Long-term trends in carpal tunnel syndrome

Abstract

Objective:

To assess temporal trends in carpal tunnel syndrome (CTS) incidence, surgical treatment, and work-related lost time.

Methods:

Incident CTS and first-time carpal tunnel release among Olmsted County, Minnesota, residents were identified using the medical records linkage system of the Rochester Epidemiology Project; 80% of a sample were confirmed by medical record review. Work-related CTS was identified from the Minnesota Department of Labor and Industry.

Results:

Altogether, 10,069 Olmsted County residents were initially diagnosed with CTS in 1981–2005. Overall incidence (adjusted to the 2000 US population) was 491 and 258 per 100,000 person-years for women vs men (p < 0.0001) and 376 per 100,000 for both sexes combined. Adjusted annual rates increased from 258 per 100,000 in 1981–1985 to 424 in 2000–2005 (p < 0.0001). The average annual incidence of carpal tunnel release surgery was 109 per 100,000, while that for work-related CTS was 11 per 100,000. An increase in young, working-age individuals seeking medical attention for symptoms of less severe CTS in the early to mid-1980s was followed in the 1990s by an increasing incidence in elderly people.

Conclusions:

The incidence of medically diagnosed carpal tunnel syndrome (CTS) accelerated in the 1980s. The cause of the increase is unclear, but it corresponds to an epidemic of CTS cases resulting in lost work days that began in the mid-1980s and lasted through the mid-1990s. The elderly present with more severe disease and are more likely to have carpal tunnel surgery, which may have significant health policy implications given the aging population.

GLOSSARY

CI

= confidence interval;

CTS

= carpal tunnel syndrome;

MESA

= Marshfield Epidemiologic Study Area.

Carpal tunnel syndrome (CTS) is a common condition with potentially high social and economic costs, especially if it requires surgical treatment or interferes with one’s ability to work.^1–4 Several studies have investigated the incidence of CTS in the general population, and rates appear to vary by study location and methodology employed.^5–10 Despite this, studies generally show an increasing temporal trend, the explanation for which is unclear. Since our earlier study of CTS incidence in Rochester, Minnesota, from 1961 to 1980,⁵ the prevalence of personal CTS risk factors, such as obesity and diabetes, has increased in the general population.^11–13 In addition, possible extrinsic CTS risk factors have increased as a result of greater labor productivity, general awareness, and more use of computers in industry, offices, and homes.^14–17 Because of these changes, as well as the influence of different CTS definitions among studies, we updated the previous Rochester study through 2005 using the resources of the Rochester Epidemiology Project and expanded it to include all Olmsted County, Minnesota, residents. The objectives were to identify temporal trends in the incidence of medically diagnosed CTS, the utilization of surgical treatment for CTS, and work-related lost time due to CTS.

METHODS

The Rochester Epidemiology Project medical record linkage system allows near complete ascertainment of diagnosed illnesses in the Olmsted County population.¹⁸ Following approval by the Mayo Clinic and Olmsted Medical Center institutional review boards, we identified new cases of carpal tunnel syndrome in all Olmsted County residents who had a diagnosis coded to rubric 357.2 (H-ICDA-8) or 354.0 (ICD-9) during the study period. To qualify as an incident case, the patient had to reside in Olmsted County for 1 year prior to the initial diagnosis (to exclude immigrants moving to the County for CTS care). In our earlier study, additional diagnostic categories (e.g., various peripheral neuropathies, nerve injuries, and selected other conditions) were screened to assure complete ascertainment of cases, but this laborious process yielded only 34 otherwise unidentified cases out of 1,016 total cases of CTS.⁵ Similarly, other investigators using electronic case identification with medical record confirmation found few cases of CTS assigned to ICD-9 diagnosis codes other than 354.0.⁶ Based on this shared experience, additional diagnostic categories were not reviewed as part of this study. Records also were not counted or reviewed of 712 potential cases (7%) who had not provided authorization to use their medical records for research.¹⁹

Because of the lack of a diagnostic gold standard, variations in clinical presentation, and differences in the diagnostic criteria used by different medical practitioners, administrative data may overestimate the incidence of CTS identified by ICD-9 code 354.0.^20,21 Due to the large number of cases, one of the authors (R.G.) reviewed data collected by trained nurse abstractors from a random sample of the charts for criteria recommended for epidemiologic studies of CTS.²² Of 194 charts reviewed, 131 (68%) met symptom criteria for a diagnosis of classic/probable CTS, defined as numbness, tingling, burning, or pain in at least two of digits 1, 2, or 3. Another 25 met criteria for possible CTS, defined as numbness, tingling, burning, or pain in at least one of digits 1, 2, or 3; 38 did not specify the location of symptoms. Thus, 156 cases (80%) met symptom quality and location criteria for CTS.²² The confirmation rate was 94%, 76%, 81%, 72%, and 86%, for each respective quinquennium. Altogether, 113 (58%) had an electrodiagnostic study (EMG); of these, 88 (45% of the whole sample and 78% of those tested) had a study result consistent with median neuropathy at the wrist (figure 1). Administrative rates were not adjusted based on these results.

Figure 1 Symptom characteristics and EMG results of 156 patients with classic/probable or possible carpal tunnel syndrome (CTS) from review of a random sample of 194 charts

Thirty-eight of the charts did not reference finger location of symptoms and were excluded. Criteria for a positive nerve conduction study (NCS) include median palmar distal latency more than 2.3 msec, median-to-ulnar palmar latency difference exceeding 0.3 msec when the palmar latency was 2.2 msec or less, or a median antidromic sensory latency more than 3.6 msec with a normal ulnar antidromic sensory latency.

Annual incidence (rate per 100,000 person-years) was estimated by dividing the number of new CTS cases observed by the entire population of Olmsted County (linear interpolation of decennial census figures from 1981 to 2000 with extrapolation through 2005), assuming everyone to be at risk.²³ Rates were directly standardized to the age and sex distribution of the US population in 2000. The 95% confidence intervals (CIs) around these rates were estimated assuming a Poisson error distribution. A Poisson regression analysis of the crude incidence rates was used to assess their association with gender, age (considered as a class predictor variable, categorized as 0–59, 60–69, 70–79, and 80 or more years), and time period of diagnosis (using the midpoints of the 5-year periods: 1981–1985, 1986–1990, 1991–1995, 1996–2000, and 2001–2005).

In a similar fashion, incidence rates for first-time carpal tunnel release operations and work-related CTS were estimated over the same period. The incidence (∼utilization) of carpal tunnel release was determined by counting the number of operations electronically identified from the surgical index of the Rochester Epidemiology Project database, which contains a record of all such procedures performed in Olmsted County. The rates of work-related CTS were obtained by counting all incident cases of work-related CTS occurring among Olmsted County residents that resulted in more than 3 days of lost work time between 1981 and 2005; this was provided by the Minnesota Department of Labor and Industry from data routinely collected from employers or insurers.

RESULTS

A total of 10,069 Olmsted County residents were diagnosed with CTS for the first time in 1981–2005, of whom 6,897 were women and 3,172 were men. The overall age- and sex-adjusted incidence of CTS was 376 per 100,000 person-years (95% CI, 369–384), but was much greater (p < 0.0001) in women (491 per 100,000 person-years; 95% CI, 479–502) than men (258 per 100,000 person-years; 95% CI, 249–268). The women-to-men ratio of CTS cases was 2.2:1, and of comparably adjusted incidence rates, was 1.9:1.

However, rates increased almost twofold over the study period, from an overall incidence of 258 per 100,000 person-years in 1981–1985 to 424 per 100,000 in 2001–2005 (p < 0.0001). The sharpest rise was seen in 1986–1990, but rates continued to increase in each subsequent period (figure 2A). Age-adjusted annual incidence in women increased 161%, from 337 to 542 per 100,000 during the time of the study; rates for men rose by a similar 172%, from 177 per 100,000 person-years in the first quinquennium to 303 per 100,000 in the last. These changes were accompanied by complex shifts in age distribution (table 1). Generally, the most marked increases in CTS incidence were seen in younger age groups of both sexes in the first part of the study period and among older age groups in the final decades of study.

Figure 2 Age-adjusted (women, men) and age- and sex-adjusted (both sexes combined) incidence per 100,000 person-years for carpal tunnel syndrome diagnosis (A), carpal tunnel release surgery (B), and work-related carpal tunnel syndrome with lost work days (C), among Olmsted County, Minnesota, residents, 1981–2005

For details on the process of submitting a First Report of Injury form for work-related injuries, see www.doli.state.mn.us/fr01info.html.

Table 1 Incidence (per 100,000 person-years) of first episode of carpal tunnel syndrome diagnosis among residents of Olmsted County, Minnesota, 1981 through 2005, by gender and age group

In contrast to the trends for CTS diagnosis, first-time carpal tunnel release surgery decreased slightly throughout the first four quinquennia (figure 2B). In 2001–2005, however, the utilization of this procedure increased, due mostly to increased rates of surgery in both men and women over 50 years of age. Detailed data are provided in table 2.

Table 2 Incidence (per 100,000 person-years) of first episode of carpal tunnel release among residents of Olmsted County, Minnesota, 1981 through 2005, by gender and age group

Figure 2C shows the rates of work-related CTS, which increased dramatically between 1981–1985 and 1986–1990 only to decline in the 1991–1995 quinquennium in men and in the 1996–2000 quinquennium in women; rates remained relatively stable thereafter. This temporal pattern differed considerably from those observed for CTS diagnosis or surgery.

DISCUSSION

This is the longest study to date of trends in the incidence of medically diagnosed CTS in the general population. We previously had documented a 1.4-fold increase in annual CTS incidence in this population, from 88 per 100,000 in 1961–1965 to 125 per 100,000 in 1976–1980.⁵ Rather than a true increase in CTS, the change was attributed to better recognition of CTS (detection bias) due to a gradual increase in awareness of the syndrome by medical professionals, the opening of a hand clinic in the community in 1967, and the introduction of more sensitive electrodiagnostic techniques, especially palmar median sensory stimulation after 1977.⁵ However, despite the fact that CTS has been very well known in this medical community since 1960,²⁴ the present investigation reveals that the trend to increasing rates accelerated in more recent decades, reaching an overall age- and sex-adjusted annual incidence of 424 per 100,000 in 2001–2005. This is consistent with other studies of CTS trends (table 3). Thus, the number of cases reportedly increased by approximately 90% in women and 145% in men between 1992 and 2001 in East Kent.⁸ In Sienna, CTS rates increased by 8% in women and 74% in men between 1991 and 1998,⁷ while in the United Kingdom generally, CTS increased by 13% for women and 35% for men between 1992 and 2000.⁹ Finally, in the Netherlands, the crude incidence of CTS in 1987 was 190 per 100,000 in women and 60 per 100,000 in men; in 2001, the rates were 280 and 90 per 100,000 in women and men.¹⁰ These trends may reflect a greater awareness of CTS among the general population, especially in the decades of the 1980s and 1990s, when articles on CTS began to appear in the popular press. For example, a search for articles containing the words “carpal tunnel syndrome” in The New York Times using ProQuest yields a total of 318 articles from 1981 to 2005; the number in each quinquennium during that time was 11, 43, 89, 103, and 72.

Table 3 Comparison of population studies on the incidence of carpal tunnel syndrome diagnosis

Although rates were higher in women in the earlier Rochester study, the increase was somewhat greater for men so that the women:men ratio of age-adjusted incidence rates fell from 4.0:1 to 2.5:1 between 1961 and 1980. Likewise in the present study, the rise in CTS incidence was again somewhat greater in men so that the sex ratio fell to 1.8:1 in the last quinquennium. These ratios are similar to those reported in Canterbury (2.0:1)⁸ and the United Kingdom (2.5:1)⁹ but differ from those reported in Marshfield (1.1:1)⁶ and Sienna (4.1:1).⁷ The discrepancies are difficult to explain. One possibility is that CTS may be underestimated in men in some populations,²⁵ perhaps because of the theory that men are less likely to seek care for symptoms.²⁶

Since incidence rates have changed over time and no other study has looked at medically diagnosed CTS over such a long interval, it seems most appropriate to compare rates in individual quinquennia. Our incidence of 397 per 100,000 person-years from 1991 to 1995 is comparable to the rate found in the Marshfield Epidemiologic Study Area (MESA), but higher than in Huddersfield (table 3). The MESA rate from July 1991 to June 1993 was 377 per 100,000 person-years,⁶ and in the EMG clinic in Huddersfield only 47 per 100,000 from 1991 to 1993.⁸ Our incidence rate from 1996 to 2000 was 412 per 100,000 person-years. These rates are higher than in Sienna, Canterbury, the United Kingdom in general, and the Netherlands (table 3). The rates in Sienna were 269 per 100,000 person-years from 1991 to 1998⁷; in Canterbury, 90 per 100,000 person-years from 1992 to 2001⁸; in the United Kingdom generally, 136 per 100,000 person-years from 1992 to 2000⁹; and in the Netherlands, 219 per 100,000 person-years in 2001.¹⁰ The use of administrative data containing all cases diagnosed in the community and the use of clinical criteria for confirmation of CTS in our study and the MESA study might explain the higher overall rates in comparison to some of the other studies, which were done in referral centers and required a confirmatory EMG.^7,8 Although we would agree that a positive nerve conduction study increases the likelihood of a correct diagnosis, not all individuals with classic symptoms have abnormal nerve conduction studies.²⁷ However, even after accounting for an 80% case confirmation rate in our study, the lower rates found in general practice in the United Kingdom and the Netherlands require some explanation.^9,10 There may be a true difference in incidence between the countries, or the differences could reflect different attitudes on the part of patients regarding conditions that require a doctor visit, or in the accessibility of medical care. It is interesting that the highest incidence is in the United States, which is the only country among those mentioned that does not have universal health care coverage. One might expect that our rates would be lower if the differences were due to insurance status. Thus, our findings tend to support the hypothesis that national attitudes toward disease in the general population may explain at least some of the differences in medically attended incidence of CTS between countries.

What other factors might explain the increasing incidence over time? Our results would suggest that newer or more frequent diagnostic tests are not responsible since the methods for CTS diagnosis using nerve conduction studies remained relatively similar during the period of study, and our finding that 58% of the cases selected for chart review had been referred for an EMG is similar to the 61% figure in the 1976–1980 quinquennium of the previous Rochester study.⁵ An increasing presentation of milder cases deserves consideration. Our chart review estimated the true incidence (using symptom criteria) at around 80% of the rate obtained from the administrative diagnostic data; rates of clinical confirmation in each quinquennium were 94%, 76%, 81%, 72%, and 86%, with no clear trend to indicate a systematic change in criteria for diagnosis over time. Conversely, the increase in CTS incidence in Canterbury was accompanied by a decline in the electrophysiologic severity of new cases, indicating that at least some of their increase might be due to diagnosing milder cases of median neuropathy.⁸

However, increasing CTS incidence in the earlier years of this study was not accompanied by increasing rates of carpal tunnel release, possibly suggesting that the cases were not severe enough to warrant surgery. Carpal tunnel surgery utilization in Olmsted County in 1986–1990 (104 per 100,000 person-years; 95% CI, 95–114) was comparable to the rate reported in Ontario in 1988 (109 per 100,000)²⁸ and slightly less than the 155 per 100,000 figure reported in Wisconsin in the early 1990s.²⁹ In 1993, crude rates in Maine varied by location from 82 to 287 per 100,000, with a state average of 144 per 100,000.³⁰ Increasing surgical rates were seen in the later years of this study and reflect the increasing number of elderly individuals presenting with CTS, who appear to have more advanced disease. The Canterbury investigators also noted that elderly individuals tended to present with more severe disease.⁸ These data again lend support to the hypothesis that public attitudes about CTS, and the symptoms associated with it, may have changed, resulting in more patients presenting for care from a given reservoir of disease.

While the increase in diagnosis may simply result from increased ascertainment of a common disorder,³¹ it is also possible that the increase could be due to an increased prevalence of known CTS risk factors, or the emergence of new ones.³² Compared to the incidence of CTS in Rochester in 1976–1980,⁵ the higher rate seen in MESA over a decade later was felt to represent an actual increase in CTS, as it was presumed that the area had a higher proportion of workers in supposedly “at risk” occupations,⁶ as reviewed elsewhere.³³ In addition, an apparent epidemic of work-related CTS resulting in lost work days began in the mid-1980s in Olmsted County and continued through the mid-1990s. Beginning in the late 1970s, there were dramatic increases in productivity in certain industries, and workers began noting an increase in problems involving pain and numbness in the hands.³⁴ Strikes at meatpacking facilities led to articles in several prominent newspapers across the country that brought occupational CTS to national and international attention.^14,35,36 Similar trends in productivity in offices increased awareness of the disorder among computer users.³⁷ Interestingly, the trend in CTS resulting in lost work days is very different from trends seen with medically attended CTS and surgical cases in the general population. We are unaware of any medical research explaining this, although the phenomenon has been attributed to occupational pseudo-illness.³⁸ The stigmatization of the condition in the press may play a role.³⁹ Possibilities for reduced CTS mentioned in the lay press include ergonomic changes, early recognition of symptoms with work modification before they become reportable problems, lack of faithful reporting of musculoskeletal disorders by employers, and the changing priorities of unions.⁴⁰ These possibilities deserve further study because of the health implications for workers.

A strength of this study was the use of a unique data system with a very long-term perspective. However, there were a number of corresponding limitations, including use of electronic case ascertainment and the exclusion of about 7% of potential cases because they had not authorized the use of their medical records for research.¹⁹ In addition, the study was conducted in a small community in the Midwest with a limited minority population, although the sociodemographic characteristics of the population resemble those of US whites generally except for overrepresentation of medical workers and a somewhat higher educational level.¹⁸ Notably, incidence rates for work-related CTS are underestimated: Due to ambiguities in determining the number of workers at risk, and the fact that all Olmsted County residents are at risk of developing CTS whether or not due to work, we used the whole population instead of the working population in the denominator. This allowed us to compare temporal trends directly with the CTS incidence and surgery rates but precluded reliable estimation of the actual incidence rate of CTS in the working population. Most significantly, we did not attempt to ascertain the prevalence of CTS in the population who did not seek the attention of a physician. This group could have sought care from providers not in our database, such as chiropractors, or not sought medical attention at all. Thus, ultimately, we are unable to know whether the changes in incidence that we observed are due to a true change in the underlying condition, or simply a change in how patients and physicians interact concerning it.

A new finding from this study was that, during the 2000–2005 quinquennium, the incidence of CTS in younger people appeared to decrease, while the rate in older individuals appeared to increase. This older group also contributed disproportionately to the increasing incidence of carpal tunnel release operations. In other words, the older individuals seeking medical care for CTS appear to have more severe disease and are more likely to be candidates for carpal tunnel release operations as compared to younger individuals. This finding may have significant health policy implications given the growing elderly population.

ACKNOWLEDGMENT

The authors thank Brian Zaidman from the Minnesota Department of Labor & Industry for collecting and summarizing the work-related CTS statistics; Margary Kurland, Linda Paradise, Christine Pilon-Kacir, Maria Stracke, and Chris Parisi for abstracting chart data; and Mary Roberts for help in preparing the manuscript.

Address correspondence and reprint requests to Dr. Russell Gelfman, College of Medicine, Mayo Clinic, Rochester, MN 55905 gelfman.russell@mayo.edu

Supported by a grant (R01-AR30582) from the NIH, US Public Health Service. This publication was made possible by Grant 1 UL1 RR024150 from the National Center for Research Resources (NCRR), a component of the NIH, and the NIH Roadmap for Medical Research. Its contents are solely the responsibility of the authors and do not necessarily represent the official view of NCRR or NIH. Information on NCRR is available at http://www.ncrr.nih.gov/. Information on Reengineering the Clinical Research Enterprise can be obtained from http://nihroadmap.nih.gov.

Disclosure: The authors report no disclosures.

Received April 17, 1008. Accepted in final form September 23, 2008.