Abstract
Objectives
To evaluate definitions of radiographic hip osteoarthritis (RHOA) for use in longitudinal epidemiologic studies of disease incidence in women.
Methods
We studied 5,839 women from the Study of Osteoporotic Fractures who had had serial pelvic radiographs obtained (mean of 8.3 years apart) and who were followed up (mean followup 7.1 years from the time of the second radiograph) for evaluation of clinical outcomes. Definitions of RHOA were assessed for construct validity (association with symptoms and signs at the time of the second radiograph) and predictive validity (association with total hip replacement [THR] and signs and symptoms a mean of 7.1 years later). Odds ratios (ORs) and 95% confidence intervals were calculated to assess the strength of association using logistic regression.
Results
The cumulative incidence of RHOA ranged from 2.2% to 11.7%. All definitions displayed significant construct validity; the most consistent was found for composite definitions that required the concurrent presence of 2 or more individual radiographic features and definitions based on stringent criteria for joint space narrowing. All definitions except minimum joint space ≤2.5 mm displayed consistent predictive validity. Composite definitions had the strongest associations with THR (OR 10.5–18.5) and hip pain (OR 2.6–2.9). The hips identified as having OA by each definition varied, with especially small overlap between findings using definitions based on osteophytes and those using definitions based on joint space narrowing alone.
Conclusions
Most definitions of incident RHOA display good construct and predictive validity. Composite definitions have the best overall performance, and definitions requiring the presence of both osteophytes (in particular, femoral osteophytes) and joint space narrowing would be recommended for most epidemiologic and genetic studies.
Osteoarthritis (OA) of the hip leads to substantial disability in the elderly and is responsible for significant medical costs, accounting for a high proportion of total hip replacements (THRs) (1–3). The reported estimated prevalence of radiographic hip OA (RHOA) ranges from 4% to 25% among white adults over 55 years of age (4–6). The large variation in prevalence estimates may be explained by lack of standardization of definitions, in addition to a true variation in prevalence.
In order to describe the prevalence and incidence of RHOA and to identify risk factors, standard definitions of RHOA for use in epidemiologic studies are needed. Despite the limitations of radiography in detecting early and nonosseous abnormalities, it is a practical and reliable means of assessing the frequency of OA in populations. Several approaches have been used to define RHOA. The first definition proposed was based on an atlas of radiographic examples (7) that defined 4 grades of disease and placed emphasis on osteophytes.
Radiographic definitions of hip OA based on individual radiographic features have also been developed, and have been utilized in epidemiologic studies of community-based cohorts. In this approach, osteophytes, joint space narrowing (JSN), subchondral sclerosis, cysts, and femoral head deformity are scored separately, using either an atlas of standard radiographic examples (8,9) or qualitative assessments (10). These individual radiographic features are used to produce a definition of OA based on the presence of a single feature or the combined presence of multiple features (8–10). A single measurement of minimum joint space (MJS) has also been proposed as a reproducible method for defining hip OA (10).
All of these definitions of hip OA were produced and validated for classifying subjects with prevalent hip disease, but they have been used in epidemiologic studies to estimate the incidence of hip OA, with incidence defined as the development of the prevalence criteria in individuals in whom the findings were previously absent (11–13). However, none of these definitions have been validated for this latter purpose in a prospective population-based study. Any definition of incident RHOA should be reproducible and should also be clinically relevant, i.e., associated with the clinical features of hip OA, including pain and limited range of motion. The present study was undertaken to validate and compare the performance of alternative definitions of RHOA in defining incident disease in a prospectively studied cohort of white women age ≥65 years.
SUBJECTS AND METHODS
Subjects
Subjects were participants in the Study of Osteoporotic Fractures, a multicenter cohort study of risk factors for osteoporosis and fracture in 9,704 women. Participants were age 65 years or older at the baseline examination and were recruited from population-based listings in 4 areas of the US, as previously described (14). Criteria for exclusion included nonwhite race, inability to ambulate, and bilateral THR.
At the baseline examination, a supine anteroposterior (AP) radiograph of the pelvis was obtained in all subjects. The fifth biennial visit occurred a mean ± SD of 8.3 ± 0.4 years after the baseline visit and was attended by 61% of the women (73% of surviving women) in the original study population. An additional 467 radiographs were obtained with a portable x-ray machine at the subject’s residence (those unable to come to the clinic). Evaluable pelvic radiographs from both baseline and followup visits were available from 5,987 women. For this analysis, we excluded 148 women who had one or more of the following: rheumatoid arthritis (self-reported physician diagnosis plus findings on baseline hand radiography), previous hip fracture or surgery, or findings of Paget’s disease on pelvic radiography. Thus, a total of 5,839 women were included in the analysis. The eighth biennial examination occurred a mean ± SD of 7.1 ± 0.58 years after the fifth examination and was attended by 71% of the women who attended the fifth examination.
Radiographic methods
Supine AP radiographs of the pelvis were obtained at baseline and followup using an identical protocol, i.e., a film-to-focus distance of 40 inches, with the hips in 15–30 degrees of internal rotation and the beam centered on the symphysis pubis (15). Pelvic radiographs were read for individual radiographic features of hip OA, using an atlas (8,9). Each hip was rated for JSN at 2 locations: lateral (from a point perpendicular to the femoral head to the lateral margin of the acetabular roof, excluding osteophytes) and medial (from a point perpendicular to the femoral head to the point where the medial continuation of the acetabular roof becomes indistinct). We also measured the MJS (9,10,15). JSN was rated on a scale ranging from 0 to 4. Osteophytes were assessed at 4 locations: lateral femoral, lateral acetabular, inferior femoral, and inferior acetabular. Osteophytes, subchondral sclerosis, cysts, and femoral head deformity were rated on a scale ranging from 0 to 3. The maximum JSN and the maximum osteophyte scores for each hip were calculated.
Radiograph pairs were read side by side. Readers were blinded with regard to order by masking of dates and identifiers, and the order of the baseline and followup radiographs randomly assigned. An initial reading was performed by a primary reader (NEL), and radiograph pairs designated as showing definite JSN or definite osteophytes (grade ≥2 for individual radiographic feature in any location) on either radiograph of the pair were reevaluated by readers (NEL, MCH) to reach a consensus score. The results from the initial reading were available to the readers. A consensus reading was performed on 21% of the radiograph pairs (9,10,15). A random sample of 178 radiograph pairs was also scored under blinded conditions by the second reader, to assess interreader reliability.
Definitions of incident OA
Incident RHOA was defined as fulfillment of the chosen definition at the fifth biennial visit in a hip that did not fulfill this definition at the baseline examination. We considered 2 definitions of RHOA that were based on osteophytes alone, i.e., femoral or acetabular osteophytes grade ≥2 and a new definition, femoral head osteophytes grade ≥2. We also assessed a new JSN-based definition, i.e., grade ≥2 lateral narrowing or grade ≥3 medial narrowing, as well as the definitions MJS ≤1.5 mm, MJS ≤2.0 mm, and MJS ≤2.5 mm (10).
For composite definitions, we considered JSN and osteophytes of grade ≥2 in any location to be definite findings, while other features required a grade of ≥1. We used a modified Croft definition (10,15), a summary grade based on a combination of JSN, osteophytes, subchondral sclerosis, cyst formation, and deformity of the femoral head. We considered 2 different cut points: 1) a score of 2 or greater, which required the presence of either definite JSN or osteophytes in any location, plus at least 1 other radiographic feature, and 2) a score of 3 or greater, which required the presence of definite osteophytes or JSN, plus at least 2 other radiographic features. We also assessed a definition that required both definite osteophytes and JSN; this definition is similar to the Kellgren/ Lawrence (K/L) grading (16) used in most epidemiologic studies (17). We also examined 2 novel definitions, using femoral osteophytes grade ≥2 and an MJS of ≤2.5 mm, and any osteophytes grade ≥2 and an MJS of ≤2.5 mm.
Interreader reliability for designation of prevalence at baseline, assessed with the kappa statistic, was good for all definitions used (definite osteophytes [grade ≥2 in any location] 0.71, femoral osteophytes [grade ≥2] 0.79, medial JSN [grade ≥3] or lateral JSN [grade ≥2] 0.91, MJS ≤2.5 mm 0.71, MJS ≤2.0 mm 0.61, MJS ≤1.5 mm 1.00, Croft grade ≥2 0.63, Croft grade ≥3 0.64, definite osteophytes and JSN 0.83, definite osteophytes and MJS ≤2.5 mm 0.83, femoral osteophytes and MJS ≤2.5 0.79).
Construct and predictive validation criteria
We tested the validity of definitions of RHOA that developed between the baseline and fifth examinations by assessing their association with clinical hip OA assessed at the fifth examination. In the absence of a “gold standard” definition of clinical OA, we used clinical measures that are strongly associated with hip OA. These included new hip pain (i.e., not present at the baseline examination) by self-report, change in lower limb disability between the baseline and fifth examinations, pain on internal rotation examination of the hip, and physician-diagnosed hip arthritis at the fifth examination. Since age is the most well-established risk factor for hip OA (18), it was chosen as a risk factor for study of construct validity. The following factors were used for study of predictive validity: THR performed between the fifth and eighth visits, self-report of physician-diagnosed hip arthritis at the sixth visit, change in lower limb disability between the fifth and eighth visits, and hip pain at the eighth visit.
Prevalent hip pain in the year prior to baseline and prior to visit 5 was assessed with the question, “Have you had pain in or around the hip joint, including the buttock, groin, or either side of the upper thigh, on most days for at least 1 month?” Hip pain at visits 5 and 8 was also evaluated using the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) questionnaire (19). At visit 5 only, if a woman reported pain for any activity, further questions followed regarding the location of the pain: (a) groin/inside leg, (b) outside of leg, (c) front of leg, (d) buttocks, and (e) low back. Analyses performed using cross-sectional data from visit 5 revealed that locations a and c were most strongly correlated with clinical and radiographic features of OA, with the following odds ratios (ORs): femoral osteophytes (grade ≥2) 4.3 (95% confidence interval [95% CI] 3.0–6.2) and 1.9 (95% CI 1.5–2.4), respectively; medial JSN (grade ≥3) or lateral JSN (grade ≥2) 4.0 (95% CI 2.9–5.3) and 1.9 (95% CI 1.6–3.3), respectively; pain on internal rotation 6.6 (95% CI 4.7–9.2) and 6.2 (95% CI 4.9–7.7), respectively. We therefore defined hip pain as pain of moderate or greater severity on any activity and occurring at locations a and c. Since information on location of hip pain at the eighth visit was not available, we also analyzed moderate or worse pain for the individual WOMAC item “pain while walking on a flat surface,” at both visit 5 and visit 8.
We assessed pain on internal rotation of the hip at visit 5, during a passive range of motion examination performed by a trained research nurse (20). Self-reported physician-diagnosed hip arthritis was ascertained at visits 5 and 6. A measure of lower limb disability was administered at baseline, visit 5, and visit 8 (15); we defined an increase in disability as an increase in score of ≥4 points between baseline and visit 5 and between visit 5 and visit 8 (15).
At the eighth visit, participants were asked if they had undergone a THR in the previous 5 years and if so, the year of surgery was ascertained. To exclude hip replacements likely due to fracture, all reported surgery was checked against hip fracture records. THRs that occurred <12 months before, contemporaneously with, or after a hip fracture were excluded.
Statistical analysis
Hips were eligible to develop incident OA by a given definition if they did not meet the criteria for that definition at baseline and had not been replaced between baseline and the fifth visit. Because the prevalence of hip OA at baseline varies by definition, the number of hips eligible to develop incident OA also varies by definition. Women were designated as having developed incident OA by a given definition if they did not meet the criteria for OA in either hip by that definition at baseline and had not undergone THR between baseline and the fifth visit. Person-based outcomes were analyzed for women who did not have radiographic OA in either hip by the given definition at baseline. For analyses of change in lower limb disability between 2 visits, women who had undergone THR between the visits were excluded.
A valid radiographic definition of incident hip OA should be positively associated with clinical features of hip OA and with known risk factors for incident hip OA (construct validity) as well as with the consequences of hip OA (predictive validity). ORs from logistic regression models were used to assess the strength of associations between incident RHOA by each definition and the validation criteria. The radiographic criteria with the largest ORs were considered to be the most valid. When hips were the unit of analysis, we used the general estimating equation to allow for the correlation between findings in each subject’s right hip and left hip. All analyses were performed using SAS version 9.1 (SAS Institute, Cary, NC).
RESULTS
Clinical characteristics of the study cohort at baseline and at the fifth examination are shown in Table 1. The prevalence of RHOA at baseline varied according to the definition used (Table 2). MJS of ≤2.5 mm had the highest prevalence, and a Croft grade of ≥3 the lowest. There was little overlap in the hips identified as having prevalent OA by the different definitions, such that the percentage of hips fulfilling any one of these definitions (excluding MJS ≤2.5 mm) was 10.8%. The incidence of RHOA varied according to the definition used, by ∼6-fold (between 2.2% and 11.7%) over the followup period (mean 8.3 years) (Table 2). With most definitions the incidence was between 2% and 4%, with the exception of MJS ≤2.5 mm and MJS ≤2.0 mm. Person-based cumulative incidence was ∼40–60% higher than that for hips.
Table 1.
Clinical characteristics of the study cohort*
| Visit 1 (n = 5,839) |
Visit 5 (n = 5,839) |
|
|---|---|---|
| Age, mean ± SD years | 70.7 ± 4.7 | 78.6 ± 4.7 |
| Weight, mean ± SD kg | 67.52 ± 12.16 | 66.1 ± 12.8 |
| Height, mean ± SD cm | 159 ± 6 | 158 ± 6.3 |
| Smoking, % current | 8.2 | 4.8 |
| Walks daily for exercise, % | 54 | 41 |
| Hip pain, %† | 35 | 35 |
| Lower limb disability score (range 0–15), mean ± SD | 0.84 ± 1.8 | 2.4 ± 4.1 |
| Hip pain on internal rotation, no. (%)‡ | 438 (7.5) | 658 (11.3) |
| Self-reported hip OA, no. (%) | NA | 971 (16.6) |
OA = osteoarthritis; NA = not applicable.
Assessed with the question “Have you had pain in or around either hip joint, including buttock, groin, or either side of the upper thigh, on most days for at least 1 month?”.
Assessed at visit 2 and visit 5 (mean 8.3 years later).
Table 2.
Baseline prevalence and incidence of radiographic hip OA by each definition*
| Prevalence |
Incidence |
|||
|---|---|---|---|---|
| Definition | Hips | Women | Hips | Women |
| 1. Osteophyte grade ≥2 in any location | 543 (4.8) | 441 (7.6) | 446 (4.3) | 351 (6.5) |
| 2. Femoral osteophyte grade ≥2 | 241 (2.1) | 214 (3.7) | 387 (3.5) | 301 (5.4) |
| 3. JSN grade ≥2 lateral or ≥3 medial | 225 (2.0) | 191 (3.3) | 441 (4.0) | 360 (6.4) |
| 4. MJS ≤1.5 mm | 206 (1.8) | 174 (3.0) | 385 (3.4) | 322 (5.7) |
| 5. MJS ≤2.0 mm | 707 (6.2) | 547 (9.4) | 632 (5.9) | 471 (8.9) |
| 6. MJS ≤2.5 mm | 2,352 (20.6) | 1,649 (28.2) | 1,061 (11.7) | 680 (16.2) |
| 7. Croft grade ≥2 (2 or more IRFs)† | 374 (3.3) | 312 (5.3) | 421 (3.8) | 318 (5.8) |
| 8. Croft grade ≥3 (3 or more IRFs)† | 118 (1.0) | 105 (1.8) | 250 (2.2) | 208 (3.6) |
| 9. Osteophyte and JSN grade ≥2 in any location (K/L grade ≥2) | 169 (1.5) | 147 (2.5) | 344 (3.1) | 281 (4.9) |
| 10. Osteophyte in any location and MJS ≤2.5 mm | 196 (1.7) | 172 (2.9) | 344 (3.1) | 283 (5.0) |
| 11. Femoral osteophyte and MJS ≤2.5 mm | 136 (1.2) | 120 (2.1) | 292 (2.6) | 244 (4.3) |
Values are the number (%). A total of 11,392 hips and 5,839 women were studied; the baseline prevalence of radiographic hip osteoarthritis (OA), and therefore the number of hips eligible to develop incident OA, varied by definition. MJS = minimum joint space; K/L = Kellgren/Lawrence.
The Croft grade is based on a combination of joint space narrowing (JSN), osteophytes, subchondral sclerosis, cyst formation, and deformity of the femoral head. A score of ≥2 requires the presence of either definite JSN or osteophytes in any location, plus at least 1 other individual radiographic feature (IRF); a score of ≥3 requires the presence of definite osteophytes or JSN, plus at least 2 other IRFs.
There was little overlap in the hips identified as developing incident OA by the different definitions; 12.0% of hips met one or more of the incidence definitions (excluding MJS ≤2.5 mm). Of those hips identified as developing incident hip OA as defined using JSN alone (medial JSN ≥3 or lateral JSN ≥2), only 27% would have been identified using osteophytes in any location as the definition; conversely, only 26% of hips identified based on osteophytes alone were also identified using JSN. A similar lack of overlap was seen between an osteophyte-based definition and MJS ≤1.5 mm. The agreement between composite definitions was much greater (51–84% hips identified as developing incident OA overlapped by any pair of composite definitions), as was the agreement between composite definitions and those based on either JSN alone (29–51% overlap) or osteophytes alone (34–64% overlap).
Construct validity
All of the definitions were significantly associated with moderate or more than moderate pain during activity in the previous 30 days at the fifth examination (Table 3), with ORs ranging from 2.4 (for MJS ≤2.5 mm) to 6.5 (for femoral osteophyte and MJS ≤2.5 mm). All definitions were also significantly associated with pain on internal rotation, with the lowest OR (1.4) observed for MJS ≤2.5 mm and the highest for the composite definitions, which were associated with an ∼3-fold increased odds of pain. An increase in lower limb disability between the first and fifth examinations was associated only with definitions based on JSN or MJS. Self-reported physician-diagnosed hip arthritis in the 2 years before the fifth examination was significantly associated with all of the definitions except MJS ≤2.5 mm, with the highest ORs for MJS ≤1.5 mm and Croft grade ≥3. All of the definitions with the exception of Croft grade ≥3 were significantly associated with increasing age, but the strongest associations were for those based on JSN or MJS.
Table 3.
Construct validity: associations of incident radiographic hip OA with clinical features of OA at visit 5*
| Definition | At least moderate pain in groin or front of leg with any activity (WOMAC), OR (95% CI)† |
At least moderate pain on walking (WOMAC), OR (95% CI)† |
Pain on internal rotation, OR (95% CI) |
Self-reported physician-diagnosed hip arthritis, OR (95% CI) |
Change in lower limb disability (from baseline), OR (95% CI) |
10-year age increase, OR (95% CI) |
|---|---|---|---|---|---|---|
| 1. Osteophyte grade ≥2 in any location | 4.1 (2.4–6.9) | 2.0 (1.3–3.1) | 2.5 (1.9–3.3) | 1.5 (1.2–2.0) | 1.0 (0.8–1.4) | 1.3 (1.1–1.7) |
| 2. Femoral osteophyte grade ≥2 | 5.0 (3.0–8.6) | 2.4 (1.5–3.7) | 2.7 (2.0–3.6) | 1.6 (1.2–2.1) | 1.0 (0.7–1.3) | 1.3 (1.0–1.7) |
| 3. JSN grade ≥2 lateral or ≥3 medial | 4.8 (2.3–8.0) | 3.3 (2.3–4.8) | 2.6 (1.9–3.4) | 2.0 (1.5–2.5) | 2.2 (1.7–2.8) | 1.8 (1.4–2.2) |
| 4. MJS ≤1.5 mm | 5.2 (3.0–9.1) | 3.5 (2.3–5.2) | 2.7 (2.0–3.7) | 2.5 (1.7–3.2) | 1.9 (1.5–2.5) | 1.6 (1.3–2.0) |
| 5. MJS ≤2.0 mm | 3.2 (1.9–5.5) | 2.4 (1.7–3.4) | 1.9 (1.4–2.5) | 1.6 (1.2–2.0) | 1.7 (1.3–2.1) | 1.5 (1.3–1.8) |
| 6. MJS ≤2.5 mm | 2.4 (1.5–3.9) | 1.6 (1.1–2.2) | 1.4 (1.0–1.8) | 1.2 (0.9–1.4) | 1.4 (1.1–1.7) | 1.4 (1.1–1.6) |
| 7. Croft grade ≥2 (2 or more IRFs)‡ | 3.3 (1.9–6.0) | 2.3 (1.5–3.5) | 2.4 (1.8–3.2) | 1.6 (1.2–2.1) | 0.9 (0.7–1.3) | 1.4 (1.1–1.8) |
| 8. Croft grade ≥3 (3 or more IRFs)‡ | 6.4 (3.5–11.5) | 3.2 (2.0–5.3) | 2.4 (1.7–3.5) | 2.5 (1.9–3.4) | 0.9 (0.6–1.4) | 1.2 (0.9–1.6) |
| 9. Osteophyte and JSN grade ≥2 in any location (K/L grade ≥2) | 5.7 (3.2–9.8) | 2.3 (1.4–3.7) | 2.9 (3.0–3.9) | 1.9 (1.4–2.5) | 1.1 (0.8–1.5) | 1.5 (1.2–2.0) |
| 10. Osteophyte in any location and MJS ≤2.5 mm | 5.6 (3.3–9.7) | 2.6 (1.6–4.1) | 3.1 (2.3–4.2) | 1.9 (1.5–2.6) | 1.2 (0.9–1.6) | 1.4 (1.1–1.8) |
| 11. Femoral osteophyte and MJS ≤2.5 mm | 6.5 (3.7–11.3) | 2.9 (1.8–4.7) | 3.4 (2.5–4.7) | 2.0 (1.5–2.7) | 1.6 (0.9–2.6) | 1.5 (1.1–1.9) |
Incident radiographic hip OA was defined as the percent of eligible hips in which radiographic OA developed between visit 1 and visit 5. WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index; OR = odds ratio; 95% CI = 95% confidence interval (see Table 2 for other definitions).
Excluding participants with pain at baseline.
See Table 2 for explanation of the Croft grade.
Since incident femoral osteophytes displayed somewhat better construct validity than incident osteophytes in any location, we tested the new appearance of acetabular osteophytes in hips that did not develop femoral osteophytes. These were not significantly related to any of the validation parameters and had point estimates that suggested no association (data not shown), with the exception of an association with pain on internal rotation (OR 1.9 [95% CI 1.1–3.4]).
Predictive validity
Ninety THRs (excluding hip fracture related) that occurred after visit 5 were reported by participants with radiographic data from baseline and visit 5. All definitions predicted THR, with ORs ranging from 3.9 to 18.5 (Table 4). The strongest predictors were Croft grade ≥3 (OR 18.5) and definitions requiring both osteophytes and decreased joint space. Definitions based on either osteophytes or JSN alone were weaker predictors, with ORs ranging from 6 to 11 with the exception of MJS ≤2.5 mm, which had an OR of 3.9. The sensitivity and specificity for THR 5 years after the fifth visit are shown in Table 5.
Table 4.
Predictive validity: associations of incident radiographic hip OA with total hip replacement and clinical features of OA assessed after visit 5*
| Definition | Total hip replacement (after visit 5), OR (95% CI) |
At least moderate pain on walking (WOMAC) (visit 8), OR (95% CI)† |
Self-reported physician-diagnosed hip arthritis (visit 6), OR (95% CI) |
Change in lower limb disability (visit 5 to visit 8), OR (95% CI) |
|---|---|---|---|---|
| 1. Osteophyte grade ≥2 in any location | 9.9 (6.0–16.3) | 2.5 (1.4–4.4) | 1.7 (1.1–2.6) | 2.0 (1.4–2.8) |
| 2. Femoral osteophyte grade ≥2 | 11.2 (6.7–18.6) | 2.8 (1.5–5.1) | 2.0 (1.3–3.0) | 2.0 (1.4–2.8) |
| 3. JSN grade ≥2 lateral or ≥3 medial | 8.2 (4.6–14.6) | 1.8 (0.9–3.6) | 2.7 (1.9–4.0) | 2.0 (1.4–2.9) |
| 4. MJS ≤1.5 mm | 10.3 (6.0–17.8) | 1.4 (0.6–3.1) | 2.9 (2.0–4.3) | 1.8 (1.2–2.5) |
| 5. MJS ≤2.0 mm | 6.4 (3.8–11.0) | 1.0 (0.5–2.1) | 2.3 (1.6–3.2) | 1.8 (1.3–2.4) |
| 6. MJS ≤2.5 mm | 3.9 (2.3–6.7) | 0.9 (0.5–1.5) | 1.2 (0.9–1.6) | 1.6 (1.2–2.1) |
| 7. Croft grade ≥2 (2 or more IRFs)‡ | 10.5 (6.3–17.6) | 2.5 (1.4–4.5) | 1.8 (1.2–2.8) | 1.9 (1.3–2.7) |
| 8. Croft grade ≥3 (3 or more IRFs)‡ | 18.5 (11.1–30.9) | 2.6 (1.1–6.0) | 3.2 (2.0–5.0) | 2.1 (1.3–3.2) |
| 9. Osteophyte and JSN grade ≥2 in any location (K/L grade ≥2) | 13.8 (8.3–23.0) | 2.8 (1.4–5.6) | 2.3 (1.5–3.5) | 2.3 (1.5–3.3) |
| 10. Osteophyte in any location and MJS ≤2.5 mm | 15.0 (9.1–24.7) | 2.7 (1.3–5.4) | 2.1 (1.4–3.3) | 2.0 (1.4–2.9) |
| 11. Femoral osteophyte and MJS ≤2.5 mm | 16.4 (9.8–27.3) | 2.9 (1.8–4.7) | 2.6 (1.3–5.3) | 2.2 (1.4–3.4) |
Table 5.
Predictive validity: sensitivity and specificity of incident radiographic OA for total hip replacement after visit 5*
| Definition | Sensitivity | Specificity |
|---|---|---|
| 1. Osteophyte grade ≥2 in any location | 0.34146 | 0.96252 |
| 2. Femoral osteophyte grade ≥2 | 0.30233 | 0.96978 |
| 3. JSN grade ≥2 lateral or ≥3 medial | 0.30233 | 0.96941 |
| 4. MJS ≤1.5 mm | 0.25000 | 0.97372 |
| 5. MJS ≤2.0 mm | 0.35714 | 0.94980 |
| 6. MJS ≤2.5 mm | 0.45946 | 0.89650 |
| 7. Croft grade ≥2 (2 or more IRFs) | 0.30000 | 0.96732 |
| 8. Croft grade ≥3 (3 or more IRFs) | 0.31818 | 0.98136 |
| 9. Osteophyte and JSN grade ≥2 in any location (K/L grade ≥2) | 0.31818 | 0.97513 |
| 10. Osteophyte in any location and MJS ≤2.5 mm | 0.34884 | 0.97477 |
| 11. Femoral osteophyte and MJS ≤2.5 mm | 0.31111 | 0.97892 |
See Table 2 for definitions and for explanation of the Croft grade.
The strongest predictors of walking pain at visit 8 were definitions based on osteophytes alone or a composite of individual radiographic features, while those based on joint space were not significantly associated with this measure of hip pain. A physician diagnosis of hip arthritis at visit 6 was associated with all definitions except MJS ≤2.5 mm. The strongest predictors were the composite measures, especially Croft grade ≥3, and MJS ≤1.5 mm. Worsening of lower limb disability was associated with all definitions except MJS ≤2.5 mm.
DISCUSSION
In this large cohort study of elderly women, we have demonstrated strong associations of a number of different definitions of incident RHOA with symptoms of hip OA, the development of disability, and risk of THR. This is the first study to provide evidence of the validity of both commonly used, as well as several novel, definitions of incident RHOA. Composite definitions that are based on the concurrent presence of more than one individual radiographic feature had the best overall construct and predictive validity, although definitions based on stringent criteria for joint space narrowing and osteophytes alone, and especially femoral osteophytes, also demonstrated good validity. Use of a definition based on either joint space or osteophytes alone can lead to substantial discordance in hips identified as developing incident RHOA, compared with a definition based on the other feature alone.
RHOA according to the definitions we considered, with the exception of the one based on MJS ≤2.5 mm, had 8-year cumulative incidences ranging from ∼2% to 6% of eligible hips (∼4–8% of women with 2 hips at risk). Our findings are consistent with the limited existing data on incident RHOA. Reijman et al studied 2,852 men and women (mean age 66 years), defining hip OA as a K/L grade of ≥2 (17). They reported a cumulative incidence of 3.9% over a mean of 6.6 years of followup (13). In a study of incident clinically diagnosed hip OA in a health maintenance organization population of women ages 60–89 years, Oliveria et al found an annual incidence of 0.4% (11). This is close to the lower range of incidence in our study, as would be expected when comparing incident RHOA with newly diagnosed clinical hip OA.
The composite definitions performed well across all of the construct validation measures; their performance was as good as or better than that of the definitions relying on joint space only, while osteophyte-based definitions performed somewhat less well. The definitions based on atlas grades of JSN or on MJS were the only ones that were significantly associated with worsening lower limb disability. An important reason to ascertain incident RHOA is to predict future pain, disability, and need for surgery, and as such, the predictive model is the most appropriate one. All definitions were strongly associated with THR following the visit at which incident RHOA was assessed; the composite definitions, and the Croft grade of ≥3 in particular, displayed the strongest associations. However, THRs were uncommon, occurring in only 1.5% of hips in the women followed up until visit 8. In women who did not have a THR, all definitions were also significantly related to pain, physician-diagnosed OA, and worsening lower limb disability after the followup radiograph.
A definition of incident RHOA should identify RHOA over a range of severity, but still have strong construct and predictive validity. Our results are consistent with there being a trade-off between incidence rate according to a given definition and the strength of association with validation parameters, with the least common end points having stronger associations, as indicated by larger ORs. These definitions will identify more severe radiographic findings and will be associated with higher specificity; however, this will be traded-off by reduced sensitivity. Thus, for association studies, the definitions with lower incidence rates may perform as well as or better than those with higher rates.
We found that composite definitions that required the concurrent presence of 2 or more individual radiographic features using either the atlas-standardized assessment or measurement of MJS displayed the greatest validity. Croft grade ≥3 had the greatest validity, but also the lowest incidence. Composite definitions requiring concurrent JSN and osteophytes (which we consider similar to a hip K/L grade of ≥2) performed nearly as well, but with an incidence over 8 years that was 20–40% higher, suggesting that this approach identifies additional cases with only slightly less severe disease. An MJS of ≤2.5 mm and osteophytes of grade ≥2 also performed well. Not all investigators interpret a hip K/L grade of 2 as requiring both osteophytes and JSN (21), and that definition may thus vary between studies. We therefore recommend that composite definitions of RHOA use atlas-standardized assessments of osteophytes and either atlas-assessed JSN or measured MJS.
Our results support the notion that incident RHOA can be defined in terms of atlas grades of JSN alone or MJS alone, but only when a relatively stringent cut point is used. Because of the high frequency of medial JSN in our sample (15), we used a definition that required grade 3 medial narrowing or grade 2 lateral narrowing, which performed well in this population. Use of an MJS cut point of 2.5 mm, proposed as a definition of prevalent OA (22), produced an incidence that was too high to be credible and only weakly associated with validation parameters, and could not be recommended. This cut point was selected and validated in studies using intravenous pyelogram radiographs in men (17), and there were also differences in the film acquisition between studies (most notably, radiographs in the present study were obtained with the subject in a supine position and were centered on the symphysis pubis while those in other studies were weight bearing and centered on the umbilicus, which may increase MJS by up to 10% [23]). In a previous report, it was suggested that a threshold of 1.7 mm may be more appropriate for assessing MJS in women (24).
Osteophytes are included in many definitions of RHOA (25), but these definitions generally do not distinguish between femoral and acetabular osteophytes. In our study, RHOA according to definitions based on osteophytes alone had an incidence similar to that of RHOA according to those based on JSN alone, and only slightly less robust associations with validation criteria. We have previously shown that an isolated finding of acetabular osteophytes did not demonstrate structural progression over 8 years (15), possibly because the anatomy of the lateral and inferior acetabular region of the hip makes small osteophytes difficult to identify. In this study, new acetabular osteophytes in the absence of femoral osteophytes were not associated with most validation parameters. Restricting the definition to femoral osteophytes improved performance to some degree, with only modest differences in frequency compared with a definition based on the occurrence of any new osteophytes. Thus, we recommend using femoral osteophytes to define OA when osteophytes are the only feature required in the definition.
The different definitions we tested identify different groups of hips with new disease, especially comparing those based on joint space or osteophytes alone. Whether this is important depends on the aims of the study: if the study aim is to describe the incidence of RHOA, the choice does not appear to have a major effect on the results since the range of incidence estimates is fairly narrow except for those obtained with the definition MJS ≤2.5 mm. If, however, the study aim is to identify risk factors for incident RHOA, the choice of definition may have a major impact on the results, because risk factors may differ between different phenotypes of RHOA (i.e., hypertrophic and atrophic) (12,26,27). Recent work also suggests that genetic susceptibility loci may be associated preferentially with changes predominantly in JSN (28) or osteophytosis (28,29). Thus, a composite measure such as the presence of both osteophytes and JSN, which displays a greater overlap with the other definitions, would appear to be a good option for identifying the complete spectrum of risk factors for hip OA.
This study has important strengths. It is the first to evaluate the validity of definitions of incident RHOA using a large cohort with a long duration of followup, including a mean of 7 years after the second radiograph, allowing the evaluation of predictive validity.
The study also has several potential limitations. The cohort comprised elderly white women, and the results cannot be directly applied to men or women of other racial groups. Because the study population was relatively healthy at the time of enrollment, baseline prevalences of RHOA may be lower than in the general population. However, the incidence rates we identified are similar to those in other population studies. The pelvic radiographs obtained in our study were non– weight bearing, and some studies suggest that this may result in lower sensitivity to reduced joint space. However, they were acquired with the feet in 15–20 degrees of internal rotation, which may mitigate the effect of non–weight bearing (30). We cannot exclude the possibility that some THRs may have been performed for primary avascular necrosis; however, a very large Norwegian database containing records of 53,698 primary THRs showed that only 0.8% were performed for avascular necrosis (31), indicating that this would be unlikely to have had a significant impact on our results. Finally, the present study formally validates definitions of incident hip OA using existing prevalent definitions in addition to novel definitions. Ideally, studies exploring both incident and prevalent hip OA should use the same definition; however, the novel definitions described need to be formally validated for prevalent hip OA.
In conclusion, we have demonstrated that a number of definitions of incident RHOA display good construct and predictive validity for clinical hip osteoarthritis. Based on their overall superior performance, composite definitions that require the presence of both osteophytes (in particular, femoral osteophytes) and loss of joint space are recommended for most epidemiologic and genetic studies.
Footnotes
AUTHOR CONTRIBUTIONS
Dr. Arden 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. Lane, Nevitt.
Acquisition of data. Lane, Hochberg, Nevitt.
Analysis and interpretation of data. Arden, Lane, Parimi, Javaid, Lui, Hochberg, Nevitt.
Manuscript preparation. Arden, Lane, Javaid, Lui, Hochberg, Nevitt.
Statistical analysis. Arden, Parimi, Lui.
REFERENCES
- 1.Furnes O, Lie SA, Espehaug B, Vollset SE, Engesaeter LB, Havelin LI. Hip disease and the prognosis of total hip replacements: a review of 53,698 primary total hip replacements reported to the Norwegian Arthroplasty Register 1987–99. J Bone Joint Surg Br. 2001;83:579–586. doi: 10.1302/0301-620x.83b4.11223. [DOI] [PubMed] [Google Scholar]
- 2.Frankel S, Eachus J, Pearson N, Greenwood R, Chan P, Peters TJ, et al. Population requirement for primary hip-replacement surgery: a cross-sectional study. Lancet. 1999;353:1304–1309. doi: 10.1016/S0140-6736(98)06451-4. [DOI] [PubMed] [Google Scholar]
- 3.Kurtz S, Mowat F, Ong K, Chan N, Lau E, Halpern M. Prevalence of primary and revision total hip and knee arthroplasty in the United States from 1990 through 2002. J Bone Joint Surg Am. 2005;87:1487–1497. doi: 10.2106/JBJS.D.02441. [DOI] [PubMed] [Google Scholar]
- 4.Felson DT. Epidemiology of hip and knee osteoarthritis. Epidemiol Rev. 1988;10:1–28. doi: 10.1093/oxfordjournals.epirev.a036019. [DOI] [PubMed] [Google Scholar]
- 5.Lawrence RC, Helmick CG, Arnett FC, Deyo RA, Felson DT, Giannini EH, et al. Estimates of the prevalence of arthritis and selected musculoskeletal disorders in the United States. Arthritis Rheum. 1998;41:778–799. doi: 10.1002/1529-0131(199805)41:5<778::AID-ART4>3.0.CO;2-V. [DOI] [PubMed] [Google Scholar]
- 6.Centers for Disease Control and Prevention. Estimated number of US population with physician-diagnosed osteoarthritis. Atlanta: Centers for Disease Control; 1980. [Google Scholar]
- 7.Kellgren JH, Lawrence JS. Radiological assessment of osteo-arthrosis. Ann Rheum Dis. 1957;16:494–502. doi: 10.1136/ard.16.4.494. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Altman RD, Hochberg M, Murphy WA, Wolfe F, Lequesne M., Jr Atlas of individual radiographic features in osteoarthritis. Osteoarthritis Cartilage. 1995;(3 Suppl A):3–70. [PubMed] [Google Scholar]
- 9.Lane NE, Nevitt MC, Genant HK, Hochberg MC. Reliability of new indices of radiographic osteoarthritis of the hand and hip and lumbar disc degeneration. J Rheumatol. 1993;20:1911–1918. [PubMed] [Google Scholar]
- 10.Croft P, Cooper C, Wickham C, Coggon D. Defining osteoarthritis of the hip for epidemiologic studies. Am J Epidemiol. 1990;132:514–522. doi: 10.1093/oxfordjournals.aje.a115687. [DOI] [PubMed] [Google Scholar]
- 11.Oliveria SA, Felson DT, Reed JI, Cirillo PA, Walker AM. Incidence of symptomatic hand, hip, and knee osteoarthritis among patients in a health maintenance organization. Arthritis Rheum. 1995;38:1134–1141. doi: 10.1002/art.1780380817. [DOI] [PubMed] [Google Scholar]
- 12.Lane NE, Gore LR, Cummings SR, Hochberg MC, Scott JC, Williams EN, et al. for the Study of Osteoporotic Fractures Research Group. Serum vitamin D levels and incident changes of radiographic hip osteoarthritis: a longitudinal study. Arthritis Rheum. 1999;42:854–860. doi: 10.1002/1529-0131(199905)42:5<854::AID-ANR3>3.0.CO;2-I. [DOI] [PubMed] [Google Scholar]
- 13.Reijman M, Pols HA, Bergink AP, Hazes JM, Belo JN, Lievense AM, et al. Body mass index associated with onset and progression of osteoarthritis of the knee but not of the hip: the Rotterdam Study. Ann Rheum Dis. 2007;66:158–162. doi: 10.1136/ard.2006.053538. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Cummings SR, Black DM, Nevitt MC, Browner WS, Cauley JA, Genant HK, et al. the Study of Osteoporotic Fractures Research Group. Appendicular bone density and age predict hip fracture in women. JAMA. 1990;263:665–668. [PubMed] [Google Scholar]
- 15.Lane NE, Nevitt MC, Hochberg MC, Hung YY, Palermo L. Progression of radiographic hip osteoarthritis over eight years in a community sample of elderly white women. Arthritis Rheum. 2004;50:1477–1486. doi: 10.1002/art.20213. [DOI] [PubMed] [Google Scholar]
- 16.Kellgren JH, Lawrence JS. Radiological assessment of osteoar-throsis. Ann Rheum Dis. 1957;16:494–501. doi: 10.1136/ard.16.4.494. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Reijman M, Hazes JM, Pols HA, Bernsen RM, Koes BW, Bierma-Zeinstra SM. Validity and reliability of three definitions of hip osteoarthritis: cross sectional and longitudinal approach. Ann Rheum Dis. 2004;63:1427–1433. doi: 10.1136/ard.2003.016477. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Hochberg MC. Do risk factors for incident hip osteoarthritis (OA) differ from those for progression of hip OA? J Rheumatol Suppl. 2004;70:6–9. [PubMed] [Google Scholar]
- 19.Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW. Validation study of WOMAC: a health status instrument for measuring clinically important patient relevant outcomes to anti-rheumatic drug therapy in patients with osteoarthritis of the hip or knee. J Rheumatol. 1988;15:1833–1840. [PubMed] [Google Scholar]
- 20.Birrell F, Croft P, Cooper C, Hosie G, Macfarlane G, Silman A. Predicting radiographic hip osteoarthritis from range of movement. Rheumatology (Oxford) 2001;40:506–512. doi: 10.1093/rheumatology/40.5.506. [DOI] [PubMed] [Google Scholar]
- 21.Spector TD, Cooper C. Radiographic assessment of osteoarthritis in population studies: whither Kellgren and Lawrence? [editorial] Osteoarthritis Cartilage. 1993;1:203–206. doi: 10.1016/s1063-4584(05)80325-5. [DOI] [PubMed] [Google Scholar]
- 22.Borenstein D. Epidemiology, etiology, diagnostic evaluation, and treatment of low back pain. Curr Opin Rheumatol. 1998;10:104–109. doi: 10.1097/00002281-199803000-00003. [DOI] [PubMed] [Google Scholar]
- 23.Auleley GR, Rousselin B, Ayral X, Edouard-Noel R, Dougados M, Ravaud P. Osteoarthritis of the hip: agreement between joint space width measurements on standing and supine conventional radiographs. Ann Rheum Dis. 1998;57:519–523. doi: 10.1136/ard.57.9.519. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Nevitt MC, Xu L, Zhang Y, Lui LY, Yu W, Lane NE, et al. Very low prevalence of hip osteoarthritis among Chinese elderly in Beijing, China, compared with whites in the United States: the Beijing Osteoarthritis Study. Arthritis Rheum. 2002;46:1773–1779. doi: 10.1002/art.10332. [DOI] [PubMed] [Google Scholar]
- 25.Reijman M, Hazes JM, Koes BW, Verhagen AP, Bierma-Zeinstra SM. Validity, reliability and applicability of seven definitions of hip osteoarthritis used in epidemiological studies: a systematic appraisal [review] Ann Rheum Dis. 2004;63:226–232. doi: 10.1136/ard.2003.010348. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Nevitt MC, Lane NE, Scott JC, Hochberg MC, Pressman AR, Genant HK, et al. the Study of Osteoporotic Fractures Research Group. Radiographic osteoarthritis of the hip and bone mineral density. Arthritis Rheum. 1995;38:907–916. doi: 10.1002/art.1780380706. [DOI] [PubMed] [Google Scholar]
- 27.Felson DT, Lawrence RC, Dieppe PA, Hirsch R, Helmick CG, Jordan JM, et al. Osteoarthritis: new insights. Part 1: the disease and its risk factors. Ann Intern Med. 2000;133:635–646. doi: 10.7326/0003-4819-133-8-200010170-00016. [DOI] [PubMed] [Google Scholar]
- 28.Greig C, Spreckley K, Aspinwall R, Gillaspy E, Grant M, Ollier W, et al. Linkage to nodal osteoarthritis: quantitative and qualitative analyses of data from a whole-genome screen identify trait-dependent susceptibility loci. Ann Rheum Dis. 2006;65:1131–1138. doi: 10.1136/ard.2005.048165. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Lane NE, Lian K, Nevitt MC, Zmuda JM, Lui L, Li J, et al. Frizzled-related protein variants are risk factors for hip osteoar-thritis. Arthritis Rheum. 2006;54:1246–1254. doi: 10.1002/art.21673. [DOI] [PubMed] [Google Scholar]
- 30.Auleley GR, Duche A, Drape JL, Dougados M, Ravaud P. Measurement of joint space width in hip osteoarthritis: influence of joint positioning and radiographic procedure. Rheumatology (Oxford) 2001;40:414–419. doi: 10.1093/rheumatology/40.4.414. [DOI] [PubMed] [Google Scholar]
- 31.Furnes O, Lie SA, Espehaug B, Vollset SE, Engesaeter LB, Havelin LI. Hip disease and the prognosis of total hip replacements: a review of 53,698 primary total hip replacements reported to the Norwegian Arthroplasty Register 1987–99. J Bone Joint Surg Br. 2001;83:579–586. doi: 10.1302/0301-620x.83b4.11223. [DOI] [PubMed] [Google Scholar]