NOTE: This page is archived for historical purposes and is no longer being maintained or updated.
C3.1 Overview of Case-crossover Methods for Injury Research-Mittleman MA, Sorock GS, Lombardi DA
In applying the case-crossover design in injury research, one of the major challenges is in identifying the person-time at risk of the specific injury under study. This key element sets observational injury research apart from the study of most chronic conditions and acute illnesses. For example, unlike myocardial infarction that can occur at any time, an occupational injury requires special circumstances to occur. Thus, only a subset of an individual's person-time ought to be considered eligible for study. In fact, if this is not taken into account, effect estimates can be biased in any direction.
A second major challenge in applying this technique in the injury field is that exposure information is often not available prospectively. This is not an inherent feature of the design, but rather a fact that arises because even the most common injuries are still relatively rare, making prospective data collection uneconomical in many settings. Furthermore, there are only limited circumstances where preexisting records collected for administrative or other purposes are available for ad hoc studies. As a result, many studies must rely on retrospective patient self-report for exposures that may be etiologically relevant.
Finally, a third major challenge in conducting these studies is in identifying control periods that are at once similar to the time at which the injury occurred with respect to the baseline injury risk, and yet for which the exposure(s) of interest are uncorrelated.
In summary, the case-crossover design allows for a novel approach to evaluate factors that may lead to injury at the workplace. This approach leads to freedom from confounding by differences between workers that would be difficult if not impossible to overcome with more traditional approaches. Implementing such studies involves overcoming several important challenges just as in any other observational design.
C3.2 Animal-Related Injuries to Veterinarians: Application of the Case-Crossover Design-Gabel CL, Gerberich SG, Maldonado G
Injury epidemiology presents some unique study design difficulties, in addition to those usually encountered in observational epidemiology. Besides defining the relevant hazard period, the acute and rapidly changing nature of this hazard period can make it difficult to measure exposures accurately.
The case-crossover design has been proposed as a way to study events related to acute exposures. This design enables each case to serve as its own control and facilitates comparison of exposures that vary over time for an individual, while controlling for those factors that are invariant. In contrast, a case-control design enables analysis of both types of factors by comparing case and control exposures.
In a recent study of risk factors for injuries among all Minnesota practicing veterinarians, several questions were included in the first phase comprehensive questionnaire to enable conduct of a case-crossover study. A nested case-control study was also conducted in a second phase of the overall effort. This provided an opportunity to contrast findings, pertinent to several potential risk factors, between the two study designs. A conceptual model provided the basis for the factors that were included in the study and the relevant analyses. Factors, that were addressed in this portion of the effort, were related to fatigue.
Exposure data for the case-crossover study were collected for the night prior to the event and for the month preceding the injury; data for the case-control (1:2 ratio) study were collected for the month prior to the injury (cases) or a randomly selected month (controls). Both univariate and multivariate analyses were conducted, relevant to the respective designs.
Results of this effort provide insight into the application of the case-crossover method. Advantages and limitations of both the case-crossover and the case-control design in injury epidemiology are identified.
C3.3 Transient Risk Factors for Occupational Acute Traumatic Hand Injury-Sorock GS, Lombardi DA, Hauser R, Eisen EA, Herrick R, Racine R, Mittleman MA
The National Electronic Injury Surveillance System reported that the most frequent occupational injury treated in hospital emergency departments in the United States in 1996 was a lacerated finger or hand. Potential transient risk factors for acute traumatic injury to the fingers or hand were studied using a case-crossover design. Patients were recruited from 21 occupational health clinics and three manufacturing companies in New England. A total of 1,128 subjects (874 men and 254 women) were interviewed a median of 1.3 days after the injury. The mean age was 37 +/- 11.4 (s.d.) (range 18-77). Of the total 1,360 injuries incurred by 1,128 subjects, 62% were lacerations, 13% crushes, 8% avulsions, 6% punctures, 5% fractures, 1% amputations, and 5% others. The occupations of the subjects were machine trades (33%), services (15%), construction (15%), manual materials handling (9%), bench work/assembly (9%), managerial (8%) and others (11%).
The case-crossover design uses cases as their own controls to assess the change in risk of a sudden-onset event during a brief "hazard" period (e.g. 10 minutes before the injury) as compared to an earlier "control" period. The control period selected for this analysis was the average exposure (frequency times duration) in the work-month prior to the injury. We found relative risks and 95% CIs as follows: using malfunctioning equipment or materials RR=12.0, (10.2-14.1), doing a task using an unusual work method, RR=11.1, (9.2-13.4), doing an unusual task RR=6.6, (5.6-7.8), being distracted RR=5.4, (4.7-6.2), rushing RR=2.4, (2.1-2.8) and glove use, RR=0.6, (0.5-0.7). The case-crossover design perfectly controlled for differences between subjects like age, gender, and occupation.
These results suggest that certain work equipment, practices and environment factors significantly increase the risk of a hand injury and glove use significantly protects against hand injury.
C3.4 Temporal Factors and the Risk of Occupational Acute Hand Injury-Lombardi DA, Sorock GS, Eisen EA, Herrick RF, Hauser RB, Racine R, Mittleman MA
Both mental and physical fatigue have been considered potential risk factors in studies of fatal and non-fatal injuries in the workplace. To evaluate associations between traumatic injuries to the hand and potentially modifiable risk factors, we enrolled 1,128 patients in a case-crossover study of occupational hand injuries. Patients were recruited over two years from 24 occupational health clinics in New England. Four temporal factors were evaluated. These factors included 1) the time of day of the injury, 2) the time since start of shift, 3) working overtime, and 4) hours of sleep before the injury relative to usual number of hours of sleep.
The median age of the subjects was 36 years and 77.5% were male. The majority of hand injuries occurred in the morning between 09:00-12:00 (43.2%), peaking from 10:00-12:00 (29.8%). The median time of injury was 3.5 hours into the work shift. The average hours worked per month was 193, including 22.5 hours of overtime, representing 11.7% of the total person-time at risk. In contrast, only 4.0% of injuries occurred while working overtime. The mean and median difference in sleep time (hours) was -0.1 and 0, respectively.
Hand injuries tend to occur early into the workday and sleep time does not appear to be a risk factor. The apparent lowered risk during overtime requires further evaluation of worker schedules and the availability of health care clinics used for data ascertainment during that time period. Alternatively, work tasks may be different during overtime as compared with regular working hours. Additional analyses are also needed to determine if these findings are modified by workplace, individual factors, or break periods during the work day.
|NOIRS 2000 Menu Page|
|NIOSH home page||CDC home page||NORA Traumatic Injury home page|