The present study supports partly the findings of previous review study of Shephard (1996) and systematic meta-analysis of Conn et al. (2009) that workplace-based physical exercise interventions have small to moderate effects on the physical functioning [2
]. According to previous studies there were small changes in anthropometric measures [2
], small [9
] to moderate [2
] changes on maximum oxygen uptake and small changes on muscle strength and flexibility [9
]. In our study among office workers the light resistance training significantly increased both subjective physical condition and upper extremity extension strength, but had no effect on body fat percentage, spinal flexibility, hang grip strength, upper extremity flexion strength and lower extremity strength.
The small or zero effects obtained in our study are probably due to the ceiling effect, as the subjects were middle-aged healthy volunteers whose level of physical functioning was already high at baseline. We can also assume that the 30% 1RM load seems to be too low to increase strength and hypertrophy
of muscles, and physical exercise intervention was not high and/or prolonged enough to affect body composition, or specific enough to improve spinal flexibility in this study population. Also Conn et al. (2009) estimated average that workplace interventions training times were higher than in our study. Median of supervised exercise session was 50 minutes and median of frequency three times per week. These numbers were based on 44-47 reports in Conn et al. meta-analysis of workplace physical activity interventions [2
]. Respectively the mean difference between treatment and control subjects´ on VO2max was 3,5 mL/kg/min and on BMI -0.3 units.
Jordan et al (2010) recommended that a standard validated measure of exercise adherence should be used consistently in future studies [44
]. Among the office workers (n=90) various measures of adherence were conducted [45
]. Adherence to exercise in the light resistance training and also training guidance was satisfactory, although we did not take individual holidays into account, which would have boosted the light resistance training adherence values. The adherence values were similar to those found earlier in the critical review by Proper et al. (2003) among worksite physical activity programs [7
The minimal training intensity threshold for improvement in maximum oxygen uptake (VO2max) is approximately 40-50% of the maximum oxygen uptake reserve (VO2R) or maximum heart rate reserve (HRR) or 55-65% of the maximum heart rate (HRmax) [17
]. VO2max values correspond to the MET values used in this study. MET reserve is METc - METrest (1MET). In our study the average relative physical strain of the light resistance training was 26.4% and 38.9% of METc and 45.2% and 29.2% of MET reserve among men and women, respectively. In our study population the average cardiovascular response during one light resistance training session, especially among men, did not reach a level that is known to improve aerobic fitness, although the participants reported better subjective physical condition during the physical exercise intervention. We can assume that 30% 1RM light training is not sufficient to transfer training to cardiovascular endurance. Although in our previous pilot study (n=11) similar training among a similar sedentary occupation population reached the threshold level that is known to improve aerobic fitness. In this pilot study, during one light resistance training session the average HR was 63% (SD 7) HRmax. The light resistance training investigated here could act as an appropriate and safe introduction to more intensive training among sedentary population. It could also be included in the exercise prescription.
In further studies measurements of maximum oxygen uptake should be determined more objectively, for instance with maximal bicycle ergometry, and the progressiveness of training should be taken care of more carefully in account, although this is more difficult to attain in real life than in a laboratory setting. Future studies should also investigate the efficacy of more progressive training loads (for example 30%, 60%). However, from the psychosocial point of view a 30% load is both supportive and agreeable. According to our previous analyses of our workplace physical exercise intervention might be more important and specific for decreasing musculoskeletal symptoms [46
] than increasing psychosocial functioning [48
] or physical functioning.
Strazdins and Bammer (2004) investigated Public Service employees (73% women and 73% clerical workers) and found gender differences in risk factors. Women’s working conditions were more likely to involve physically repetitive work demands. Women were also more likely to work in poorly designed and uncomfortable environments and, in addition, women spent considerably less time than men exercising or relaxing during their leisure-time [49
]. Also in our study population men had higher TWA and maximum intensity of LTA than women, but there were no gender differences in AT (work, commuting and LTA combined), but the physical exercise intervention was not more powerful among women office workers than in our originally study population [50
The methodological quality of study, assessed according to the Physiotherapy
Evidence Database (PEDro) quality score and Cochrane Collaboration Criteria [51
], were high. In our CRTs the methodological quality was lowered by the fact that neither the subjects nor the therapists were blinded and the assessors were only in some measurements blinded. In physical exercise studies, it is very difficult or even impossible to blind subjects by including a placebo treatment, because it is not easy to develop a good and trustworthy placebo [9
]. In physical exercise interventions, in particular, the subjects cannot be made unaware that they have received treatment. Therefore, we maintain that physical exercise studies always include some non-specific effects, attraction or placebo effects. In the standard cross-over design the order of the interventions is randomized for each cluster and a time period (called the “washout” period) is often allowed between the two interventions so that the first intervention does not affect the second. In our study we did not have a washout period between the two treatment periods, but we analysed the carry-over effects. Where signs of carry-overs appeared, the results were analysed according to the first treatment period.
The strengths of our physical exercise intervention study were the randomisation by clusters in the natural working environment, controlling for possible confounding factors, such as other physical activity and department, learning, and seasonal effects [13
], and careful documentation of the training dose. Moreover, a cross-over trial is ethically more acceptable and statistically more efficient than similar-sized parallel group trials. Furthermore, in this study gender differences in physical functioning measurements at baseline and their possible influences on the study results were controlled by using baseline physical functioning measurements in the data analysis as covariates.
The intention-to-treat approach is often inadequately described and inadequately applied. Authors should describe the handling of deviations from randomised allocations and missing responses and discuss the potential effect of any missing response. The absence of an intention-to-treat analysis in intervention studies can lead to bias since subjects may drop out because for reasons that make them non comparable with the group that completes the study with respect to the outcome variable. If the intention-to-treat analysis is not performed, the effectiveness of treatment may be overestimated [8
]. In their critical review Proper et al. found that only one out of eight workplace intervention studies had included an intention-to-treat analysis, and this study was a controlled trial [8
]. In our CRT cross-over study the intention-to-treat analysis meant that all the subjects, who were randomly assigned to the two treatment sequence groups (Treatment Group 1 and Treatment Group 2), were analysed together, regardless of whether or not they completed or received the physical exercise intervention treatment. An advantage of mixed models [37
], compared to ANOVA (or MANOVA) for repeated measures, is that there is no technical need to exclude subjects with incomplete data from the analysis. Instead, all the available observations, whether from completers or non-completers, contribute to the statistical inference (parameter estimation, significance testing) by virtue of the likelihood-based estimation method. In the mixed model approach we assume that each single observation obeys the same specified model, even where the observation is lacking no matter if it is observed or not. The model compensates for the missing data. The validity of this model assumption (and inference) requires, however, that the possible drop-out mechanism is random. In our study the number of non-completers was small and we can also assume that the drop-out were random. Thus we do not see any problems of bias due to incompleteness of the data. Although, there was a low level of systematic error during the physical exercise intervention (selection bias, information bias, confounding factors), care must be taken in generalizing the study results beyond the target population (office workers, sedentary workers) as we were not able to control for all non-specific effects, and as the study sample was relatively small. To confirm assumptions of the effectiveness of physical exercise interventions, more randomized and controlled follow-up studies among different sedentary occupations and workplaces are required. There is also a need to study the effects of different training doses and movements, using different training tools and methods of guidance, to clarify the role of PA for physical functioning. Follow-up studies of long duration are also needed to explore the possibility that even slight positive changes could be important in preventing impairments in people’s functioning, work ability and general well-being.
In summary, on average 1.12 metabolic equivalent hours per week during the working day had a positive effect on subjective physical condition and upper extremity extension strength among the office workers studied. However the changes in physical functioning were probably not large enough to be of clinical importance. Controlling for training dose, other physical activity and confounding factors provides for a better understanding of the effectiveness of exercise interventions on physical functioning.