Reply: We appreciate the opportunity to reply to the letter to the editor by the Association of American Railroads (AAR) consultant team of Dr Thiese, Dr Hegmann, Mr Page, and Mr Weames,1 which raised questions about our article published in JOEM2 and two other publications3,4 from our study of working conditions and musculoskeletal symptoms among railroad maintenance-of-way (MOW) workers. Similar letters were sent to the other two journal editors. We conducted the study jointly with two other research teams who were studying mortality, and the social and economic impact of diseases and injuries among MOW workers. We followed guidance from our respective Institutional Review Boards (see publications). We reported the data in great detail and acknowledged in our publications the potential limitations of our study, including use of workers’ self-report of symptoms and workplace exposures, and a low survey response rate. Our study questionnaire and research protocol were reviewed and approved by an international outside panel of experts, including Prof C. Hulshof (Amsterdam UMC, Netherlands), Prof M. Bovenzi (University of Trieste, Italy), Dr R. Dong (NIOSH, USA), and Dipl Ing. U. Kaulbars (IFA/DGUV, Germany). Our questions on vibration exposure and symptoms were based on validated instruments, primarily the collaborative European VIBRISKS project5 and earlier research.6 SELF-REPORTED WORK EXPOSURES Thiese et al argue that “Many studies have concluded that self-reporting of exposure to workplace factors is not accurate.”1 Such a blanket statement does not accurately summarize the literature. All physical work exposure assessment methods have their strengths and limitations. Observational methods “may lead to exposure misclassification by not accounting for all variation in exposure between workers or within multi-task jobs during limited periods of observation.”7,p.1247 Uncertainties and systematic errors of direct measurements and observation may be introduced by the selection and characteristics of the measured vehicle or tool and the particular job site and duties.8 One study found that the overestimation of actual driving time by surveyed drivers was small, in the range of “5–13%, mean 11%.” In that study, Bovenzi concluded that there existed a “good agreement between reported and observed duration of exposure.”9 The use of health and symptom questionnaires is a common practice in medical surveillance and to monitor health status and diagnose vibration-induced disorders.10 Thiese et al misrepresent the conclusions of the 2005 review by Stock.11 Stock did find that the validity of self-reports for repetitive movements and for hand or finger use were “poor to fair.” However, overall, she found that “Questions evaluated for both reproducibility and validity that performed well in both sets of studies” included “presence and duration of whole-body vibration” as well as “duration or presence of sitting and standing posture, the presence of walking, kneeling or squatting postures, duration or frequency of hands above shoulders, manual handling of more than or less than 10 kg, general level of physical effort…, and duration of the use of visual display terminals.”11,p.409. Similarly, Gardner found “self-reported and observer-rated exposures showed moderate to good levels of agreement for some upper extremity exposures, including lifting, forceful grip, use of vibrating tools, and wrist bending.”12,p.547 A related important issue is whether measurement error from self-reports is primarily random or systematic? Is it primarily non-differential (biased mainly toward the null) or differential (biased toward or away from the null) depending on level of symptoms? Some studies have shown “differential reporting of self-reported exposures by workers with musculoskeletal symptoms… who either overestimated or underestimated their physical work exposures.” However, other studies “have shown no systematic differences in exposure reporting related to musculoskeletal complaints.”7,p.1247 Stock reported a similar variation of study results.11 Dale found that “Upper extremity symptoms did not predict greater disagreement between self-reported and observed exposures.”13,p.1 Similarly, Buchholz found that “Workers with hand/wrist symptoms provided ratings that were more strongly related to the directly measured exposures than those without symptoms.”14,p.1064 DIRECTION OF POTENTIAL SELECTION BIAS The AAR consultant team acknowledges one method we used to assess the potential degree and direction of selection bias. That is, we compared survey respondents to a random sample of non-respondents, who agreed to complete a 10-question version of the survey by telephone.2 However, Thiese et al failed to mention that we also compared survey respondents to the national membership of the Brotherhood of Maintenance of Way Employes Division (BMWED), the union representing MOW workers, on available demographic information. They also failed to mention the direction of potential bias. Using both methods, compared with non-respondents, active members who completed the survey were younger, had slightly better working conditions, and tended to be healthier (except for back pain). These results suggest that our analyses may underestimate associations between working conditions and musculoskeletal symptoms among MOW workers.3 ACCESS TO DATA The AAR consultants also fail to mention that they were denied access to the data because the sponsor of the study, the BMWED, retains “proprietary possession of raw data.” The union felt that confidentiality and protection of the research subjects was of great importance, in order to encourage participation in the study and to “prevent possible misuse in other contexts” (letter from Zachary Voegel, General Counsel, BMWED, October 23, 2020). The union's concern was based upon a history of retaliation by railroad employers against employees who report injuries. Documentation of that history includes: A 1989 General Accounting Office report which found “substantial underreporting and inaccurate reporting of injury and accident data.”15 Amendments to the Federal Railway Act in 1996 which specifically state that “harassment or intimidation of any person that is calculated to discourage or prevent such person from receiving proper medical treatment or from reporting such accident, incident, injury, or illness will not be permitted or tolerated…” (49 CFR § 225.33). August 3, 2007 amendments to the Federal Railroad Safety Act (FRSA), 49 U.S.C. § 20109, to transfer authority for railroad carrier worker whistleblower protections to the Occupational Safety and Health Administration (OSHA), and to include new rights, remedies, and procedures, including protection from discrimination against employees reporting safety concerns or injuries.16 October 16, 2008 amendments to FRSA in the Rail Safety Improvement Act (Public Law 110-432) to specifically prohibit discipline of employees for requesting medical treatment or for following medical treatment orders.16 A January 16, 2009 letter by the Federal Railroad Administration administrator to a major rail carrier criticizing “widespread harassment and intimidation” and “discouraging injured employees from receiving proper medical treatment and retaliating against employees who reported injuries.”17 A $1.1 million fine by OSHA of another major rail carrier for terminating three workers for reporting injuries.18 This history may have been a contributing factor in reducing our survey response rate. In addition, to ensure that the identity of all survey participants would be legally protected from discovery, the researchers obtained a Certificate of Confidentiality from the National Institutes of Health. UNPUBLISHED MOW EXPOSURE ASSESSMENTS The AAR consultants provide two references to MOW exposure assessments that Weames et al (including two of the authors of the letter to the editor) conducted (Thiese et al's references 34 and 35) between 2004 and 2013, which appear to have been published as conference proceedings but not subsequently as journal articles. These studies cannot be found through commonly used broad-coverage academic search engines including PubMed, Google Scholar, or Scopus, nor were they provided to us. Thus, it is ironic that Thiese et al chastise us for not mentioning those papers in our articles. (The proceedings mentioned in Thiese’ references 34–35, while not appearing in academic search engines or journal sites, nor in the subscriptions of our institutions, appear instead to be behind a paywall available on line for $225.) To be properly evaluated, Weames et al's research needs to be submitted to journals and undergo the same rigorous peer-review process, and perhaps post-review critique, that our three MOW publications have gone through. At the beginning of our study, in 2016–2017, a number of BMWED officials made inquiries with local railroad officials regarding access to work sites and to MOW equipment in order to allow us to measure vibration and other biomechanical exposures as part of our study. Those requests were denied. BMWED also requested that the Railroad Retirement Board (RRB) provides aggregate, de-identified sickness (medical), and disability claim data for BMWED members; however, those data were unavailable, as the RRB does not categorize medical and disability data by craft or class titles or designations. DAILY DURATION OF TOOL USE The AAR consultants also criticize our 2020 paper4 for its use of estimates of daily duration of tool use. The weights we applied to the response options to the question “About how often do you use this (these) tools in a typical day?” may be less precise than some other measurement methods. These weights were “always” (1.0), “often” (0.75), “sometimes” (0.50), “rarely” (0.25), or “never” (0). These response options and applied weights led to the mistaken implication in our 2020 paper that, for each of 9 tools, a majority of workers (of those reporting use of that tool) used that tool for at least 4 hours/day. An alternative interpretation of those response options would be as relative categories of exposure rather than absolute hours of tool use. Therefore, as a sensitivity analysis, we reduced our weights for daily tool use by half to “always” (0.5), “often” (0.375), “sometimes” (0.25), “rarely” (0.125), or “never” (0). As a result, all statistically significant associations between power tool use and upper extremity symptoms remained statistically significant, and had larger adjusted prevalence ratios. Reducing such weights further, to the range of hours of use suggested by the AAR consultants, would increase estimates of associations with symptoms even further. Job tasks, exposure duration, and situations of MOW workers in the field are highly variable and dependent on many factors, which limits the validity of taking a few observer measurements, as suggested by Weames et al, to be the proper approach. Large mechanized MOW production gangs (such as tie renewal, and rail laying/gauging) often have pre-arranged track occupancy “windows” established in advance of the production crews’ arrival. These large production crews will often have track occupancy windows of 8 to 10 consecutive hours for 4 to 8 consecutive days before off duty rest days. Local “section maintainers” and other MOW employees are often assigned to work with and supplement the production crews, or are tasked to perform ancillary track/bridge/grade crossing work within the production crews’ occupancy “window.” Additionally, when not supporting production crew activities, “section maintainers” and “track welders” often perform work under “traffic conditions,” meaning that they perform track maintenance activities without the need for “track occupancy” authority. Weames et al's job task analysis suggests that 80% of MOW employees’ on-duty time is time when the MOW employee's hands are idle, which is not consistent with the workers’ job descriptions. The results we report in our other papers are not inconsistent with findings from other studies. In an earlier Finnish study of railway track maintenance workers, the hand-held tamping machines caused most of the vibration. Many of the workers also used other vibrating tools, as also found in our study. In the Finnish study, symptoms of vibration-induced white finger and hand numbness increased significantly with the total duration of maintenance work.19,20 CONTROL OF CONFOUNDERS We agree that control of confounders is important in epidemiological research. However, the AAR consultants fail to mention that the confounders that we were able to control for in Table 22 were the variables available in national data sets—age, region of the country, race/ethnicity for the NHIS-OHS, and age and race/ethnicity for the NIOSH QWL. We controlled for additional confounders in our analyses in our other MOW papers.3,4 However, given that survey questions were added by all three MOW study teams, while at the same time the length of the survey needed controlling, we focused in our other papers on the main available confounders. The AAR consultants present a narrow list of potential psychosocial confounders (job satisfaction, anxiety, depression), which may be considered, at least in part, as outcomes of work organization.21 They leave out workplace psychosocial or work organization variables, which we did assess in another paper.3 For example, in exploratory analyses, we examined associations between symptoms and safety climate, job decision latitude, supervisor support, and staffing.3 While beyond the scope of our earlier papers, we intend to conduct future explorations of the combined role of biomechanical and work organization variables among MOW workers. Other research has found, for example, that biomechanical and work psychosocial exposures were independent risk factors for incident carpal tunnel syndrome, with only a slight amount of confounding.22 In fact, modern models of the role of psychosocial factors emphasize the key role of the organization of work in shaping exposure to both biomechanical and psychosocial risk factors.23 STATISTICAL METHODOLOGY The AAR consultants criticize our statistical methodology while failing to mention that our 2019 paper in JOEM2 was mostly descriptive. Tables 1, 3–6 presented only descriptive data on demographics, working conditions, and musculoskeletal symptoms.2 Only Table 22 (comparison of active male MOW workers with employed US men), text comparing active male MOW workers to US men on self-reported carpal tunnel syndrome diagnosis, and supplemental Tables 1 and 22 (our assessment of potential selection bias) used statistical tests. Thiese et al provide no specific critique of the statistical methodology we used. The AAR consultants also fail to mention that we did explain how missing data was handled. We stated that “The prevalence of specific working conditions and measures of ill health from survey respondents were computed by dividing the number of respondents reporting that working condition, symptom, or diagnosis, by the number of respondents who completed at least one question on that page of the survey, as the response rate tended to decline in later pages of the survey.”2,p.587 DEFENSE EXPERTS FOR RAILROAD CORPORATIONS Finally, Thiese et al fail to mention in the text of their letter that they are long-term designated defense experts for major American railroad companies. Mr Page, author along with Mr Weames of the unpublished studies (Thiese et al's references 34 and 35), testified in 2014 that he had previously worked for the AAR, and, since he founded Page Engineering, the firm has received about $13 million from the AAR and major railroads, including payments for evaluations and testimony in lawsuits.24 Thiese et al state they received only “partial support” from the AAR, an industry trade group representing the major railroads of North America. However, the authors of the letter fail to mention that, in an earlier request for our full study data, Dr Thiese stated that “we were asked by the Association of American Railroads to read your articles and provide our opinion” (Dr M. Thiese, personal email, October 16, 2020, available upon request). OPPORTUNITIES FOR COLLABORATION AND JOINT RESEARCH The BMWED informed us that, in 2019, they provided the full text of the reports from the three MOW study research teams to the AAR and to individual railroad companies. This was done “in anticipation of meaningful collaboration to eliminate or mitigate work-related health risks among MOW workers. However, to date our offer for mutual collaboration and cooperation on risk reduction efforts has gone unanswered by the industry” (letter from Zachary Voegel, General Counsel, BMWED, October 23, 2020, available upon request). Comprehensive studies of ergonomic working conditions and vibration exposure of workers from railroad equipment have long been recommended by the Federal Railroad Administration and other agencies.25–27 Greater scientific understanding of the exposure and health issues identified in our studies could occur if Thiese et al encouraged their sponsor, the AAR, and the railroad companies, to accept the BMWED's invitation and conduct a major collaborative study. Researchers from NIOSH have offered technical assistance for any further follow-up intervention studies. Such a study could include, along with self-reports, observer measurements of biomechanical risk factors, vibration measurements of tools and equipment, and a review of anonymous medical claim and disability data.