We performed a retrospective cohort study on data provided by an American manufacturing facility, whose production jobs involve hand-intensive assembly line work. Beginning with the opening of the facility, job applicants underwent a post-offer pre-placement (POPP) screening that included a physical examination, functional testing, and nerve conduction velocity study (NCV). Data used in this study were obtained from results of the POPP screening, personnel records including dates of hire and termination, facility medical records, and physical job demands analyses conducted by the employer. Data were available for five calendar years since the opening of this facility. All employee data records were linked by an individual employee identification number assigned by the employer. Workers hired by the company were identified in the dataset, though reasons for not being hired following POPP examination were not available. POPP of potential workers began during the first year the facility was opened, but for the initial 21 months after the start of POPP examinations, all workers who received screening tests were hired regardless of POPP NCV results. This provided the opportunity to follow workers with a range of NCV test results for several years after hire, to assess the predictive value of screening nerve conduction tests in a setting where these tests were performed but were not used in hiring decisions. The Institutional Review Board of Washington University School of Medicine provided the ethical approval for this study.

All job candidates received several screening tests including a physical examination, nerve conduction velocity studies, and functional screening tests addressing strength, coordination, and repetitive motion relevant to the work activities. The NCV evaluated the motor response of the median nerves across both wrists using the NeuMed Brevio^® nerve conduction monitoring system. Testing was performed in a single on-site clinic by several trained physical and occupational therapists, according to the device manufacturer’s guidelines. Skin temperature was measured prior to testing; if below 34°C, the candidate’s hands were warmed using hot packs; skin temperature was not recorded. Testing with the Brevio device used preconfigured electrodes applied to the skin. Electrodes for median nerve motor latency testing were placed around the thenar eminence, the dorsum of the hand, and the thumb interphalangeal joint. A nerve stimulator was held in place by the tester on the volar wrist, seven centimeters proximal to the electrode on the thenar eminence. Impulses were delivered manually by the tester until a maximum value for the nerve latency and amplitude was observed for four impulses. NCV results available for this analysis included average readings of bilateral median nerve distal motor latencies (DML) and amplitudes. The test procedure used a median nerve DML of 4.5 milliseconds (ms) or greater of one or both hands as the criterion for seeking additional diagnostic nerve conduction testing. The results of all testing were reviewed by an onsite medical provider in making a determination of medical fitness for duty.

Demographic data on workers was obtained from facility medical and personnel records including age, gender, body mass index (BMI), past medical history including diabetes or rheumatoid arthritis, date of hire and date of termination if no longer employed, and the job title into which the candidate was hired. Individual work hours were not available but the employer provided average monthly hours for all production workers for each month, for all years of the study.

We used the physical job demand scores provided by the employer to estimate physical exposures relevant to carpal tunnel syndrome. Occupational health providers scored the frequency of 35 physical job demands in each work process (scored from 1= rarely, 2=infrequently, 3= occasionally, 4= frequently, 5= constantly). We selected four physical job demands that were most closely related to hand-intensive work: “firm grasping,” “fine hand manipulation,” “wrist flexion/extension/deviation,” and “vibratory hand tools.” We assigned the value for each exposure that we believed best represented workers’ exposures on each production line. If the exposure for any single process occurred “constantly”, the exposure was considered “high” for the line. If exposures in all processes of a line occurred less frequently (not constantly), the exposure for the line was considered “low.” Each worker was assigned the exposure value of their production line for their job title. Exposure assignments were blinded to CTS case or POPP test status of workers.

We received a de-identified version of the medical records for all visits to the onsite medical provider. Because the available medical records did not contain standard diagnostic codes for CTS, potential cases of CTS were identified based on information in the medical provider notes, including symptoms typical of CTS in the hand, wrist, or fingers, or a CTS diagnosis by the medical provider. Based on text string searches of three fields in the medical records (diagnosis, body part, and comments fields), workers meeting one of the following criteria were classified as potential CTS cases:

The diagnosis or other text fields included one of the following terms: “cts,” “carpal tunnel syndrome,” or “ct syndrome.”

All potential cases thus identified were reviewed and coded independently by two occupational therapists (AMD and BG), using the scoring criteria described in Table 1 for definite CTS, possible CTS, and no CTS. Raters also coded the date of CTS diagnosis and the side(s) affected (right, left, or bilateral). Date was coded as the date when the CTS diagnosis was first assigned by the physician. If the date of diagnosis was not clearly stated in the record, date was assigned based on the CTS outcome scoring criteria as the earliest date of the positive NCV results, referral to hand specialist, or surgery. Cases assigned “possible CTS” were given the earliest date when CTS was suspected as the diagnosis. Date of CTS diagnosis was noted separately for the right and left hands, if the dates differed. If date of determination of the second side was unclear, the same date was assigned for both hands. The raters were blinded to workers’ POPP NCS results, personal, work, and medical information, with the exception of the information provided in the text record of the medical treatment. Discrepancies were resolved by consensus ratings for the CTS outcome, side affected, and the date(s) of diagnosis for each side.

Descriptive statistics were calculated for the demographic characteristics of the cohort, the percentage of workers with abnormal POPP NCV results, and the percentage of workers by physical exposure level (high/low hand-intensive work). For this study, the main outcome measure was a diagnosis of CTS as previously defined. We calculated incidence rates of CTS by person-time during the 5-year study period.

To evaluate the potential predictive value of nerve conduction testing results in hiring decisions, we calculated incidence rates of CTS among production workers using Poisson regression models. The incidence rate accounts for the workers’ time at risk that began with the date of hire and ended with either the first date of CTS diagnosis, the earliest termination date if the worker was no longer working for the company, or at the end of the study (end of year 5) if the worker was still working. Workers who were terminated and then rehired at a later date during the study period (n=119) were censored at the first termination date; none of these workers developed CTS during the rehire period. We ran multivariable Poisson regression models predicting the likelihood of CTS and included abnormal POPP NCV results (median DML greater than or equal to 4.5 ms), BMI, age at time of hire, gender, and “high” exposure job in the model. Medical conditions were not included due to their low prevalence. We also examined the effects of work hours to adjust for variation in the number of overtime hours by worker.

Importantly, we explored CTS cases occurring among those workers hired during a period when the NCV test results were not used in hiring decisions, and compared future CTS rates among workers with normal or abnormal POPP NCV at time of hire. We ran sensitivity and specificity analyses by varying the threshold for defining an “abnormal” POPP NCV result, to determine whether the predictive value of the POPP NCV screening would differ with different test thresholds. We calculated the number needed to screen to detect a single case of CTS and the number of workers that would have been denied employment among the number screened. All analyses used SAS version 9.4.²²

There are several limitations to this study that may affect the results and conclusions. The POPP NCV study tested the distal motor latency (DML) of each hand rather than using the more sensitive test of the distal sensory latency or median-ulnar distal sensory latency difference that requires testing the ulnar nerve of across the wrist. This study represented an evaluation of nerve conduction velocity testing as applied in current practice at the studied employer, and by other employers our study team is familiar with. Similarly, body temperature of the tested region was not recorded, and may have affected test results. The majority of the CTS cases were identified by the physician’s diagnosis and did not require standardized nerve conduction testing. This use of the physician’s notes and available text from the medical records to define cases of CTS was a limitation of our study, but also reflects real world clinical practice. An appreciable fraction of workers were missing production line assignment in the available records, and thus could not be assigned job exposures, although there is no reason to believe the missing data were non-random. Those with job title information may have received misclassified exposure levels given our method of exposure assignment, which relied on 5 point exposure scales estimated by the employer, and assigned the same exposures to all workers with the same job title and same production line.

This study had several strengths. We had data on all workers in a newly-opened manufacturing facility and were able to follow the workforce for five years. We had results of the screening (nerve conduction velocity studies and physical examination) and job exposures as well as the medical notes on all workers with hand or wrist complaints that were seen in the on-site medical department. Few studies have both baseline and follow-up data available on a large sample of newly employed workers with five years of follow-up information. Because the employer did not use the NCV results in making hiring decisions, all tested workers could be followed into the workplace, thus creating the opportunity to conduct a “natural experiment” to determine the rates of future CTS cases in relation to baseline testing.

A similar serendipitous event led to a classic paper demonstrating that POPP screening with low back radiographs did not predict low back injuries.³⁰ A high demand for workers in the forestry products industry caused the employer to hire workers despite “high risk” radiographs that would normally have excluded them from being hired; follow-up showed that these workers subsequently had a lower incidence of back injury than those with “normal” findings. Subsequent studies have confirmed the lack of utility of routine low back radiographs at predicting future work-related back pain. While low-back radiography is thankfully no longer routine in occupational health practice, other types of POPP screening to prevent musculoskeletal disorders are ubiquitous, despite the absence of high quality studies evaluating such assessments as an effective preventive measure.

Consistent with results of previous studies,^{10, 13} our study found that POPP screening in the current sample was an ineffective preventive strategy for CTS. There are no published studies supporting the use of POPP screening using NCV to prevent CTS; this is the third study to demonstrate that such screening is a poor use of employers’ health and safety resources. Employers and occupational health providers should apply similar scrutiny to the effectiveness of other common POPP screening practices.³¹ As a profession, there is a need for occupational health professionals to re-examine current practices in light of existing evidence, and to seek additional evidence on the utility – or lack of utility – of commonly performed screening that may determine future employment.^{31, 32}