Risk assessment of manual handling operations at work with the key indicator method (KIM-MHO) — determination of criterion validity regarding the prevalence of musculoskeletal symptoms and clinical conditions within a cross-sectional study

Symptoms and disorders of the musculoskeletal system (MSDs) are the most common reasons for absence from work in Germany [1]. 18.9 million days of sick leave were caused by symptoms and disorders of the musculoskeletal system and connective tissue in 2015. This is equivalent to 26.8% of all days off work [2‐4]. The resulting economic loss amounted to 19.2 billion euros in 2011 [1]. Upper extremity musculoskeletal disorders are still common in the working population [5‐7], for example carpal tunnel syndrome (CTS) [8], and medial or lateral epicondylitis [9].

In addition to manual handling of heavy loads, exerting high forces, working in awkward postures and repetitive hand/arm movements are the most frequently discussed work-related physical risk factors [10, 11]. Many different occupational groups are associated with CTS, e.g. in manufacturing, fishing or construction. Those at risk include workers on assembly lines, meat packers, poultry workers, gardeners, musicians, farmers, mechanics and factory workers [12].

According to the European Council Directive 89/391/EEC of 12th June 1989 on the introduction of measures to encourage improvements in occupational safety and health (OSH), the employer must perform an assessment of the risks to safety and health at work, including those to which specific groups of workers are exposed [13]. Long-lasting, dynamic repetitive exposure of the hand-arm system during manual handling operations (MHO) alone or in combination with static and postural effort are known as causes of musculoskeletal symptoms and disorders [14‐19]. But based on a systematic review, da Costa et al. claimed, that studies with an adequate methodology designed to clarify the incidence and prevalence of WRMSDs are sparse and raise concerns in regard to the causal relationship [16]. According to the BIBB/BAuA Employment Survey in 2012, based on interviews of a random sample of 20,000 employees in Germany, 42.5% of the employees were exposed to MHO, and 18.2% suffered complaints from this exposure [20]. An ergonomic design and organization of workplaces should focus on reducing harmful exposures to the musculoskeletal system and thus the associated symptoms and disorders. Therefore, the assessment of manual work tasks is crucial to estimate health risks of exposed employees and to implement conditional prevention strategies in the working area to reduce such exposures.

Currently, a number of risk assessment methods for the evaluation of work-related physical exposures exist, e.g. Hand-Arm-Risk-Assessment Method (HARM) [21], Manual Tasks Risk Assessment Tool (ManTRA) [22], Assessment of Repetitive Tasks of the upper limbs (ART) [23], Job Strain Index (SI) [24], American Conference of Governmental Industrial Hygienists Threshold Limit Value for Mono-Tasks Handwork Hand Activity Level (ACGIH TLV-HAL) [25]. Takala et al. [26] analyzed existing methods systematically according to concurrent and criterion validity, repeatability, and aspects related to utility. However, the authors stated that many of these methods underwent only insufficient validation processes [26]. This is a relevant limitation, since the explanatory power of the method is not clear. What kind of risk (e.g. pain, accidents, job satisfaction, or fatigue) is assessed and what does the result mean? In addition, validated assessment tools are one prerequisite for the development and implementation of evidence-based OSH measures [27]. Only the SI and the HAL were tested against all quality criteria indicated above. Both methods are widely used internationally. The SI has the disadvantage that the data on frequencies of movements and force levels are assessed relatively broadly. In the case of the HAL, no activities lasting less than 4 h could be considered.

A new method for the assessment of working conditions involving manual handling operations was developed by the Federal Institute for Occupational Safety and Health (BAuA) and released as a draft in the year 2007 [28]. This draft of the so-called Key Indicator Method for Manual Handling Operations (KIM-MHO) was developed analogous to the existing KIMs for Lifting, Holding and Carrying (KIM-LHC) and for Pulling and Pushing (KIM-PP) of loads [29]. The draft of KIM-MHO was evaluated and modifications for its improvement were made [30]. Considering the requirements mentioned above, the method was validated regarding different criteria, including face validity, criterion validity, reliability and further aspects related to utility. The revised KIM-MHO and the results of the validation in regard to criterion validity are described in this article. An advantage of the KIM-MHO compared to the other methods mentioned above is that the force exertions can be considered more differentially. Furthermore, the method is designed to be applicable for any duration of an activity. Thus, even minor improvement measures (e.g. reducing the force level of individual motions) or elimination of unnecessary motions can be visualized and utilized for implementation of even minor improvements.

The KIM-MHO was developed for the risk assessment of manual handling operations. This paper aimed to present the level of criterion validity. The KIM-MHO was also validated regarding additional quality criteria, including face validity, reliability and further aspects related to utility [33]. A cross-sectional study design was chosen to compare the workplace exposures, determined by work analyses and described by the risk category score of the KIM-MHO, to clinimetric parameters (MSDs as outcome criteria), determined by interviews and physical examinations of the employees. The relation of exposures and outcomes was assessed using log-linear Poisson regression models to provide adjusted prevalence ratios.

The results of the evaluation of the criterion validity support significant associations with a dose-response relationship between the level of exposure to MHO (KIM-MHO categories 3 and 4) and the frequency of musculoskeletal symptoms (predominantly in the hand/wrist and elbow/forearm regions, as well as in the shoulder region and a tendency of increase in the neck region). Physical examinations of the exposed employees showed a tendency of increase of relevant diagnoses in the upper extremities and neck regions in those employees with high exposures according to KIM-MHO scores.

In the present study, criterion validity was determined in a cross-sectional design. Therefore, it was only possible to show correlations between the predicted risk of prevalent adverse health outcomes by KIM-MHO at workplaces with MHO and the real prevalence of pre-existing musculoskeletal symptoms, but not in relation to incident adverse health events. In order to achieve a high-quality testing of methods, homogeneous job demands and employee groups in the occupational setting of manual work processes are necessary. However, homogeneous conditions are only rarely present in the work areas. Typically, order-related changes in work requirements and flexible staff assignments occur. This complicates the definition of the level of occupational exposure. Longitudinal studies are usually of higher value, such as in the validation of exposure assessment by HAL to predict carpal-tunnel syndrome [42‐45]. However, in practice, a longitudinal method of looking at the possible link between occupational exposures and health related outcomes can hardly be realized, taking into account the necessity of accurate documentation of the changing occupational exposures during follow-up and the mostly recurrent characteristic of the health outcomes of interest. Since health related outcomes are individual and multifactorial, a direct association between working conditions and work-related symptoms is detectable only at relatively high exposures. These high exposures, however, were observed only at a few workplaces. While in epidemiological research the coherence of large groups of employees with similar workloads apply, the practical use of the KIM refers to specific working conditions. This basic contradiction must be considered. Also it has to be taken into account, that not all exposures automatically cause direct damage.

The aim of the present study was to determine the criterion-related validity by comparing the results of the KIM MHO with the prevalence of musculoskeletal symptoms in employees within a cross-sectional study design in terms of concurrent validity. The risk score provided by KIM-MHO can be interpreted as an external criterion. It provides hypotheses based on the concept, that higher values of the risk score are related to a higher probability of specific work-related adverse health outcomes. Considering the terminology of the COSMIN checklist (COnsensus-based Standards for the selection of health Measurement INstruments) [31], for evaluating the measurement properties of instruments, the validation of the risk score of KIM-MHO could also be seen as a part of the construct validity regarding the quality of hypotheses testing.

Prevalence of symptoms in the upper limbs region were used as the main outcome. Generally, the observed prevalence of symptoms in the upper extremities, neck and lower back for unexposed subjects of office workers is comparable to the prevalence in the German working population [46] and in other populations of office workers [47, 48]. Remembering earlier symptoms in a retrospective period could involve a recall bias. In order to minimize the bias and to get a more detailed overview, we asked for the past 7-days prevalence and the 12-months prevalence of symptoms. For most body regions, both 7-day and 12-months prevalences were associated with the exposure levels of the KIM. In addition, a physical examination was carried out, which found trends of an increasing number of diagnoses in the employees working at work places classified to the higher exposure levels of KIM-MHO. The observation of these three outcomes together might have reduced the recall bias.

The risk category of the KIM-MHO only describes the additional risk of MSDs according to the exposure to MHO. This does not mean, that employees in risk category 1 have no MSDs. As shown in Tables 2 and 4, there is already a base prevalence of symptoms in subjects with no or low exposure risk due to MHO. This high baseline prevalence of symptoms in the upper limbs and neck is seen in unexposed subjects as well [46‐48]. This is especially true for the neck region, where the 12-months prevalence of symptoms in females in the no/low exposed risk category is already 66%. This might be the reason why no dose-response relationship could be described for higher risk categories due to MHO for the neck region.

The regression analyses consider age, gender, body mass index, and body height to reduce the risk of confounding by these risk factors. The incidence and prevalence of symptoms and diseases of the musculoskeletal system are age-dependent and differ between men and women [49, 50]. Body mass index is a surrogate for the risk factors overweight and obesity, and is correlated with individual muscle strength and fitness [51]. Body height reflects the differences in the individual working height and the physical workload to the upper extremities [52]. The bias caused by confounding is estimated as low considering the marginal differences between the raw and the adjusted effect estimates.

More than 40 observational tools have been developed to assess biomechanical exposures [28], but only some of these tools have been tested in a systematic manner for validity, repeatability, and aspects related to their practical use [26]. In their review, Takala et al. (2015) [26] identified 8 eligible observational methods addressing workload on upper limbs. Five of them were tested for associations with MSDs within cross-sectional studies. Among them, only two (Strain Index [24] and Hand Activity Level [25]) were tested additionally for prediction of MSDs in longitudinal studies. As far as comparable, the present validation of KIM-MHO shows a dose-response relationship in the same direction and dimension for exposures to MHO as in the validation studies mentioned above.

In regard to other forms of bias, it is important to emphasize that the unexposed subjects (office workers) and the exposed subjects were examined in different projects, but with the same methodology. We do not expect relevant effects from the time lag between the investigations. The participation of the employees in the study was voluntary. We do not have information about nonparticipants. The participation rate per workplace was high, but it is not possible to provide more detailed information about this. Selection bias cannot be excluded. If employees with complaints especially under exposure to MHO preferred to participate, the effects could be overestimated.

Nevertheless, from the perspective of the developers of the KIM, a broader view of the validity of the method is necessary. The criteria that determine whether the assessment results are correct (i.e. whether possible physical overuse is predicted accurately enough) may depend not only on the method itself. It is also of fundamental importance that valid input data are available. The experience gained from this project and the long-time experiences with the other KIMs repeatedly point to imprecise input data. Inaccurate time estimates, non-representative and superficial descriptions of exposures seem to be the most common faults. One conclusion is, therefore, that correct data assessment is imperative.

The results of the evaluation process suggest satisfactory criterion validity of KIM-MHO, as the KIM risk score correlates significantly with the prevalence of MSDs in shoulder, elbow and hand/wrist regions and shows a dose-response relationship. Therefore, workplace risk assessment should use KIM-MHO in the case of exposures to MHO. If KIM-MHO indicates higher risks, measures of prevention such as job redesign, organizational changes, and provision of preventive occupational medical care should be considered. The usage of adverse health outcomes as parameters (clinimetric parameters) is useful, as long as the prevalence of symptoms in the reference group is low. If prevalence is high in the reference group (e.g. neck symptoms in office workers), only slight contrasts to high exposed groups of employees can be shown.