Prevalence of Doctor-Diagnosed Thumb Carpometacarpal Joint Osteoarthritis: An Analysis of Swedish Health Care




While the prevalence of radiographic thumb carpometacarpal (CMC1) osteoarthritis (OA) is well-described, little is known about clinically symptomatic disease presenting to physicians for care. We sought to determine the prevalence of doctor-diagnosed CMC1 OA.


Using health care data from Skåne in southern Sweden (population 1.24 million), we identified all adults ages ≥20 years who consulted a physician at least once and received a diagnosis for CMC1 OA (International Classification of Diseases, Tenth Revision, code M18). Data from the 15-year period 1998–2012 were analyzed. Using cross-referencing with the Swedish population register to exclude subjects who were deceased or had relocated, we obtained point estimates of the proportion of the population consulting for CMC1 OA.


The prevalence of doctor-diagnosed CMC1 OA in adults was estimated at 1.4% (2.2% in women and 0.62% in men). The mean ± SD age in the prevalent CMC1 cohort (n = 11,111) was 67.7 ± 11.4 years; 78.5% of diagnoses were in women. Prevalence peaked in women ages 70–74 years with an estimate of 5.3% and in men ages 80–84 years with an estimate of 1.7%. Age at initial diagnosis also differed, with women presenting between ages 60–69 years and men presenting between ages 70–79 years.


The clinically important prevalence of CMC1 OA is 3 to 4 times higher in women than men. By the end of 2012, more than 1 in 20 elderly women had consulted a physician for CMC1 OA over the last 15 years. The high prevalence of this subset of hand OA is a concern in an aging population.


The thumb carpometacarpal (CMC1) joint is a common site of hand osteoarthritis (OA), thought to occur secondary to high forces imparted to this joint with grip and pinch in daily activity ([1]). Symptomatic patients with CMC1 OA complain of pain, weakness, and decreased function. Radiographic changes include trapeziometacarpal joint space narrowing, osteophyte formation, and subluxation of the joint. Multiple conservative and operative treatment options for CMC1 arthritis have been described ([2]).

The epidemiology of CMC1 OA has been previously studied, but is predominantly based on radiographic evidence alone. The prevalence of radiographic CMC1 OA reported in the literature ranges from 4–33% for middle-aged and elderly people ([3-8]). However, the correlation between symptomatic OA and radiographic change has been shown to be inconsistent. Dahaghin et al noted a modest correlation between general hand pain and radiographic CMC1 OA ([7]). In the Framingham study, Niu et al reported that in a cohort of persons with hand joint radiographic OA, only 8.3% of men and 16.7% of women had self-reported clinical symptoms ([9]). In studies of the impact of OA of the hand joints on function, investigators have noted that CMC1 OA had a higher impact on self-reported function and pain than interphalangeal joint OA ([5, 10]).

The majority of studies of symptomatic OA in the hand have used pain diagrams ([10, 11]) or interview questions ([7, 12]) to determine whether subjects have painful joints. A minority have reported physician examinations to evaluate CMC1 OA ([4]). With an aging population and changing health care, we need better and more up-to-date estimates of clinically important CMC1 OA and the actual burden on the health care system.

Hence, in order to more accurately report clinically meaningful CMC1 OA prevalence, we evaluated a comprehensive population-based health care register in Sweden, where the diagnostic codes from physicians are retrieved directly from the electronic medical records. The purpose of this study was to examine the prevalence of doctor-diagnosed, i.e., symptomatic, CMC1 OA within a defined population and report the related demographic factors of age and sex.

Box 1. Significance & Innovations

  • Symptomatic thumb carpometacarpal (CMC1) osteoarthritis (OA) is a notable disease, which in Sweden has led to consultation and diagnosis by physicians in 3.2% of the population age ≥55 years using the last 15 years of health data.
  • Women have a prevalence of symptomatic CMC1 OA 3 to 4 times higher than men, and present for clinical care earlier in life than men do.

Patients and methods

Using the Skåne Health Care Register in southern Sweden, we identified all residents ages ≥20 years who consulted a physician between 1998 and 2012, and were given a diagnosis of CMC1 OA. The diagnoses are chosen by the physicians themselves. We also noted the proportion of patients diagnosed by a specialist in orthopedics or hand surgery. We queried the register using the International Classification of Diseases, Tenth Revision, code M18 (including the subsets M18.0, M18.1, M18.2, M18.3, M18.4, M18.5, and M18.9). We then cross-linked all cases with the population register via the patients' personal identification numbers to exclude those who were deceased or had moved out of the Skåne region by December 31, 2012. The Skåne regional population by December 31, 2011 was used as the denominator.

These data are therefore estimates of doctor-diagnosed CMC1 OA prevalence by 2012, i.e., estimates represent the proportion of the population who are living and still resident, and who have received a physician-made CMC1 OA diagnosis over the last 15-year period. As codes are individually linked, only 1 diagnostic code of CMC1 OA was counted per patient. Approximately 30% of all outpatient medical care is provided by private health care in the Skåne region. To account for attrition due to patients seen in private practice whose diagnoses are not automatically forwarded to the Skåne Health Care Register (only the other details of the consultations are), the population denominator was reduced by 20% (using a deduction less than 30% because most orthopedic specialists and hand surgeons work in public care) ([13]).

We also calculated the 2012 incidence proportion of CMC1 OA, i.e., the number of patients consulting for CMC1 OA in 2012 but having no prior record (1998–2011) of a CMC1 OA diagnosis in the register divided by the adjusted Skåne region population ages ≥20 years. We used these data to also evaluate the age of first consultation. The 95% confidence intervals (95% CIs) for proportions were calculated with an exact binomial method.

As a sensitivity analysis, we estimated the expected point prevalence of CMC1 OA using the 2012 incidence proportion multiplied with the mean disease duration. The mean disease duration was considered equal to the remaining number of years of life expectancy at age 65 years (mean age of first consultation for both men and women). Data on age-specific life expectancy are available from Statistics Sweden.

The study was approved by the Regional Ethical Review Board of Lund University.


Based on census data, the total population in the Skåne region was 975,255 individuals ages ≥20 years by the end of 2012, with the reduction of 20% resulting in a population denominator of 780,204. During the 15-year time period, a total of 11,111 patients (of the 2012 prevalent population, i.e., still alive and resident in the region) had been diagnosed with M18-related codes for CMC1 OA by a medical doctor. In this cohort, the mean ± SD age was 67.7 ± 11.4 years and 78.5% of those given a diagnosis of CMC1 OA were women, with men making up 21.5%.

We calculated the overall prevalence estimate of doctor-diagnosed CMC1 OA in adults (ages ≥20 years) to be 1.42% (95% CI 1.40–1.45). The prevalence of doctor-diagnosed CMC1 OA in women was 2.19% (95% CI 2.15–2.24), compared to 0.62% (95% CI 0.60–0.65) in men. The highest prevalence was noted in women ages 70–74 years at 5.3%, with the lowest prevalence in women in those ages 20–24 years (0.03%). In men, the prevalence peak was in the age group of 80–84 years, calculated at 1.7%. Similar to women, the youngest age group (20–24 years) in men had the lowest prevalence at 0.01%. When stratified by decade, women and men both demonstrated the highest percentage prevalence between ages 70–79 years, with female prevalence at 4.4% and men at 1.5% (Figure 1). The prevalence in persons ages ≥70 years was 3.2% (4.3% in women and 1.7% for men) (Table 1).

Figure 1.

Point prevalence of doctor-diagnosed thumb carpometacarpal osteoarthritis by 2012, i.e., the proportion of the year 2012 population alive in 2012 who had been diagnosed by a medical doctor in the last 15 years (1998–2012).

Table 1. Prevalence of doctor-diagnosed thumb carpometacarpal osteoarthritis by year 2012 (the proportion of the year 2012 population who has been diagnosed by a medical doctor in the last 15 years, 1998–2012)*
Age, yearsWomenMenTotal
  1. Values are the percentage (number). Total denominator population ages ≥20 years = 780,204.
20+2.2 (8,726)0.6 (2,385)1.4 (11,111)
25+2.4 (8,716)0.7 (2,379)1.6 (11,095)
30+2.7 (8,708)0.8 (2,376)1.7 (11,084)
35+2.9 (8,692)0.8 (2,365)1.9 (11,057)
40+3.3 (8,653)1.0 (2,340)2.1 (10,993)
45+3.7 (8,535)1.1 (2,305)2.5 (10,840)
50+4.2 (8,273)1.3 (2,223)2.8 (10,496)
55+4.7 (7,762)1.4 (2,093)3.2 (9,855)
60+5.0 (6,798)1.6 (1,860)3.4 (8,658)
65+5.1 (5,370)1.7 (1,486)3.6 (6,856)
70+5.0 (3,679)1.8 (1,037)3.6 (4,716)
75+4.5 (2,311)1.9 (670)3.5 (2,981)
80+4.0 (1,362)1.9 (375)3.3 (1,737)
85+3.6 (661)2.0 (186)3.1 (847)

Of the prevalent CMC1 OA cohort, 63% had received a diagnosis of CMC1 OA at least once by a specialist in orthopedics or hand surgery, while 34% were diagnosed exclusively in primary care.

The 2012 (1 calendar year) incidence proportion (or cumulative incidence) in women was 228 per 100,000 persons, and in men was 71 per 100,000 persons. When evaluating first-time consulters with CMC1 OA, the peak rate (51 per 10,000 persons/year) occurred in women ages 60–69 years, while men had the highest consultation rate at approximately ages 70–79 years (approximately 20 per 10,000 persons/year) (Figure 2). The expected point prevalence obtained from sex-specific incidence and mean disease duration was 4.8% in adult women and 1.3% in adult men.

Figure 2.

Incidence proportion of doctor-diagnosed thumb carpometacarpal osteoarthritis by age and sex for 2012.


Our knowledge of the epidemiology of CMC1 OA is primarily derived from census data with subjects self-reporting a diagnosis of arthritis ([9, 11]), or from large radiographic studies ([12, 14]). Previous studies of CMC1 OA epidemiology have examined the relationship between radiographic OA and clinical symptoms, with examination performed by trained interviewers ([7]) or by physiotherapists ([5]). In the Mini-Finland Health Survey data reported by Haara et al, “specially trained” physicians examined 3,434 subjects using a standardized protocol, without further details on the background or specialty of the physicians ([4]). Our epidemiologic study is distinct in that we have captured diagnostic coding performed by general and specialty physicians seeing patients in the normal course of care and treatment. Diagnosis of a clinical entity by a physician has the advantages of integrating clinical symptoms, physical examination, and radiographic evidence. In Sweden, the patient's self-reported symptoms and findings from the physician's physical examination form the primary sources of diagnosis, but many use radiographic examinations to verify the diagnosis. Unfortunately, the current register data do not allow us to ascertain the methods used to assign diagnoses.

Similar to previous studies, we noted definite sex differences, with women having nearly 4 times greater point prevalence of physician-diagnosed CMC1 OA compared to men. This disparity has been noted as well in radiographic studies of CMC1 ([4, 14]). The reason for this finding is not known. Possible explanations include greater genetic susceptibility to the development of CMC1 arthritis in women, perhaps due to differences in reproductive hormones and their impact on the stabilizing ligaments of the joint, or biomechanical differences in joint loading ([1]).

We found that the highest point prevalence by age, indicating persons who had ever received a diagnosis of CMC1 OA and were still alive by the end of 2012, was in women between ages 70–74 years, with a later peak in men between ages 80–84 years. When evaluating first time presentation prevalence in 2012, the mean age was lower by a decade in both men and women. This suggests that women typically present with complaints of CMC1 OA between ages 60–70 years. Men present later, between ages 70–80 years. The point prevalence measures persons with the diagnosis of CMC1 OA in their medical record, but does not reflect those who gained improvement from nonoperative treatment or went on to surgery. The Finnish study had similar findings in men, with higher clinical and radiographic prevalence, but noted higher prevalence of symptomatic CMC1 OA in women age >75 years, at approximately 14% compared to 11% in women ages 65–74 years ([4]). Marshall et al also noted a significant association between age ≥70 years in both sexes and radiographic and clinically painful CMC1 OA ([5]). It is plausible that most elderly persons with CMC1 OA symptoms consult physicians due to other diseases (cardiovascular, pulmonary, etc.), with the more acute issues taking precedence when these patients consult and the diagnosis of CMC1 OA is not addressed. It is also likely that some elderly have had their CMC1 OA treated with arthroplasty, arthrodesis, or other surgery before 1998, and since then have been asymptomatic, hence, not captured in our data search. This is supported by the notably higher estimate of prevalence derived from the 2012 incidence and mean disease duration. The discrepancy between these 2 methods to estimate point prevalence likely has several explanations, including surgical treatment. It may also be explained by the steadily increased life expectancy in Sweden as well as a potential trend for increased consultations of clinically meaningful CMC1 OA in the last decades. It is therefore likely that the incidence-based calculation is a more reliable future prevalence estimate. The drop in prevalence of doctor-diagnosed CMC1 OA in the most elderly may also be explained by the possibility of CMC1 OA being associated with increased mortality ([4, 15]).

Other limitations of this study include the loss of patients to private practice in the Skåne region, although we reduced the population denominator to account for this. Additionally, the coding does not provide information on left or right, so we cannot identify the affected side or proportion of patients with bilateral disease. Finally, the differential diagnosis of hand pain in the region of the thumb is broad, including scaphotrapeziotrapezoid joint OA, de Quervain's tenosynovitis, and fracture or injury of the metacarpophalangeal joint. Several of these diagnoses have clearly different physical examination and radiographic findings than CMC1 OA. It is certainly possible, however, that the provider might fill in the code for generalized hand OA, which might lead to underestimation. We doubt that the opposite is true, as CMC1 OA is a fairly specific diagnosis as opposed to the wider diagnoses covered by general hand OA.

We present estimates of prevalence of physician-diagnosed CMC1 OA in a geographically well-defined Swedish county region using 15 years of consultation data. We noted the clinically important prevalence of CMC1 OA to be 3 to 4 times higher in women than in men. More than 1 in 20 elderly women in the prevalent population have consulted for CMC1 OA. These data can be used in health care planning and allocation of resources as the aging population continues to grow, with expectations of higher function and physical activity.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Wolf had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Wolf, Turkiewicz, Atroshi, Englund.

Acquisition of data. Turkiewicz, Atroshi, Englund.

Analysis and interpretation of data. Wolf, Turkiewicz, Atroshi, Englund.