O8D.7 The effect of right truncation bias on biomechanical factor risk estimates for CTS. (April 2019)

• Record Type: Journal Article
• Authors: Harris-Adamson, Carisa; Eisen, Ellen; Rempel, David; Gerr, Fred; Bao, Stephen; Hegmann, Kurt; Kapellusch, Jay; Meyers, Alysha
• Is Part Of: Occupational and environmental medicine. Volume 76(2019)Supplement 1
• Page Start: A76
• Page End: A77
• Publication Date: 2019-04
• Subjects: Medicine, Industrial -- Periodicals; Environmental health -- Periodicals; 616.980305
• DOI: 10.1136/OEM-2019-EPI.206
• Languages: English
• ISSNs: 1351-0711

Abstract

Abstract : Carpal tunnel syndrome (CTS) is a potentially disabling occupational illness with high incidence rates in certain occupations. Prospective workplace studies have identified associations between personal, psychosocial and biomechanical factors and incident CTS. However, biased estimates of association due to the healthy worker survivor effect (HWSE) have not been fully investigated. The HWSE results in left truncation bias when affected workers are lost from the population of more highly exposed workers before study participants are sampled. Right truncation bias occurs when affected workers are lost disproportionately from higher exposure strata after enrollment in a prospective cohort study. This study evaluated the effects of right truncation bias on risk estimates for biomechanical factors. Between 2001 and 2010 five research groups in the US conducted coordinated prospective studies of incident CTS among workers from various industries. Hazard ratios (HR) were estimated using Cox proportional regression analyses with adjustment for personal and workplace psychosocial factors. Inverse probability censoring weight (IPCW) values were created using a cross-validated approach analogous to the lasso method for variable selection and parameter regularization. The weighting model predicted the probability of exit from the cohort in the next 3 months. The final models included 2520 workers and 166 incident CTS cases. IPCW models with adjustment for right truncation bias were used to predict incident CTS. Effect size estimates for the workplace biomechanical factors of Peak Force, HAL Scale, Forceful Repetition Rate and the ACGIH TLV for HAL increased by 11% to 71% in comparison to models without adjustment for right truncation bias. After adjusting for right truncation bias the predicted incidence of CTS in highly exposed workers increased up to 3-fold. Bias from symptomatic workers disproportionately leaving the workplace was identified in this cohort. Adjusting for this bias has likely produced more accurate dose-response relationships than those previously reported.