Review Article

Sports Medicine II (I); 20-51 , 1991 0112-1642/ 91 / 0001-0020/$16.00/0 © Adis International Limited All rights reserved. SPORT2354

Health Effects of Recreational Running in Women Some Epidemiological and Preventive Aspects

Bernard Marti Institute of Social and Preventive Medicine, University of Zurich, Switzerland

Contents

Summary .. .. ........... ................. ............. ............. ... ... .......... ... ........ ...... .. ... ........... ....... ........ ........... 20 I. Physiological Aspects of Female Runners ................ ...... ......................... .. ...... .. ... ........ .. ..... 22 2. Exercise, Physical Fitness and Prevention of Cardiovascular Diseases in Women .... .... 25 3. Running and the Blood Lipid Profile ........ ........ .......................................... .... ........ .. ... .. ... .. 26 4. Ph ysical Fitness, Exerci se, and Hypertension .................. ............ ........ ............ .. ........ ......... 28 5. Exercise, Insulin, and Diabetes .................................. ........ .. .. .. ............ .. .. ....... ..................... 29 6. Running and Weight Control .. .. ............. .. .... ............. .......... .... ... .. ................ .. ............ .......... 29 7. Running and Amenorrhoea ......... .... .... ..... .. .. ....... ... ..... ....... ..... ......... .. ... ......... .. .. ... ..... .. ....... . 30 8. Running. Menstrual Function, and Bone Density ........ .. ......... ........ .... .... ..... ........ .. ............ 32 9. Running-Related Injuries and Complaints ............ .. .. .... .... .... .......... .. .... ........ .... ...... .. .. ........ 35 10. Running and Iron Status .. .. .. ................ .. .. .............. .... .. .... .. ....... .. ............ .. .......... ..... .... ...... 39 II. Exercise and Cancer .. ........ ........ ................ .... ............ .. .......... .. .......................... ........ .......... 40 12. Exercise and Psychological Well-Being .. .. .. ........ .. ........ .. .. .. .. .... .. .. .. ........ ..... .. .. .... ...... .. .. ..... 41 13. Running. Smoking. and Other Health Habits .............. ................................ .... ................. 42 14. Ph ysician Visits of Runners .. .......... .. .............................. .. .................... .. ........ .............. ...... 43 15. Limitations of Data ...... .............. ... ...... .... ........ .. .. .... .......... .. ........ ........ .. .. .. ...... .. .. ...... .......... 43 16. Conclusions ....... .. .. ..... .. ..... ........ ......... ......... ..... .. .. ......... ....... ... .... ..... ... ..... ... .. .. ..... .. .. ..... ....... 45

Summary

Estimated maximum oxygen uptake of middle-aged nonelite road race entrants is around 45 to 50 ml/ kg/min , which is 40 to 100% higher than values from the female general population. Endurance training, low bodyweight, and non smoking of runners explain part of, but not the whole, difference in aerobic capacity observed between athletes and the general population. Sedentary women can improve cardiorespiratory fitness through aerobic exercise programmes, and the women with the lowest level of initial fitness have the highest proportional improvement following training. Regularly exercising women have a significantly reduced risk of fatal and nonfatal coronary events, and low cardiorespiratory fitness is associated with an increased risk of death and nonfatal stroke. The influence of habitual running on the female blood lipid profile is not clear. Cross-sectional studies have found elevated HDL cholesterol concentrations in distance runners, but intervention studies on the effect of jogging on lipid and lipoprotein levels have prov ided equivocal results. A higher level of physical fitness is associated with a lower risk to subsequentl y develop hypertension. Experimental studies have shown that moderate intensity aerobic exercise (40 to 60% V02max) is able to reduce blood pressure significantly in hyperte nsive subjects. An athletic lifestyle may be associated with a reduced risk of adult-onset diabetes mellitus (via an exercise-induced increase in insulin

Running and Health in Women

21

sensitivity), and with a reduced risk of cancers of the reproductive system, breast, and colon. Recreational running is also correlated with better weight control. Surveys of recreational and elite distance runners show a great variability in the prevalence of secondary amenorrhoea, between I and 44%. Environmental factors determining the risk of amenorrhoea in runners are low body fat content, mileage, and nutritional inadequacy, with low intakes of calories, protein, and fat. Amenorrhoeic athletes in their third and fourth decade have lower vertebral bone density, which is improved after resumption of menses but does not completely reach age-specific average values. Regardless of menstrual status, the effectiveness of exercise to maintain bone mass throughout life is an important issue. Habitual exercise is associated with increased bone density of the spine both in premenopausal and postmenopausal women. Several controlled training studies suggest that postmenopausal women may at least retard their bone loss with regular aerobic exercise. Running-related injuries and complaints are common in recreational joggers, even though the reported I-year incidence, varying between 14 and approximately 50%, depends on injury definition. Mileage and a history of previous running injury are known risk factors. Overweight, irregular menses, and absence of oral contraceptive use have been identified as risk factors in single studies. Female gender itself does not seem to be a major risk factor of running injuries among habitually active subjects, but it may be a relevant factor for sedentary subjects taking up jogging. Regarding the effect of habitual running on the development of osteoarthritis in weight-bearing joints, available data suggest that reasonable recreational exercise, carried out within limits of comfort, putting joints through normal motions, without underlying joint abnormality, even over many years, is unlikely to lead to significant joint injury. Habitual exercise is associated with reductions in anxiety and depression as well as increased self-esteem. The latter is an empirically supported outcome of exercise, and programmes of aerobic exercise seem also to be effective in reducing state anxiety and symptoms of mild depression. The prevalence of anorexia nervosa among competitive distance runners is not higher than among the general population, but it is the best runners who are most likely to be anorectic. Nonsmoking is highly prevalent among runners, and habitual runners who smoke have a quit rate of roughly 75%, with the rate of smoking cessation being related to mileage. Compared with the general population, age-matched runners have significantly fewer medical consultations, and probably less missed work days. Little data on the health effects of recreational, in contrast to competitive, running is available, and most epidemiological studies on prevention through exercise suffer from methodological shortcomings that hamper the ability to evaluate the health potential of aerobic exercise in an unbiased way. Nevertheless, there is a broad consensus that an energy expenditure of at least ISO to 400 kcal/daY (corresponding to jogging 2.5 to 6 km/ day) at a moderate intensity, should be the goal for health-oriented exercise.

With regard to historical aspects of fitness in the modern world, it was recently stated that 'the pursuit of physical fitness has undergone a transformation from being an able-bodied male concept achieved by a narrow range of activities, to one that embraces both sexes, all ages, and is achieved by an incredibly wide variety of activities. The female factor in this progress must feature prominently in any review, and it is ail the more impressive because of the unfavourable circumstances

against which this contribution has been made' (Redmond 1988). Evidence suggests that there has been a significant increase in leisure-time exercise during the 1970s and 1980s in several developed countries (Marti et aI. I 988a; Stephens 1987). Women may even have increased their activity levels more than men (Stephens 1987). Competitive running has attracted considerable attention over the years as an obvious growth area. Admittedly, recent data from the United States on race

22

participation suggest that marathons as well as 10km events have been losing popularity in both men and women since 1983/1984 (Stephens 1987), while in Europe the mean age of joggers has been found to increase - a phenomenon pointing to a possible cohort effect (Marti & Vader 1987). On the other hand, very little is known on the prev~ alence of regular, noncompetitive, recreational jogging, and certainly it is too early to write the obituary of running and jogging as a mass participation sport. However, some authors even question whether there has been a true increase in population-wide levels of exercise during the 1980s, given the possibility of a potentially large recall and social desirability bias in questionnaire surveys (Brooks 1988). Due to increased public awareness of the health benefits of exercise and increasing social or peer pressures to participate, the sedentary respondents may inflate their reported participation level of physical activity. Examination of an objective indicator of physical activity can help validate the self-reports of current exercise participation. Blair and coworkers (1987a) calculated mean annual maximal treadmill stress test times of healthy Cooper Clinic men and women at first visit and found a fairly consistent, solid increase in maximal treadmill time for both men and women from 1975 through 1985. This indicates that at least in predominantly white, well-educated, middle to upper socioeconomic class people the self-reports of physical participation are reasonably accurate and that the overall trend is towards an increasingly active population. Nevertheless, according to 1985 survey data from the United States, only 7% of females and 8% of males aged 18 to 65 years were regularly and appropriately physically active, i.e. in exercise which involves large muscle groups in dynamic movement for periods of 20 minutes or longer, 3 or more days/week and which is performed at an intensity of 60% or greater of an individual's cardiorespiratory capacity (Caspersen et al. 1986). The corresponding figures may be slightly higher for Canada (Schoenborn & Stephens 1988; Shephard 1988) but for most other industrialised

Sports Medicine II (I) 1991

countries no data of similar specificity are available. The idea that regular physical activity may be necessary for optimal health is receiving increasing attention in both the medical and lay literature. Physical activity has been associated with the prevention and control of numerous medical conditions, such as coronary heart disease (CHD), hypertension, non-insulin-dependent diabetes mellitus (NIDDM), osteoporosis, obesity, and mental health problems, but also with a number of specific risks, such as injuries and diseases of the musculoskeletal system. For example, concerns have been raised that running might accelerate the development of osteoarthritis in weightbearing joints, and that runners in a sense might be sacrificing their joints to save their hearts. This overview summarises the evidence linking regular exercise (with a particular emphasis on recreational running) to a variety of female health benefits and risks. The purpose is not to provide an exhaustive discussion of the whole literature dealing with different health aspects of female exercisers; rather, this review focuses on a restricted number of recent communications.

1. Physiological Aspects of Female Runners It is commonly accepted that there are typical physiological and morphological gender differences which are responsible for differences in athletic performance between women and men. However, it is interesting to note that there are probably no relevant differences between genders in central or peripheral cardiovascular adaptations to aerobic training (Lewis et al. 1986). This is reflected by the fact that many important training studies have investigated the physiological adaptation processes in mixed male and female samples. For example, Hickson et al. (1981) showed that the effect of a high intensity daily exercise programme on maximum oxygen uptake (V02max) is achieved within as little time as 3 weeks and that any further increase in V02max requires an increase in the training stimulus. Hoppeler et al. (1985) described sim-

Running and Health in Women

ilar exercise-induced, ultrastructural changes of skeletal muscles, such as an increased volume density of mitochondria, in women and men undergoing an endurance training. Correspondingly, a recent study (Moore et al. 1987) found no gender differences in the effects of training and detraining on substrate utilisation responses during submaximal exercise. In a study from the Netherlands (Verstappen et al. 1989), 18 female and 60 male untrained subjects participated in a long term endurance training programme to be completed after 1.5 years with running a marathon. These authors confirmed the rapidity of adaptation OfV02max to a given training intensity (Pedersen & Jorgensen 1978) but they also found a steady, long term increase of about 10% in running velocity at a blood lactate concentration of 4 mmol/L which tended to occur earlier during the programme in female than in male runners. Although a genetic basis of endurance capacity cannot be discounted (Bouchard et al. 1988), there is ample evidence from training studies that sedentary women can improve their cardiorespiratory fitness through aerobic exercise programmes. Drinkwater (1989) reviewed 13 training studies and found that those women with the lowest levels of initial fitness often had the highest proportional improvement following training. On average, 181 women aged 18 to 29 years with pretraining V02max less than 40 mlfkgfmin exhibited a training-related increase OfV02max of 19.5% (from 35.6 to 42.5 mlfkgfmin), while the corresponding improvement in the 105 women with pretraining V02max greater than 40 mlfkgfmin was only 9.1 % (from 44.4 to 48.4 mlfkgfmin). Hagberg and coworkers (1989) found a 22% increase in V02max after 26 weeks of training 3 times per week for 40 minutes at 70 to 80% V02max in 47 men and women aged 70 to 79 years. They concluded that healthy men and women in their seventies can respond to prolonged endurance exercise training with adaptations similar to those of younger individuals. It must be borne in mind, however, that there is a considerable variation in the trainability ofhumans, especially regarding individual sensitivity to

23

long term training for aerobic performance, which is largely inherited (Bouchard et al. 1988). Several detailed, cross-sectional descriptions of physiological profiles of elite women distance runners are available (Pate et al. 1987; Sparling et al. 1987; Upton et al. 1984; Wilmore & Brown 1974), but only relatively little data on recreational runners (table I). Wilmore and Brown (1974) drew attention to the high average V02max ("" 60 mlfkgf min) and the low body fat content ("" 15%) of 11 elite long distance runners from the United States. In a study of middle-aged women older than 30 years by Upton et al. (1984), V02max was 56 mlf kgfmin in marathoners, 49 mlfkg/min in 10km runners, and 31 mlfkg/min in sedentary controls. A recent investigation on female elite and good local runners from the United States (Pate et al. 1987; Sparling et al. 1987) reported even higher V02max values, 67 mlfkgfmin for elite runners and 59 mlf kgfmin for good local runners. In a survey of female participants in a popular 16km road race in Switzerland, V02max equivalents were estimated from age-adjusted running-times; these values appear to be somewhat lower; for example, around 50 mlfkgfmin for those training 25 km/week (Marti 1988). These values of maximum oxygen uptake observed in female runners are considerably higher than those reported from the Swiss general population (Gutzwiller et al. 1985), a population sample in New England cities (Siconolfi et al. 1985), a nonrepresentative sample of 212 women US recruits (Vogel et al. 1986) and those recently reported by Drinkwater (1989), as shown in table I. What is the relative importance of factors determining female aerobic capacity? In a cross-sectional investigation of 109 women aged 10 to 68 years, Drinkwater and coworkers (1975) found apparently little effect of aging on V02max from ages 20 to 49 years, but a clear decrease from age 50 onwards. On the other hand, there was a wide interindividual variation in V02max in the 20 to 49 age range that was at least to some extent determined by habitual activity. In the Swiss study of 16km road race participants (Marti 1988), habitual running activity, relative weight, smoking

24

Sports Medicine 11 (l) 1991

Table I. 'il02max of female runners and reference populations, by runners' status or by age Study

Subjects

n

Age

Mean distance 'ilO 2max run (ml/kg/min) (km/week)

32 ± 5 43 ± 12

No data 37

59 ± 7 43 ± 11

38 ± 33 ± 39 ± 27 ± 29 ± 358

74 40

56 ± 49 ± 31 ± 67 ± 59 ± 47a

Runners Wilmore & Brown (1974) Kavanagh & Shephard (1977) Upton et al. (1984)

Sparling et al. (1987) Pate et al. (1987) Marti (1988)

Elite runners Master athletes

11 7

Marathoners 10km runners Sedentary controls Elite runners Local good runners 16km runners (nonelite)

42 10 37 16 14 428

6 3 6 5 3

104 63 10 25 50

6 3 4 4 5

50 54

Reference populations Siconolfi et al. (1985)

Random sample of 2 New England cities

Gutzwiller et al. (1985)

Residents of a Swiss city

Vogel et al. (1986) Drinkwater (1989)

Female US recruits Nonathletic women b

86 54 24 41 103 95 127 212 135 136 117

a b

18-29 30-39 40-49 50-59 30-39 40-49 50-59 17-25 20-29 30-39 40-49

29 25 21 18 36 33 29 38 39 31 27

All 'il02max values were estimated with linear regression of 'ilo2max on km/week and adjusted to age 35 years. Data compiled from 21 studies (weighted means).

status and age explained 51 % of the variance observed in 16krn running times. Based on multivariable regression analyses, it was estimated that running, low relative weight and frequent nonsmoking explained part, but not the whole, of difference in aerobic capacity observed between these female runners and the general population. This was in contrast to findings from the male participants in the same 16krn event, where the entire difference in estimated V02max between male recreational runners and the general population could be attributed to training and lifestyle characteristics (Marti et al. 1988b). It was concluded from these analyses that female, but not male, participants in that specific road race, held in Central Europe in 1984, represented to a certain degree a

selection of endowed subjects with above-average fitness. With respect to physiological differences between women and men relevant to exercise, after puberty the V02max (normalised for bodyweight) of the female is around 70 to 75% of the male value at an equivalent age (Wilmore 1984). When evaluating athletes, these differences between sexes are reduced considerably. Wilmore and Brown (1974) were among the first to emphasise the importance of differences in body composition when trying to explain the sex differential in V02max per kg of bodyweight. In their study of 11 elite women distance runners, they reported V02max values which were 16% lower than equally trained male distance runners when the values were expressed relative to

Running and Health in Women

bodyweight, but only 7.8% difference when expressed relative to lean body mass. Based on multivariate analyses of data from 34 women (V02max 52 ± 5 ml/kgjmin) and 34 men (61 ± 5 ml/kgj min), Sparling and Cureton (1984) concluded that differences among similarly trained male and female recreational runners in 12-minute run performance are associated primarily with differences in body fat, to a much lesser degree with differences in cardiorespiratory capacity, and only minimally with differences in running economy. Thus, the word 'inferior' is used too often in comparing women's athletic performance or aerobic power with values for males (Drinkwater 1989). Such judgmental phrases may lead to the wrong assumption that men's higher scores in physical fitness confer an added health benefit. In fact, mortality statistics on cardiovascular diseases suggest otherwise. Therefore, women's physical fitness and athletic performance should be interpreted in terms of their physiological capacity and their health concerns.

2. Exercise, Physical Fitness and Prevention of Cardiovascular Diseases in Women Multiple well-designed cohort studies have shown a strong association between physical activity and the prevention of coronary heart disease (CHD) in men. The evidence regarding women has been inconclusive although women have not been extensively studied (Harris et al. 1989). In 1987, a comprehensive review of all published studies that provided sufficient data to calculate a relative risk or odds ratio for CHD at different levels of physical activity found only 4 studies out of 43 that had investigated women (Powell et al. 1987). In 2 cohort studies, in women from Framingham (Kannel & Sorlie 1979) and from Gothenburg, Sweden (lapidus & Bengtsson 1986), there was no significant protective effect of increased .Ieisure activity, which may be due to use of assessment tools that are more appropriate for 'male' than 'female' activity patterns. A Finnish cohort study found a modest but statistically significant excess risk of is-

25

chaemic heart disease in women with low leisuretime exercise which, however, became nonsignificant after adjustment for cigarette smoking, serum cholesterol concentration and blood pressure level (Salonen et al. 1982, 1988). A case-control study from the Netherlands (Magnus et al. 1979) revealed a strong protective effect of moderate-intensity leisure exercise such as walking, cycling, and gardening, provided the activity was maintained during at least 8 months of the year; the relative risk of an acute coronary event in female exercisers compared to nonexercisers was only around 0.2. This inverse association was considerably stronger in women than in men, to the extent that the upper 95% confidence limit for women (0.4) was still lower than the lower 95% confidence limit for men (0.5). Similar findings have been reported from a casecontrol study in New Zealand women (Scragg et al. 1987), where the age-adjusted odds ratio for myocardial infarction was 0.3 (0.2 to 0.6), and that of sudden coronary death even 0.1 (0.01 to 0.4) in women with regular exercise (defined as at least weekly aerobic activity which induced shortness of breath). In addition, these authors described an inverse relation between the period a person had been exercising regularly and the risk of coronary heart disease, with the most prominent risk reduction occurring only after more than 10 years of exercise. A further recent case-control study (O'Connor et al. 1987) confirmed the reduction in risk of nonfatal myocardial infarction in physically active women, with odds ratios around 0.6 for the 3 upper quartiles of daily caloric expenditure, as compared to the bottom quartile. In that study, controlling for high density lipoprotein (HDL2 and HDL3), low density lipoprotein, total caloric intake, saturated fat intake or alcohol consumption did not materially change the risk estimate, which suggests that physical exercise decreases the risk of myocardial infarction independent of other risk factors. An observational study in a Californian retirement community (Paganini-Hill et al. 1988) described an interesting, fairly strong inverse relation between recreational physical activity and death from stroke: age-adjusted mortality from stroke was more than 2 times higher in women exercising less

26

than 30 minutes daily than with those exercising at least 1 hour per day. Very recently, Blair and coworkers (1989a) reported a large prospective study in 3120 healthy women demonstrating a strong and consistent inverse relation between physical fitness (mea:;ured by a maximal treadmill exercise test) and all-cause mortality. Age-adjusted all-cause mortality rates declined significantly across physical fitness quintiles from 39.5 per 10 000 person-years in the least fit women to 8.5 per 10 000 person-years in the most fit women, representing a relative risk of 4.7 for all-cause mortality for the least fit women. Interestingly, the major reduction in the risk of death occurred between the lowest and the second-lowest fitness quintile (RR 4.7 to 2.4), suggesting that even a modest improvement in fitness level among the most unfit may confer a substantial health benefit. The trend oflower all-cause mortality with increasing physical fitness remained after statistical adjustment for smoking habit, cholesterol level, systolic blood pressure, fasting blood glucose, parental history of CHO and follow-up interval. In another report from the same sample, a case-control study of nonfatal stroke, Blair et al. (1989b) found significantly shorter treadmill times in cases than in controls which suggests that low cardiorespiratory fitness may be associated with an increased risk of nonfatal stroke. Thus recent research tends to show that regular exercise may also protect women from cardiovascular disease, although this effect needs to be replicated in additional prospective cohort studies. Suggested benefits of regular exercise on the incidence of CHO should be weighed carefully against potential cardiovascular hazards of acute physical exertion (Eichner 1983; Shephard 1986). Even though the incidence of a cardiac emergency during vigorous exercise is very low in absolute terms, several epidemiological studies have shown an unequivocal increase in the acute risk of sudden cardiac death during strenuous physical activity, including jogging and running, by a factor of the order 5 to 50 (Marti et al. 1989a; Siscovick et al. 1984; Thompson et al. 1982; Vuori 1986). Plainly, the observation that habitual exercise may protect

Sports Medicine 11 (1) 1991

overall from CHO does not preclude the possibility that there is a transient increase in the risk of sudden cardiac death during the act of exercise itself. Surprisingly enough, these investigations refer to men only, and nothing can thus be said about a potential, exercise-induced cardiac risk of female runners. However, it appears that this risk must be negligibly small, since checking more than 100 published cases of sudden death during running (not shown) provided one single female case which died from a reason other than CHO (Thompson et al. 1979). Vuori (1986), in his series encompassing all sudden deaths that occurred during physical exertion in the total Finnish population from 1970 to 1978, found that the number of events was 14 times less among women than among men, despite approximately equal proportions of women and men participating in recreational exercise with similar frequency. This author concluded that the relative risk of exercise-related sudden cardiac death seems to be much lower in women than in men.

3. Running and the Blood Lipid Profile Abnormal blood profiles of lipids and lipoproteins constitute a major risk factor for coronary atherosclerosis and subsequent heart disease. It has been demonstrated that elevated total cholesterol and low density lipoprotein (LOL) cholesterol are directly related to CHO incidence, whereas high density lipoprotein (HOL) cholesterol concentration has an inverse association with CHO. American reviewers (Bush et al. 1988) argue that HOL cholesterol appears to be the single most important lipid risk factor in women, while Scandinavian authors (Johansson et al. 1988; Lapidus et al. 1985) also emphasise the importance of high serum triglycerides as an independent risk factor of CHO in women. Early cross-sectional investigations found that women joggers have elevated HDL cholesterol levels when compared to their sedentary counterparts (Moore et al. 1983; Nakamura et al. 1983; Wood et al. 1977). For example, Moore et al. (1983) investigated 45 long-distance runners (average training distance 50 km/week), 49 joggers (20 km/week),

Running and Health in Women

and 47 inactive controls, all around age 40. They found a consistent, highly significant gradient in HDL cholesterol concentration, 78, 70, and 62 mgJ dl, respectively. Multivariable analysis showed that this difference could not be attributed to differences in dietary intake, and even after adjustment for body fat content, significant differences between exercise groups remained for HDL cholesterol. A more recent, cross-sectional study of sedentary women (n = 16), recreational (n = 14), good local (n = 14) and elite women distance runners (n = 16) also reported a consistent increase in HDL cholesterol level (55, 59, 69 and 69 mgJdl, respectively), with most of the increase attributable to the HDL2 subfraction (Durstine et al. 1987). Furthermore, apolipoprotein A-I concentration (which is inversely related to coronary risk) was significantly higher in the elite and the good local group than in the recreational and in the sedentary group of the same study (Durstine et al. 1989). Other workers, however, have been unable to demonstrate significant differences in lipid profiles according to physical fitness. Perry et al. (1989) argue that nonobese, premenopausal inactive females may have the same beneficial lipid profile as their more active counterparts. Indeed, Upton and coworkers (1984) were among the first to propose that even in female runners lipid and lipoprotein values may be influenced more by body fat content than by activity level. On the other hand, Sallis et al. (1986a) described significant positive associations of stringently defined vigorous leisure-time exercise with HDL cholesterol as well as the HDL/LDL ratio in a population-based sample of 809 women aged 20 to 35 years from Stanford, California. In a crosssectional analysis of 2067 Lipid Research Clinic Study women, Haskell et al. (1980) found a relatively weak but statistically significant direct association between self-reported strenuous physical activity and HDL cholesterol concentration in those aged 20 to 39, but no association of either treadmill test time or heart rate response to submaximal exercise with HDL cholesterol level. Several investigators have failed to show significant modifications of the serum lipid profile after endurance training in women (Farrell & Bar-

27

boriak 1980; Frey et al. 1982; Moll et al. 1979; Wynne et al. 1980). Possibly, the duration oftraining in these studies, 6 to 10 weeks, was too short. Brownell et al. (1982) found a borderline significant 4% reduction in total cholesterol concentration and a nonsignificant 4.3% decrease in LDL cholesterol after a 10-week exercise programme of 37 women but no changes in HDL cholesterol; these workers therefore suggested that lipoprotein alterations in women after endurance training may differ from that in men. Prospective data from Rotkis and coworkers (1981) indicate that in female recreational runners (n = 22) a substantial increase in weekly mileage [30 miles (48km) per week] for several months is followed by a significant, additional increase in HDL cholesterol even in subjects with previously elevated HDL levels. A randomised, controlled study of20 weeks ofjogging in 16 women in Minnesota found a considerable net effect (i.e. change in intervention group minus change in control group) of jogging on HDL concentration (+ 6 mgJdl), but suffered from lack of statistical power (Santiago et al. 1987). Recently, brisk walking at a pace of 6 to 7 km/h for a weekly distance of 16 to 17km over 1 year was reported to be associated with a net increase in HDL of 12 mgfdl in a controlled study of 44 middle-aged British women (Hardman et al. 1989). The effect seen in walkers could not be explained by concomitant changes in body fatness (Hardman et al. 1990). Another similar study of the effects of 40 weeks of vigorous walking on blood lipids suggested that, in the long run, habitual brisk walking may reverse tendencies for a decline over time in HDL levels associated with the progressive increase in fatness observed in sedentary women (Santiago et al. 1989). A recent meta-analysis of 27 training studies published by the end of 1987 (Lokey & Tran 1989) found marginally significant training-related decreases in total serum cholesterol (-4 mgJdl) and total triglycerides (-9 mgJdl), but no significant change in HDL cholesterol in 379 women. However, since the majority of studies included into meta-analysis were uncontrolled, and since the cholesterol-lowering effect of exercise appeared to be greater in those women with elevated pre-ex-

Sports Medicine 11 (1) 1991

28

ercise cholesterol levels, regression-to-the-mean may account for part of the observed effect of exercise on the lipid profile. A lack of significant change in lipids after endurance training among women in some reports may also have reflected pre-exercise low risk lipoprotein profiles. Lipoprotein lipase activity is generally higher among women than men, such that it is still higher in sedentary women than male long distance runners (Nikkilii et al. 1978). The finding of more favourable lipid and lipoprotein risk factors in women may be due to increased lipoprotein lipase activity, and a reduced lipolytic response of the adipose tissue in women has been suggested as the underlying difference between genders after exercise conditioning (Goldberg & Elliot 1987). Canadian researchers have shown that catecholamine-stimulated lipolysis of adipose tissue increased among men but not (Despres et al. 1984a), or to a lesser degree (Despres et al. I 984b), for women after 20 weeks of endurance training. Further prospective randomised studies, controlled for confounding variables such as diet, alcohol consumption, body composition, and physical fitness, are necessary before the somewhat inconsistent effects of endurance exercise on the serum lipid profile and associated coronary risk in women are more clearly understood.

4. Physical Fitness, Exercise and Hypertension The role of physical fitness and exercise training in the prevention and treatment of elevated blood pressure has been extensively reviewed in recent years (Hagberg 1990; Hagberg & Seals 1986; Seals & Hagberg 1984), and it shall only briefly be addressed here. Physical fitness has repeatedly been reported to be inversely associated with blood pressure in normotensive female subgroups ofthe population, for example in a cross-sectional investigation of young Israeli women (Saar et al. 1986). More importantly, a large cohort study of 6039 middle-aged men and women, visitors of the Cooper Clinic in Dallas and normotensive at baseline, found a rel-

ative risk of 1.52 for development of hypertension during a median follow-up time of 4 years in persons with low levels of physical fitness when compared with those highly fit at baseline (Blair et al. 1984). The estimate of relative risk of hypertension was adjusted for sex, age, baseline blood pressure, and baseline body mass index. Most population-based, cross-sectional investigations of levels of habitual exercise and blood pressure have shown only weak and nonsignificant associations after adjustment for body mass index (Folsom et al. 1985; Marti et al. 1987; Sallis et al. I 986a). Sallis et al. (1986b) reported significantly lower diastolic blood pressure values for Stanford women aged 20 to 34 years and engaging regularly in moderate intensity activities such as walking and climbing stairs, but these workers did not adjust blood pressure values for the significantly lower relative weight of the exercisers. Experimental studies have confirmed that exercise training may be a useful nonpharmacological way to lower blood pressure in both men and women (Jennings et al. 1986; Nelson et al. 1986a). At the same time, several studies have indicated that exercise intensity may be a key characteristic in the blood pressure lowering effect of physical training (Hagberg 1990). Roman et al. (1981) had previously observed that systolic and diastolic blood pressure of 30 hypertensive females was lowered by the same extent, around 20/15mm Hg, by 3 and 12 months oflow intensity training and 12 months of high intensity training, even though these programmes resulted in different increases in aerobic capacity. Kiyonaga et al. (1985) found relatively large reductions in both systolic and diastolic blood pressure (around 15/lOmm Hg) in male and female essential hypertensives following a mild to moderate 10- to 20-week training programme. However, this latter study was uncontrolled, and the same group of researchers replicated their observations in a well-controlled study 2 years later (Urata et al. 1987). These Japanese workers feel that mild exercise intensity and a concurrent depletion in blood volume are relevant to the depressor mechanism of physical training. Hagberg and Seals (1986) also conclude that it is possible

29

Running and Health in Women

that lower intensity exercise training (40 to 60% V02max may be as, or more, efficacious in reducing blood pressure in hypertensive populations. Plainly, a differential effect of recreational vs competitive running would be of considerable public health importance, but unfortunately no appropriate data are available to address this issue.

5. Exercise, Insulin and Diabetes A questionnaire study of the prevalence of diabetes mellitus among 5398 survivors of 7559 college alumnae, residing throughout the United States, found a reported prevalence of 0.6% (15 of 2622) among the former college athletes, compared with 1.3% (37 of 2776) among the former nonathletes (Frisch et al. 1986). The difference in prevalence of diabetes mellitus was not altered substantially when only cases occurring after age 20 were taken into account. When all cases of gestational diabetes were omitted, the relative risk of diabetes mellitus in nonathletes vs athletes was 3.4 (1.3 to 8.7). Admittedly, one might question the term 'athletes' in this investigation, since very few women during 1930 to 1970 were athletes by today's standards. Even though the study by Frisch and her collaborators is a single observation that definitely needs to be confirmed in other populations, its findings are plausible given the well-known acute and long term metabolic effects of physical exertion on glucose regulation. Healthy trained subjects, regardless of gender, have an improved insulin action, due to increased sensitivity to insulin and have a smaller plasma insulin response to an identical glucose stimulus than untrained individuals (King et al. 1987, 1988). However, it is controversial whether endurance training has the ability to specifically improve normal glucose tolerance. For example, in 11 men and women aged 60 to 69 years participating in 12-month endurance training, glucose tolerance (which was normal initially) was not significantly changed after training, while the total area for insulin was 8% lower after lowintensity training, and 23% lower after high-intensity training, compared with before training (Seals

et al. 1984). On the other hand, Hersey et al. (1989) recently reported that 6 months of walking/jogging, 3 days/week for 40 minutes at approximately 75% of maximum heart rate reserve, significantly improved glucose metabolism in 22 healthy women aged 70 to 79 years. After training, postglucose challenge areas under the curve were reduced by 7% for total plasma glucose, and by 20% for insulin. Holloszy and coworkers (1986) have concluded that exercise appears to be effective in normalising glucose tolerance only in patients who still have an adequate capacity to secrete insulin, and in whom insulin resistance is the major cause for abnormal glucose tolerance. The amount of exercise required to normalise glucose tolerance in such patients, however, appears to be in the range of 25 to 35 km/week of jogging. Richter and Galbo (1986) add that physical training cannot be recommended as a means of improving metabolic control in insulindependent diabetes mellitus. The group of Krotkiewski and Bjorntorp (1986) in Sweden have drawn attention to the fact that the pattern of body fat distribution may substantially influence the effect of physical training on metabolic control. Only women with abdominal (android, 'male') type of fat distribution were able to improve insulin sensitivity by 3 months of aerobic exercise. Women with 'female' (gynoid) type of fat distribution did not improve insulin sensitivity after training, but these women were less insulinresistant than those with male fat patterning at outset.

6. Running and Weight Control Habitual leisure-time exercise is inversely related to female bodyweight in some industrialised countries (Marti et al. 1986, 1988a) but this is not consistently seen (Blair et al. 1985). The cross-sectional nature of these observations precludes any inference on the direction of cause and effect and of the temporal sequence of both characteristics involved in the association. In a survey of participants in a 16km road race, at all ages female runners training more than 25 km/week (n = 114) had

30

a bodyweight approximately 3kg lower than runners training less than 25 km/week (n = 307), and all runners were clearly leaner than the women from the general population, with the difference the greater the older the subjects (Marti 1988). Of these 421 runners, 14% reported to have lowered their bodyweight on average by 7kg since they took up jogging, with the amount of weight lost being only weakly related to weekly distance run. Similar results were reported by Koplan and coworkers (1982) from an epidemiological survey of 580 female participants in a 10km road race in Atlanta, Georgia. Among 416 female runners still training 1 year after the event, 67 (16%) reported a weight loss of 5kg or more since they took up aerobic training. Weight loss was greater among persons who had higher relative weight when they began running and among persons who began running to lose weight than among those who began for other reasons. In that study, weight loss was independent of years of running but correlated significantly with the number of kilometers run per week. A comparative, cross-sectional study of elite (n = 15) and good (n = 12) female distance runners estimated body composition with underwater weighing (Graves et al. 1987). Elite runners, training an average distance of 104 km/week, had 14.3% body fat, and good runners, training 63 km/week, 16.8% body fat, while reference values for equally old, untrained women are 23 to 24% body fat. However, prospective research examining the effect of endurance training on body composition of nonobese women has yielded less impressive results (Epstein & Wing 1980). In an 18-month training study terminated by running a marathon, 9 previously untrained women lost only 0.9kg of bodyweight and reduced body fat content nonsignificantly from 24.9 to 23.6% (Janssen et al. 1989). On the other hand, heavy persons lose more weight at the same exercise intensity than light persons (Epstein & Wing 1980). In addition, it seems that different forms of aerobic exercise tend to produce different weight loss. In a randomised study of 3 types of exercise (60 minutes every day during 6 months, with roughly comparable energy expenditure in all 3 modes of exercise) in 29 middle-aged

Sports Medicine 11 (1) 1991

women, subjects walking briskly lost 11 % of initial weight (or 7.7kg), and those cycling lost 13%, while those swimming lost no weight at all (Gwinup 1987). In a Finnish study of 97 overweight (BMI > 25 kg/m 2) women in a 17-month programme of 30 to 60 minutes of walking and crosscountry skiing at least 3 times per week, 43 (44%) dropped out while the remaining 54 women significantly reduced their bodyweight by 3.3kg, with most of the decrease occurring during the first 5 months of the study (Kukkonen et al. 1982). In addition, exercise may help to maintain lean body mass, when weight loss is principally attempted by dietary restriction. In a study of 8 obese women, both exercising and nonexercising dieters lost approximately 8kg of bodyweight during 5 weeks. However, nonexercising women lost significantly more lean body mass (approximately 3.5kg) than exercising ones (Hill et al. 1987). It has been generally observed that addition of aerobic exercise to weight loss programmes of energy restriction helps to preserve lean body mass and may help to accelerate body fat loss (Walberg 1989). At older age, endurance exercise training may not only help to control bodyweight and to reduce body fat, but it may even be more important to counteract the age-related loss of fat-free mass (Kohrt et al. 1989), by maintaining or increasing muscle fibre area (Cress et al. 1989). Metabolic effects and possible health consequences of repeated weight loss and gain, as it may occur in specific athletes and only seasonally active joggers, are discussed elsewhere (Brownell et al. 1987).

7. Ru""i", a"d Ame"orrhoea The aetiology, complications and management of athletic amenorrhoea has recently been extensively reviewed (Highet 1989; Loucks 1990; Shangold et al. 1990). Runners who start training before menarche have a later onset of menarche, and Frisch et al. (1981) estimated that this delay is in the order of 5 months for each year of previous training. Frisch and McArthur (1974) proposed that a critical percentage of body fat is required for the

Running and Health in Women

onset of menarche (l7%) and its maintenance (22%). Later, Vandenbroucke et al. (l982) suggested that thinness and intensive sports activity may act synergistically to delay menarche. The percentage body fat theory proposed by Frisch and McArthur has been criticised for methodological reasons, and particularly by the finding that many eumenorrhoeic runners have body fat levels clearly lower than at which menstrual dysfunction is believed to develop (Noakes & van Gend 1988). For example, in a carefully designed study of 14 runners using hydrostatic weighing and matching running distance and somatotype, body fat content was around 17% in both the 7 runners with secondary amenorrhoea and the 7 regularly menstruating runners (Sanborn et al. 1987). Recently, however, a new statistical explanation for later menarche in athletes was offered by Stager et al. (l990). These investigators used a computer to simulate retrospective sampling from a large population of adolescent athletes in whom age of menarche and age at initiation of training were independently distributed. They then drew samples from the resulting subpopulations in which age at initiation of training occurred before and after menarche, and found 2 surprising results. First, even though there was no relationship whatsoever between age at initiation of training and age of menarche in the population as a whole, these 2 variables were significantly correlated in the subpopulations. Second, the age of menarche was significantly later in the subpopulation in which training began before menarche. These are exactly the same findings reported in support of the idea that athletic training delays menarche, and they prove that any retrospective sampling in former studies is statistically biased. At present, it is correct to say that the average age of menarche is later in athletes than in nonathletes but there is no experimental evidence that athletic training delays menarche. Other studies suggest that chronic energy malnutrition may play an essential role in the development of menstrual dysfunction in runners. Drinkwater and her coworkers (l984) found that despite running 67 km/week in training, 14 ame-

31

norrhoeic runners had a lower daily energy intake (l622 vs 1965 kcal) than 14 eumenorrhoeic runners who trained only 40 km/week. Marcus et al. (1985)

also reported a lower daily caloric intake in II amenorrhoeic runners than in their cyclic running peers (1272 vs 1715 kcal). Kaiserauer et al. (1989) examined the dietary composition in equally lean and equally active runners, and found that amenorrhoeic runners consumed 36% less calories, 41 % less protein, and 51 % less fat than their regularly menstruating peers. These authors concluded that nutritional inadequacy, not exercise or body composition, would appear to separate amenorrhoeic from eumenorrhoeic runners. Others have observed a high frequency of vegetarianism in amenorrhoeic runners (Brooks et al. 1984). In a group of II amenorrhoeic runners, 9 (82%) were vegetarians, while in 15 regularly menstruating runners, only 2 (l3%) were vegetarians. Similarly, Slavin et al. (1984) found a prevalence of secondary amenorrhoea in middle-aged vegetarian exercisers 7 times higher than that in exercisers under a traditional North American diet. Surveys of recreational and highly-trained distance runners show a great variability in the prevalence of secondary amenorrhoea, between 1 and 44%, which can in part be explained by different definitions of amenorrhoea (Loucks 1990; Loucks & Horvath 1985), but which is also influenced by training load. A number of studies have shown a direct relationship between the incidence and severity of menstrual dysfunction and the average distance run per week (Loucks & Horvath 1985; Noakes & van Gend 1988). Feicht et al. (1978) found that as much as 43% of athletes running an average distance of 115 km/week had menstrual dysfunction, and there was a practically linear relationship between distance run and the prevalence of amenorrhoea. Feicht-Sanborn et al. (1982) expanded their cross-sectional study by inclusion of female cyclists training up to 400 km/week and found an average prevalence of amenorrhoea of 12%, but no dependence on training distance at all. In a study of female runners which yielded a relatively low prevalence of amenorrhoea in the sample as a whole, the prevalence was highest, 17%, in

32

the subgroup running the greatest distance per week, 80 km/week (Lutter & Cushman 1982). However, when evaluating this apparently high incidence in runners, it must be kept in mind that, for example, approximately 70% of female US Military Academy cadets indicated menstrual irregularities during the first year of training (Welch 1989). There is one prospective study of the induction of menstrual disorders by running, with and without concomitant weight loss, in 28 untrained women (Bullen et al. 1985). Subjects were expected to run 6 km/day, progressing to 16 km/day by the fifth week. Only 4 subjects had a normal menstrual cycle during this extremely vigorous training, 3 out of 12 in the weight-maintenance group, and lout of 16 in the weight-loss group. Within 6 months of termination of the study, all subjects were again experiencing normal menstrual cycles. These authors concluded that strenuous exercise of a few months' duration, particularly if compounded by weight loss, can reversibly disturb reproductive function in previously eumenorrhoeic women. However, a questionnaire survey of 394 entrants to the New York City marathon revealed that the best predictor of a woman's menstrual pattern during training was not her running history, but her pretraining menstrual pattern (Shangold & Levine 1982). The precise, endocrinological abnormalities associated with exercise-induced menstrual dysfunction have yet to be fully clarified (Noakes & van Gend 1988). Loucks and coworkers (1989) suppose that the degree to which the hypothalamic-pituitary-adrenal axis is disturbed in cyclic and amenorrhoeic athletes is associated, perhaps causally, with their hypothalamic-pituitary-ovarian axis abnormalities (as described by Cumming et al. 1985). Ding et al. (1988) have found that among amenorrhoeic athletes those who did not have elevated serum cortisol levels spontaneously regained menstrual cycles within 6 months, while those with elevated cortisol levels remained amenorrhoeic. As a consequence, disruption of reproductive function in female runners may not be directly associated with factors in athletic lifestyle itself, but rather with the degree to which the hypothalamic-pituitary-

Sports Medicine Jl (J) 1991

adrenal axis is disturbed in specific women. Taken together, the data indicate that the abrupt initiation of strenuous aerobic training can disrupt the menstrual cycle in at least some women, but these women may be more susceptible to reproductive disruption than others, and some aspects of athletic training other than exercise (such as caloric deficiency) may be responsible for the observed disruption (Loucks 1990).

8. Running, Menstrual Function and Bone Density Although it is recognised that exercise, diet, oestrogen levels, and aging all interact in affecting changes in women's bone mass, the- relative contribution of each factor is unknown. One important question is whether in the hypoestrogenemic state of amenorrhoeic runners and postmenopausal women, exercise can counterbalance bone loss. Recent evidence suggests that amenorrhoeic athletes in their third and fourth decade have lower vertebral bone density, in spite of much higher levels of physical exercise. In a cross-sectional study by Lindberg et al. (1984) a lower bone density was observed in the radius and lumbar vertebrae of women with definite exercise-induced amenorrhoea and this was most pronounced in trabecular bone. Cook et al. (1987) confirmed that it was preferentially trabecular bone density of the lumbar spine that was reduced in oligomenorrhoeic runners. Marcus et al. (1985) demonstrated that mineral density of lumbar spine in 11 amenorrhoeic runners was lower than in 6 running eumenorrhoeic controls, but was higher than that in runners with secondary amenorrhoea who were less physically active [as described earlier by the same workers (Cann et al. 1984)]. Mineral density of the radius was normal both in amenorrhoeic and cyclic runners, but running-related stress fractures were more frequent among the amenorrhoeic women. Others (Linnell et al. 1984) have found reduced bone mineral content also at radial sites in amenorrheic runners, but only in combination with extreme leanness, i.e. less than 10% body fat. Drinkwater et al. (1984) found a decreased vertebral, but

Running and Health in Women

not radial, mineral density in 14 amenorrhoeic runners, as compared to eumenorrhoeic peers. Interestingly, 2 years later Drinkwater and her team (1986) reported follow-up data of 7 amenorrhoeic athletes who had regained menses since the initial test. Six of these 7 runners exhibited a marked increase in vertebral bone mineral density, and for all 7 the significant mean increase was + 6.3%, alongside a concomitant average bodyweight increase of 1.9kg. In cyclic runners (n = 7), vertebral bone mineral density did not change while 2 runners who remained amenorrhoeic during follow-up continued to lose bone (-3.4%). However, the increase in bone density observed in the runners regaining menses subsequently slowed and then ceased, so that 4 years after resumption of normal menses lumbar bone density for these women remained well below the average of their age group (Drinkwater et al. 1990). The group of Cann, Martin and coworkers has made relevant observations on the natural history of bone loss due to runninginduced amenorrhoea. They showed (Cann et al. 1985) that the rate of spinal trabecular bone loss is most rapid during the first 2 years after cessation of menses (around 4% per year), while in long term amenorrhoeic runners bone loss is not significant. In addition, recent results from 208 women suggested that menstrual history is a crucial determinant of trabecular bone density in runners, since prior amenorrhoea as short as 3 years (and untreated with estrogens or oral contraceptives) led to irreversible bone loss in athletic women (Cann et al. 1988). Based on a cross-sectional study of 97 young athletes, Drinkwater et al. (1990) also suppose that extended periods of oligomenorrhoea/ amenorrhoea may have a detrimental residual effect on lumbar bone density. A recent, cross-sectional investigation (Ding et al. 1988) of 19 amenorrhoeic and 35 eumenorrhoeic athletes found significantly lower vertebral bone mineral density, lower estrogen levels, and higher cortisol levels in the former. These authors felt that bone health in women with exercise-induced amenorrhoea may be threatened from 2 sources - diminished estrogen levels and elevated cortisol levels - even though no effect of cortisol

33

on bone density that was independent of the effect of hypoestrogenaemia could be demonstrated in that study. Several cross-sectional studies have examined the relationship of diet, running, and menstrual status with bone mineral density. Nelson et al. ( I986b) found reduced bone density and lower daily energy intake, but no difference in the calcium intake of 11 amenorrhoeic runners compared to 17 normally menstruating running peers. In a study of collegiate athletes, Lloyd et al. (1987) found no difference in the diet between eumenorrhoeic and oligomenorrhoeic athletes, with the single exception of a significantly higher intake of dietary fibre in the latter. Dietary composition thus may be a relevant modifier of the effect of running-induced menstrual irregularities on bone density. The group of Sanborn and her collaborators (Berning et al. 1984) suspected that distance runners, whether amenorrhoeic or regularly menstruating, tend to have a critically low calcium intake which in turn leads to low bone density. Dalsky (1990) recently concluded that where there are strong stimuli (such as estrogen and calcium deficiencies) toward bone resorption, exercise may not be sufficient to counteract these tendencies toward negative calcium balance. Regardless of menstrual status, one question of crucial importance is how effective exercise may be in maximising bone mass in the early adult years and in maintaining that mass throughout life. The old observation that bone adapts to increased mechanicalloading by increasing mass, referred to as Wolff's Law [1892, cited in Montoye (1987)] has led to the hypothesis that exercise may actually prevent osteoporosis (Smith & Raab 1986). Crosssectional studies have attempted to show a relationship between habitual exercise and bone mineral content. In 2 studies, investigating 83 women aged 30 to 85 years, and 299 women aged 19 to 91 years, respectively, Stillman et al. (1986, 1989) found a weak but statistically significant association between the level of physical activity and radial bone mineral content, after adjustment for age and menstrual status. Plainly, the hypothesised effects of dynamic..and/

34

or weightbearing exercise on bone mineral content may be largest in the lumbar spine and in the legs. Pocock et al. (1986), indeed, described a correlation between aerobic capacity and bone mass in the femoral neck in 84 pre- and postmenopausal women. Moreover, in a case-control study of 60 to 70-year-old women Swedish workers found that cases with a fracture of the neck of the femur had been significantly less physically active during middle-age than controls (Astrom et al. 1987). Determining physical activity by a motion sensor in 24 premenopausal nonathlete women (mean age 39 years), Aloia et al. (1988) found a significant correlation of total physical activity with bone mineral density of the spine (r = 0.41) as well as total body calcium (r = 0.51), but no association between physical activity and radial bone density. Others have confirmed that in eumenorrhoeic women in their third decade, V02max and isokinetic knee strength is significantly related to bone mineral content of the femoral neck (Snead et al. 1989). Based on a cross-sectional study of 60 women aged 24 to 35 years, Kanders and Lindsay (1985) proposed that even though exercise and dietary calcium intake are both significant and independent determinants of bone density in young women, a high calcium intake appears necessary for the benefits to be expressed. According to Lane and coworkers (1986), master runners may have higher than average bone densities. The first lumbar vertebra of women in an 'over-50' runners' club had a 35% greater bone density than that of nonrunners of the same age. However, a subsequent report from the same workers (Michel et al. 1989) revealed that, even though a strong correlation existed between lumbar bone density and the amount of exercise for up to 5 hours/week, the 5 women (out of 28) exercising 5 hours/week or more had surprisingly low bone density, not explained by other factors such as age, BMI, age at menopause, estrogen replacement or calcium intake. Tentatively, these authors conclude that moderate weightbearing exercise may increase lumbar bone density, but extremely vigorous exercise may possibly be detrimental to bone density in individuals after age 50. Finally, Nelson

Sports Medicine II (1) 1991

et al. (1988) were unable to detect lower bone minerai density in lean, hypoestrogenaemic, endurance-trained postmenopausal women, which led them to hypothesise that endurance-trained women may have improved calcium absorption as a result of higher carbohydrate intake and higher serum 1,25-dihydroxyvitamin D levels. Several controlled intervention trials suggest that postmenopausal women may at least retard their bone loss by doing any of several types of physical exercise for 30 to 60 minutes 3 times per week (Harris et al. 1989). Women with a mean age of 61 years who walked, ran and did callisthenics for 1 hour twice weekly for an 8-month period increased vertebral bone density by 3.5%, while sedentary controls lost 2.7% (Krolner et al. 1983). Even women averaging 81 years of age and exercising only 30 min/day 3 times/week designed around a chair for 36 months increased radial density by 2.3%, while sedentary controls decreased by 3.3% (Smith et al. 1981). A randomised, controlled study of 48 women aged 50 to 62 years found significantly greater bone mass in exercisers than in controls after I year of training (Chow et al. 1987). The increase in bone mass was slightly greater in the subgroup of exercisers who added 10 to 15 minutes of strengthening exercises to the standard programme of 3 weekly sessions of 40 minutes of aerobic exercise. However, not all investigators have been able to demonstrate significant effects of physical training on bone mineral content. A randomised study of 229 women with a target activity of llkm walking per week over 3 years found significantly increased levels of physical activity in the intervention group, but no difference in the rate of loss of bone density between walkers and controls (Cauley et al. 1986). Walking alone was possibly too Iowa stimulus to induce significant bone remodelling at radial sites in this investigation. Another recent, interesting study assessed the effect of weightbearing exercise training and subsequent detraining on lumbar bone mineral content in 35 postmenopausal women aged 55 to 70 years (Dalsky et aI.1988). Walking, jogging, and stair climbing at 70 to 90% of V02max for 50 to 60 minutes 3 times weekly

Running and Health in Women

increased bone mineral content by 5.2% (2.0 to 8.4%) after 9 months of training, and 6.1 % (3.9 to 8.3%) after 22 months of training. However, after 13 additional months of decreased activity, bone mass reverted to a density only 1.1 % above baseline value in the detraining group. Nevertheless, it must be emphasised that even maintaining, instead of losing, bone mass at postmenopausal age by the virtue of intermittent exercise may be very relevant to female health at older age. This also applies to the most recent results from Dalsky et al. (1989) showing that exercise training was as effective as hormone replacement therapy in maintaining (not increasing) lumbar bone mass when initiated early in menopause. Unfortunately, no prospective studies addressing the role of physical training in preventing osteoporosis in premenopausal women have been published so far. Therefore, some authors feel that the evidence is not yet firm enough to indiscriminately counsel individuals to undertake programmes of exercise with the hope of preventing osteoporosis (Block et al. 1987).

9. Running-Related Injuries and Complaints Typically, case reports and case series present data on clinical characteristics and relative frequencies of various running injuries. Such studies may provide important information regarding the natural history of injuries as well as successful treatments (Renstrom & Johnson 1985), and they may help generate hypotheses about the causes of running injuries. However, as Walter et al. (l985) and Powell et al. (1986) have pointed out, even very large case series (e.g. Pagiiano & Jackson 1989) cannot provide incidence rates, identifY high risk runners or establish risk factors for running injuries, because they do not provide sufficient information about the population from which the injured runners came. In the absence of randomised experimental studies, which are likely to be rare in the area of running injuries, causality can be inferred from epidemiological studies if the results are consistent,

35

plausible, and appropriately sequenced. Table II presents an overview of II epidemiological studies of running injuries with definitions of the population at risk and of injuries, incidence rates of running injuries and salient findings concerning potential risk factors. Apparently, definitions of running injuries are so different that they have a strong influence on reported incidence data. The study that used the most rigorous definition of a running injury also reported the lowest I-year incidence, 14% (Marti 1988). When using comparable definitions, the 2 largest published surveys of female road race participants found a similar 1year incidence of running injuries and complaints, around 35% (Koplan et al. 1982; Marti 1988). Both studies found an unequivocal dependence of injury risk on habitual mileage, with an approximately 2fold increase in the risk of injury when going from low (i.e. less than 15 km/week) to high (i.e. more than 50 km/week) distance run. With a few exceptions (Lloyd et al. 1987; Macera et al. 1987, 1989), most studies have confirmed distance run per week as a risk factor for injury. At present, mileage seems to be the only firmly established risk factor for running-related injuries in women. However, it is interesting to note that although the point prevalence of running injuries increased throughout an 18-month prospective study on 73 untrained Dutch men and women (where mileage was gradually increased from zero at baseline to more than 50 km/week at the end), the injury incidence per 1000 hours of running decreased from 12 in the first 6 months of the study to 7 in the last 6 months (Bovens et al. 1989). This reduction is suggestive for the adaptative potential of the musculoskeletal system to increasing training load. In another study from the Netherlands, Vrencken et al. (l988) determined the I-year incidence of sports injuries in 405 female subjects aged 40 years and over. Using a sportsmedical screening setting, they found jogging to be a low risk sport. The annual incidence of jogging injuries was 12% (16% in men), with a rate of 1.2 injuries per 1000 hours of participation (men 1.3/1000 hours). Nevertheless, the injury incidence in these 2 Dutch studies, or in a survey of female· participants in a 16km road race,

Table II. Epidemiological studies of running-related injuries in women Reference

Study design

t...>

0\

Definition of runner

n

Runs ~ 10 km/wk

730

Decreased mileage, 12 or took medications, or consulted physican because of running injury

96 (+ 355 men)

Decreased mileage [stopped running] because of running injury

Koplan et al. (1982)

Retrospective cohort of entrants to 10km road race

Jacobs & Berson (1986)

Cross-sectional 10km race survey of entrants entrant to 10km race

Blair et al. (1987b)

Cross-sectional survey of health club members

~ 16 km/wk at least once within a 3month period

Blair et al. (1987b)

Controlled prospective study of worksite health promotion

Macera et al. (1987)

Definition of injury

Duration (months)

Injury incidence (%)8 mean

lowest

highest

38

29 ( I: CIl .... :::I "0 .!!!

>co

m::. co::.

:::1-

.... 0 I: II

"0

o.CIl

0

.,

~aG;

a.

.~ .~

gz

....

0.0;';:::::0.

!~ I: I: 8.

CIl E CIl 8. :::I '" I: 0 o~8§

a; co

&.

.,00 .! '. ~ [@

'6 .5

.52

~

~CD

~~