Catherine Trask

Measuring posture for epidemiology: Comparing inclinometry, observations and self-reports

The objective of this study was to use and evaluate three postural assessment methods for epidemiological studies of back disorders. The methods were: (1) a data-logging inclinometer; (2) observations by trained observers; (3) self-reports by employees. All methods were feasible in 50 heavy industry worksites. Inclinometry provided quantitative measures of flexion–extension (mean 17°, SD 11.2°), lateral flexion (mean 8.5°, SD 2.6°) and trunk movement speed (mean 14.3° per second, SD 4.9° per second). Observations and self-reports provided estimates of time spent in various trunk angles, general postures, materials handling and vehicles. Compared to observations, self-reports under-reported less common tasks, but over-reported task durations. In statistical modelling to determine if observations or self-reports could be used to estimate measured postures, observations accounted for 30 to 61% of the inclinometer measurement variance and self-reports for 33 to 40%. A combination of inclinometry and observations would be an ideal option to provide both depth and breadth of data on postures and other physical exposures for epidemiological research.

Data collection costs in industrial environments for three occupational posture exposure assessment methods

BackgroundDocumentation of posture measurement costs is rare and cost models that do exist are generally naïve. This paper provides a comprehensive cost model for biomechanical exposure assessment in occupational studies, documents the monetary costs of three exposure assessment methods for different stakeholders in data collection, and uses simulations to evaluate the relative importance of cost components.MethodsTrunk and shoulder posture variables were assessed for 27 aircraft baggage handlers for 3 full shifts each using three methods typical to ergonomic studies: self-report via questionnaire, observation via video film, and full-shift inclinometer registration. The cost model accounted for expenses related to meetings to plan the study, administration, recruitment, equipment, training of data collectors, travel, and onsite data collection. Sensitivity analyses were conducted using simulated study parameters and cost components to investigate the impact on total study cost.ResultsInclinometry was the most expensive method (with a total study cost of € 66,657), followed by observation (€ 55,369) and then self report (€ 36,865). The majority of costs (90%) were borne by researchers. Study design parameters such as sample size, measurement scheduling and spacing, concurrent measurements, location and travel, and equipment acquisition were shown to have wide-ranging impacts on costs.ConclusionsThis study provided a general cost modeling approach that can facilitate decision making and planning of data collection in future studies, as well as investigation into cost efficiency and cost efficient study design. Empirical cost data from a large field study demonstrated the usefulness of the proposed models.

Development and evaluation of an observational Back-Exposure Sampling Tool (Back-EST) for work-related back injury risk factors

We developed and evaluated an observational Back-Exposure Sampling Tool. A literature review suggested 53 exposure variables; these were reduced to 20 following field trials. Kappas for agreement beyond chance between six observers assessing exposures in 72 photos ranged from 0.21 to 1.0, with the highest values for posture type, trunk angle, manual materials handling, hands on item, and load weight. Intraclass correlations for agreement between pairs observing 17 workers once per minute for a full-shift were >0.74 for most postural, trunk angle, and manual materials handling variables. In validity testing, the proportions of shifts in flexion/extension and lateral bending observed for 169 full-shifts were compared to inclinometer measurements. Pearson correlations were 0.42 for 45-60 degrees flexion and 0.9 for >60 degrees flexion, but only 0.11-0.19 for lateral bending and trunk flexion less than 45 degrees . When lower flexion angles were collapsed to include trunk extension, correlations increased to >0.5.

Assessing whole body vibration exposure for use in epidemiological studies of back injuries: measurements, observations and self-reports

Improved assessment of whole body vibration exposure is needed for epidemiological studies investigating the causes of low back disorders. Vibration was measured on 54 worker-days in five heavy industries, with data collected on observed and self-reported driving conditions, demographics, and vehicle characteristics. Variables significant at p < 0.1 in simple linear regressions (20 of 34) were retained for mixed effects multiple regressions to determine the best prediction of rms vibration level and 8-h equivalent vibration exposure. Vibration was measured, on average, for 205 min per work shift (SD 105). Means and standard deviations in m · s−2 were: x-axis 0.35 (0.19); y-axis 0.34 (0.28); z-axis 0.54 (0.23); vector sum 0.90 (0.49); and 8-h equivalent vector sum 0.70 (0.37). The final three regression models retained only 2 or 3 of the 34 variables (driving speed (<20 km/h and/or 20–40 km/h) and industry and/or vehicle type and explained up to 60% of the variance (R 2 = 0.26–0.6). Practitioner summary The purpose of the project was to create a model that can predict whole body vibration exposure from a number of observed or self-reported variables. This could eliminate the need for costly and time-consuming field measurements of WBV in epidemiological studies. Despite a large number of variables included in the model (34) and 54 worker-days of WBV measurement, the final models contained only two or three variables, and explained 60% of the variance. While this is an improvement over use of job title in epidemiological studies, it still leaves a considerable amount of WBV variance unexplained.

How Long Is Long Enough? Evaluating Sampling Durations for Low Back EMG Assessment

Few ergonomic measurement tools explicitly state when and how to sample exposures. Traditional ergonomic sampling has used short, task-based or worst-case measurements, but these may misrepresent exposures, since they neglect the temporal variations throughout the workday. Understanding the representativeness of data from shorter measurement durations compared with full-shift measurements allows for optimization of measurements resources. This study compared a variety of low back electromyography (EMG) exposure metrics measured over a full-shift with the same metrics sampled over shorter durations to identify whether shorter durations provide representative measures of exposure. Portable EMG devices were used to measure low back EMG for 138 full work shifts in a range of jobs in heavy industry. Using a random start time, each full shift of data was resampled for 4 hr, 2 hr, 1 hr, 10 min, and 2 min. Exposure metrics from each duration were compared with the full shift using absolute and percent error, bias, and limits of agreement. Comparisons between one full shift and two full shifts were made for the subset of 35 workers with two measured workdays. Compared with full-shift data, bias is very low at all sampling durations. However, as sampling durations decreased from a full-shift to a few min, the absolute error, percentage error, and limits of agreement for exposure estimates show more deviation from full-day estimates. Estimates of mean and 90th percentile exposure averaged 8% error for 4-hr durations and 14% error for 2-hr durations. The errors for 4-and 2-hr measurement durations may be acceptable for many applications, particularly if the trade-off is measuring more subjects. Sampling durations of 1 hr or less seem likely to produce very large errors over all exposure metrics, particularly for the range and peak exposures. Depending on the purpose of measurement and the detail required, 4 hr or even 2 hr appears to be long enough to reasonably estimate full-shift exposure.

Data processing costs for three posture assessment methods

BackgroundData processing contributes a non-trivial proportion to total research costs, but documentation of these costs is rare. This paper employed a priori cost tracking for three posture assessment methods (self-report, observation of video, and inclinometry), developed a model describing the fixed and variable cost components, and simulated additional study scenarios to demonstrate the utility of the model.MethodsTrunk and shoulder postures of aircraft baggage handlers were assessed for 80 working days using all three methods. A model was developed to estimate data processing phase costs, including fixed and variable components related to study planning and administration, custom software development, training of analysts, and processing time.ResultsObservation of video was the most costly data processing method with total cost of € 30,630, and was 1.2-fold more costly than inclinometry (€ 26,255), and 2.5-fold more costly than self-reported data (€ 12,491). Simulated scenarios showed altering design strategy could substantially impact processing costs. This was shown for both fixed parameters, such as software development and training costs, and variable parameters, such as the number of work-shift files processed, as well as the sampling frequency for video observation. When data collection and data processing costs were combined, the cost difference between video and inclinometer methods was reduced to 7%; simulated data showed this difference could be diminished and, even, reversed at larger study sample sizes. Self-report remained substantially less costly under all design strategies, but produced alternate exposure metrics.ConclusionsThese findings build on the previously published data collection phase cost model by reporting costs for post-collection data processing of the same data set. Together, these models permit empirically based study planning and identification of cost-efficient study designs.

A Biopsychosocial Profile of Adult Canadians with and without Chronic Back Disorders: A Population-Based Analysis of the 2009-2010 Canadian Community Health Surveys

Chronic back disorders (CBD) are a significant public health concern. Profiling Canadians with CBD and the associated biopsychosocial factors at a national population level is important to understand the burden of this condition and how clinicians, health systems, and related policies might address this potentially growing problem. We performed a secondary analysis of the 2009 and 2010 Canadian Community Health Surveys to calculate prevalence and to better understand the differences between people with and without CBD. An estimated 20.2% of the adult Canadian population reports having back problems lasting for 6 months or more. Among people with CBD, there was significantly greater likelihood of living in a more rural or remote location, being Aboriginal, being a former or current smoker, being overweight, having other chronic health conditions, having greater activity limitations, having higher levels of stress, and having lower perceived mental health. People who were single/never married or had an ethnicity other than Caucasian or Aboriginal were less likely to report having CBD. These results contribute to a growing body of research in the area that may assist with strategic prioritization and tailoring of health promotion efforts and health services for people with CBD, particularly among vulnerable groups.

Turbulent Touch: Touchscreen Input for Cockpit Flight Displays

Touchscreen input in commercial aircraft cockpits offers potential advantages, including ease of use, modifiability, and reduced weight. However, tolerance to turbulence is a challenge for their deployment. To better understand the impact of turbulence on cockpit input methods we conducted a comparative study of user performance with three input methods -- touch, trackball (as currently used in commercial aircraft), and a touchscreen stencil overlay designed to assist finger stabilization. These input methods were compared across a variety of interactive tasks and at three levels of simulated turbulence (none, low, and high). Results showed that performance degrades and subjective workload increases as vibration increases. Touch-based interaction was faster than the trackball when precision requirements were low (at all vibrations), but it was slower and less accurate for more precise pointing, particularly at high vibrations. The stencil did not improve touch selection times, although it did reduce errors on small targets at high vibrations, but only when finger lift-off errors had been eliminated by a timeout. Our work provides new information on the types of tasks affected by turbulence and the input mechanisms that perform best under different levels of vibration.

Prevalence of Musculoskeletal Disorders Among Saskatchewan Farmers

ABSTRACT The extent of the musculoskeletal disorder (MSD) problem is not well understood among Canadian farmers, and little too is known about their epidemiology. The purpose of this study was therefore to (1) determine the prevalence of MSDs among farmers in one Canadian province; and (2) describe the types and severities of these disorders and patterns in their occurrence. This cross-sectional analysis was conducted using baseline survey data from the Saskatchewan Farm Injury Cohort Study. Reports of MSDs, demographic and health-related variables, reports of farm-related injuries, and economic conditions of individual farms were available for 2595 adult participants from 1212 farms in Saskatchewan, Canada. Relationships between MSDs and time spent doing farm work were investigated using tests of association. The participation rate was 48.8%. Most (85.6%) of participants reported having musculoskeletal pain in at least one body part over the past year. The lower back was most frequently affected (57.7%), followed by shoulders (44.0%), and neck (39.6%). More serious pain prevented 27.9% of respondents from performing regular work activities. MSD prevalence did not vary by sex, commodity type, or by total hours of farm work completed; prevalence was significantly (P < .05) related to time spent performing biomechanically demanding tasks such as heavy lifting and working with arms overhead. The most common MSD site in farmers was the low back, followed by the upper and then lower extremities. Although this study aimed to identify high-risk groups, lack of differences between demographic groups suggests that the majority of farmers are at risk for MSDs.

Using observation and self-report to predict mean, 90th percentile, and cumulative low back muscle activity in heavy industry workers

Occupational injury research depends on the ability to accurately assess workplace exposures for large numbers of workers. This study used mixed modeling to identify observed and self-reported predictors of mean, 90th percentile, and cumulative low back muscle activity to help researchers efficiently assess physical exposures in epidemiological studies. Full-shift low back electromyography (EMG) was measured for 133 worker-days in heavy industry. Additionally, full-shift, 1-min interval work-sampling observations and post-shift interviews assessed exposure to work tasks, trunk postures, and manual materials handling. Data were also collected on demographic and job variables. Regression models using observed variables predicted 31-47% of the variability in the EMG activity measures, while self-reported variables predicted 21-36%. Observation-based models performed better than self-report-based models and may provide an alternative to direct measurement of back injury risk factors.