Free Occupational Injury & Illness Incidence Rates - Kansas

Kansas Division of Workers Compensation

Annual Statistical Report

2006

Appendix A
Technical Notes: Occupational Injury and Illness Incidence Rates and Closed Claims Study

Kansas Division of Workers Compensation

Annual Statistical Report

2006

Occupational Injury and Illness Incidence Rates
BLS Survey of Occupational Injuries and Illnesses: The Bureau of Labor and Statistics (BLS), with the help of the state agencies, selects a non-proportional stratified probability sample of employment establishments and mails them questionnaires. Employers are instructed to record all nonfatal employee injury and illness incidents, number of days away from work for each recorded injury/illness, the number of employee hours worked and the establishment's average employment. Participants in the annual survey consist of employers who maintain Occupational Safety and Health Administration (OSHA) records on employee injuries and illnesses on a regular basis under federal law and smaller employers who are exempt from OSHA record keeping requirements. The data collection process differs for the former and the latter. The former are mailed a questionnaire in February, following the survey year, and are asked to transfer from their records all injuries and illnesses incurred as well as demographic and hours worked data. The latter, exempt employers (those with fewer than 11 employees and those designated as "low-hazard industries" by OSHA) are notified in December of the prior year (e.g., contacted in December of 2000 to record injuries for the 2001 survey) that they have been chosen to participate in the survey and must keep records of all employee injuries. The participating state agencies are responsible for collecting data from employers within their jurisdiction and for submitting these questionnaires to BLS for analysis. The BLS uses its incidence rates as a benchmark by which to compare the frequency of injuries and illnesses occurring within jurisdictions, industries or specific occupations for a calendar year. The variable "Total Injuries and Illnesses per 100 Full-time workers" (the most widely quoted measure) is calculated as follows: Formula: IR = (N/EH) x 200,000 IR = Incidence Rate N = Total number of occupational injuries and illnesses EH = Total hours worked by all private industry employees during the calendar year 200,000 = Base for 100 equivalent full-time workers - 40 hours per and 50 weeks per year Kansas Occupational Injury and Illness Incidence Rates: The division collects data on the entire population of workplace injuries and illnesses in the state of Kansas through its first report of injury form and stores it in its relational database. Every employer covered under the Workers Compensation Act that has workplace injuries must submit first reports of injury. The severity of each occupational accident or illness and the industrial classification code are mandatory data elements that must be reported by employers to the state. The severity of each accident or illness is exhibited by the numerical code representing the following severity categories: 0-No time lost, 1-Time lost and 2- Fatality. The division's analysts utilized the BLS statistical formula (see above) to calculate the incidence of injury for each severity classification for Kansas's nonfederal employment hours for the past ten fiscal years. Data used in the calculation of incidence rates was obtained from the Kansas Labor Market Information Services and division databases.

192

Kansas Division of Workers Compensation

Annual Statistical Report

2006

Kansas Closed Claims Study (CCS) Methodology
The following is a description of the methodology used by the Technology and Statistics section of the division for the 2006 Closed Claims Study (CCS). Sample Design: The division consulted with a professor of statistics from Washburn University in order to achieve both efficiency and effectiveness in the CCS Study. Rather than collecting data from the entire population of claims for a calendar year, which would be impractical (as it would result in very large data sets), extremely expensive and labor intensive, the division's researchers used a random sample from the population and make valid inferences about its characteristics using reliable and credible statistical techniques. The Workers Compensation division must have a sample that is sufficiently large and accurately representative of the population in order to perform relevant statistical inference. The sample must also preserve the power of equal probability associated with simple random sampling for statistical purposes. This enables the researchers to process the statistics without having to weigh different variables differently. Simple random sampling will not work with this data because carriers with higher paid losses tend to have a higher proportion of paid loss claims, making the distribution of paid loss claims unevenly distributed. In order to sample a larger percentage of those carriers, the division utilizes a two-stage type of probability sampling procedure known as "disproportionate stratified sampling." That procedure ensures that different groups within the population will be adequately represented in the sample. The general strategy employed is to first create strata (subsets of the total population) that are more homogeneous than the population as a whole, and then to sample a different fraction of each strata. Then, when recombined through analysis, the resulting sample will be reasonably representative of the more heterogeneous total population. The population was stratified according to paid losses. The specific variable used to stratify the population was a percent of total paid losses for all workers compensation claims in the state of Kansas. Subsequent to stratification, the division selected carriers from each stratum by utilizing a random number generator. All carriers in any particular stratum had the same chance of being selected as any other carrier in the same stratum. The selected carriers were asked to randomly select claims from their own databases that met the CCS study criteria. Data Collection: The organizations included in the study were then asked by the division to randomly sample from their databases approximately 200 (or less, if they did not have 200) claims for the specified calendar year. The sample was to be taken from each entity's pool of claims, including both medical and indemnity payments. Each claim in the sample also was required to have been open at least one day during the period of January 1, 2005, to December 31, 2005. The division secured permission from the National Council on Compensation Insurance to print and use the Detailed Claim Information (DCI) survey instrument in order to create data definitions and structure for the Kansas Closed Claims Study survey. DCI is a national standard for reporting comprehensive claim data from insurance carriers. Adhering to the DCI structure, programmers in the Kansas Department of Labor created two software packages to assist reporting entities. One package was a manual entry system; the other application allowed a text 193

Kansas Division of Workers Compensation

Annual Statistical Report

2006

file to be imported electronically. Both products included editing limitations on inputs to certain data fields. In addition, the division's analysts performed data scrubbing on the data sets to ensure that accurate aggregate statistics were reported to the legislature. Response Rate: Non-response bias is always a threat to the accuracy of a sample because nonrespondents may differ significantly from survey respondents. Typically, in any study such as this one, certain organizations do not respond due to various circumstances, including, but not limited to, bankruptcy, refusal to answer or lost forms. This year, however, all participants responded, thus the statistics do not need to be adjusted to reflect a non-response rate. Adjusting for Inflation: In order to compare claim costs over time the division utilized standard statistical procedures to adjust past costs for the effects of inflation. Every year, the value of the dollar to the average American consumer is driven down by inflation. This makes comparison of costs from year to year difficult to accurately interpret because the value of the dollar does not remain constant. In order to remedy this situation, the Bureau of Labor Statistics (BLS) has produced multiplying factors that are meant to "even out" cost data. These factors are referred to as Consumer Price Indexes and when used properly in an inflation adjustment formula, allow the researcher to represent each year's costs with a consistent dollar value. Over the last 25 years, the BLS has made numerous improvements to the CPI-U (Consumer Price Index for All Urban Consumers), making it a more accurate conversion factor for comparing costs over time. However, because the CPI-U is tied to federal income tax brackets, Social Security benefits, wage levels specified in collective bargaining agreements, government programs and private contracts, it would be impracticable for the BLS to update older CPI-Us to reflect the changes. Therefore, the researcher is not able to represent costs over time in a consistent dollar value by using the CPI-U. The CPI-U is thus better situated for converting specific costs for simple comparisons and not for analyzing time series data. In order to remedy the situation for the researcher, the BLS has developed the Consumer Price Index for All Urban Consumers Research Series (CPI-U-RS). This series, built exclusively for researchers doing time series analyses, allows the researcher to represent cost figures that reflect estimates of what inflation would have been if it had been measured using current CPI-U methods since 1978. The CPI-U-RS incorporates most of the improvements made to the CPI-U over time, however it has some limitations. The estimates are based on research covering a short time and extrapolated to a longer time period. Additionally, some of the improvements haven't been included for various reasons. In spite of these limitations, the CPI-U-RS is the most detailed and systematic estimate available of a consistent CPI series. CPI-U-RS is utilized by the Division of Workers Compensation for adjusting all costs for inflation from 1998-2004.

194