RULA uses diagrams of posture for classification the biomechanical exposure.18 The postures are evaluated in two groups, upper limbs (group A) and neck, trunk and legs (group B). Each group has partial scores range from 0 to 6 points. Then, for each group (A and B) the muscle use and strength/load are scored, which is added to the partial scores. Finally, it is obtained a total score from groups A and B, their values crossed in Table C (Appendix 1) to obtain the final score (from 1 to 7 points). The level of exposure and investigation priorities are classified as acceptable posture if not maintained or repeated for long periods (1 or 2 points); more investigations are needed and changes may be required (3 or 4 points); investigation and changes are required soon (5 or 6 points); investigation and changes are required immediately (7 points).13

RULA uses diagrams of posture for classification the biomechanical exposure.18 The postures are evaluated in two groups, upper limbs (group A) and neck, trunk and legs (group B). Each group has partial scores range from 0 to 6 points. Then, for each group (A and B) the muscle use and strength/load are scored, which is added to the partial scores. Finally, it is obtained a total score from groups A and B, their values crossed in Table C (Appendix 1) to obtain the final score (from 1 to 7 points). The level of exposure and investigation priorities are classified as acceptable posture if not maintained or repeated for long periods (1 or 2 points); more investigations are needed and changes may be required (3 or 4 points); investigation and changes are required soon (5 or 6 points); investigation and changes are required immediately (7 points).13

SI evaluates six exposure variables: intensity of exertion, duration of exertion, exertion per minute, hand/wrist posture, speed of work, and the duration of the task per day.8,9,19 Each of these variables has five classification criteria, with different weights (multiplier factors).19 This score is used to sort the task into three categories: probably safe tasks (<3); tasks associated with risk for the disorder in distal extremities from the upper limb (>5) and task that is probably dangerous (≥7)19 (Appendix 2).

SI evaluates six exposure variables: intensity of exertion, duration of exertion, exertion per minute, hand/wrist posture, speed of work, and the duration of the task per day.8,9,19 Each of these variables has five classification criteria, with different weights (multiplier factors).19 This score is used to sort the task into three categories: probably safe tasks (<3); tasks associated with risk for the disorder in distal extremities from the upper limb (>5) and task that is probably dangerous (≥7)19 (Appendix 2).

The creators of the methods allowed the adaptation to Portuguese-Brazilian and Professor Alan Hedge from Cornell University authorized the use of the RULA protocol graph. The process of cross-cultural adaptation was carried out based on the guidelines of Beaton et al.,20 which consisted of translation, synthesis of translation, back-translation, revision by a committee of experts, pre-test, and the presentation of documentation for analysis by the Experts Committee.20

Two hundred ninety five physical therapists with knowledge, training or job in the area of occupational health and ergonomics were invited via email to perform the pre-test of the pre-final version of the RULA and SI methods through the analysis of 3 video-tasks along with instructions for applying. On the first attempt, the majority of physical therapists did not respond to a request via e-mail, some of whom refused to attend, justifying lack of time. The respondents (n=23; 8%) reported difficulty in using the methods (RULA: 24.7%; SI: 21.7%), due to incomprehension at the data crossing in tables A, B and C from RULA, calculations from steps 2 and 3 from SI.

All the difficulties reported were forwarded to the experts committee and the pre-final versions were revised.21 The second pre-test was performed after a theoretical and practical training in groups of the 20–30 participants using video-tasks with duration of 1 hour already with additional instructions included for each method reviewed by experts committee. In a period of 5–7 days after the training the physical therapists received the new pre-final version of the two observational methods (RULA and SI) and 3 video-tasks to analyze. The new RULA and SI versions were applied in a new sample of 156 (RULA) and 107 (SI) physical therapists from both gender, with an average age of 31.2±7.2 years (ranging from 24 to 45 years).

Two experienced raters (physical therapists three and five years using of biomechanical exposure assessment in the workplace) analyzed independently one hundred and sixteen video-tasks in a randomized order. The raters received training together, which consisted of explanation of the methods and theoretical/practical application of 80minutes in video-tasks that did not take part of this analysis. After training the raters received 116 video-tasks to analysis (test) and reanalysis (retest) in separate. Playback of the videos could be done as often as needed and they could use a stopwatch and calculator.

The tasks assessed were performed in the textile industry, electronics industry, assembling line, tinsmith and sawmills, self-employed workers (hairdresser, dentist, beautician, woodworker, butcher, bricklayer, cook and upholster). The inclusion criteria were tasks which workers were exposed to biomechanical requirement in upper limbs, who remained in static posture and/or repetitive movements from 4 to 6h a working day. In addition, the repetition movements of the upper limb should last for more than 50% of the time of the work cycle.22

The measurement properties tested for RULA and for SI, were reliability, test and retest (inter and intra-rater), tests of agreement, internal consistency (rater 1, test data), construct validity and interpretability of methods. The measurement properties were based in studies by Terwee et al.10,12 and Mokkink et al.13,23 (COSMIN – Consensus-based Standards for the selection of health Measurement Instruments).

The retest was performed with an interval of 7 days. The video-tasks were analyzed in a random order.10,13,23 The construct validity between RULA and SI was tested using data from the retest (rater I) for the total scores of the two methods, besides the partial scores regarding the wrist/hand-wrist posture, muscle required/repetition of the movement and strength/intensity of exertion.

The hypothesis tested for the adapted versions were: (1) the measurement properties of the methods would be acceptable; (2) both methods would identify tasks with higher risk and that there was a positive correlation from moderate to substantial among them.

The data were analyzed using the Statistical Package for Social Sciences – IBM® (SPSS, version 19.0). The statistical tests used can be seen in Table 1.

In the stage of SI adaptation, we identified an error in the formula that was corrected. In addition, the calculation formula has been revised. The terms “exertion” and “efforts” used in SI were referred in Portuguese to “esforço”.

The methods also needed additional instructions. The versions in Brazilian-Portuguese of RULA and SI were semantically understandable and culturally appropriate after adjustments, allowing for testing their psychometric properties.21

The intra-rater 1 reliability ranged from moderate to almost perfect (k=0.57–0.86; 95% CI=0.22–1.00) for RULA, with agreement percentage above 75%. The same was not observed for the intra-rater 2, who presented reliability ranging from very poor to moderate (k=0.14–0.53; 95% CI=0.00 to 0.80) in most of the steps. Only to force or load in arms the reliability was almost perfect (k=0.86–0.93; 95% CI=0.64–1.00). The agreement ranged between 42 and 99%. The inter-rater reliability for RULA was very poor (k=−0.12 to 0.13; 95% CI=−0.22 to 0.26) and the agreement ranged from 18% to 96% (Table 2).

The intra-rater 1 reliability was excellent for all steps of SI with ICC2,1 ranging between 0.91 and 0.99. The agreement was very good with ranging from −1.03% to 0.01%. The reliability from intra-rater 2 ranged from poor to almost perfect (ICC2,1 0.05 to 0.92; SEM from −0.06% to 1.98%. The inter-raters’ reliability ranged from poor to moderate (ICC2,1 0.00–0.54) with SEM between −0.17% and 1.51% (Table 3).

The English version of the Brazilian-Portuguese RULA showed appropriate internal consistency (Cronbach's α=0.88), whereas SI had low internal consistency (Cronbach's α=0.65).

The construct validity between RULA and SI was poor for the total scores (Spearman rho=0.06; p=0.49) and for the partial scores regarding muscular requirement/repetition of movement (Spearman rho=0.04; p=0.59). There was a good correlation for the assessment of the wrist/hand-wrist posture (Spearman rho=0.61; p=0.00) and strength/intensity of exertion (Spearman rho=0.39; p=0.00) (Table 4).

The interpretability showed for total score to the highest (ceiling effect) and the lowest (floor effect) respectively 15.5% and 10.1% for the RULA, and 9% and 4% for SI. There was only a ceiling effect for the RULA (higher than 15%).

RULA and SI cross-cultural adaptation process was conducted in accordance with Beaton et al.20 The first pre-test showed a low number of questionnaires returned due to high percentage of doubts in interpreting questions, which required a second pre-test.21 According to the respondents, the main problems found in the first pre-test of the pre-final version were incomprehension of the criteria for analysis and calculation methods. However, these problems were solved after the elaboration of additional instructions and previous training face-to-face.

The physical therapists in the second pre-test no longer reported difficulties in applying observational methods resulting in the final version suitable for testing the measurement properties.

RULA showed ceiling effect only a little above the threshold of 15%. This finding indicates that the method has certain difficulty in distinguishing the level of risk exposure between the high-risk tasks, and may interfere in the classification for decision-making regarding the immediate or delayed intervention. SI did not exhibit floor-ceiling effect, which indicates that the method can discriminate between the tasks with higher and lower risk, increasing the reliability of the method.10