Prevalence of specific types of arthritis and other rheumatic conditions in the ambulatory health care system in the United States, 2001–2005†
The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention.
To estimate the overall prevalence of medically-treated arthritis and other rheumatic conditions (AORC) for adults, the prevalence of specific medically-treated conditions, and the overall annual number of visits for these conditions in the ambulatory health care system.
We used data from the 2001–2005 National Ambulatory Medical Care Survey and 2001–2005 National Hospital Ambulatory Medical Care Survey to estimate annual ambulatory health care visits for the International Classification of Diseases, Ninth Revision, Clinical Modification codes thought to represent AORC. Using data on the number of prior annual visits per patient per condition, we converted the visit estimates into prevalence estimates of adults age ≥18 years with medically-treated AORC overall and for specific conditions.
The overall prevalence estimate of adults with medically-treated AORC was 29,150,000 adults (95% confidence interval [95% CI] 26,473,000–31,826,000) and accounted for 77,887,300 ambulatory care visits (95% CI 71,266,000–84,508,000). The top 5 most prevalent conditions were osteoarthritis and allied disorders, unspecified joint disorders, peripheral enthesopathies, unspecified arthropathies, and other disorders of synovium, tendon, or bursa.
The advantage of our approach is that it uses existing rather than expensive new surveys for tracking the prevalence of medically-treated AORC overall and tracking the prevalence of difficult to measure specific conditions. The estimates are data based and national in scope. More relevantly, they better estimate the numbers of persons whose AORC impacts on the ambulatory health care system.
It is estimated from a national self-report survey that 46.4 million American adults have doctor-diagnosed arthritis (1). However, estimating the national prevalence of specific types of arthritis and other rheumatic conditions (AORC) is difficult. The main reason is that self-report of specific conditions is unreliable (2, 3), so that available national surveys using self-reported data for specific conditions would be subject to misclassification and not provide credible estimates. More reliable prevalence estimates of specific conditions depend on a more standard and objective method of case definition, such as those emanating from providers with medical training. A recent study conducted by the National Arthritis Data Workgroup (NADW) reviewed the literature and arrived at population prevalence estimates for some specific conditions, but found that these were largely based on “a few small studies of uncertain generalizability” because their geographic or demographic populations might not be representative of the nation as a whole (4, 5). The NADW concluded that given the dearth of available data, further studies were needed to clarify the national prevalence of these conditions. The purpose of this study was to provide such national prevalence estimates for specific conditions using 2001–2005 national ambulatory care data.
MATERIALS AND METHODS
The NADW developed an International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code set that has been used to analyze and report AORC burden for adults for more than 15 years (6). Using these NADW-defined ICD-9-CM codes (Table 1), we studied data from the 2001–2005 National Ambulatory Medical Care Survey (NAMCS) and National Hospital Ambulatory Medical Care Survey (NHAMCS) (7) to estimate annual visits for AORC to ambulatory care facilities. These surveys sample visits to physician offices, hospital outpatient departments, and emergency departments. Visit data include up to 3 ICD-9-CM diagnostic codes assigned by medical coders using text provided by health care professionals.
Table 1. International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code-based definition of arthritis and other rheumatic conditions (AORC) developed by the National Arthritis Data Workgroup
|Selected ICD-9-CM codes from chapter 13 (710–739)|
| 710 Diffuse diseases of connective tissue|
| 711 Arthropathy associated with infections|
| 712 Crystal arthropathies|
| 713 Arthropathy associated with other disorders classified elsewhere|
| 714 Rheumatoid arthritis and other inflammatory polyarthropathies|
| 715 Osteoarthritis and allied disorders|
| 716 Other and unspecified arthropathies|
| 719 Other/unspecified joint disorders (excluding 719.1)|
| 720 Ankylosing spondylitis and other inflammatory spondylopathies|
| 721 Spondylosis and allied disorders|
| 725 Polymyalgia rheumatica|
| 726 Peripheral enthesopathies and allied conditions|
| 727 Other disorders of synovium, tendon, or bursa|
| 728 Disorders of muscle, ligament, fascia (excludes 728.4 and 728.5)|
| 729.0 Rheumatism, unspecified and fibrositis|
| 729.1 Myalgia and myositis, unspecified|
| 729.4 Fasciitis unspecified|
|Selected ICD-9-CM codes from other chapters|
| 095.6 Syphilis of the muscle|
| 095.7 Syphilis of synovium, tendon, bursa|
| 098.5 Gonoccocal infection of joint|
| 099.3 Reactive arthritis|
| 136.1 Behcet's syndrome|
| 274 Gout|
| 277.2 Other disorders of purine/pyrimidine metabolism|
| 287.0 Allergic purpura|
| 344.6 Cauda equina syndrome|
| 353.0 Brachial plexus lesion/thoracic outlet syndrome|
| 354.0 Carpal tunnel syndrome|
| 355.5 Tarsal tunnel syndrome|
| 357.1 Polyneuropathy–collagen vascular disease|
| 390 Rheumatic fever without mention of heart involvement|
| 391 Rheumatic fever with heart involvement|
| 437.4 Cerebral arteritis|
| 443.0 Raynaud's syndrome|
| 446 Polyarteritis nodosa and allied conditions|
| 447.6 Arteritis, unspecified|
| 696.0 Psoriatic arthropathy|
Data from 2 sources (physician offices and outpatient clinics) also included information on the number of past visits within the last 12 months for the patient represented by the record (categorized as 0, 1–2, 3–5, or ≥6, excluding the current visit). We used those prior visit data to estimate the prevalence of adults (age ≥18 years) with AORC overall and for specific conditions in the manner suggested by Burt and Hing (8) and used previously by us to estimate pediatric arthritis prevalence (9). As we noted before, the method assumes that the weighted number of visits for 1 record divided by the number of annual visits for the patient reflected in the record (visits/patient) equals the number of patients with that condition (i.e., visits/[visits/patient] = patients).
Thus, if a sampled visit was for a new patient with gout who had a current visit only and the weighted number of visits estimated for that record was 5,000, then 5,000/1 = an estimated 5,000 patients with gout where 1 is the total number of annual visits for that patient (1 current visit + 0 prior visits). Similarly, for a sampled visit of a patient with gout who had been seen 1–2 times previously during the past 12 months, if the weighted number of visits estimated for that record was 5,000, then 5,000/2.5 = 2,000 estimated patients with gout (2.5 is the total number of annual visits for that patient: 1 current visit + 1.5 prior visits [the midpoint of 1–2]). Using the midpoint for the number of prior visits, the total number of visits was calculated as follows: 0 prior visits = 1 total visit; 1–2 prior visits = 2.5 total visits; 3–5 prior visits = 5 total visits; and ≥6 prior visits = 8 total visits.
We used SAS, version 9.1, (SAS Institute, Cary, NC) for data extraction and analysis. We first estimated the weighted number of visits for each year for each diagnostic code for each ambulatory care site. We then converted those estimates to persons and then averaged the estimates to produce annualized 5-year estimates. For the number of persons with each specific condition, we made these estimates when an AORC code for that condition was a listed diagnosis anywhere on the record. However, for the overall burden of AORC, we counted a qualifying record only once. Thus, if a record listed osteoarthritis and gout in different fields, we counted that person only once to estimate AORC overall, but for specific conditions we counted that person once for osteoarthritis and once for gout.
We calculated 95% confidence intervals (95% CIs) and SEs for both the average annual visit and prevalence estimates based on the complex survey design variables as before (9, 10). National estimates were rounded to thousands. Because NAMCS and NHAMCS estimates based on fewer than 30 records or with relative SEs >30% are considered unreliable (11), we did not report estimates below these limits. We calculated prevalence rates per 100,000 population using census data for adults age ≥18 years for 2003, the midpoint year of the study period (12).
The overall annualized prevalence estimate of adults with medically-treated AORC in the ambulatory health care system in 2001–2005 was 29,150,000 (95% CI 26,473,000–31,826,000), accounting for 77,887,300 annual outpatient encounters (95% CI 71,266,000–84,508,000). The top 5 most prevalent medically-treated specific conditions were osteoarthritis and allied disorders, unspecified joint disorders, peripheral enthesopathies, unspecified arthropathies, and other disorders of synovium, tendon, or bursa (Table 2).
Table 2. ICD-9-CM code specific annualized prevalence and visits for AORC, by specific condition, United States, 2001–2005*
|715||Osteoarthritis and allied disorders||7,674 (6,805–8,542)||3,532||20,892 (18,783–23,001)|
|719||Other/unspecified joint disorders||5,540 (4,889–6,191)||2,550||15,357 (13,687–17,027)|
|726||Peripheral enthesopathies||4,396 (3,867–4,925)||2,024||10,450 (9,314–11,586)|
|716||Other and unspecified arthropathies||2,632 (2,275–2,990)||1,212||7,370 (6,537–8,203)|
|727||Other disorders of synovium, tendon, bursa||2,550 (2,185–2,916)||1,174||5,690 (5,025–6,356)|
|354.0||Carpal tunnel syndrome||2,240 (1,840–2,641)||1,031||4,569 (3,887–5,251)|
|729.1||Myalgia and myositis, unspecified||1,807 (1,530–2,083)||832||5,545 (4,856–6,234)|
|714||Rheumatoid arthritis and other inflammatory polyarthropathies||1,481 (1,154–1,808)||682||4,068 (3,296–4,841)|
|728||Disorders of muscle, ligament, fascia (excludes 728.4 and 728.5)||1,236 (1,045–1,428)||569||3,417 (3,006–3,828)|
|721||Spondylosis and allied disorders||1,069 (887–1,251)||492||3,039 (2,486–3,592)|
|710||Diffuse diseases of connective tissue||825 (570–1,080)||380||2,594 (1,584–3,604)|
| Systemic lupus erythematosus (710.0)||280 (238–321)||129||1,032 (867–1,196)|
| Scleroderma (710.1)||55 (33–76)||25||196 (140–251)|
| Sjögren's syndrome (710.2)||319 (167–471)||147||850 (292–1,409)|
|274||Gout||792 (649–936)||365||2,232 (1,933–2,530)|
|725||Polymyalgia rheumatica||262 (186–337)||121||894 (664–1,123)|
|720||Ankylosing spondylitis and other inflammatory spondylopathies||194 (133–256)||89||537 (423–651)|
| Ankylosing spondylitis (720.0)||61 (34–88)||28||194 (130–258)|
|696.0||Psoriatic arthropathy||138 (88–189)||64||319 (155–484)|
|443.0||Raynaud's syndrome||121 (64–177)||56||241 (158–323)|
|446||Polyarteritis nodosa and allied conditions||119 (50–188)||55||339 (249–428)|
| Giant cell arteritis (446.5)||95 (23–167)||44||244 (170–317)|
|447.6||Arteritis, unspecified||65 (55–76)||30||179 (136–223)|
|711||Arthropathy associated with infections||49 (34–64)||23||193 (161–225)|
|353.0||Brachial plexus lesion/thoracic outlet syndrome||35 (28–41)||16||125 (109–141)|
|Remaining AORC codes‡||112 (89–135)||52||348 (269–426)|
|Total||29,150 (26,473–31,826)||13,419||77,887 (71,266–84,508)|
We estimated that from 2001–2005 an average of 29,150,000 patients made AORC-related visits to physician offices or hospital outpatient departments. That overall estimate of arthritis differs with the NADW estimate that 46.4 million adults had doctor-diagnosed arthritis in 2003–2005 (4). There are several plausible reasons for this difference. First, the target population is different; the NADW estimate is for the larger civilian, non-institutionalized population, rather than the smaller population seen in physician offices or outpatient departments. Second, the time frame of reference is different; the NADW estimate is based on a “yes” answer to the following self-report question in the National Health Interview Survey: “Have you ever been told by a doctor or other health care professional that you have some form of arthritis, rheumatoid arthritis, gout, lupus or fibromyalgia?”(1); whereas the NAMCS/NHAMCS asked about visits in the previous 12 months. This fact may exaggerate differences for component conditions that are episodic (e.g., gout), less symptomatic (e.g., osteoarthritis and allied disorders), or cured (e.g., carpal tunnel syndrome). Third, the information source is different; the NADW estimate is based on self-report, which is considered adequate for estimates of overall arthritis as any misclassification is likely to occur within that broad rubric, whereas our estimates are based on more restrictive health care provider assessment.
Fourth, our estimates do not include emergency department visits, which might not affect our estimates if all emergency department visitors are assumed to see other ambulatory care sites within 12 months, but might increase our estimates from 405,000 (assuming each emergency department visit had the upper range [n = 8] of prior visits for conversion of visits to adults) to 3,242,000 (assuming that each such emergency department visit represented the only annual visit for that adult).
Fifth, our data sources required contact with the health care system; accordingly, asymptomatic persons with AORC not seeking care were likely undercounted, as were symptomatic persons with AORC but no source of health care. Moreover, the proportion of those with AORC deciding to seek care likely varies by diagnoses. For example, persons with osteoarthritis, soft tissue, or other ill-defined conditions may choose not to see a physician after receiving a diagnosis and recommendation for acetaminophen on an earlier occasion. On the other hand, a visit may be far more likely for someone with more serious and rarer conditions such as rheumatoid arthritis (RA), other kinds of inflammatory arthritis, and connective tissue diseases for which medical treatment is necessary. It is noteworthy that for these conditions, the estimates in this article are more similar to the numbers generated from the NADW. These are also the conditions that because of their relative rarity are more difficult to monitor with population data.
Prevalence estimates were also made for over 20 specific AORC conditions, far more than the NADW was able to address (4, 5). While some of our estimates, e.g., systemic lupus erythematosus, scleroderma, and RA, were similar to the NADW study, many estimates were strikingly different.
Our specific condition estimates are likely to be underestimates for many of the reasons above and some below. As we noted previously (9), converting visits to persons assumes accurate prior visit data. In 2001 and 2002, data on the number of past visits in the last 12 months were missing ∼6–7% of the time for NAMCS and 9–14% of the time for outpatient departments and were not imputed. Starting in 2003, the National Center for Health Statistics began imputing missing data for number of past visits as well as for whether the patient was new or established (the latter variable generally had a very low nonresponse rate).
Given the large number of conditions coded unspecified or other, lack of specific diagnostic categorization and assignment by the providers can lead to specific condition estimate errors. Even for experienced physicians, many AORC-related conditions are difficult to diagnose, especially in the early stages (4). On the other hand, even assignment of a specific diagnosis may still reflect uncertainty because it is preliminary and awaits later confirmation through testing or observation of disease evolution. One approach to improving ambulatory health care disease-based estimates has been to apply algorithms that search for multiple visits for the same condition over time in the same person (13). Unfortunately this is not possible in NAMCS/NHAMCS data given the short-term nature of the sampling at a site and the lack of individual identifiers.
Our estimates are based on diagnoses listed in any of 3 diagnostic fields, as opposed to only the primary diagnosis for the visit. Although an AORC code may not have been the primary diagnosis, by virtue of it being listed we assume it impacted on the visit in some way; moreover NCHS instructions call for listing the physician's primary diagnosis plus up to 2 additional diagnoses that are related to the current visit, including chronic conditions. For example, if a person with RA went to a physician for an acute infection, the infection might be coded as the primary diagnosis for the visit and RA as a secondary diagnosis. We did not want to miss counting such persons. If we were to use only the first-listed (i.e., primary) diagnosis, the estimates would be lower. Additionally, because only 3 diagnoses could be captured, it is possible there were more AORC that were not listed, especially in the elderly and those with multiple comorbid conditions.
Despite these limitations, the main advantage of our method is that it uses existing surveys for tracking the prevalence of specific AORCs. Further, the estimates are data based, national in scope, include a health care provider assessment of the diagnosis, are unchanging in methods over time, address specific AORC conditions in a way not done in the past, and can be tracked over time to assess temporal changes. More relevantly, they better estimate the numbers of persons whose specific AORC impacts the health care system. 3
Table 3. Comparison of the prevalence estimates for specific conditions in the current study and the NADW*
|715||Osteoarthritis and allied disorders||7,674,000||27,000,000|
|354.0||Carpal tunnel syndrome||2,240,000||4,000,000–10,000,000|
|729.1||Myalgia and myositis, unspecified||1,807,000||5,000,000|
|710.0||Systemic lupus erythematosus||280,000||161,000–322,000|
|446.5||Giant cell arteritis||94,800||228,000|
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. Sacks 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. Sacks, Luo, Helmick.
Acquisition of data. Luo, Helmick.
Analysis and interpretation of data. Sacks, Luo, Helmick.