney Rank Sum test or Chi-Square test. The effects of physical loading on cardiac arrhythmia and ST segment depressions among both sexes were assessed using two-way ANOVA with repeated measures. An alpha level of 0.05 was set to mark statistical significance. The statistical software package used was SigmaStat program version 2.0 (Jandel Scientific Corp, USA).

Effort tolerance in elderly people with different physical activity backgrounds (Study 1)

Background data
The exercised physically active women (PA) and control women (CO) were comparable in age. PA were leaner than CO (Table 14). Submaximally exercised PA were somewhat older than maximally exercised PA. Cardiovascular and musculoskeletal problems were common in the groups studied especially among CO in submax. group (Table 15). Three of the exercised controls used digitalis medication and 4 used beta-blocker. Among the physically active elderly women 4 used digitalis and 2 used beta-blocker, respectively.

Cycle ergospirometry and aerobic capacity
Twenty-two of the 42 PA terminated the ergometer exercise before objective exhaustion, and of the CO only 3 out of the 22 went to the maximum. The most common reasons for termination among PA were the subject's own wish to stop or abnormal cardiovascular reactions. Among CO tiredness of the legs was also a common reason for termination (Table 16). In submaximal exercise PA differed from CO in aerobic performance characteristics only when expressed in relation to body weight. In maximal exercise the difference in peak heart rate and rate pressure product (peak RPP) reached statistical significance (Table 17).

Aerobic capacity expressed in peak power and peak oxygen uptake was higher among PA than CO.

This difference was more pronounced when the aerobic capacity figures were expressed in relation to body weight. The RER values tended to be lower among CO than PA (Table 17).

Assessment of cardiac load during exercise by C-ANP and N-ANP
Both C-ANP and N-ANP increased during physical loading in submaximal and maximal exercise (Figure 4a and Figure 4b). No significant differences between the groups were found in the effect of exercise on either C-ANP or N-ANP (ANOVA with repeated measures). The basal level of N-ANP among PA before submaximal exercise was, however, significantly higher than before maximal exercise among PA and higher than before submaximal exercise among CO (Figure 4a and Figure 4b).

Exercise test outcome and mortality (Study 2)

The survival of the subjects
Of the 282 subjects, 117 (42%) died during the 9-year follow-up. In the non-exercise test group 49 subjects (62%) died, in the exercise test termination group 32 (34%) and in the exercise test completion group 36 (33%). The most frequent causes of death were cardiac-related and ranged between 28 and 49% of cases in the different groups. The figures of deaths from cancer ranged between 12 and 36%. The non-exercise test group had significantly higher mortality than the exercise test termination group. A similar difference was found between the non-exercise test group and exercise test completion group. No statistically significant difference was found in survival between the exercise test termination group and exercise test completion group either over the whole 9-year follow-up or after 4, 5, and 6 years of observation.

Cox proportional hazards models for mortality
The multivariate hazard ratio (HR) for death was 1.87 times higher among the subjects in the non-exercise test group than in exercise test completion group. A similar difference was found between the non-exercise test group and exercise test termination group (HR 1.97, CI 1.24 - 3.13). No statistically significant difference was found in the risk for death between the exercise test completion group and exercise test termination group.

The high peak cycling power in W·kg^-1 body weight was associated with reduced risk for death in Model 3 (multivariate HR 0.20; CI 0.08 - 0.50). In both models a high number of chronic diseases per subject was independently associated with an increased risk for death. Males were at increased risk for death in Model 3. Physical inactivity was independently associated with increased risk for death in Model 2. In this model the risk for death in relation to physical inactivity was defined as risk for death among the physically inactive subjects (category 1) compared to physically very active subjects (category 3). The risk for death among the intermediately physically active subjects (category 2) compared to risk for death among the physically inactive subjects (category 1) did not reach statistical significance.

The hazard ratio for death among subjects whose tests were terminated for cardiovascular reasons was 1.185 (CI 0.71 - 1.99) compared to the exercise test completion group in a non-adjusted model (data not shown). When both gender and cycling power were controlled the corresponding hazards ratio was 0.78 (CI 0.45 - 1.36). In the latter model male gender resulted in a hazard ratio of 2.33 (CI 1.36 - 3.98) and cycling power in W/kg of 0.31 (CI 0.17 - 0.55).

Improving cardiovascular fitness by endurance and strength training (Study 3)

Background data
The medications, health characteristics, and physical activity level of the subjects at the baseline are presented in Table 13. No significant differences were found in these variables or in the baseline values of physical characteristics between the groups. Beyond the training induced in the trial, the study groups did not differ with respect to the initial level of physical activity, which remained constant throughout the experiment. Body fat decreased significantly among the subjects in the strength training group compared to the controls (Table 18).

Effects of training on aerobic capacity
There was a significant increase in peak power in Watts (W) in the strength training compared to control group (Table 19). Peak power in Watts·kilogram^-1 body weight increased in both the endurance and strength training groups compared to controls. No interaction of group by time was found for peak oxygen uptake (Table 19). The respiratory exchange ratio (peak) did not differ between the baseline and after the 18-week intervention in any of the groups (Table 19). The mean percentage changes in peak oxygen uptake (ml·kg^-1·min^-1) in the endurance, strength, and control groups, respectively, were +6.8 %, +9.4 %, -6.2 %.

The mean respective percentage changes in power (in W·kg^-1) were +3.8 %, +8.1 %, and -5.9 %. There was wide inter-individual variation in all groups in percentage change in peak power and peak oxygen uptake (Figure 5a and Figure 5b).

Health problems emerged during the exercise intervention
Six subjects (19%) in the exercise groups withdrew from the study because of health problems. One subject in the endurance group dropped out of the intervention because she was unwilling to continue. Two subjects in the exercise groups were not able to the intervention in the exercise groups.

One subject in the strength group died because of large myocardial infarction 8 weeks into the intervention. The symptoms of the myocardial infarction started two days after the exercise session. No cardiac problems, or hospital contacts because of heart problems before the intervention were documented. The initial exercise ECG showed no cardiac ischemia. Another subject in the strength group sustained an unstable angina pectoris at 4 weeks into the intervention starting 2 days after the exercise session. She was operated successfully for three- vessel coronary heart disease. In her medical history she had been treated in the hospital 2 years before the exercise intervention for a chest pain attack, but this event was considered non-cardiac. A small painless and horizontal 1mm ST segment depression in leads V₅ and V₆ was detected in her initial exercise ECG at peak exercise.

One person in the strength group began to suffer from occasional angina pectoris and dyspnoe when walking at 8 weeks into the intervention. No actual ST segment changes were seen in her resting ECG and the chest X-ray was normal. ST segment depressions were detected in her initial exercise ECG neither. A nitrate was prescribed for evident angina pectoris and the symptoms disappeared. Her exercise intervention was terminated.

A member of the endurance group had slight weakness of the limbs in the right side together with mild dysartria 2-3 hours after the strength measurements at 9 weeks into the intervention. A neurologist made a clinical diagnosis of infarction of the brainstem. The symptoms gradually disappeared. Her exercise intervention was terminated. In the last medical screening (after the 18 weeks endurance training) a suspected abdominal aortic aneurysm was found in one asymptomatic woman. An abdominal aortic aneurysm with diameter of 5 cm was confirmed by ultrasonography. The subject was excluded from the final measurements and she was successfully operated after the intervention.

Six cases of musculoskeletal problems emerged in the exercise groups, but these complaints were minor and did not lead to premature termination of the exercise intervention. One subject in the strength group, however, fractured her clavicle when she fell off a bicycle during her free-time.

Cardiac adverse effects and acute exercise (Study 4)

Cardiac disturbances during cycle ergometer test and cardiac adverse events after the test
In the population group (PG) the cycle ergometer exercise (CEE) was terminated in 23.4% of the men and 6.4% of the women because of cardiac arrhythmia or deep ST segment depression (gender difference, χ² = 10.93, p < 0.001). Almost twelve percent (11.7%) of the men were stopped because of deep ST segment depression compared to 5.6% of the women (gender difference, χ² = 1.70, p = 0.192). An increasing number of ventricular premature beats or multifocal ventricular premature beats was the reason for exercise cessation in 6.5 and 5.2% (men) and 0.8 and 0.8% (women) of the subjects (gender difference for the arrhythmia, χ² = 9.90, p= 0.002). No episodes of nonsustained ventricular tachycardia were detected in our subjects. One subject in the PG was referred to hospital immediately after the cycle ergometer exercise because of acute atrial fibrillation while performing it. The subject was treated successfully in the local hospital.

No other subject in the PG was referred to hospital within the 24 hours following the laboratory examinations. Seven performances (3 men and 4 women) out of 23 in the HG were submaximal. Five out of these seven were terminated for medical reasons, 3 cases because of deep ST segment depression and 2 because of bundle branch block.

Cardiac disturbances in the 24-hour ambulatory ECG recordings
As expected, the sum duration of the ambulatory ECG recordings was similar in time-matched pre-exercise and post-exercise periods in both the men and the women (holter goup, HG). Similarly, no gender differences were found in the minimal, average or maximal heart rates.

The men had a lower mean number of isolated ventricular premature beats in the post-exercise than pre-exercise period, whereas among the women the opposite was true (HG). However, the effect of exercise on isolated ventricular premature beats (p = 0.309) and isolated supraventricular premature beats (p = 0.251) did not differ significantly between the sexes in the HG.

Also, there were no statistically significant intragroup differences in isolated ventricular beats between the pre- and post-exercise recording periods among either in men (p = 0.665) or in women (p= 0.555) in the HG (Figure 6a). The same was true in isolated supraventricular premature beats (p = 0.795 and p = 0.896) respectively (Figure 6b). A corresponding result was found with the number of ventricular couplets, runs, and supraventricular runs, respectively (data not shown). Ventricular runs occurred in 2 men and 2 women having 1-5 episodes of ventricular runs with 3-5 consecutive beats at a rate of 121-147 beats per minute.

Totally of forty-five ST segment depressions (range -1.0 - - 7.1mm and 1.10- 41.50 minutes) meeting the criteria mentioned earlier were found among three men and four women in the HG during the two recordings. No angina pectoris was reported in relation to these ST segment depressions. The number of ST segment depressions was comparable between these two recordings, with 24 in the pre-exercise and 21 in the post-exercise period. In four subjects (three women and one man) the number of depressions, however, increased in the post-exercise period. In two subjects the number of ST segment depressions decreased (one woman and one man) in the post-exercise period and remained the same in one subject (a man).

Exercise had no significant effect in either gender on the variables related to ST segment depressions (Table 20). Men in the HG showed significantly more severe ST segment depressions than women as expressed in peak absolute ST depressions and ST-deviation time. Men tended also to have longer periods of ST segment depressions than women. There also was a tendency among men and women to have deeper ST segment depressions during the pre-exercise period than the post-exercise period (Table 20).

When looking at the time period surrounding the acute physical exercise (from 0,1, 2 hours pre-and post-exercise) the number of isolated ventricular premature beats increased considerably in relation to exercise in 5 cases in the men and in 4 cases among the women in the HG (data not shown). The number of isolated supraventricular premature beats increased in 6 cases among the men and 1 case among the women. Furthermore, the ST-deviation time was more pronounced in 7 cases among the men during the observation period (0, 1, 2 hours pre and post-exercise); a similar effect was seen in 2 cases among the women (data not shown).

In general, the cases of more marked arrhythmia and ST segment depressions occurred usually in the exercise hour, and during the pre- or post-exercise hour in both men and women. According to the two-way ANOVA, no significant gender differences were found in the HG in the effect of exercise on these variables. Differences in these variables between the recording hours and sexes did not reach statistical significance either.