Muscle power, which is the product of the velocity and force of muscle contraction, is needed for performing daily habitual tasks and activities. Muscle strength declines with advancing age, starting at the beginning of the sixth decade, and the power-type strength, i.e. the capacity to produce explosive muscle force, declines more drastically than the maximal muscle strength (Izquierdo et al., 1999; Anton et al., 2004). Mechanisms contributing to this development may include the loss of Type II fast-twitch motor units (Lexell et al., 1988), or intrinsic changes in muscle force and power production capacity (Frontera et al., 2000). The age-related strength decrease has been previously reported to be faster in lower extremities than in the upper body (Asmunssen and Heeboll-Nielsen, 1962; Bemben et al., 1991). Recently, Anton et al. (2004) demonstrated similar age-related declines both in the arm and leg muscles.

Strength and power-type strength training are recommended for middle-aged and even elderly people for the purpose of maintaining the functional capacity (Häkkinen et al., 1998; Izquierdo et al., 1999; Jozsi et al., 1999). This is important especially with increasing age, in connection with daily activities and even in prevention of falling (Bassey et al., 1992; Skelton et al., 2002). People are commonly engaged in and familiar with endurance training and resistance training. In natural human movements, however, several physiological functions interact simultaneously, and therefore, all the components of muscular performance should be trained equally. It has been suggested (Häkkinen et al., 1998) that strength training in combination with some explosive types of exercises be recommended as a part of overall physical training to maintain the functional capacity in middle-aged and elderly people. For explosive muscle performance, the underlying factors are muscle fibre type, muscle hypertrophy and enzymatic and neural adaptations.

It is also important to investigate the impact of power-type strength training on the low back and knee muscles and joints, as well as the injury risks and adherence and motivation to training. For being effective in improving the explosive muscle performance, training programmes should be designed so as to be motivating, easy to achieve, effective concerning the time spent in exercises, low in expenses, and they should give consideration to the exercise history and present exercise activity, health status and musculoskeletal symptoms and diseases of the individual. Even the socio-economic status and the social and economic environment should be taken into account when evaluating the actual possibilities for completing the planned programme. The exercises should be integrated in everyday life and take place on a regular basis.

Both in physical training and in the rehabilitation of middle-aged people, the endurance type training is commonly used, e.g. walking, jogging, cycling or swimming. The effects of endurance exercises are well known, and various training modes are established and widely adopted by non- athletic people. However, in everyday life the explosive muscle qualities are also needed in various tasks and reactions, e.g. prevention of falls. Training that affects the explosive muscle qualities should therefore not be ignored, especially when it is known that explosive type strength declines with ageing more drastically than maximal muscle strength. However, physical training has been shown to be effective in preventing the decline of muscle power provided that the intensity, duration and frequency of training is sufficient.

For decades, resistance training has been used for the purpose of achieving strength and power, but this type of training needs special training facilities and equipment. The purpose of this study was to find out an alternative method for exercising the explosive muscle characteristics that would use no or very little equipment, be simple and effective, and feasible for middle-aged sedentary people. The programme should also motivate the participant to continued physical activity after the intervention.

Power-type strength in leg and trunk muscles
Muscle actions are either isometric or dynamic. In isometric actions the muscle length does not change, while dynamic contractions affect the length. Dynamic muscle contractions can further be classified into concentric and eccentric. In concentric contraction the muscle length decreases and in eccentric contraction it increases. Human movement is seldom based on purely isometric, concentric or eccentric muscle contraction. Body segments are periodically organised to impact forces, for instance, in running or jumping, where external forces lengthen the muscle. In these phases, muscles act eccentrically, and the concentric action follows for achieving positive work (Cavagna et al., 1968). A combination of eccentric and concentric muscle actions constitutes what is called stretch shortening cycle (Komi, 1984; Cavanagh, 1988). The eccentric action influences the subsequent concentric phase so that the final contraction is more powerful than a concentric action alone would have been (Komi, 1984). Strength is defined as the maximal amount of force a muscle can generate in a specified movement at a specified movement velocity. The power of muscle contraction is a measure of the total amount of work that a muscle can produce in a given time period. This is determined by the strength of the muscle contraction, by the distance of contraction and the number of contractions in a time period. A performance of daily activities requires both strength and power-type strength, and therefore, muscle conditioning and strength training should be supplemented by exercises with higher velocities. Typical performances requiring explosive power-type strength include various jumps, where the maximal strength level must exceed the load to be moved (i.e. own body). The power-type strength is needed also in high-velocity training requiring acceleration, fast running, and rapid changes of direction (e.g. football, tennis).

Trunk muscles protect the spinal structures against potentially harmful loads and sudden movements (Floyd and Silver, 1955; Troup, 1986). The measurements of trunk muscle velocity, acceleration and torque are important for investigating the stress components of the spine (Beimborn and Morrissey, 1988). Muscle biopsies from diskectomy patients have revealed selective atrophy of fast-twitch fibres in low back muscles (Mattila et al., 1986; Zhu et al., 1989), with physical inactivity presented as one of the possible explanations. Poor trunk muscle function is a potential risk factor for low back disorders (Suzuki and Endo, 1983; Lee et al., 1995).

Measurements in power-type strength training studies
Force production and velocity of the neuromuscular system are the major elements of power-type strength. Vertical jump tests are widely used to evaluate the power-type strength of lower extremities. Measurement of the vertical jumping height is a simple and reliable (reproducibility r = 0.92) method for measuring the explosive force of leg muscles (Bosco et al., 1982; Bosco et al., 1983). The height of vertical jump correlates with 60 m sprint running (Bosco et al., 1983), and also with the maximal power of Wingate cycling test (Maud and Shultz, 1986). Margaria's (Margaria et al., 1966) staircase running test is another simple and reliable test of anaerobic power. Standing long jump has also been widely used in sports research in measuring horizontal explosive force of leg muscles (Bosco et al., 1983; Vandewalle et al., 1987; Manning et al., 1988; Moir et al., 2004). Twenty-metre sprint running is recommended as one of the methods to measure maximal anaerobic performance (Rusko and Nummela, 1996; Moir et al., 2004). Rusko et al. (1993), Rusko and Nummela (1996), and Nummela (1996) developed a method that allows the evaluation of several determinants of maximal anaerobic performance, including the changes in the force of leg muscles and relative to speed in sprint running. Isokinetic knee dynamometers have also been used to test the power of lower extremities (Moffroid et al., 1969; Osternig et al., 1977; Madsen, 1996).

Several studies (Parnianpour et al., 1989; Rytökoski et al., 1994; Hutten and Hermens, 1997) have shown the isoresistive dynamometer measurement of trunk muscle flexion and extension strength and velocity to be reliable and valid.

Perceived health and fitness were assessed by using a five-point Likert scale (poor, fairly poor, average, fairly good, good) that has previously been used by, i.e., Moum, 1992 and Wolinsky and Johnson, 1992. This method has shown to be reliable and consistent with the assessed medical health and its functional consequences (Lundberg and Manderbacka, 1996; Manderbacka, 1998).

Musculoskeletal disorders were inquired about by using the standardised Nordic musculoskeletal questionnaire, which has shown to be a reliable and valid method for that purpose (Kuorinka et al., 1987).

Effects of power-type strength training on leg muscles
Proteins are the major component constituting the contractile apparatus of the muscle. There is a continuous process of protein synthesis and degradation in the body (although the structure of the body is stable). The half-life of proteins determines the rate of adaptation to physical exercise training. The range of variation of the half-life of proteins is from less than one hour to several weeks (Maughan et al., 1997). The contraction velocity of a muscle fibre is determined by the isoform pattern of the contractile proteins. The muscle proteins (i.e myosin heavy chains) Type I, Type IIa and Type IIb are the prime determinants of the muscle contraction velocity. Type I represents the slow and fatigue-resistant muscle contractions, while Type IIa represents the fast, oxidative and fatigue-resistant muscle contractions and Type IIb the fast, fatigable muscle contractions (Staron, 1997). Upon initiation of training, changes in the types of muscle proteins begin to take effect within a couple of training sessions (Staron et al., 1994). Heavy-resistance training promotes hypertrophy in all three fibre types (I, IIa and IIb). The greatest growth is usually seen in Type IIa, followed by Type IIb, and the least growth in Type I fibres. Training with high velocity and at low loads does not lead to hypertrophic changes in fibres. Transitions appear to occur within the Type II subtypes, but there is no convincing evidence of transitions between Types I and II (Deschenes and Kraemer, 2002).

Muscle training is the main contributor to strength and power gains (Coyle et al., 1981; Behm and Sale, 1993). The influence of training is reflected both in neural adaptation and muscle fibre composition (Komi, 1973; Komi et al., 1978; Moritani and DeVries, 1979; Sale, 1988). Ross et al. (2001) also speculated in their review that the nerve conduction velocity might reflect the adaptation of nerve structure, with increased diameter of axon and myelination. This adaptation may decrease the refractory period of the nerve, which possibly allows increased impulse frequency and potentially increased muscle activation. A major part of the improvements in untrained subjects during the initial weeks in power-type strength training is probably due adaptations of the neural system, such as increased motor unit firing frequency, improved motor unit synchronization, increased motor unit excitability, and increase in efferent motor drive. Also, a reduction of the antagonist and improved co-activation of the synergist muscles may explain part of the changes (Häkkinen, 1994). In a study of Aagaard et al. (2002), the major part of the training induced improvements after 14 weeks of resistance training were explained by increases in efferent neural drive.

Power-type strength performance can be improved almost by means of any training method, provided that the training frequency and loading intensity exceed the normal activation of the muscle (Kaneko et al., 1983; Moritani et al., 1987, Häkkinen and Häkkinen, 1995; Kraemer, 1997; Häkkinen et al., 1998; Izquierdo et al., 1999; Jozsi et al., 1999; Häkkinen et al., 2000; Marx et al., 2001). In investigating the strength and muscle power output in upper and lower extremities in athletes engaged in various sports, Izquierdo et al. (2002) found that the maximal power output was produced at higher load condition in lower extremities (45-60% of 1 repetition maximum) than in upper extremities (30-45%). They suggested that the sports-related differences might be explained, in addition to training background, by differences in muscle cross-sectional area, fibre type distribution, and by the different muscle mechanisms of the upper and lower extremities. Kawamori and Haff discussed this finding in their review (2004) and suggested that another possible explanation for the differences may be the fact that during lower extremity exercises a larger part of body mass must be lifted up, compared with the upper extremity exercises. Several studies have shown enhancements in middle-aged and in older subjects in maximal and fast force production (Häkkinen and Häkkinen, 1995; Häkkinen et al., 1998; Izquierdo et al., 2001), in explosive jumping performances (Häkkinen et al., 1998; Häkkinen et al., 2000), and isotonic muscle power output in lower extremities (Jozsi et al., 1999).

Cavagna et al. (1971) were the first who showed that the elastic component of leg muscles provides the additional power that is required for sustaining the maximal velocity during sprint running. Furthermore, in studies of Mero et al. (1981) and Chelly and Denis (2001) multi-jump performances correlated highly with sprint running in young subjects. Consequently, Mero et al. (1981) proposed the drop jump test to be useful in predicting maximal running speed. Also, Young et al. (1995) found a high correlation between concentric squat jump performance and maximal running speed. Sprint running and initial acceleration represent a complex movement where the stretch-shortening cycle is dependent of the adaptation of the neuromuscular system and strength (Mero et al., 1981; Mero and Komi 1986, Cronin et al., 2000). Cronin et al. (2000) found that for stretch-shortening cycle actions of short duration, such in sprint acceleration, the greater maximal strength will lead to greater instantaneous power production. The same authors pointed out that in concentric actions which need high initial power production, such as vertical squat jump, the neuromuscular ability to produce the highest amount of power per time unit is more important than maximal strength. Stone et al. (2003) concluded that improved maximal strength was the primary component in improving the jumping power. Explosive exercises (Linnamo et al., 2000) and sprint training (Sleivert et al., 1995) also seem to facilitate the neuromuscular system.

Three times a week of resistance training is generally recommended for achieving enhancements in muscle strength and power in extremities (Pollock et al., 1998; Feigenbaum and Pollock, 1999). Previous reports indicate that, to achieve training effects, the minimum training frequency should be at least twice a week (Pate et al., 1995; DeMichele et al., 1997; Feigenbaum and Pollock, 1999; Kraemer et al., 2002). Previous studies also show that detraining leads to a decrease in strength and loss of training effect within a few weeks (Häkkinen and Komi, 1983; Narici et al., 1989; Häkkinen et al., 2000).

One of the major exercise methods has been the use of heavy loads to induce recruitment of high-threshold fast Type II motor units by the size principle (Sale, 1988). Another exercise method is to use light loads to maintain the specificity of the exercise velocity and to maximise the mechanical power output. Kaneko et al. (1983) reported that 30% of maximal load resulted in the greatest improvement in maximal mechanical power. There are several studies indicating the specificity of power training (Komi et al., 1982; Häkkinen and Komi, 1985; Scutter et al., 1995). Power-type strength training with lighter loads and higher shortening velocities has been shown to increase the force output at higher velocities, as well as the power development (Häkkinen and Komi, 1985). Muscular power increased significantly when high training volume and high-velocity exercises were used in training (Häkkinen and Häkkinen, 1995; Kraemer, 1997; Marx et al., 2001).

Previous reports support specificity of exercise type, i.e. the greatest training effects are achieved when the same type of training is used both in training and testing (Caiozzo et al., 1981; Kanehisa and Miyashita, 1983; Häkkinen and Komi, 1985; Ewing et al., 1990; Colliander and Tesch, 1990; 1992; Morrissey et al., 1995).

Experimental studies examining the effects of power-type strength training in middle-aged, sedentary men and women have usually compared the pre and post training effects of resistance training. Most of the intervention studies evaluating the effects of power-type strength and resistance training are conducted with younger and physically active subjects. Moreover, randomised, controlled studies in this field are sparse. Especially few are training interventions evaluating both the training effects and the feasibility aspects, including injuries, adherence and motivation. A summary of previous studies with power-type strength training programmes in the training protocol is presented in Table 1.

Effects of power-type strength training on trunk muscles
Despite the large number of different exercise protocols for trunk muscles, scientific research investigating the specific effects of power-type strength training on trunk muscle velocity in healthy subjects is lacking. However, several studies concerning the exercise effects in low back patients have shown that improved muscular fitness, trunk muscle strength and power or spinal flexibility may prevent future low back pain and spinal disorders (Biering-Sorensen, 1983; Suzuki and Endo, 1983; Mayer et al., 1985; Lahad et al., 1994; Harreby et al., 1997; Abenheim et al., 2000). Trunk muscles should be trained by various types of exercises (aerobic, strength and power training) in order to provide many-sided and sufficient loading for lumbar muscles. In a recent study of Pedersen et al. (2004) the authors showed that exercises which focused on reactions to various expected and unexpected sudden trunk loadings together with coordination exercises can improve the response to sudden trunk loading in healthy subjects, without an increase in pre-activation and associated trunk muscle stiffness. Lumbar exercises are recommended in chronic and even in sub-acute low back pain, but not in acute phase (Abenheim et al., 2000). According to previous reports, it appears that a training frequency of 1-2 times a week elicits optimal gains in strength and power in trunk muscles (Graves et al., 1990; Tucci et al., 1992; DeMichele et al., 1997; Pollock et al., 1998). Previous studies (Graves et al., 1990; Pollock et al., 1989; Tucci et al., 1992) have investigated the effects of training frequency on increased strength of lumbar extension muscles, which, unlike the other muscle groups, have a large potential for strength gains. Improved lumbar extension strength can be maintained up to 12 weeks with a very low training frequency (1 session per 2 or 4 weeks), when the volume, type and intensity of training are constant (Tucci et al., 1992).

Feasibility of power-type strength training in middle-aged subjects
Ageing leads to a loss in muscle mass, a decrease of strength and a decline of contractile velocity (Aniansson et al., 1981; Frontera et al., 1991). The main reason for age-related decrease in strength is muscle fibre atrophy (Lexell et al., 1988) and the decreased contractile velocity may be related to a reduction of the relative proportion of fast Type II muscle fibres (Lexell et al., 1988; Proctor et al., 1995). This process accelerates in the beginning of the sixth decade both in men and women (Lexell, 1988; Häkkinen, 1994). In a study among men and women aged between 20 and 84 years, Akima et al. (2001) estimated that the leg extension and flexion strength declined by 8% on decade in women and by 12% in men. Metter et al. (1997) reported that the decrease of muscle power is 10% faster than decrease of strength in ageing men. Savinainen et al. (2004) investigated the changes in physical capacity (hand-grip-, trunk flexion and extension strength and aerobic capacity) during a 16 year follow-up period and found a greater decrease of physical capacity in men (ranging from 11.6% to 33.7%) than in women (ranging from 3.3% to 26.7%).

Muscle strength and the ability of the leg muscles to produce force rapidly are of importance, especially with increasing age, in connection with daily activities, and even in prevention of falling (Bassey et al., 1992; Skelton et al., 2002). Samson et al. (2000) found that the decline of leg muscle strength and functional mobility accelerated in women from the age of 55 years onwards; in men the decline was more gradual. In healthy urban population of 35-, 45- and 55-year-old men and women, the vertical jumping height was 25% greater in 35-year-old men than in 55-year-old men, but the 35-year-old men were only 15% stronger in trunk muscles than 55-year-old men (Viljanen et al., 1991; Era et al., 1992). The average vertical jumping height was at least as good in physically active subjects as in those who were 10 years younger but physically inactive (Kujala et al., 1994). In the same study, the authors observed that mixed training with varied types of exercises for the neuromuscular system enhanced the jumping height most. Korhonen et al. (2003) showed in their recent study that the age-related deterioration in sprint running in former sprint athletes was associated with reduced stride length and increased ground contact time.

For the purpose of maintaining functional capacity, strength and power-type strength training are recommended for middle-aged and elderly people (Häkkinen et al., 1998; Izquierdo et al., 1999). In strength training the minimum of two sessions a week is recommended for the adult healthy population (ACSM, 1998; Feigenbaum and Pollock, 1999; Kraemer et al., 2002). Probably the same frequency is also needed for maintaining and enhancing the power-type strength characteristics.

Previous studies on supervised resistance training programmes (Tsutsumi et al., 1997), controlled circuit weight training programmes (Norvell and Belles, 1993) and anaerobic training programmes (Norris et al., 1990) indicate that these training modes are beneficial both for physical and psychological health. The perception of physical ability and perceived fitness have improved in physical training interventions in adults, independently of the type of activity (Caruso and Gill, 1992; Bravo et al., 1996). Studies evaluating the effects of power-type strength training programmes in middle-aged and older subjects are sparse (see Table 1).

For being effective in enhancing explosive muscle performance, the training programmes designed for middle-aged and older subjects should take into consideration, in addition to age and gender, the existing musculoskeletal symptoms, previous injuries, and exercise history. A population survey (Uitenbroek, 1996) showed that exercise-related injuries constitute a high proportion of all injuries, particularly in men. The amount of previous injuries and exposure time may also increase the risk for injuries (Van Mechelen et al., 1992; 1996). Poor physical condition increases the risk of training induced injuries (Lysens et al., 1991) and highly intensive fitness programmes may even have non-beneficial effects on physical health among less fit subjects in the form of injuries, increased muscle pain, muscle soreness and other training-related inconveniences (Egwu, 1996). When an injury occurs, athletic and well-trained subjects suffer more of post-injury mood disturbances (caused by the loss of active training time) than less trained people (Little, 1969; Smith, 1996).

Exercise programmes should to be safe enough for the exercisers to avoid injuries and musculoskeletal consequences. This is especially important in programmes designed for middle-aged, sedentary men and women. Injuries and musculoskeletal symptoms also influence the exercise motivation. Approximately 30% of adult population in Finland (Helakorpi et al., 1998) and the United States (Caspersen and Merrit, 1995) are sedentary. Physical activity generally declines with age, with a temporary increase in activity at the time of retirement (Bouchard et al., 1994). The decline is greatest when the activity is vigorous and unorganised, and the decrease is greatest in men. Also, men are engaged more often in vigorous physical activities than women (Caspersen et al., 2000; Sallis, 2000).

In Finland, physical activity declines in early adulthood and begins to increase again at the age of 45-54 years (Helakorpi et al., 1998). Physical activity can be promoted by various kinds of interventions. In group-based exercise programmes the adherence has been highest in interventions of short duration (Bij et al., 2002), but the effects are usually temporary (Dishman and Buckworth, 1996). In an aerobic exercise programme, the dropout rate was approximately 50% within six months (Robison and Rogers, 1994). Adherence to physical activity is a complex interaction of personal, behavioural and environmental conditions, including perceived health and fitness, marital status, smoking, obesity, lack of time, previous exercise behaviour, socio-economic status and neighbourhood (Grzywacz and Marks, 2001; Trost et al., 2002). The adherence is lower in high-intensity training, but high training frequency is necessarily not associated with low adherence (Perri et al., 2002). Future adherent behaviour in supervised training programmes is positively influenced by previous physical activity, perceived health and fitness, the spouse's support, agreements and training facilities (Dishman et al., 1985).

Aims of the study
The general purpose of this study was to investigate the effects of a power-type strength training programme on leg and trunk muscles, and to examine the training responses in men and women with high, moderate and low training activity. Additionally, the feasibility of the power-type strength training programme for middle-aged, sedentary men and women was evaluated. The following qualities were set for the programme design: the programme should be simple and practical, and it should encourage and motivate middle-aged men and women to increase their overall physical activity by getting accustomed to and adopting power-type strength training.

The individual studies were performed to specifically answer the following questions:

1. Does the use of light external loading (totalling 2.2 kg) in lower extremities increase the efficiency in power-type strength training exercises? (Study 1)
2. Which training frequency is needed for improved angular velocity of the trunk muscles in power-type strength training in middle-aged men and women? (Study 2 and 3)
3. What is the influence of power-type strength training on perceived health, musculoskeletal symptoms and injuries in middle-aged men and women? (Study 4)
4. What is the adherence rate in men and women, and what are the reasons for dropping out from the power-type strength training programme? (Study 5)
   

Study design
Two hundred and fifty-two (252) subjects volunteered to the study. A total of 171 participants completed the training programme, and 55 subjects dropped out during the training programme (Study 5). The control group consisted of 26 non-exercising volunteers (Figure 1).

For evaluating the impact of light loads attached to the lower extremities, the exercisers were divided into two subgroups, one with light loads and one without any loads. The results of those exercisers whose training attendance was at least twice a week (high training group) were included in the analysis (Study 1).

For evaluating the training frequency vs. training response, the participants were classified into three training frequency groups according to their attendance at the exercises, and to a non-exercising control group. The subjects with training attendance rate > 67% (2-3 times a week) were classified as female and male high training groups; the subjects with training attendance rate between 33% and 67% (1-2 times a week) were classified as female and male moderate training groups, and the subjects with attendance rate < 33% (less than once a week) or with at least six weeks of detraining period at the end of the intervention were classified as female and male low training groups. The numbers of subjects participating in different physiological measurements are presented by the training attendance groups in Table 2.

For analysing the effects of training frequency on trunk muscles, the participants were classified into two training frequency groups (Study 3). The design of the study is presented in Figure 1.

The physical performance measurements were performed and questionnaires were answered one week before the training programme started, and same procedure was carried out one week after the training programme ended. The study was completed within two years: a new controlled and supervised exercise class started once 10-20 subjects had been measured, and the group continued training together for the whole training period of 22 weeks. The training programme included three progressive periods. The orientation period consisted of basic strength and conditioning exercises (6 weeks). The second period consisted of training for explosive strength and velocity (10 weeks), and the last period consisted of velocity training (6 weeks).

Subjects

To be eligible for the intervention, the participants should be middle-aged, healthy and sedentary. All participants were examined by a physician to be qualified to participate. Medical screening included cardiovascular, neurological and musculoskeletal examinations. The physical activity level was assessed by interviewing the subjects (those who trained sports regularly, at least three times a week, or had been training in the past five years were excluded from the study).

Participants (n = 252) were recruited among the staffs of the local university and polytechnic institutes, secondary schools and private companies, or among the participants of retraining courses and the members of a local association of the unemployed. The recruiting information was the same for all. Brochures about the training intervention were attached on billboards providing the following information: "The aim of this study is to develop in practice power-type strength exercises that are simple to perform and feasible for anybody. Exercise sessions are supervised, and various physiological measurements are carried out before and after the intervention". The groups were also reached by e-mail and by visiting people at their jobs, course centres and institutes with the purpose of recruiting volunteers to the intervention. The subjects in the training group (men n = 86, women n = 140) and in the non-exercising control group (men n = 11, women n = 15) were healthy and middle-aged, and most of the subjects were sedentary (Table 3 and Table 4). All subjects were informed of the purpose of the study before they gave their written consent to participate in the study. The Ethical Committee of the Research and Development Centre of the Social Insurance Institution approved the study protocol.

Measurements
After the medical examination the following measurements were carried out on four different days: Vertical squat jump, Trunk muscle performances and all inquiries (Day 1); Maximal anaerobic cycling power (Day 2); Maximal oxygen uptake (Day 3); and Standing long jump and 20 metre running time (Day 4).

Vertical Squat Jump (Study 1)
Vertical Squat Jump (VSJ) (cm) was used to measure the explosive force of leg muscles before and after the intervention. Participants had three attempts in vertical squat jump, with 1-3 minutes' rest between the attempts. The best value (cm) of the three trials was included in the statistical analysis. VSJ was measured by using a contact mat (Newtest powertimer^®, Finland). The recorded flight time (s) was transformed to centimetres (cm) (Bosco et al., 1982; 1983). Participants were barefoot, with knees flexed at 100 degrees, and they held a wooden stick behind the neck to standardise the position of arms and upper body.

20 metre Running Time (Study 1)
20 metre Running time (20mRT) (s) was measured with a flying start, and the first 5 metres were omitted from the calculation of the running time. Participants had three attempts in running speed, with 1-3 minutes' rest between the attempts. The best result was used in the statistical analysis.

Standing Long Jump (Study 4)
Standing Long Jump (SLJ) (cm) was used to measure the explosive force of leg muscles in horizontal direction. Subjects jumped from a standing position, swinging of arms and leg counter-movements were permitted. Participants had three attempts. The best result was used in the statistical analysis.

Maximal Anaerobic Cycling Power (Study 1)
The Maximal Anaerobic Cycling Power (MACP) (W) test is a cycle ergometer modification of an anaerobic power test on a treadmill (Rusko et al., 1993; Rusko and Nummela, 1996). MACP consisted of 3-8 cycling bouts lasting 20 s each with a 100 s recovery between the bouts. The pedalling frequency was constant, 90 rpm for men < 40 yrs and 86 rpm for women and men > 40 yrs. The work rate of the initial bout was determined by the subject's body weight and estimated physical fitness, supposing that sedentary subjects were within the range of average maximal oxygen uptake of population (or less). The load was increased by 30-60 W in general, depending on the subject's age, gender and physical fitness. The work rate was increased after every recovery in equal increments throughout the test. Cycling power, pedalling moment and pedalling frequency were recorded and saved on a computer. A cycling bout was accepted if the pedalling rate was not decreased by 5% or more from the target speed. The subject continued the test until he or she could not cycle at the target rate. The test ended at the moment when the pedalling rate was decreased by 5%. To be acceptable, the final bout was not to be shorter than 12 s. The maximum of the moving average over the 5 s period of cycling power was applied for describing the maximal anaerobic cycling power of the subject. The cycle ergometer used in the test was RE 820 (Rodby Elektroniks AB, Södertälje, Sweden), which was modified to give power output of 1000 W with high pedalling rate.

Maximal oxygen uptake (Study 1)
Maximal oxygen uptake (VO₂max) (ml·kg^-1·min^-1) was measured to evaluate the subject's endurance capacity. A 2-min incremental exercise test on the electromagnetically controlled cycle ergometer (Rodby Ergometer RE 820^®, Södertälje, Sweden) until volitional exhaustion or fatigue of the lower limbs was employed for measuring the VO₂max. The subjects pedalled at a constant frequency of 60 rpm. The test was preceded by a 4-min warm-up at 30 W to become familiar with the pedalling frequency, mouthpiece and nose clips. Thereafter, work rate was increased every 2nd min, with equal increments throughout the test. The increments were individually determined (10-25 W) on the basis of the subject's physical fitness to reach the maximum work rate in approximately 12-15 min. The test continued until the subject was unable to maintain pedalling frequency above 45 rpm. Respiratory gas exchange variables were determined continuously with a breath-by-breath method suing the SensorMedics Vmax 229^® equipment. The VO₂ values were averaged over the breath-by-breath values of the 30-second intervals. VO₂max was recorded as the highest averaged value at the maximum work rate. The corresponding heart rate and work rate were recorded and represented their maximums. Subjects rated their perceived exertion using the Borg scale 6-20 (Borg, 1982) and the amount of fatigue in their lower limbs on scale 1-5 every 4 minutes at the beginning and every 2 minutes later on during the test and at the end of the test in order to evaluate subjective feelings along the whole exercise test and the character of subjective maximum.

Isometric and dynamic trunk Flexion and Extension torques and angular velocities (Study 2, 3)
Isometric trunk flexion (IsomFL) (Nm) and extension (IsomEX) (Nm) torques, and the trunk flexion (FLTorq) (Nm) and extension (EXTorq) (Nm) torques during dynamic actions, and the angular velocities during flexion (FL_vel) (deg·s^-1) and extension (EX_vel) (deg·s^-1) were measured by using a triaxial, isoresistive lumbar dynamometer (Isostation B-200^®, Isotechnologies, Hillsborough, NC, USA). The system allows simultaneous measurement of the velocity, angular position and torque of the three spatial axes of the body spine.

Questionnaires (Study 1, 4, 5)
Questionnaires were used to inquire about the physical activity, smoking, employment status, motivation for exercising, and perceived health, fitness and musculoskeletal disorders. The participants filled in a questionnaire asking yes or no questions about the present and previous physical activity (excluding school-time sports activities), smoking (yes or no) and employment (yes or no).

Perceived health and fitness were assessed by using a five-point Likert scale (poor, fairly poor, average, fairly good, good) used by, among others, Moum (1992) and Wolinsky and Johnson (1992). This method has shown to be reliable and consistent with the assessed medical health and its functional consequences (Lundberg and Manderbacka, 1996; Manderbacka, 1998).

For the assessment of musculoskeletal disorders, the standardised Nordic musculoskeletal questionnaire (Kuorinka et al., 1987) was used. The subjects were inquired about the presence of neck, shoulder, low back, hip, knee and ankle symptoms during the preceding six months. Further, the participants were in advance instructed to report the instructor about any injuries occurring during the training programme, and to describe the injuries in detail. In order to minimise the number of missing reports, the participants were given a questionnaire form for reporting the injuries. They were also asked to evaluate whether the injury was acute or a result of overuse.

Training
The power-type strength training programme was based on the general training principle with the exercises performed with low loads, but with high movement velocities. The aim was to activate the muscles subject to training by various exercises to a high or maximal degree, with a short activation time. This type of training leads to improvements primarily in the earlier force portion of the force-time curve or the higher velocity portions of the force-velocity curve (Häkkinen, 1994).

Training sessions were supervised and controlled by a qualified instructor. The duration of the training programme was 22 weeks, including 52 training sessions, which lasted 60 minutes each. The targeted exercise frequency was three times a week. The training programme (described in detail in original Studies 1, 3 and 5) was progressive, with an emphasis on power-type strength training. The programme included the following three periods: the first period of 6 weeks consisted of basic physical exercises, the second period of 10 weeks consisted of power-type strength training and the third period of 6 weeks consisted of power-type strength and velocity training. The purpose of the first period was to familiarise the exercisers with physical training and to enhance muscle strength and co-ordination skills. The second period consisted of power-type strength training with submaximal and maximal intensity. The third period consisted of power-type strength and high-velocity training with maximal intensity. During the first exercise week the intensity of training was determined individually for each participant on the basis of maximal Number of Repetitions subjects performed during 60 s (NR). The maximal Number of Repetitions during 60 seconds was calculated for various types of exercises. The exercises focused on leg (approximately 60%) and trunk muscles (approximately 40%). Training was carried out in male and female exercise classes of 10-20 subjects. After the first 6 week period exercisers were divided into Light Load (LL) or into No Load (NOL) groups. Exercisers in the LL group had 1.1 kg weights in each ankle during all exercises. Each exercise class consisted of either LL or NOL exercisers.

Statistical analyses
The General Linear Models Procedure (GLM) of the Statistical Analysis System (SAS/STAT, 1989) was applied to compare the changes between the groups and to evaluate possible interaction between group and gender, and for multiple comparisons between the groups. Mean changes and lower and upper 0.95 confidence limits of the outcome variables in three different training activity groups and the control group were calculated by gender. Means, standard deviations and correlation coefficients were calculated by standardised methods.

For the comparison of the changes in the Light Load (LL) and No Load (NOL) groups (Study 1), the individual data for VSJ, 20mRT, MACP and VO₂max at the baseline and after the intervention were presented in scatter-plots. The chi-square test was applied to examine the distribution of the type of previous exercise activity (four categories: endurance type, power-type, no exercise history, and other leisure activity than endurance or power-type), and the pre-training shoulder-neck, low back, hip, knee and ankle symptoms. The t-tests were used to analyse the differences between the mean values of the baseline measures and the changes in the LL and NOL groups.

The Linear Structural Relationships (LISREL) model was used to analyse the flexion and extension movement velocities and the reliability of the measurement (Study 2). The LISREL model facilitated understanding the nature of the measured flexion and extension movement, movement velocity and range.

For the analysis of the trunk muscle performances (Study 3), one-way analysis of variance (Procedure GLM of the Statistical Analysis System) was applied to compare the changes of the outcome measures between two (high vs. low training) groups by gender. The chi-square tests were applied to investigate the differences in low back symptoms, perceived health, physical activity, and smoking between the groups. Mean changes and lower and upper 0.95 confidence limits of the outcome variables in three different training activity groups and the control group were calculated by gender. Means, standard deviations and correlation coefficients were calculated by standardised methods.

The chi-square test and the GLM procedure analysis were applied to examine the distribution of the musculoskeletal symptoms, smoking, employment, perceived health, perceived fitness, overall physical activity (present and previous) and smoking between the groups. To investigate whether there were any changes in perceived health, in perceived fitness, or in the incidence of knee and low back symptoms during the training programme, the marginal probabilities of two-dimensional contingency tables were used and analysed by gender using Proc Catmod SAS/STAT (Study 4). The chi-square tests and the GLM Procedure analysis were applied to investigate the associations with training activity and employment, smoking and age (Study 5).

Study subjects and training effects on leg muscle performances in exercisers and non-training controls
One hundred and seventy-one (171) participants (64 men and 107 women) completed the training programme. The control group consisted of 26 non-exercising volunteers (11 men and 15 women). Of the initial group, 55 dropped out (22 men and 33 women) during the training programme. The overall dropout rate in this study was 24%. The overall training activity was 63% for those (n = 171) who completed the programme. The baseline characteristics of the subjects are presented in Table 3 and Table 4, including age, body mass index, perceived health and fitness, jump performance, and knee and low back symptoms. At the baseline, the exercisers (n = 171) did not differ from the non-exercising controls (n = 18) or dropouts (n = 55).

In performances requiring power-type strength the most visible training effects were observed in vertical squat jump with 18% improvement in exercisers (15% in men and 20% in women), while in controls the increase was 1% (no change in men and 2% increase in women). Trunk flexion velocity improved in exercisers by 14% (13% in men and 15% in women), whereas in controls the increase was 3% (5% in men and 1% in women). The improvement in extension velocity was 16% (15% in men and 17% in women) in exercisers, while the increase in controls was 5% (7% in men and 3% in women). The exercisers improved their results in standing long jump by 4% (1% decrease in controls), 20 metre running time by 5% (no change in controls) and maximal anaerobic cycling power by 6% (1% increase in controls). In maximal oxygen uptake, which was measured for individual endurance capacity, a 4% improvement was observed in exercisers (2% decrease in controls). The changes in the non-training control group were not significant in any of the measurements. The pre- and post-intervention values and the percentage changes of the vertical squat jump (cm), standing long jump (cm), 20 metre running time (s) and maximal anaerobic cycling power (W) for the different training activity groups in men and women are presented in Table 5 and Table 6.

There were significant differences in changes between the groups in vertical squat jump (F = 19.33, df = 3, p = 0.0001), in standing long jump (F = 4.20, df = 3, p = 0.007), in maximal oxygen uptake (F = 3.10, df = 3, p = 0.03), in 20 metre running time (F = 11.35, df = 3, p = 0.0001), and in maximal anaerobic cycling power(F = 4.83, df = 3, p = 0.0003). In vertical squat jump the changes were higher in all of the training groups compared with the controls (p < 0.05). In 20 metre running time, the changes were greater in the high and moderate training groups compared with the controls (p < 0.05). In standing long jump, in maximal anaerobic cycling power and in maximal oxygen uptake the changes were greater in the high training group compared with the controls (p < 0.05). In maximal anaerobic cycling power the changes were greater in the high training group compared with the low training group (p < 0.05).

Mean changes and 0.95 confidence limits in vertical squat jump (Figure 2a), 20 metre running time (Figure 2b), standing long jump (Figure 2c), maximal anaerobic cycling power (Figure 2d), and maximal oxygen uptake (Figure 2e) are shown for the three training activity and control groups and by gender.

No significant gender differences were observed in the changes of vertical squat jump, standing long jump or in the maximal oxygen uptake after the training programme. Women achieved greater changes after the training in 20 metre running time (F = 10.62, df = 1, p = 0.01), while men achieved greater changes after the training in maximal anaerobic cycling test (F = 5.86, df = 1, p = 0.02).

Effects of external light load vs. no load on muscle power in lower extremities (Study 1)
No significant differences were found between the light load and no load groups concerning the type of previous exercise activity, perceived health and fitness, and shoulder-neck, low back, hip, knee and ankle symptoms at the baseline (Study 1), or immediately after the intervention. No significant differences between the groups were observed in body weight after the intervention. There were no differences in exercise induced injuries between the light load and no load groups. At baseline, no differences between the groups were observed in vertical squat jump, 20 metre running time, maximal anaerobic cycling power or in maximal oxygen uptake values (Study 1). After the intervention, subjects in the light load group (with 2.2 kg external loading in ankles) improved vertical squat jump by 23% (p = 0.03) and maximal anaerobic cycling test by 12% (p = 0.05). The changes are significant compared with the no load group (16% increase in vertical squat jump and 5% increase in maximal anaerobic cycling power) (Study 1). No differences were observed in 20 metre running time between the light load and no load groups.

Measurement of trunk flexion and extension velocities (Study 2)
The analysis of the repetitive trunk muscle flexion and extension velocities at three angular phases showed that the peak velocities of the second phases of these movements (between 15° and 35° in flexion and between 20° and 0° in extension) correlated highly (r = 0.99) with the peak velocity of the whole movement (from -5° to 55° in flexion and from 40° to 20° in extension) both in flexion and extension. Correlations were high both before and after the 22-week intervention. The LISREL analysis showed high reliability in peak flexion (r = 0.78) and extension (r = 0.81) velocities between the pre- and post-intervention values (Study 2).

Effects of power-type strength training on trunk muscle performances (Study 3)
The age, weight and height or lumbar spine measurements at baseline of women and men did not differ between the groups (Study 3), and no difference between the groups were found in self-reported low back symptoms, perceived health and fitness, physical activity and smoking at baseline (Study 3).

Differences were observed in training induced changes of peak flexion velocity between the female and male high training groups vs. female (F = 7.54, p = 0.008) and male (F = 4.86, p = 0.03) low training groups, and in peak extension velocity correspondingly (F = 9.07, p = 0.003 for women, and F = 12.31, p = 0.001 for men). The training induced change in peak flexion velocity was 13 deg/s greater both in the female and male high training groups than in the corresponding low training groups (p < 0.05). The change in peak extension velocity was 15 deg·s^-1 higher in the female high training group than in the female low training group (p < 0.05), and 18 deg·s^-1 higher in the male high training group than in the male low training group (p < 0.05).

Effects of training on perceived health and fitness (Study 4)
Both male and female exercisers perceived that their physical fitness improved (p < 0.01 for men and p < 0.0001 for women) during the intervention period. Perceived physical health improved in female exercisers only (p < 0.001).

The male dropouts showed a significantly poorer perceived health than the exercising men (p < 0.01). Men attended 62 ± 23% (mean ± SD) and women 66 ± 18% of the scheduled training sessions. Twelve men and 25 women attended 80% or more of the scheduled training sessions. Those with a training attendance > 50% showed improved perceived fitness; in women the change was significant (p < 0.05). While significant improvements occurred in perceived physical fitness (men and women) and in perceived physical health (women), the control subjects (n = 18) did not show any changes in either of these variables.

Men with improved vertical squat jump performance showed improved perceived health (p < 0.05) and women with improved standing long jump performance showed increased perceived fitness (p < 0.05). No such trends were observed in the controls.

Knee and low back symptoms, and training induced injuries during the intervention (Study 4)
In exercisers the number of men reporting no low back or knee symptoms increased from 20 at the baseline to 25 after the intervention, and in women the corresponding values were 49 and 55. The frequency of low back symptoms decreased by 13% (p = 0.06) in men and by 10% (p = 0.06) in women. Knee symptoms increased by 2% (p = 0.8) in men and by 5% (p = 0.35) in women. Among the controls, low back symptoms decreased by 11% and knee symptoms increased by 6%. Exercising men who reported more knee symptoms after the intervention had higher body mass index (28 ± 3, p < 0.05) than men on average (26 ± 3). The same was not observed in women. Of those participants who had no knee symptoms before the intervention, 17% reported symptoms in their knees during the programme, and 14% of the participants who had knee symptoms before the intervention reported after the programme that their symptoms had relieved.

The injury rate during training sessions was on average 10% (n = 16); 19% for men (n = 10) and 6% (n = 6) for women. The injuries included non-specific knee pain (19%), sprain or strain in thigh (37%) and calf muscles (13%), twisted ankle (19%), muscle cramp in low back (6%) and strained shoulder muscles (6%). Five participants sustained overuse injuries during the intervention, including non-specific knee pain (n = 2), low back pain (n = 2) and pain in calf muscle (n = 1).

Adherence to training programme (Study 5)
The analysis of the data concerning all the participants (n = 226) who started to exercise showed that the training activity was associated with unemployment (F = 15.2, p < 0.0001), smoking (F = 5.21, p = 0.02) and age (F = 3.88, p = 0.05) with the younger subjects having lower adherence to the programme. No association was observed between training activity and gender, body mass index, shoulder, neck, low back, knee or ankle symptoms, perceived health or fitness. Twenty-two of the dropouts interrupted because of lack of motivation, 18 because of lack of time, 8 because of an exercise induced musculoskeletal symptom, and 7 because of other reasons.

The subjects' age and body mass index, the distribution of smoking, previous and present physical exercise activity, the rate of physical leisure activity, perceived health, perceived fitness and musculoskeletal symptoms are presented in Table 2 for men and in Table 3 for women.

Of all female smokers 57% dropped out of the training programme, while only 15% of female non-smokers dropped out (p < 0.05). Of all female participants the unemployed women smoked significantly more (p < 0.01); this was not observed in men.

Among the subjects in age groups < 40 years, 40-49 years and > 50 years who completed the training programme, the significantly lowest training attendance (%) was found in women under 40 years of age (58 ± 19%) (p < 0.05). The attendance rate was 66 ± 19% in women > 50 years and 69 ± 16% in women aged 40-49 years.

The overall unemployment rate was 21%. The unemployment rate was 47% among the dropouts, while it was 8% in high training, 16% in moderate training and 17% in low training groups. Nineteen (19) percent of dropouts perceived their fitness good, whereas 48% of exercisers had good perceived fitness (p = 0.02). Most of the subjects trained both for physical and mental well-being (approximately 43%), the second frequent motive for physical training was mental well-being (approximately 25%).