Recruitment procedure

Participants were recruited from cleaning work places in the urban, rural and metropolitan regions of Denmark. Recruited participants were required to be employed at least 20 hours/week at the workplace, and primarily work during day hours. Their main work task had to be cleaning, but their job could also involve other service tasks such as washing, kitchen work or attending patients. All eligible employees (N = 758) were invited to an information meeting during their working hours, asked to answer a screening questionnaire, and to give consent if willing to enrol in the study. For employees who did not attend the information meeting, managers subsequently handed them written information on the project and screening questionnaires with a stamped addressed envelope. Among the eligible employees, 394 consented to participate, and these were invited to answer a questionnaire. Among the consenters, 363 (48%) entered the intervention study. A large fraction of immigrants are employed at cleaning workplaces in Denmark, but Danish language skills are required. Therefore, all written and oral communication was conducted in Danish. Language and reading difficulties were dealt with through guidance from the research personnel or collegial assistance. Further details of the recruitment procedure has previously been reported [14].

Randomisation procedure

For the cluster randomisation procedure, each workplace was considered a stratum. Clusters depended on work teams where possible or were made up from groups either in which employees had lunch, groups where they worked in close proximity to each other, or groups who reported to the same manager. Clusters were matched on sex, age and job seniority. The randomisation was made by lot conducted by blinded staff. To obtain a homogeneous group and due to the high proportion of females in the job group, the 69 males were excluded from further analyses in the present study (PCT = 25 males, CBTr = 22 males, REF = 22 males).

Design

The female cleaners were randomised into PCT (n = 95), CBTr (n = 99) or REF (n = 100). Both data collection and interventions were conducted during the cleaners' working hours. Questionnaires were collected and physical tests conducted at baseline and after three months intervention.

Interventions

The PCT was instructor-guided and offered 3 times a week during 12 weeks in sessions of 20 minutes duration at the workplace during working hours. Six intensive physical coordination exercises with 2-4 levels of progression were performed. The exercises were four point kneeling, prone plank, bridge, vertical plank, bodyblade and horizontal side support and are described in detail elsewhere [15]. Each training session began with a short warm up consisting of ballistic stretching. During the first 1 or 2 weeks, only 3-4 exercises were presented at each session until the performance was technically correct. Thereafter each exercise was performed 2 times 25 seconds in a circuit programme. The exercises all involved a high demand of coordination and were performed at the maximal level of progression, challenging the performer optimally during the 25 seconds with a work load corresponding to 60-80% of their maximal voluntary muscle strength. One or two instructors supervised the participants' performances and offered corrections when necessary.

The CBTr was instructor-guided and offered every second week during 12 weeks in two-hour sessions at the workplace during working hours. The programme was a modified version of the programme developed by Linton [16, 17]. With respect to the hypothesis of this paper, the purpose of the programme was to reduce kinesiophobia and thereby diminish negative effects of kinesiophobia. Therefore, the programme focused on improving the participants' understanding of pain, the experience of pain, and the anticipation of pain by performing cognitive exercises on how physical activity not necessarily leads to pain. Cognitive and behavioural exercises were performed to train the ability to function despite pain (i.e. pain coping, increasing health behaviour, adapt skills to daily life). Moreover, the positive long-term effects from appropriate pain coping were discussed [17]. Every session was planned with the same general outline: feedback on homework from previous session, introduction to today's theme of the session, a short lecture on the theme, problem-solving exercises and training of new skills (i.e. applied relaxation training) [18]. To enable the participants to train the applied relaxation programme during the workday, they received an mp3 with the necessary instructions.

Measurements and analyses

Statistics

To determine if differences between the three groups had happened by chance in the randomisation, descriptive data regarding the variables age, height, weight, body mass index, fat %, job seniority, distribution of immigrant and native cleaners, muscle strength, balance and kinseiophobia were reported. Since no differences appeared from these reports, the analyses on intervention effects were unadjusted. The effects of intervention on muscle strength, balance and kinesiophobia were evaluated in an intention-to-treat analysis using one-way analysis of variance (ANOVA) on the difference between baseline and follow-up, followed by Bonferroni corrected post-hoc-tests when relevant. Due to missing observations, observations were carried backward and forward at baseline and follow-up, respectively, thereby avoiding bias of non-random drop-out [26, 27]. In addition, to display the true observations and to avoid false negative results, data were analysed exhaustive by performing the same analysis procedure only with true observations on both outcome measures at baseline and follow-up. This procedure was termed true-observations analysis. Finally, to obtain further power to the dataset, if no difference between REF and the other intervention group existed, both analyses were further explored in an aggregated analysis, collapsing REF with the inefficient intervention. For example, the PCT intervention effect on muscle strength and balance was tested against CBTr and REF combined, and the CBTr intervention effect on kinesiophobia was tested against PCT and REF combined. IBM SPSS statistics version 19 was used for all statistical analyses.

Employee flow

There were 33, 40 and 26 cleaners who dropped out of the study from PCT, CBTr and REF, respectively, as given in Figure 1. Those who dropped out did not attend the follow-up measurements. Seventeen and eighteen percent did not receive the intervention at any time in the PCT and CBTr, respectively. Mean adherence rates (including those with zero adherence) were 37% and 49% in PCT and CBTr, respectively.

Baseline characteristics

Baseline characteristics of the participants in the three intervention groups are given in table 1. No differences between the three groups were found at baseline. Both mean for the participants at baseline, the intention-to-treat baseline populations (with missing variables carried backwards) and the true observations baseline population (those with outcome measures both at baseline and follow-up) are displayed. A large fraction of the study population was immigrants, as given in Table 1. Among the immigrants, 49 different ethnic identities were present with the 8 most prevalent being Turkey, Marcedonia, Pakistan, Thailand, Philippines, Morocco, India and Serbia, representing 70% of the immigrants in the study population. Previously, studies on the immigrant population in the current study has been published [20, 28, 29].

Physical capacity

Kinesiophobia

The TSK had a high internal consistency, with Chronbach's alpha of 0.80. The CBTr significantly reduced their kinesiophobia after the intervention in comparison with PCT (p = .012), but not with respect to REF (p = .457). However the difference was significantly different in the aggregated analysis, CBTr versus PCT and REF (p = .013). Furthermore, these differences were highly significant in the true observations analysis both against PCT (p = .008) and against PCT and REF aggregated (p = .01). The true observations revealed a ~16% reduced kinesiophobia in CBTr.

Main findings

This 3-month randomised controlled trial among female cleaners from 9 workplaces in Denmark significantly improved their individual physical and cognitive behavioural resources. The PCT intervention improved trunk muscle strength and balance, and the CBTr intervention reduced kinesiophobia. In the following, implications and perspectives on the improvements of these resources for the prevention of deterioration among cleaners will be discussed.

Comparison with other studies

Our study is the first RCT among workers with high physical work demands demonstrating a workplace training intervention to improve muscle strength in an ITT-analysis. It is well recognized that the dose of a work task is relative to the capacity of the performing worker [30, 31]. Therefore, the relative physical work exposure on the musculoskeletal system of the worker can be considered reduced when strength improvements are obtained. Generally, the strength level of the cleaners in the current study was comparable to reports from previous studies with a representative sample of the Danish population [32] and a sample of Danish cleaners [8], although trunk strength was below the previous reports. High muscle strength has been shown to characterize senior (> 8 years) cleaners without muscle pain in comparison with cleaners with similar seniority with muscle pain [8]. Therefore, the increased muscle strength from the PCT may improve the cleaners' tolerance for high work loads and possibly reduce the risk for deterioration, i.e. musculoskeletal disorders.

The PCT was tailored to improve both strength and coordination of the cleaners. Accordingly, the PCT resulted in an improved postural balance. The PCT included training of the bracing manoeuvre, which produces a global co-activation of the muscles of the abdominal wall [33]. In our study, the only instruction that was given during the balance test was "stand as still as possible" and the test leader was blinded to the randomisation. Therefore, the improved balance may imply that the cleaners in PCT were able to transfer and use the improved strength and coordination of abdominal wall muscles in tasks not related to the training. Some studies suggest that poor stabilisation may predispose injury and musculoskeletal disorders [34–36]. Thus, the improvements in strength and balance in PCT may prevent deterioration of health among female cleaners in the longer term.

CBTr was shown to decrease kinesiophobia compared to both REF and PCT. Pain-related fear of movement is closely related to measures of disability and catastrophising [9, 37–39]. Furthermore, kinesiophobia is shown to predict long-term recovery from pain-related functional disability among males with chronic non-specific low back pain [40] and improved kinesiophobia among work-disabled pain patients have shown positive effects on return to work [41]. Thus, reductions in kinesiophobia may reflect reduced pain-related fear of movements related to work tasks.

Baseline values of kinesiophobia among the working cleaners were on average 32-34. These values are comparable to a sample of the Dutch general population [25] and less than the average of 38-40 reported from patients seeking care due to musculoskeletal pain [40, 42, 43]. Although baseline levels were lower in the current study population, significant improvements were still found. This introduces CBTr as a possible valuable prevention strategy to reduce kinesiophobia in workplace interventions as well as in rehabilitation.

No effects on kinesiophobia, strength and balance were seen across the interventions. This finding indicates that contamination was successfully avoided by the cluster randomisation. However, five out of seven measures of strength and balance were numerically impaired in the CBTr-group, and kinesiophobia was numerically increased in the PCT-group. These changes were not significantly different from REF, and conclusions can not be drawn on these aspects. However, for improvements to be fully reflected in the ability to tolerate high physical work demands, interventions to counteract the reduced resources and thus improve both physical and cognitive-behavioural resources would probably be optimal. We therefore suggest that future interventions should integrate both PCT and CBTr in one initiative in the prevention on physical deterioration among workers. Future research is needed to verify this recommendation.

Strengths and limitations

This study tested new approaches to prevention of health deterioration among 294 female cleaners from nine representative workplaces [14]. The interventions were thoroughly developed and tailored to the specific job group of cleaners. Although adherence rates were rather low, they were not lower than intervention studies in similar job groups [44]. However, the development of the interventions primarily built on a theoretical rationale derived from efficacy studies. Efficacy studies differ from effectiveness studies by being conducted in a context that gives optimal conditions for implementation [45]. Efficacy is necessary to, but not sufficient for effectiveness [45] and implementation is suggested to be thought of as interacting with the efficacy to determine effectiveness [46]. Thus in an effectiveness study, implementation plays an important role in obtaining results. In the current study, two specific efforts were made to support implementation of the interventions. First, workplaces adopting this intervention study were obliged by contract to provide time for the intervention during working hours. Second, each training session was guided by an instructor to personalise the interventions. Although adaption to the workplace setting was performed and pilot studies conducted, the practical rationale behind the interventions could have gained strength, if they had derived from feasibility studies among cleaners. Both inadequate efficacy as well as implementation is a possible reason for the lack of effect in some of the strength parameters. Thus, further efforts to improve implementation and adherence rates of workplace intervention studies in job groups with low socio-economic resources and among workers with low influence on work schedules should be implemented in future study designs.

By the conservative intention-to-treat analyses (with forward and backward carrying of missing observations), a tendency to underestimate the variances appears. It should be mentioned though, that the analyses reported in the results section of this paper follow the standards of the consort statement [26] and are conservatively designed to false positive finding due to avoid bias associated with non-random drop-out. Even with the relatively large drop-out, intention-to-treat analyses were able to reveal significant intervention effects, supporting our hypotheses of the interventions. Nevertheless, given the large standard deviations shown in table 2 it is likely that this study suffers from impaired power and some false negative results may be evident.

A non-significant increase in kinesiophobia was seen in the PCT-group. Since the increase was non-significant, it cannot be ruled out, that it happened by chance. However, according to the fear-avoidance theory of pain, one reaction to an expected painful stimulus may be avoidance behaviour [47]. That is, the individual has certain expectations on the painful consequences of an activity, which lead to avoiding the activity. It is well known, that physical training in itself can introduce an acute pain response [48]. Thus a confirmation of the fearful expectation to physical training may exacerbate the fear of movement and result in the increased kineisophobia seen in the current study. Graded activity has been suggested as treatment method for pain patients with high kinesiophobia. With graded activity, loads are introduced gradually and thereby producing disconfirmations of expectations of pain and harm and actual consequences of the activity [47]. In the PCT, exercise intensity was increased gradually and instructors carefully informed the participants that some pain and soreness could be experienced after training. However, no cognitive-behavioural or operant exercises were included. Actually, this was avoided to reduce overlapping interventions. In spite of the insignificance, the numerical increase in kinesiophobia in PCT may encourage that future training interventions corporate such kinesiophobia preventive exercises prior to or concurrent with the training.