Skip directly to search Skip directly to A to Z list Skip directly to page options Skip directly to site content

NIOSHTIC-2 Publications Search

Search Results

Left truncation, susceptibility, and bias in occupational cohort studies.

Authors
Applebaum-KM; Malloy-EJ; Eisen-EA
Source
Epidemiology 2011 Jul; 22(4):599-606
NIOSHTIC No.
20040112
Abstract
BACKGROUND: Left truncation occurs when subjects who otherwise meet entry criteria do not remain observable for a later start of follow-up. We investigated left truncation in occupational studies due to inclusion of workers hired before the start of follow-up in a simulation study. METHODS: Using Monte Carlo methods, we simulated null and positive associations between exposure (work duration) and mortality for 500 datasets of 5000 subjects, assuming the absence and presence of heterogeneity in susceptibility to disease and to the effect of exposure. We examined incident hires (followed since hire) and left-truncated prevalent hires (those hired before baseline and remained employed at baseline). We estimated the association (&OV0404;1*) as the mean slope using Cox proportional hazards with a linear term for exposure, under scenarios with and without susceptibility. RESULTS: With homogeneous susceptibility, there were no differences between incident and prevalent hires. Introducing only disease susceptibility did not change results. However, with heterogeneous susceptibility to the effect of exposure, downward bias was observed among prevalent hires under both the true null and positive exposure-response scenarios. The bias increased with time between hire and baseline (null: &OV0404;1* = 0.05 [SD = 0.08], &OV0404;1* = -0.08 [SD = 0.24], &OV0404;1* = -0.18 [SD = 0.98] if hired <15, 15 to <30, and = 30 years before baseline, respectively), coincident with a decreasing percentage of susceptible subjects. CONCLUSIONS: Prevalent hires induce downward bias in an occupational cohort. This occurs because subjects who are less susceptible to the exposure remain exposed the longest, thereby underestimating the association.
Keywords
Models; Epidemiology; Mortality-data; Statistical-analysis; Analytical-processes
Contact
Katie M. Applebaum, Department of Epidemiology, Boston University School of Public Health, 715 Albany St, T322E, Boston, MA 02118
CODEN
EPIDEY
Publication Date
20110701
Document Type
Journal Article
Email Address
kappleba@bu.edu
Funding Type
Grant
Fiscal Year
2011
NTIS Accession No.
NTIS Price
Identifying No.
Grant-Number-K01-OH-009390; B01182012
Issue of Publication
4
ISSN
1044-3983
Source Name
Epidemiology
State
CA; DC; MA
Performing Organization
Boston University Medical Campus
TOP