1. ABSTRACT Objectives: In occupational epidemiologic research, data on current job are often used as surrogate for longest held job and its exposures. We evaluated concordance between current and longest held jobs in a large, population-based, representative sample of US workers. Methods: Cohen’s Kappa was used for analyses of combined data from the 1986 NHIS Supplement on Longest Held Job and the 1988 NHIS Supplement on Occupational Health. Included were over 49,000 workers, age 18 years and over, in occupational groups with an estimated employment of 100,000 persons or more. Analyses were performed for different occupational categories by gender, race, and ethnic subgroups. Results: Statistically significant percent Kappa values (± SE) were observed for all occupational groups and ranged from 31.7 ± 1.1 to 71.2 ± 0.4 when 13 broad occupational categories were considered; from 25.6 ± 2.3 to 85.6 ± 1.4 for 41 more detailed categories; and from 9.2 ± 3.1 to 92.7 ± 2.6 for 206 detailed categories. Statistically significant differences were observed between occupational groups, and between gender and race-specific Kappa values within certain occupational groups. Conclusions: The high levels of agreement found in this large, population-based, representative sample of US workers indicate that for most occupational categories current or most recent occupation and industry could be used as a surrogate for longest held job (and its exposures), particularly when lifetime work histories are not available. However, given the rapidly changing employment patterns in the US and elsewhere, confirmatory analyses using more current data are warranted in the future. Longest Held Job in US Occupational Groups: 1986 & 1988 National Health Interview Surveys Orlando Gómez-Marín MSc PhD, Lora E. Fleming MD PhD, William LeBlanc PhD, David J. Lee PhD, Terry Pitman BS, Alberto Caban MPH University of Miami School of Medicine Departments of Epidemiology & Public Health, and Pediatrics Study Website www.rsmas.miami.edu/groups/niehs/niosh; This study was funded in part through a grant from the National Institute of Occupational Safety and Health (R01 OH03915); occupational photos by Dr David Parkerwww.rsmas.miami.edu/groups/niehs/niosh METHODS The National Health Interview Survey, conducted annually by the National Center for Health Statistics, is a continuous, multipurpose, and multistage area probability cross-sectional survey of the US civilian non-institutionalized population living at addressed dwellings. Two occupational health related supplements, the 1986 "Longest Job Worked" supplement (n=37,917) and 1988 "Occupational Health" supplement (n=44,233), provided information on: current job, job during the previous 12 months, and longest held job. Concordance was assessed between reported current and longest held occupation. Several measures of agreement were considered, including Burnett's approach (i.e. sensitivity), Dice's Coincidence Index, and Cohen’s Kappa. Cohen’s Kappa was found to be most appropriate for the current analysis due to its ease of interpretation, robustness, extensive use in the literature, and availability of confidence intervals. Those 206 occupations with an average of 100,000 US workers/year in the period 1986-1994 were evaluated for concordance, as well as 13 and 41 larger categories. As well, several subgroups were analyzed: 2 genders; 3 ethnic subgroups (Whites, Blacks, Others); 2 Hispanic subgroups (Hispanics, non Hispanics). RESULTS The present study found moderate concordance between current and longest held job in a large, population-based, representative sample of US workers, with variations in this concordance by gender, race-ethnicity and occupational subgroups. The majority of the 13 and 41 occupational grouping categories had Kappa > 50.0. Some occupational categories and subgroups reached levels as high as 71.2 ± 0.4 when 13 broad occupational groups were considered; 85.6 ± 1.4 for 41 more detailed occupational categories; and 92.7 ± 2.6 for 206 detailed occupational categories. There was a clear pattern of increased concordance for higher paying jobs (such as professional specialties) and decreased concordance for low paying jobs (such as laborers), as seen previously in other studies. Statistically significant differences were observed between occupational groups, and between gender and race-specific groups within certain occupations. CONCLUSIONS In general, moderate to high levels of agreement between current and longest held occupation were observed using data from a large, population-based, representative sample of US workers. The present results indicate that current occupation can often be used as a surrogate for longest held job (and its exposures) when analyzing the NHIS data as well as other occupational databases. However, specific occupational subgroups, including gender and race- ethnicity occupational subpopulations, may have substantially decreased concordance, particularly among the less skilled and lower paying occupations. Finally, given the rapidly changing employment patterns in the US and elsewhere, confirmatory analyses using more current data are warranted.