OA is the most common type of arthritis. We estimated prevalence for each of the most commonly affected joints (knees, hips, and hands) as well as for overall OA.
Estimating the prevalence of OA is difficult because the structural changes of the disease occur in most persons as they get older, but these changes may not be accompanied by symptoms. Furthermore, prevalence estimates vary considerably depending on whether only moderate and severe radiographic changes are counted or mild changes are also included.
Clinically defined OA.
Study examiners characterize a person as having OA on the basis of symptoms and physical examination findings. The National Health and Nutrition Examination Survey I (NHANES I) (2) showed that 12.1% of the US population ages 25–74 years had clinically defined OA of some joint.
Radiographically defined OA.
Generally, researchers grade radiographs according to the Kellgren/Lawrence scale (3), which defines OA on the basis of the presence of osteophytes (outgrowths of bone at the margin of the joint). We summarized prevalence data primarily from 3 recent US population-based studies: the NHANES III, the Framingham Osteoarthritis Study, and the Johnston County Osteoarthritis Project. In these studies, participants had to appear in person to undergo radiographic testing, and the validity of the estimates could be compromised if only individuals who had symptoms (instead of all individuals) attended. High participation rates for radiography (>70%) in all 3 studies make this unlikely.
In phase 2 of the NHANES III (1991–1994), prevalence of knee OA was assessed in adults age ≥60 years; this was the only study to use non–weight-bearing radiographs, a method that minimizes joint space narrowing evident with weight bearing (4). The Framingham Osteoarthritis Study was a survey of knee and hand OA in ∼2,400 adults age ≥26 years from suburban Boston, Massachusetts (5, 6). The Johnston County Osteoarthritis Project was a study of hip and knee OA in ∼3,000 African American and white adults age ≥45 years in a rural county in North Carolina (7). The prevalence of knee OA in adults age ≥45 was 19.2% in Framingham and 27.8% in Johnston County, and the prevalence among adults age ≥60 was 37.4% in the NHANES III (Table 1). The prevalence of hip OA was high (27.0%) in Johnston County adults age ≥45, but in another US community-based study of 4,855 women age ≥65 years, prevalence was found to be only 7.2% (8). In the latter study, hip OA was defined based on individual features, but the discordance among study results leaves uncertainty regarding to the prevalence of hip OA. The prevalence of hand OA in Framingham adults was 27.2% overall and reached ≥80% among older adults, but only a minority of persons with radiographic OA have pain in these joints.
Table 1. Prevalence of radiographic OA in the hands, knees, and hips, by age and sex, from population-based studies*
|Hands, ≥26||Framingham OA study (6)||25.9||28.2||27.2|
|Knees†|| || || || |
| ≥26||Framingham OA study (5)||14.1||13.7||13.8|
| ≥45||Framingham OA study (5)||18.6||19.3||19.2|
| ≥45||Johnston County OA Project (7)||24.3||30.1||27.8|
| ≥60||NHANES III (4)||31.2||42.1||37.4|
|Hips, ≥45||Johnston County OA Project (10)||25.7||26.9||27.0|
OA prevalence increased with age and affected the hands and knees of women more frequently than men, especially in persons age ≥50 years. In Johnston County and in the NHANES III, African Americans were more likely than whites to have radiographic knee OA. A study of perimenopausal women in Michigan also demonstrated that African Americans were more likely than whites to have radiographic knee and hand OA (9). In the NHANES III, the prevalence of radiographic knee OA was significantly higher in non-Hispanic African Americans than in non-Hispanic whites or Mexican Americans (52.4%, 36.2%, and 37.6%, respectively) (4). In Johnston County, the prevalence of radiographic hip OA was comparable in African Americans and whites (10).
Persons are considered to have symptomatic OA if they have frequent pain in a joint and radiographic evidence of OA in that joint, although sometimes this pain may not actually emanate from the arthritis seen on the radiograph. Most prevalence surveys require that a person have pain in a joint on most days of a recent month, to meet the definition for presence of symptoms.
The prevalence of symptomatic knee OA was 4.9% among adults age ≥26 years in the Framingham study, 16.7% among adults age ≥45 in the Johnston County study, and 12.1% among adults aged ≥60 in the NHANES III study (Table 2). The prevalence of symptomatic hip OA was 9.2% among adults age ≥45 in the Johnston County study. The prevalence was slightly higher among women than among men for both of these outcomes (Table 2).
Table 2. Prevalence of symptomatic OA (symptoms and radiographic changes of OA in the symptomatic joint) in the hands, knees, and hips, by age and sex, from population-based studies*
|Hands, ≥26||Framingham OA study (6)||3.8||9.2||6.8|
|Knees|| || || |
| ≥26||Framingham OA study (5)||4.6||4.9||4.9|
| ≥45||Framingham OA study (5)||5.9||7.2||6.7|
| ≥45||Johnston County OA Project (7)||13.5||18.7||16.7|
| ≥60||NHANES III (4)||10.0||13.6||12.1|
|Hips, ≥45||Johnston County OA Project (10)||8.7||9.3||9.2|
The prevalence of symptomatic hand OA in the Framingham subjects (6) was 6.8% overall and was especially high in older adults. Among individuals age ≥71 years, prevalence was 26.2% in women and 13.4% in men.
Using the Framingham data on age/sex prevalence among persons age ≥26 years and the corresponding 2005 population estimates from the Census Bureau, we estimated that 9,267,000 adults have symptomatic knee OA and 13,054,000 adults have symptomatic hand OA. The generalizability of the Framingham estimates to the US population has not been determined.
For estimating overall OA, we calculated weighted age/sex-specific prevalence estimates of clinical arthritis for persons ages 25–74 from the NHANES I, the only published national source of data on OA at multiple anatomic sites and the source we used in our 1998 report (11). Using the corresponding 2005 population estimates from the Census Bureau and additionally applying the NHANES I estimate for those ages 65–74 to the Census population age ≥75 years, we estimated that 26.9 million adults age 25 and older have clinical OA of some joint. Whether the 1971–1975 NHANES I estimates reflect the 2005 US population prevalence is uncertain.
Polymyalgia rheumatica (PMR) and giant cell (temporal) arteritis (GCA).
PMR and GCA, two closely related syndromes that occur almost exclusively in persons age ≥50 years, appear to be much more common in whites than in other racial or ethnic groups (12). In the US, prevalence is highest among persons of northern European descent (13, 14).
The only population-based study of PMR and GCA in the US is from Olmsted County, Minnesota, where the prevalences of PMR and GCA were derived from cumulative incidence rates. GCA was diagnosed according to the American College of Rheumatology (ACR) criteria (15). Among persons age ≥50 years in 2000, the prevalence of PMR was 739 per 100,000 (95% confidence interval [95% CI] 674–808) (16) and the prevalence of GCA was 278 per 100,000 (95% CI 192–268) (17). For each, the prevalence was higher in women than in men and increased dramatically with age (for PMR, from 21 per 100,000 among persons ages 50–54 years to 4,070 per 100,000 among those age ≥90 years) (Table 3).
Table 3. Prevalence of polymyalgia rheumatica and giant cell arteritis in Olmsted County, Minnesota, by age and sex*
In addition to these data, case series of PMR and GCA suggest that these conditions are common throughout the US. In Europe, incidence rates in populations in the northern area of the continent are similar to those of Olmsted County, Minnesota, whereas rates among southern European populations are lower (18–21).
In summary, PMR is common in older adults. GCA, which is approximately one-third as common as PMR, is the most common form of vasculitis in the population over 50 years of age. Using the Olmsted County age/sex prevalence rates and the corresponding 2005 estimates from the Census Bureau, we estimated that 711,000 Americans have PMR and 228,000 have GCA. With the aging of the US population, these estimates are likely to increase in coming years. These Olmsted County estimates are generalizable to the white US population, but their generalizability to other racial/ethnic populations is uncertain.
Gout is an inflammatory arthritis that results from phagocytosis of monosodium urate monohydrate crystals within the joint and is usually associated with an elevated concentration of uric acid in the blood, i.e., hyperuricemia. Among criteria developed to classify gout, the ACR criteria (22) have been used in recent epidemiologic studies (23, 24).
Our previous National Arthritis Data Workgroup (NADW) report (11) reviewed the population-based Tecumseh Community Health Study (25), the Framingham Heart Study (26), and the Sudbury Study (27) (Table 4), all of which were relatively small studies of gout conducted in confined geographic regions. All of these studies took place before the ACR criteria were developed, the age populations included were not consistent among studies.
Table 4. Prevalence of gout in the United States*
|Regional population studies|| || || || |
| Tecumseh Community Health Study, 1960/“Rome” (25)†||≥20||720||480||ND|
| Framingham Heart Study, 1964/arbitrary (26)‡||≥42 (mean 58)||2,850||390||1,480|
| Sudbury Study, 1972/Rome and New York (27)||≥15||660||100||370|
|National survey studies|| || || || |
| NHIS, 1988/self-report (1-year||≥18||ND||ND||850|
| prevalence) (64)§||18–44||290||90||310|
| NHIS, 1992/self-report (1-year||≥18||ND||ND||840|
| prevalence) (65)§||18–44||440||30||380|
| NHIS, 1996/self-report (1-year||≥18||ND||ND||940|
| prevalence) (30)§||18–44||340||20||180|
| NHANES III,||≥20||3,800||1,600||2,600|
| 1988–1994/self-report (lifetime||≥20–29||200||500||400|
| prevalence) (32)¶||30–39||2,100||100||1,100|
The prevalence of gout also has been estimated using self-reported information from the National Health Interview Survey (NHIS) and the NHANES (Table 4). Because these self-reports were not verified, the data may produce an overestimate of prevalence. For example, investigators in the Sudbury Study (27) could validate only 44% of self-reported cases using Rome (28) or New York (29) criteria, and in a study of health professionals (23), only 70% of cases could be validated by ACR criteria. However, in one study of physicians (24), it was reported that 100% of self-reported cases could be validated by ACR criteria and medical record review. In the most recent NHIS survey on gout (1996), the prevalence for the 1-year period was 940 per 100,000 adults age ≥18 years in the US (30). Prevalence increased with age, was higher in men than in women at all ages, and among those age ≥45 years was higher in African Americans than in whites.
One-year period prevalence estimates derived from the NHIS over time can be compared directly because the instrument has not changed. In this survey, the presence of gout is recorded if a respondent answers “yes” to the question, “Have you or any member of your household had gout within the past year?” From 1969 to 1985 the prevalence more than doubled, with the steepest increase occurring between 1969 and 1976 (30), but the increase later slowed between 1992 and 1996 (840 and 940 per 100,000, respectively) (Table 4). In a recent study from a US managed care population, the overall prevalence of gout or hyperuricemia requiring a gout or urate-lowering medication increased by 80% from 1990 to 1999 (31); however, this increase may reflect treatment change.
The lifetime prevalence estimate from the NHANES III (1988–1994) (32), based on the question, “Has a doctor ever told you that you had gout?,” was 2,600 per 100,000 overall for those age ≥20 years, with a low of 400 per 100,000 in adults ages 20–29 years and a peak of 8,000 per 100,000 in adults ages 70–79 years (Table 4). Gout was reported more often by men than by women overall, but prevalence increased with age for both, especially among women after menopause.
In summary, gout appears to be increasing in frequency. Using 1996 NHIS and NHANES III age/sex prevalence data and the corresponding 2005 population estimates from the Census Bureau, we estimated that 3.0 million adults age ≥18 years had gout in the past year, and 6.1 million adults age ≥20 have ever had gout. Both are likely overestimates because they are based on self-reported data.
The 1990 ACR criteria for the classification of fibromyalgia require the presence of widespread pain for at least 3 months and pain on palpation in at least 11 of 18 anatomic sites (33). Despite its acceptance, a number of problems occur when the ACR criteria set is used to define fibromyalgia in populations, leading to difficulties in estimating prevalence (34–36). Also, fibromyalgia may be more common among persons with other medical conditions (37), so prevalence estimates of primary fibromyalgia may be lower than estimates that do not differentiate primary fibromyalgia from fibromyalgia secondary to other disorders.
Wolfe et al conducted the only study of the prevalence of primary fibromyalgia in the US, in Wichita, Kansas in 1993 (38). At the time of the study, the population of Wichita was ∼88% white, different from that of the overall US population. From a random sample of 3,006 adults age ≥18 years, 193 individuals with chronic widespread pain were examined and 36 cases of ACR-defined fibromyalgia were confirmed. The overall prevalence among adults was ∼2% (95% CI 1.4–2.7); prevalence was higher among women than among men (3.4% versus 0.5%). In women, the prevalence of fibromyalgia rose sharply in middle age, to a maximum of 7.4% in the 70–79-year age group, and then dropped off. Prevalence in men similarly peaked in the eighth decade of life, but was only slightly more than 1% among men in this age group. Fibromyalgia was associated with depressive and anxiety symptoms, current and past depression, and a history of depression in the family. Other characteristic symptoms, i.e., “pain all over,” subjective swelling, paresthesias, stiffness, sleep disturbance, fatigue, and irritable bowel syndrome, were also associated with fibromyalgia (38). Women experienced more of these associated symptoms than men (34). Other factors associated with fibromyalgia included reduced income and education, higher rates of divorce, and application for disability benefits (38).
Although no recent prevalence studies of fibromyalgia in the US have been published, a population survey of 3,395 randomly selected adults (≥18 years of age) in Ontario, Canada showed 100 cases of fibromyalgia, yielding an even higher overall age/sex-adjusted prevalence of 3.3% (95% CI 3.2–3.4%) (4.9% in women and 1.6% in men) (39). As in the US study, the prevalence of fibromyalgia in women rose with age, to a peak of 7.9% in the 55–64-year age group, and declined thereafter. The prevalence in men also increased with age, and peaked at 2.5% in the 45–54-year age group.
Using the Wichita age/sex prevalence and the corresponding 2005 population estimates from the Census Bureau, we estimated that ∼5.0 million adults age ≥18 years have primary fibromyalgia. The generalizability of the Wichita estimates to the US population is uncertain.
Carpal tunnel syndrome (CTS).
Because there is no gold standard for the diagnosis of CTS and no standard definition of the syndrome, and because some cases can be transient, challenges arise in determining the prevalence of this condition. The clinical examination findings associated with CTS have sensitivities and specificities in the range of 45–80% (40). The typical location of symptoms in the median nerve distribution has sensitivity and specificity in the range of 60–70% (41, 42). On median nerve conduction testing, >20% of asymptomatic individuals exhibit abnormalities consistent with CTS (42, 43).
The most well–accepted definitions of CTS involve combinations of symptom, electrophysiologic, and physical examination findings. Three population-based prevalence studies have been performed in Europe, involving a combination of history, physical examination, and nerve conduction assessments. Atroshi and colleagues (44), in a Swedish study, inquired about symptoms typical of CTS (40, 42) and studied symptomatic and asymptomatic subjects further by history-taking, physical examination, and nerve conduction studies. They estimated the prevalence of symptoms typical of CTS, of symptoms accompanied by examination findings indicative of CTS, of symptoms accompanied by positive electrophysiologic results, and of symptoms accompanied by positive findings on all tests. The prevalence of CTS was consistently higher in women than in men, and rose with age. CTS prevalence ranged from 2% to 4% in men and 3% to 5% in women, depending on the stringency of the case definition.
Ferry and colleagues (45) administered a hand symptom diagram (42) to a random population sample in the UK. Persons with and persons without typical symptoms of CTS underwent physical examination and electrophysiologic testing. The prevalence of positive nerve conduction findings did not differ between subjects with typical CTS symptoms and asymptomatic subjects. Thus, although the authors reported a prevalence of electrophysiologic evidence of CTS on the order of 10%, it is difficult to draw conclusions about the prevalence of symptomatic CTS.
In an earlier study, de Krom and colleagues (46) evaluated the prevalence of CTS in 715 persons from The Netherlands. Of the respondents, 1.6% had diagnosed CTS and 9% had nocturnal finger paresthesias. The latter underwent neurologic and electrophysiologic evaluation. The overall prevalence of CTS (either diagnosed or detected via this 2-stage screening process) was 5.8% in women and 0.6% in men.
Studies using a combination of self-report, physical examination, and nerve conduction testing have not been performed in the US. Tanaka et al analyzed data from the 1988 NHIS (47). Respondents were asked if they had CTS. An estimated 1.55% of 170 million adults (2.65 million) self-reported CTS. CTS was twice as common among women as among men (47). The self-report nature of the case definition makes these data difficult to interpret.
In summary, the prevalence of electrophysiologically confirmed, symptomatic CTS, based on studies conducted outside the US, is ∼1–4% in men and 3–5% in women, with prevalence increasing with age. Using these adult prevalence estimates and the corresponding 2005 population estimates from the Census Bureau, we estimated that 1–4 million men and 3–6 million women in the US have CTS.
Back and neck pain.
Low back pain affects most adults at some time. Because back pain is rarely permanent, terms such as incidence and prevalence may be ambiguous. Although neck pain is less common than low back pain, it shares many similar characteristics. The thoracic spine's mobility is limited by the rib cage, and is less often a cause of back pain except in osteoporosis. Thus, the prevalence of thoracic spine pain is rarely reported.
Back pain is a symptom, not a disease. For many, a precise pathoanatomic diagnosis is impossible (48, 49). There is no definitive imaging or diagnostic test to determine its prevalence. Investigators depend on patient self-report, and many patients have few objective physical findings.
In epidemiologic surveys, back pain has been defined in many ways. These include counting any episode of back pain, counting pain lasting a certain length of time, counting only “severe” pain, or counting only pain that results in work disability. We focused on population-based estimates of self-reported symptoms, separating neck and low back pain, and attempting to distinguish reports of any back pain, frequent or persistent back pain, and pain with symptoms of radiculopathy (nerve root irritation, usually manifesting as sciatica).
In the 2002 NHIS, respondents were asked whether they had low back pain or neck pain during the past 3 months. Respondents were instructed to report pain that lasted a whole day or longer (50). Approximately one-fourth of all adults in the US reported experiencing low back pain in the past 3 months, and the prevalence of neck pain was approximately half that estimate (Table 5). The prevalence was similar among men and women and among most racial groups, with the exception of a high prevalence among American Indians and Alaskan Natives, and a low prevalence among Asian Americans. Prevalence declined with increasing levels of education.
Table 5. Prevalence of neck pain and various categories of low back pain, by race, age, and sex
|Race|| || || || || |
| White||14.2 (13.7–14.8)||27.4 (26.6–28.2)||59||19||14|
| African American||11.7 (10.5–12.9)||23.9 (22.2–25.6)||46||19||11|
|Age, years|| || || || || |
| 18–44||11.9 (11.3–12.6)||23.7 (22.9–24.5)||61||15||10|
| 45–65||16.9 (16.0–17.8)||29.8 (28.7–30.9)||53||20||12|
| 65–74||14.2 (12.7–15.7)||28.8 (26.9–29.8)||56||21||17|
| ≥75||14.0 (12.5–15.5)||28.7 (26.7–30.7)||49||18||16|
|Sex|| || || || || |
| Male||11.7 (11.0–12.4)||24.3 (23.4–25.2)||53||15||14|
| Female||15.7 (15.0–16.4)||28.3 (27.5–29.1)||57||20||13|
From an analysis of the 1997 NHIS, it was estimated that 3.2% of all persons experienced activity limitations due to chronic back conditions (51). Among the poor, activity limitation was nearly 3 times more frequent than in middle- and high-income adults.
Although about half of adults report low back pain during a given year and about two-thirds report low back pain at some time in their lives (52–60), only 15–21% of the adult population reported frequent low back pain (56) and only 14% reported an episode of low back pain lasting longer than 2 weeks at any time in their lives. Pain lasting beyond 3–6 months occurs in only 5–10% of patients with back pain. Approximately 1–2% of adults report having received a diagnosis of a herniated disc.
Computed tomography and magnetic resonance imaging studies in small samples indicate that disc degeneration, fractures, herniated discs, and spinal stenosis are all common among asymptomatic persons (61, 62). Thus, the prevalence of radiographic findings is substantially different from the prevalence of clinically important symptoms.
In summary, back pain is common. The etiology is often unclear and classification is controversial, but most episodes probably originate in muscles or ligaments, or are the consequences of degenerative changes in the intervertebral discs and adjacent vertebrae. These changes include osteoarthritic changes in the facet joints and similar degenerative changes in and around the intervertebral discs. Back pain remains a leading cause of work disability.
Using 2002 NHIS age-specific prevalence and the corresponding 2005 population estimates from the Census Bureau, we estimated that 59.1 million adults age ≥18 years have had low back pain “in the past 3 months.” Using the 1997 NHIS age-specific prevalence and the corresponding 2005 population estimates from the Census Bureau, we estimated that 7.1 million adults age ≥18 have activity limitation due to chronic back conditions. Using the same sources, we estimated that 30.1 million adults age ≥18 have had neck pain “in the past 3 months.”
As discussed in the companion article (1), the burden of a chronic condition can be measured in various ways. The NADW has chosen to focus on national disease prevalence as an important measure of burden.
The prevalence of clinical osteoarthritis has grown to nearly 27 million, up from our estimate for 1995 of 21 million (11), as would be expected for such a strongly age-related disease. Gout appears to be increasing in frequency as well, with a 1-year prevalence of 3.0 million adults, a higher frequency than in the earlier study. These increases in such common conditions suggest they will have a growing impact on the health care and public health systems in the future, one that needs to be anticipated in order to provide the early diagnosis and interventions that can help reduce that impact.
We have provided estimates of prevalence and numbers of persons affected for overall arthritis and for selected rheumatic conditions and given a rough snapshot of current burden. These estimates have been made by recognized disease experts using the best data available, but, as noted in many of the sections, must be interpreted with several limitations in mind, including those detailed in the companion report (1). Given the large and growing burden of arthritis and other rheumatic conditions, we hope this work will inspire studies that better address these gaps and limitations and provide a better understanding of the burden of these conditions.
Dr. Helmick 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 design. Lawrence, Felson, Helmick, Choi, Gabriel, Hunder.
Acquisition of data. Felson, Helmick, Choi, Hochberg, Hunder, Jordan, Katz, Maradit Kremers, Wolfe.
Analysis and interpretation of data. Felson, Helmick, Arnold, Choi, Deyo, Gabriel, Hunder, Jordan, Katz, Maradit Kremers, Wolfe.
Manuscript preparation. Lawrence, Felson, Helmick, Arnold, Choi, Deyo, Gabriel, Hirsch, Hochberg, Hunder, Jordan, Katz, Maradit Kremers, Wolfe.
Statistical analysis. Helmick, Choi.
Project initiation and organization. Lawrence.