This study was proposed to evaluate the NIOSH equations of 1981 and 1991 to determine their ability to predict back injury. The proposed study called for 2000 cases; however, only data for 444 cases was collected for which LIs (lifting index) were calculated. Only 406 out of the 444 cases had positive exposure hours. In the descriptive analysis, all 444 were included. For relating outcomes to predictors, only 406 subjects with positive exposure hours were included. Prior to carrying out the study as proposed, a pilot study was conducted to develop data collection methods, the associated questionnaires and other forms, and to identify the problems that may be encountered in the process. One of the major problems encountered is the discovery that the 1991 equation applied to a small percent of jobs identified. This resulted in the revision of the data collection protocol to include a higher percentage of manual handling jobs, particularly complex jobs such as those commonly found in storage and distribution sectors. While the data are sparse, a statistically significant prospective relationship between LI (both 81 and 91) appears to be confirmed. Due to sparseness of the data, statistically significant modifiers of LI effects are not found. The only variable that comes close to significance is subject weight, which shows a decreasing effect on injury probability. Due to the small number of injuries, the ability to make firm conclusions was hampered. In two formal analyses, the lifting index 1981 equation (L181) model appeared to be a better predictor, while in the simple logistic regression analysis, the lifting index for the 1991 equation (LI91) model appeared to be a better predictor. It was also noted that mild evidence of non-liner increases in injury probability as a function ofLI91 also noted by Waters et al (1999). Such nonlinear effects when using the LI81 index was not noticed in this study. The primary findings include: (1) Many jobs were excluded because of the selection criteria and the requirements outlined in the users' manual. (2) The equation parameter ranges were often violated. In such cases, the Recommended Weight Limit (RWL) was zero which resulted in an LI of infinity. (3) Even though the sample of jobs was biased towards jobs amenable to analysis with the revised NIOSH equation, the majority of jobs studies technically violate restrictions on the maximum shift length (8 hours) and that no other significant Manual Materials Handling (MMH) tasks be performed. (4) Difficulties in measuring some of the parameters, particularly angles of twist and task frequencies were noted. (5) Jobs containing multiple tasks created difficulty in calculating the combined Li' s as proposed by the equation. This often resulted in large values for LI.
Texas Tech University, Industrial Engineering Department, Lubbock, Texas