Donna L. Mohr

Low-back injuries in a heavy industry. I. Worker and workplace factors.

The costs and circumstances of low-back impact injuries, and non-low-back injuries among field employees of an offshore petroleum drilling company, 1979-1985, were compared. The objectives were to identify worker and workplace factors associated with low-back injuries, to identify factors differentially associated with lost-time injuries, and to formulate recommendations for the control of low-back injuries. Low-backimpact injuries resulted largely from falls. Efforts to prevent falls would have a potential to reduce other serious consequences as well as back injuries. Workers performing the heaviest physical labor were at highest risk of low-back strains. Based on activities precipitating the injury, modifications of work site, equipment, and procedures to help reduce low-back strains are recommended. Only job was a predictor of whether a low-back strain was likely to be associated with lost time. Even this association was lacking for low-back impact injuries. Cost control by preventing the small proportion of high cost injuries may not be feasible. Rather, subsets of low-back injuries defined, for example, by work site or activity can suggest options for intervention.

Low-back injuries in a heavy industry. II. labor market forces

Trends in rates of low-back strains, low-back impact injuries, and non-low-back injuries among field employees of a petroleum drilling company, 1979–1985, were examined to investigate the relationship between economic factors and the incidence of low-back and other injuries. Economic indicators included the rate of resignations, a surrogate for turnover, and the rate of layoffs. Only lost-time low-back strain rates increased during times of worker layoffs. Non-low-back injury rates were highest during periods of high turnover and no layoffs. Although the increasing age of the work force and the anxiety generated by an industry-wide depression may have played a role, it is likely that the increase in lost-time low-back strain injuries was a worker response to possible layoff.

Bayesian identification of clustered outliers in multiple regression

We propose a Bayesian model for clustered outliers in multiple regression. In the literature, outliers are frequently modeled as coming from a subgroup where the variance of the errors is much larger than in the rest of the data. By contrast, when a cluster of outliers exists, we show that it can be more informative to model them as coming from a subgroup where different regression coefficients hold. We can explicitly model the clustering phenomenon by assuming that the probability of an outlier is a function of the explanatory variables. Fitting proceeds via the Gibbs sampler, using the Metropolis-Hastings algorithm to produce variates from the more unusual distributions. Initialization uses a least median of squares fit, and in some ways this method can be viewed as a Bayesian version of the many algorithms that use this fit as a start to some more efficient estimator. This method works very well in a variety of test data sets. We illustrate its use in a data set of sailboat prices, where it yields information both on the identity of the outliers and on their location, spread, and the regression coefficients inside the minority subgroup.

The "accident prone" worker: an example from heavy industry

A study of workers in a heavy industry shows that the proportion of workers with excessive injuries in two successive time periods did not exceed that expected by chance. While accident repeaters may exist, they were not a stable component of this workforce. Removing individuals with excessive injuries in a given time would not appreciably reduce the number of injuries in succeeding periods. Unlike earlier studies, this analysis was based on medically attended injuries only and controlled for job hazards, exposure, age, and changes in job and location.

Confidence limits for estimates of totals from stratified samples, with application to medicare Part B overpayment audits

Abstract Superpopulation models are proposed that should be appropriate for modelling sample-based audits of Medicare payments and other overpayment situations. Simulations are used to estimate the coverage probabilities of confidence intervals formed using the standard Stratified Expansion and Combined Ratio estimators of the total. Despite severe departures from the usual model of normal deviations, these methods have actual coverage probabilities reasonably close to the nominal level specified by the US governments sampling guidelines. An exception occurs when all claims from a single sampling unit are either completely allowed, or completely denied, and for this situation an alternative is explored. A balanced sampling design is also examined, but shown to make no improvement over ordinary stratified samples used in conjunction with ratio estimates.

A calcium model with random absorption: A stochastic approach

Absorption of calcium, or any mineral, by the body is subject to the random fluctuations typical of diffusion through membranes. In this paper we consider the absorption of calcium from the gut as a white noise process added to the deterministic model of Sen & Mohr (1990, J. theor. Biol. 142, 179-188). The first two moments for the amount of calcium in the extracellular fluid (ECF) have been derived using the Ito Calculus. A confidence interval for the total amount of calcium in the ECF is constructed. The equations for the first two moments of the fraction of dose calcium in the ECF are also given. Suggestions are made for the collection of experimental data in a form which should be helpful in investigating the magnitude of the stochastic effect.

A kinetic model for calcium distribution

Models for the distribution of minerals in the body are of interest as they allow researchers to trace the effect of a dose on mineral levels in plasma, storage and other compartments. Limited models are available in the literature for tracing the distribution of a calcium dose through a short time period. We propose a more general kinetic model which includes both limited absorption through the gut and loss of calcium via excretion. This new method has the advantages of giving reasonable results over moderate time periods, and allowing the extrapolation of calcium levels in extracellular fluid and storage. We fit the model to published data in order to obtain typical parameter values. These values are then used to analyze the implications of the model regarding the effect of calcium dose on calcium levels in various compartments.

A statistical model of tracking

A method is proposed to model individual patterns of growth over time by linear combinations of optimally chosen weighted orthogonal vectors. The goal is to distinguish individuals who track from nontrackers. Nontrackers are defined as those who follow different, usually more complex, growth patterns than trackers. Thus, nontrackers require more vectors than do trackers in modeling their longitudinal observations. A method of specifying the class-specific vectors and individual weights is demonstrated. When the proportion of nontrackers in the population is small, a modified form of the Akaike maximum entropy criterion is used to select the number of vectors appopriate for each person and also to classify each person into a tracking category. When the proportion of nontrackers is large, the modified Akaike criterion together with scatterplots of the growth curve weights are needed to distinguish trackers from nontrackers. The apprach is illustrated with longitudinal observations of height measured in an e...

Planning sample sizes for before-after accident comparisons.

Before-and-after safety studies are becoming more desirable in lieu of traditional cross-sectional studies in establishing crash modification factors, especially after the introduction of the first version of the Highway Safety Manual (2010). We present a simple method for estimating necessary sample sizes to obtain a target precision or power when the effect is represented as a proportional change in a Poisson rate. We also show that necessary sample sizes are not very different when full Bayesian models are implemented.

Testing for a ‘within-subjects’ association in repeated measures data

‘Within-subjects association’ is a tendency for a subjects personal highs in a variable Y to be associated with personal highs (or lows) in a variable X. We show that the within-subject Spearman correlations, individually of little reliability, can be aggregated across subjects in a repeated measures data set to effectively detect such an association. Observed significance levels can be assigned using an approximate randomization procedure even in small samples with variable numbers of observations per subject and ties in the ranks. We investigate the power of this method under the assumptions that the ranks within subjects follow a distribution proposed by Henze. On the basis of these results, we make recommendations for sample designs.