Carpal tunnel syndrome (CTS) is a common work related upper extremity musculoskeletal disorder and has the longest time away from work and the highest associated direct costs among upper extremity work-related injuries and musculoskeletal disorders^1–4. Direct medical costs are estimated to exceed $1 billion per year^{1, 3}. CTS can also cause significant impairment in functional ability for workers in both work and daily activities^{2, 5, 6}. Some employers routinely use post-offer pre-placement (POPP) screening including nerve conduction studies (NCS) to identify workers at higher risk of developing CTS, so that these workers will not be hired into hand-intensive jobs at higher risk of CTS, thus reducing the employer’s injury rates and workers’ compensation costs⁷. It is difficult to estimate the number of employers currently utilizing POPP NCS to make hiring decisions; to our knowledge there are no published scientific reports describing the prevalence of POPP testing by employers. Despite a lack of clear scientific evidence that pre-placement, post-offer screening with nerve conduction studies is sufficiently predictive of future carpal tunnel syndrome, this practice appears to be widespread based on publications in trade journals, advertisements by healthcare facilities offering “carpal tunnel testing or screening,”^8–17 and promotion by device manufacturers.

Studies of active industrial workers have found a high prevalence of asymptomatic nerve conduction abnormalities, up to 15–20%^18–26. Available studies indicate that asymptomatic workers with abnormal nerve conduction studies are at a higher risk of developing CTS than asymptomatic workers whose nerve conduction studies are normal^{18, 27, 28}. However, the magnitude of the increased risk conferred by nerve conduction abnormalities, the cost benefit of doing such screening, and the effectiveness of different work placement strategies in preventing carpal tunnel syndrome all remain to be defined.

Another potential limitation of most existing studies is the possibility of a survivor bias resulting from selection of subjects who have worked for years in a hand-intensive industry; these studies may have evaluated those workers who remained asymptomatic despite work demands and abnormalities of nerve conduction²⁹. The predictive value of nerve conduction studies (NCS) may thus be different among job applicants than among active workers in hand intensive industries. Only one study to date has screened new employees at the time of hire and followed their development of CTS longitudinally. This study by Franzblau et al (2004)²⁸ studied workers in a single manufacturing plant who received NCS prior to hire, but were hired regardless of the results. Results from this study showed that abnormal NCS conferred a higher risk for a future workers’ compensation claim for CTS; however, the majority of the claims came from workers whose screening NCS were normal at the time of hire, and the cost of worker testing exceeded the potential savings that would have resulted from not hiring workers with abnormal NCS.

For the purposes of screening, it is unclear whether it is preferable to measure median nerve latency via sensory nerve latency, motor nerve latency, or in comparison to ulnar nerve latencies. Different testing techniques and different placement of electrodes can alter the results obtained^30–32. Another important issue in screening studies of asymptomatic persons is the need to define what constitutes an “abnormal” test result for working populations. The appropriateness of current normative values is questioned by studies showing higher prevalence of “abnormal” values among asymptomatic populations of active workers than the general population^{33, 34}. It is not clear if the populations from which normative values were drawn are truly representative of the worker populations in which the tests are being used.

The aim of our study was to determine if nerve conduction studies as part of POPP screening for new hires correctly identifies people at risk for CTS across a wide range of industries. This study examined the hypothesis that workers with baseline abnormalities of median nerve conduction would have a higher incidence of CTS than those with normal nerve conduction. In addition, we tested how the prediction yield varied across different case definitions for determining normal and abnormal nerve conduction studies. Finally, we estimated the cost-benefit of screening from the perspective of the employer.

The Predictors of CTS study (PrediCTS), is a prospective, longitudinal study that recruited 1107 subjects from eight employers and three trade unions between July 2004 and October 2006. Subject recruitment took place during company post-offer screenings, new employee orientations, or training classes depending upon each employer’s established hiring procedures. Eligible subjects were newly hired or had become benefits eligible within the prior 30 days, were at least 18 years of age, worked a minimum of 30 hours weekly, and were able to speak English. Subjects were excluded if they had a prior diagnosis of CTS or peripheral neuropathy, were pregnant at the time of enrollment, or had a contraindication to nerve conduction testing. Subjects were recruited from both low and high hand intensive jobs and represented a range of industries: construction (carpenters, floor layers, sheet metal workers), healthcare (laboratory and hospital technicians), service (food service, housekeeping) and clerical work (computer and clerical workers). Detailed information about recruitment and data collection methods used for this prospective study has been described in several previous manuscripts^35–38. The Washington University School of Medicine and the University of Michigan Institutional Review Boards approved this study. All subjects provided written informed consent and were compensated for their participation.

Of the original 1107 newly hired workers screened in the PrediCTS study cohort, 1,027 (92.8%) completed at least one follow-up survey. Five subjects were excluded because baseline NCS results were indeterminate for median DML, DSL, and MUDS. Of the 1,022 subjects remaining, 35 had partially missing NCS data and were excluded from analyses requiring the missing parameters as seen in Tables 2–4. The majority of workers were male (64.5%) with a mean age of 30.3 years and a mean BMI of 28.5 (Table 1). A quarter of workers had abnormal POPP NCS at baseline. Subjects with abnormal NCS at baseline were significantly older, more likely to be male, and to work in the construction trades. The mean length of follow-up for the cohort was 3.7 years (range 0.4–6.2). Over the study period 33 workers reported having received a diagnosis of CTS with a mean time to first report of diagnosis of 2.4 years (range 0.4–5.1). Workers with any NCS abnormality at baseline were significantly more likely to report a diagnosis of CTS at follow-up. The overall incidence rate of CTS diagnosis in the cohort was 8.7 per 1000 person years (PYs). The rate of CTS diagnosis was higher among workers with abnormal POPP NCS versus workers with normal POPP NCS (22.2 cases per 1000 PYs versus 4.0 cases per 1000 PYs, rate ratio of 5.5 (95% CI: 2.6–11.5)).

Baseline NCS abnormality was a statistically significant predictor of future CTS diagnosis for all nerve test parameters using our pre-defined cut-points of median DML > 4.5 ms, median DSL > 3.5 ms, or MUDS > 0.5 (Table 2). Workers with abnormal MUDS were at the highest risk of becoming diagnosed with CTS over the study period, followed by abnormal median DSL, any NCS abnormality, or abnormal median DML. Age, gender, and BMI were statistically significant predictors of CTS diagnosis. There were no substantial changes in the hazard ratios of NCS tests when we repeated these analyses adjusting for age, gender, and BMI. Using the AIC to describe the goodness of fit, MUDS had the lowest AIC value and thus was a better predictor of CTS than the screening parameters of any NCS abnormality, median DML, or median DSL.

To examine the effect of defining different NCS cut-points for abnormality on the yield of screening, including the positive predictive value (PPV) and number needed to test to avoid 1 future case of CTS, we examined the sensitivity and specificity for the cut-points shown in Table 3. As expected for a relatively rare condition, specificity and NPV were high (75.9 – 93.1%) and (97.5 – 98.8%), respectively; however, sensitivity was low (36.4–65.5%) and PPV very low (6.4–18.5%). The most sensitive cut-point, MUDS > 0.5, detected 65.5% of those who would later be diagnosed with CTS. While reasonably sensitive, this measure had a very poor PPV: only 13.1% of those with an abnormal value based on this cut-point reported a diagnosis of CTS over the study period. We repeated these analyses using the right hand only; results were similar to the findings of bilateral testing; as expected, sensitivity decreased slightly and specificity improved slightly with unilateral testing.

In order to assess the preventive effectiveness of POPP NCS for CTS, we calculated the number of job candidates who would need to be screened to avoid 1 future case of CTS among the screened workforce using the simple methods previously described by de Kort and van Dijk⁵¹. Using our pre-defined cut-points, the most sensitive screening parameter, MUDS > 0.5, only detected 65.5% of those who would later be diagnosed with CTS, and would have required testing 54 subjects to detect 1 future case of CTS in the workforce. At the same time, the lower specificity of the MUDS >0.5 parameter would have inappropriately denied employment to 7 workers among the 54 who would not have developed CTS despite having an “abnormal” NCS test.

In assessing the predictive validity of POPP NCS for predicting future CTS, the analyses reported above followed the full cohort of workers including those who changed employers, and recorded all cases of CTS during the study period (n=33). We conducted cost-benefit analysis from the perspective of the employer, and thus only the 23 CTS cases that occurred while a subject screened at baseline was still working for the original employer and had complete NCS data were relevant for inclusion in the cost models. In our base case scenario using the criterion of “any NCS abnormality” at baseline, 987 newly hired workers were screened with POPP NCS; 247 workers would have been rejected for hire, a failure rate of 25%. In order to attain a work pool of 987 workers who tested normal, a total of 1,317 workers would need to be screened (987 + 247 replacements from failed screens + 83 additional replacements due to a continuous failure rate of 25% among replacement workers screened). Sixteen of the 23 future CTS cases would have been avoided under a screening strategy, as they occurred in the population of workers testing abnormal at baseline using this criterion. Seven of 23 CTS cases occurred in the 740 workers screening normal at baseline; we used a cumulative incidence rate of CTS in the population screening normal at baseline to calculate the expected number of cases in a population of 987 workers screening normal at baseline (10 CTS cases when using the testing criterion of any NCS abnormality). These case counts and number of workers screened were entered in to our cost model using base screening costs of $150 per worker and CTS treatment cost of $20,000 per case, as shown in Table 4. The NCS failure rate and number of CTS cases that would have been avoided varied with the sensitivity of the NCS parameter used as the screening criterion. Cost benefit analysis showed that screening was favored when NCS test sensitivity was high (any abnormal NCS, DSL, and MUDS) and cost of screening was low ($150). Using the same low screening cost, but a less sensitive NCS measure (median DML >4.5 ms), a no screening strategy was favored. With a higher screening cost ($358), modeled to account for the other costs associated with hiring new workers²⁸, a no screening strategy was favored in all models, regardless of NCS parameters. These findings were also sensitive to changes in treatment cost for CTS. When the treatment cost for a CTS case was varied ($13,253 and $5605), using published workers’ compensation data from Ohio for 1999–2004⁵⁰, and Washington State ¹, a no screening strategy was favored for all NCS screening parameters except when using a selection criteria of MUDS >0.5ms with the higher Ohio State costs.

The aim of this study was to determine if NCS performed at the time of hire were predictive of a future CTS diagnosis among a cohort of workers in a variety of industries. Results showed that newly hired workers with abnormal baseline NCS were significantly more likely to report a CTS diagnosis during the study period, which is consistent with previous studies^{18, 27, 28}. Despite this finding, the predictive validity of such POPP NCS screening is at best low or modest. We tested how the prediction yield varied across different thresholds for defining normal and abnormal nerve conduction studies, and consistently found low positive predictive value across all screening parameters. POPP NCS screening appears to be widely used by employers, but our cost-benefit models of screening conducted from the perspective of the employer showed that the costs of screening did not outweigh the savings for CTS cases that would have been avoided in the majority of scenarios modeled.

The overall rate of CTS in our cohort was 8.7 cases per 1000 person years (PYs), slightly higher than the rate of 7.8 per 1000 PYs reported by Franzblau et al²⁸. The slightly higher rate of CTS observed in our study population may be partly attributable to our case definition of CTS diagnoses reported by workers rather than accepted workers’ compensation claims for CTS²⁸. The rates reported in our study and by Franzblau likely both underestimate the true occurrence of disease due to untreated or unreported cases^{33, 52}. Not all workers who have symptoms are likely to seek treatment, and of those workers who seek treatment, not all will file a workers’ compensation claim or have an accepted claim. The slightly higher incidence rate in our study may also be attributable to differences in the nerve conduction parameters used to identify abnormalities for workers. Our rate is based on the screening definition of “any NCS abnormality” (median DSL, DML, or MUDS) at baseline, whereas actual test results (latencies, amplitudes, or conduction velocities) were not available in the Franzblau study, and thus prediction models were based solely on test summaries defined as normal or abnormal. In addition, our study had a longer mean follow-up time of 3.7 years versus 2.1 years²⁸.

A potential limitation of existing studies is the possibility of a survivor bias resulting from selection of subjects who were working for years in a hand-intensive industry^{18, 27}; these studies may have evaluated those workers who remained asymptomatic despite heavy work demands and abnormalities of nerve conduction²⁹. Jobs in hand-intensive industries often have high turnover rates creating a natural selection bias against workers at greater risk of developing CTS. Our study and that of Franzblau et al (2004) avoid this potential bias that the predictive value of nerve conduction studies (NCS) may be different among job applicants than among active workers in hand intensive industries, and thus more closely models the use of NCS as a screening tool during POPP testing. Results from our study included workers from a wide variety of industries and employers, while Franzblau’s study (2004) included workers from a single employer.

To determine whether the prediction yield of screening for future CTS cases could be improved, we evaluated 6 definitions of an “abnormal” NCS result from baseline DML, DSL, and MUDS using published thresholds previously used in working populations^{46, 47}. Our results showed that the percentage of future CTS cases that were correctly predicted was highly dependent upon the definition used to define an abnormal screening result. The sensitivity and specificity and positive and negative predictive values of POPP NCS have varied, in part because different criteria were used to define normal and abnormal NCS measures^{18, 28}. The increased risk of CTS diagnosis in our study varied widely depending upon the measure chosen to define an abnormal result from a hazard ratio of 2.95 (95% CI: 1.45–6.01, p<0.01) for median DML (>4.5ms) to 11.25 (5.22–24.21, p<0.0001) for abnormal MUDS (>0.5 ms). As with any diagnostic test, there are explicit trade-offs between sensitivity and specificity as seen with use of different criteria for defining abnormal nerve conduction studies. As parameter cut-points used to predict the outcome of future CTS diagnosis were made more specific, the employers in the study would have assumed more risk of potentially hiring workers with greater risk of developing CTS, whereas when the cut-points were more sensitive, more healthy workers would have been inappropriately excluded from employment. Our results showed very different results between different nerve tests and different criteria for defining abnormality. CTS POPP screening is often based on testing only the median nerve; in our study we found the highest sensitivity and specificity using the median ulnar difference. In our experience, POPP testing programs in industry have not formally considered the effects of different cut-points or nerve conduction testing techniques in assessing the likely yield and cost-benefit of their screening programs, nor have such programs acknowledged the very low positive predictive value of POPP testing.

POPP screening with NCS appears to be widely used by employers, but only 2 published studies have evaluated its cost-benefit to employers. Evaluation of POPP screening using cost data from a manufacturing plant²⁸ found that the cost of screening outweighed potential savings from averting workers’ compensation costs for CTS. A cost-benefit modeling study⁴⁸ found that POPP screening for CTS would not be cost beneficial for the majority of employers, though could be cost beneficial under circumstances where there was a high cost per case and high incidence of CTS. Using these published models, we compared the cost of screening the workers in our cohort and not hiring those with abnormal baseline NCS, versus a policy of no screening. Our results were consistent with these previous studies, showing that POPP NCS screening would not have been cost beneficial to the employer in the majority of screening scenarios. When the cost of screening was high, a no screening strategy was favored in all cases. As Franzblau (2004) described, there are a number of other potential costs associated with screening in addition to a nerve conduction test, which will vary depending upon each employer’s hiring practices. These costs may include a medical physical examination, drug test (which may or may not be done concurrently with other POPP screening), and administrative costs associated with new hire paperwork and orientation. In addition, we assigned the same cost of screening regardless of NCS parameter, however, screening parameters which require testing of both the median and ulnar nerves (MUDS) would likely cost more than median nerve testing only (DML, DSL). Our simplified cost model did not account for screening costs in subsequent years of hiring to replace workers lost to turnover, and thus may have not have fully captured costs of pursuing a POPP screening strategy.

Cost of treatment for CTS is also highly variable. A 2005 study of work-related CTS claims in Washington State¹ found a median cost per claim of $5605 over a five-year period, though costs for individual cases ranged from $359 to $79,265, depending upon whether surgery was performed and when the CTS diagnosis occurred in the course of the claim. Our base cost of $20,000 per diagnosed case is very likely to be an overestimate of the costs of CTS to employers, as not all diagnosed cases would be claimed under workers’ compensation, and not all would be surgical cases. The lower costs seen in Ohio and Washington State are more likely to reflect true costs of CTS claims under workers’ compensation.

The main limitation of our study was the use of self-reported CTS diagnosis as the outcome. As our study was designed to replicate employer practices, CTS diagnosis was a more appropriate outcome than an epidemiological case definition, which would have included workers who had symptoms and abnormal NCS findings, but who never sought treatment or filed a workers’ compensation claim. A similar case definition of self-reported CTS has been used in several previous studies of national health surveillance data^43–45. Survival models were censored at survey date as the actual date of diagnosis was not available. The use of an absolute latency for defining an abnormal NCS consistent with CTS has been criticized because many factors such as age, temperature, co-morbidities, and BMI can influence the absolute latency of the median nerve⁴⁶. Comparison on the median and ulnar latencies controls for these confounding factors and our analysis demonstrated this model was the strongest predictor of future CTS. We included all three definitions of abnormal NCS because they are still widely used in clinical practice.

Our model showed that screening could be favorable only if the average treatment cost was high and the screening cost was low. As discussed in other papers^{48, 51}, for screening to be cost beneficial the prevalence of the disorder must be high among new hires, and the incidence high following hire. High employee turnover also increases the cost of screening and decreases potential employer cost savings as screened workers may leave employment before developing the disease of interest⁴⁸. Our results highlight the sensitivity of the cost-benefit model to the cost of screening, the cost per CTS case, and the sensitivity of the screening test. This relationship highlights the need for employers who utilize POPP screening to pick appropriate test cut-points for their workforce rather than a clinical or general population. Finally, the cost-benefit of screening should be considered within the larger societal context of not hiring otherwise qualified workers. Due to the low PPV of POPP NCS screening for future CTS, data from this study showed that each workplace case of CTS avoided for the employer would come at the cost of inappropriate denial of employment or job placement for 4–15 workers who would not develop CTS during the course of their employment. This raises another consideration for employers, the several court cases claiming that employers violated the Americans with Disabilities Act or other laws when applicants were excluded from production jobs based on POPP testing for future risk of CTS. Although some past cases were settled in favor of the employer (EEOC vs. Woodbridge Corporation, 8^th Circuit No. 01-L045, August 24^th, 2001; EEOC vs. Rockwell International Corporation, 7^th Circuit Nos. 00–1897 & 00–2034, March 8, 2001), the EEOC recently won a discrimination case against an employer who excluded a worker from employment based on perceived risk of future CTS.⁵³ While this case was based on violation of the Genetic Information Nondiscrimination Act, at least one other lawsuit in progress addresses discriminatory aspects of POPP screening by median nerve conduction studies.

Although abnormal NCS at the time of hire is strongly associated with increased risk of future CTS, the predictive value of such testing is poor, even when using optimal criteria as screening thresholds. The cost of screening and rejecting large numbers of healthy workers from employment is high, and in most cases seems to outweigh potential employer cost savings from reducing the incidence of CTS in a given workforce. The social costs of rejecting otherwise qualified healthy workers from employment should also be considered. In all scenarios described above, the vast majority of workers with “abnormal” screening results did not develop CTS, and thus would be inappropriately placed or denied employment based on a test with demonstrably poor positive predictive value for future disease. POPP screening for CTS appears to be widely used, despite ongoing uncertainty about ideal screening procedures, appropriate cut-points for screening versus diagnosis of clinical median neuropathy, and cost-benefits of the procedure for employers. Employers should be cautious in implementing any broad worker screening programs without careful consideration of the costs and benefits of such programs. Available evidence shows that POPP screening for CTS is poorly predictive of future disease, and published studies to date have failed to show that this practice is effective. Occupational health professionals should not endorse screening programs that are not based on evidence of benefit to workers’ health.