Epidemiology of Osteoarthritis

Leena Sharma

Dipali Kapoor

Introduction

Osteoarthritis (OA) is the most common form of arthritis and a leading cause of chronic disability, in large part due to knee and/or hip involvement. The societal burden of OA relates to its pervasive presence. For example, in the Rotterdam study, only 135 of 1040 persons 55 to 65 years of age were free of radiographic OA (definite osteophyte presence or more severe) in the hands, knees, hips, or spine.¹ Not all OA is symptomatic; still, the World Health Organization estimates that OA is a cause of disability in at least 10% of the population over age 60 years² and OA affects the lives of more than 20 million Americans.³ Knee OA alone was as often associated with disability as were heart and chronic lung disease.⁴ Current treatments for OA may improve symptoms but do not delay progression. Progression of OA to advanced and disabling stages is the leading indication for joint replacement.

The increase in the prevalence of symptomatic OA with age, coupled with the inadequacy of symptom-relieving or disease-modifying treatment, contributes to its impact. The number of persons in the U.S. with arthritis is anticipated to rise from 15% of the population (40 million) in 1995 to 18% of the population (59 million) by 2020.³ A better understanding of the factors that contribute to disease and disability in OA is a high priority, especially given the lack of disease-modifying treatment options.

Epidemiologic studies, in addition to incidence and prevalence data, have supplied much of what is known about the natural history of OA and predisposing or protective factors. In addition, epidemiologic investigation has provided information to aid the performance and interpretation of clinical trials; such background information is critical in a disease like OA, which is heterogeneous in its expression and variably progressive.

The following chapter will provide an overview of the areas in which epidemiologic investigation of OA has occurred or has spurred methodologic development: defining OA for study, identifying typical patterns of disease (intra-articular localization, inter-articular joint clustering), developing approaches to assess OA progression, identifying risk factors for OA development and progression, identifying factors that mediate the effect of other factors, and understanding pathogenesis in terms of both anatomic and functional outcomes. This chapter focuses on knee, hip, and hand OA: knee and/or hip OA bear most of the responsibility for the burden of OA; hand OA may also be a source of symptoms, and may be a marker of a systemic predisposition toward OA.

Defining Osteoarthritis

Consensus Definition

Over the twentieth century, the definition of OA has evolved from “hypertrophic arthritis” to the most recent current consensus definition:⁵ “OA diseases are a result of both mechanical and biologic events that destabilize the normal coupling of degradation and synthesis of articular cartilage chondrocytes and extracellular matrix, and subchondral bone. Although they may be initiated by multiple factors, including genetic, developmental, metabolic, and traumatic, OA diseases involve all of the tissues of the diarthrodial joint. Ultimately, OA diseases are manifested by morphologic, biochemical, molecular, and biomechanical changes of both cells and matrix which lead to a softening, fibrillation, ulceration, loss of articular cartilage, sclerosis and eburnation of subchondral bone, osteophytes, and subchondral cysts. When clinically evident, OA diseases are characterized by joint pain, tenderness, limitation of movement, crepitus, occasional effusion, and variable degrees of inflammation without systemic effects.”

Classification of Osteoarthritis

OA is usually classified as primary (idiopathic) or secondary to metabolic conditions, anatomic abnormalities, trauma, or inflammatory arthritis (Table 1-1).

Diagnostic Criteria

Diagnostic criteria have been developed for knee,⁶ hip,⁷ and hand⁸ OA. Recursive partitioning yielded criteria sets with the best combination of sensitivity and specificity (Table 1-2). In the studies in which these criteria were developed, comparison groups were patients with causes of joint pain other than OA. The American College of Rheumatology (ACR) criteria are intended to distinguish OA from other causes of symptoms and are best suited to recruit participants from clinical settings in which a high prevalence of other arthritides or soft tissue conditions and a higher (than the general population) likelihood of having symptomatic OA is expected. Of note, in community-based studies, the ability to distinguish OA from the absence of joint disease is paramount,⁹, ¹⁰^,¹¹ and definitions of OA for epidemiologic study have been developed with this in mind. These definitions are discussed in the following paragraphs.

TABLE 1-1 CLASSIFICATION OF OSTEOARTHRITIS

PRIMARY (IDIOPATHIC)

Peripheral joints

Spine

Apophyseal joints

Intervertebral joints

Subsets

Generalized osteoarthritis

Erosive inflammatory osteoarthritis

Diffuse idiopathic skeletal hyperostosis

Chondromalacia patellae

Hereditary

SECONDARY

Trauma

Acute

Chronic (occupational, sports)

Underlying joint disorders

Local (fracture, infection)

Diffuse (rheumatoid arthritis)

Systemic metabolic or endocrine disorders

Ochronosis (alkaptonuria)

Wilson disease

Hemochromatosis

Kashin-Bek disease

Acromegaly

Hyperparathyroidism

Crystal deposition disease

Calcium pyrophosphate dihydrate (pseudogout)

Basic calcium phosphate (hydroxyapatite-octacalcium phosphate-tricalcium phosphate)

Monosodium urate monohydrate (gout)

Neuropathic disorders (Charcot joints)

Tabes dorsalis

Diabetes mellitus

Intra-articular corticosteroid overuse

Miscellaneous

Bone dysplasia (multiple epiphyseal dysplasia, achondroplasia)

Frostbite

Patterns of Disease

Specific inter- and intra-articular patterns of OA may represent subsets that have distinctive risk factor profiles and disease course, and, in theory, respond differently to treatment.

Knee Osteoarthritis

Unilateral and bilateral knee OA may represent not different subsets as much as different stages within the same subset. Bilateral knee OA is more common than unilateral disease, affecting 5% versus 2%, respectively, of persons 45 to 74 years of age in NHANES I.¹² Having OA in one knee increases the likelihood of having OA in the contralateral knee.¹³, ¹⁴ Among Chingford study participants with unilateral knee OA, 34% developed contralateral OA within 2 years.¹⁵ In a clinic-based study of 63 patients with knee OA, 12 of 13 with unilateral OA at baseline developed contralateral OA over 11 years.¹⁶

Based on x-ray data, it is believed that tibiofemoral OA is more common than patellofemoral OA. In the Framingham cohort, patellofemoral OA was found in 5%, tibiofemoral OA in 23%, and mixed tibiofemoral and patellofemoral OA in 20%.¹⁷ In a community study in the United Kingdom, men with symptomatic OA most often had isolated medial disease (21%, vs. patellofemoral in 11%, mixed in 7%).¹⁸ In women, however, patellofemoral OA was most common (24% vs. medial in 12%, mixed in 6%).

Studies which have examined whether OA is more likely on the right or left side revealed no difference in one study¹⁹ and, in a recent study, a slightly greater prevalence of tibiofemoral OA on the right side.²⁰

Hip Osteoarthritis

In persons with hip OA, bilateral involvement was reported in 35%²¹ and 42%.²² Involvement of one hip increased the likelihood of contralateral hip OA in the Chingford study.¹⁴ Hip OA appears to be equally common on the right and left sides.¹⁹, ²⁰

Superior or lateral involvement is more common than medial involvement. In 6000 patients who had bowel x-rays, 4.7% had hip OA; of these, involvement was lateral in 50% and medial in 24%.²³ In a hospital-based study, Ledingham et al. found superior pole migration in 82%, medial/axial migration in 8%, and an indeterminate pattern in 10%.²² Superomedial and medial/axial patterns were more common in women, and superolateral patterns were more common in men.²²

Hand Osteoarthritis

There is strong evidence for clustering of hand joint involvement in OA. Having OA in either distal interphalangeal (DIP) or proximal interphalangeal (PIP) joints at baseline increased the risk of incident OA in all other hand joints.²⁴ Having thumb base OA at baseline increased the risk of developing metacarpophalangeal (MCP) OA and, to
a lesser extent, DIP and PIP OA.²⁴ After adjusting for age, the risk of hand OA was increased by having contralateral hand OA,²⁵ prevalent OA in one or more joints in the same row,²⁴^,²⁵ or prevalent OA in the same ray.²⁴^,²⁵ DIP OA was more common on the right than on the left side;²⁰ a previous study revealed no differences between dominant and non-dominant hands.²⁶

TABLE 1-2 OSTEOARTHRITIS CLASSIFICATION CRITERIA

Joint	Clinical and Laboratory	Clinical, Laboratory, and Radiographic
Knee	Knee pain AND Crepitus, and morning stiffness ≤30 minutes, and age ≥38 years OR Crepitus, and morning stiffness >30 minutes, and bony enlargement OR No crepitus, and bony enlargement	Knee pain AND Osteophytes OR OA synovial fluid (clear, viscous, WBC <2000/mm³), and morning stiffness ≤30 minutes, and crepitus
Sensitivity Specificity	89% 88%	94% 88%
Hand	1. Hand pain, aching, or stiffness AND 2. Hard tissue enlargement of 2 or more of 10 selected hand joints* AND 3. Metacarpophalangeal swelling in fewer than 2 joints AND 4a. Hard tissue enlargement involving 2 or more distal interphalangeal joints (second and third distal interphalangeal joints may be counted in both 2 and 4a) OR 4b. Deformity of 2 or more of 10 selected hand joints*
Sensitivity Specificity	93% 97%
Hip	1. Hip pain AND 2a. Hip internal rotation <14° AND 2b. ESR ≤15 mm/h (hip flexion ≤115° if no ESR available) OR 3a. Range of motion ≥15° internal rotation AND 3b. Morning stiffness of the hip ≤60 minutes AND 3c. Age >50 years	Hip pain AND At least two of the following: ESR less than 20 mm/hr Radiographic femoral or acetabular osteophytes Radiographic joint space narrowing (superior, axial, or medial)
Sensitivity Specificity	87% 75%	89% 91%
*Second and third distal interphalangeal, second and third proximal interphalangeal, and first carpometacarpal joints. ESR, erythrocyte sedimentation rate.

A study of 53-year-old men and women from a large general population sample revealed evidence of a polyarticular hand OA subset that involved the DIP, PIP, and thumb base joints.²⁷ Clustering was most apparent by row (rather than by ray) and by symmetric involvement of the same joint in both hands. There was clear evidence of clustering in the men as well, and the patterns were indistinguishable between men and women.²⁷

Clustering of Osteoarthritis Involvement of the Knee, Hip, and Hand

Multiple involvement of five joint groups—DIP, PIP, carpometacarpal (CMC), knee, and hip—occurred more frequently than could be expected by chance in the Chingford population.¹⁴ However, the association between contralateral joints was stronger than the associations between different joint groups. Knee and hip OA are each associated with the presence of hand OA.²⁸^,²⁹^,³⁰ When examined within the same population, the link between knee and hand OA appeared stronger.¹⁴^,³¹ In BLSA participants, an association was found between knee OA and DIP OA, PIP OA, and OA in two or more hand joint groups, adjusting for age and BMI (ORs 1.71 to 2.16).²⁸ Of patients who had undergone
meniscectomy, those with hand OA had more frequent and more severe knee OA in both operated and unoperated knees than did those without hand OA, adjusting for sex and age.³²^,³³ The presence of hand OA was associated with a threefold increase in the risk of hip OA.²⁹^,³⁰

Defining Osteoarthritis for Epidemiologic Study

Much effort has been devoted toward developing a definition of OA for epidemiologic study that encapsulates symptoms, disability, and joint pathology. At the heart of the difficulty is the issue that, while there is some correlation between radiographic disease severity and both symptoms and disability, the relationships are not as strong as one would expect.

As noted above, in epidemiologic studies, a key distinction is between OA and the absence of arthritis. For this reason, and the frequency of mild or intermittent symptoms, epidemiologic studies have tended to rely upon radiographic definitions of OA. Symptomatic, radiographic OA has been defined by a radiographic criterion coupled with a positive response to a question, e.g., pain, in that joint, on most days of a month within the preceding year. The use of a definition combining symptom and x-ray criteria reflects a desire to capture persons with clinically significant OA. A potential limitation of this approach is that a subset of persons with OA may have physically limiting disease but self-reported symptoms that fall below the applied “symptomatic” cut-off.

The most widely used system to grade radiographic severity continues to be the Kellgren and Lawrence grading system,³⁴ by which 1 of 5 grades is assigned with the aid of atlas reproductions, according to the following definitions: 0 for normal, 1 for possible osteophytic lipping, 2 for definite osteophytes and possible joint space narrowing, 3 for moderate or multiple osteophytes, definite joint space narrowing, some sclerosis, and possible bony attrition, and 4 for large osteophytes, marked joint space narrowing, severe sclerosis, and definite bony attrition. The K/L system is osteophyte driven; it is unclear how to handle knees with joint space narrowing without osteophytes. Also, the K/L system is limited by incorrect assumptions, including the following: that change in any one feature is linear and constant, and that the relationship between features is constant.³⁵ Most investigators assess individual radiographic features in addition to a global score.

Although x-ray continues to be used heavily, MRI is common in epidemiologic studies, and provides rich opportunities to assess articular cartilage, subarticular bone, menisci, ligaments, and, aided by contrast, synovium. An MRI-based definition of OA has not as yet been established.

Knee Osteoarthritis

The presence of definite osteophytes is the recommended definition for radiographic knee OA.³⁶^,³⁷ Further validating an osteophyte-based definition, tibiofemoral osteophytes predicted cartilage defects on MRI, whether or not radiographic joint space narrowing (as defined by <3 mm) was present, in individuals 49 to 58 years of age.³⁸^,³⁹ In the patellofemoral joint however, osteophytes predicted MRI cartilage defects only in narrowed patellofemoral joints, suggesting that osteophytes alone may not be sufficient to identify cases of patellofemoral OA.³⁹

Hip Osteoarthritis

A definition including joint space width appears to be valid and practical for epidemiologic study of hip OA.⁹^,⁴⁰ In men, an overall grade, minimal joint space width, and thickness of subchondral sclerosis were most predictive of hip pain.⁴⁰ Minimal joint space was best associated with other radiographic features, and measures of joint space were more reproducible than other indices.⁴⁰ However, there are caveats with a joint space width definition: the cut-off for what is a normal joint space width at the hip may differ between ethnic groups and change with age; it is unclear how to handle osteophytes without joint space narrowing; a less stringent cut-off increases sensitivity but sacrifices specificity.⁹ Specificity may be enhanced by requiring at least one other radiographic feature or by using a global system.⁹ Of note, minimum joint space width 2 mm or less was more closely associated (than global radiographic scoring approaches) with hip pain.⁴¹ The effect of using alternative definitions of disease on estimates of prevalence has been demonstrated.⁴²

Hand Osteoarthritis

Defining hand OA is important to advance the investigation of hand OA itself and to document its presence as a marker of a systemic predisposition towards OA. Most epidemiologic studies have relied on the presence of definite osteophytes, or K/L 2. Alternative global radiographic scoring systems were developed.⁴³^,⁴⁴ While Heberden’s node presence and DIP osteophytes had similar sensitivity, the specificity and positive predictive value of radiographic osteophytes was higher for detecting knee, CMC, and PIP OA, and OA in more than two groups of joints.⁴⁵

At present there is no agreement on the best definition of generalized OA; Cooper et al. demonstrated that thresholds could be defined for the number of involved joint groups that distinguished a polyarticular subset of OA; these thresholds varied with age and other factors.¹⁴

Incidence

Knee Osteoarthritis

In the Framingham study (participant mean age 70.8 years), 2% of women per year developed radiographic knee OA, and 1% per year developed symptomatic, radiographic knee OA, versus 1.4% and 0.7% of men, respectively.¹³ In a Dutch population-based study (participant age 46 to 66 years), about 2% of women and 0.8% of men developed radiographic knee OA per year.⁴⁶ In the Goteborg study (participant age 75 years), the incidence of knee OA was 0.9% per year.⁴⁷

Two incidence studies were restricted to patients seeking medical care with symptomatic joint disease. Oliveria et al. evaluated incident symptomatic, radiographic knee OA
rates in a large HMO in central Massachusetts, mostly involving white, blue-collar workers, and found higher rates in women (female to male ratio for hand, hip, and knee OA 2:1), and an increase in incidence with age until 80 years.⁴⁸ The age- and sex-standardized incidence rate for knee OA was 240 per 100,000 person-years (95% confidence interval (CI) 218.00–262.00). The incidence of clinical knee OA was over 1% per year in women of age 70 to 89 years. Wilson et al. found equal rates of incident symptomatic OA in men and women of Olmstead County Minnesota (mostly northern European).⁴⁹ The age- and sex-adjusted rate for knee OA was 163.8 per 100,000 person-years (95% CI 127.1–200.6). The difference in results between these two studies may relate in part to broader exclusions for secondary OA in the latter study.

Hip Osteoarthritis

Over 8 years, 3.5% to 11.9% (depending on the radiographic definition used) of women of age 65 years and older in the study of osteoporotic fractures (SOF) developed hip OA.⁵⁰ The age- and sex-standardized incidence rate for symptomatic, radiographic hip OA was reported to be 88 per 100,000 person-years (95% CI 75-101) by Oliveria et al.⁴⁸ and 47.3 per 100,000 person-years (95% CI 27.8 -66.8) by Wilson et al.⁴⁹

Hand Osteoarthritis

In the Tecumseh Community Health Study, 1.8% of participants (of age 27 to 51 years) developed hand OA per year.⁵¹ In the Goteborg study (participant age 75 years), 2.7% of participants developed DIP or PIP OA per year.⁴⁷ In the Framingham study (mean age 55 years), Chaisson et al. found that 3.6% of women and 3.2% of men developed radiographic OA in at least one hand joint per year.²⁴ Women had more incident disease than men in all hand joints except the MCP group for which rates were comparable between men and women. The most frequently affected joints were, in decreasing order, DIP-2 (57% in women, 36% in men), thumb IP, CMC-1, and DIP-5.²⁴ The higher overall rate in the older Framingham cohort versus the younger Tecumseh cohort (51) most likely reflects an increase in the incidence of hand OA with age, but also may relate to differences in how OA was defined.

TABLE 1-3 PREVALENCE OF KNEE OSTEOARTHRITIS

Study	Age Range of Participants	Age Subset	Radiographic Knee OA in Women	Symptomatic Radiographic Knee OA in Women	Radiographic Knee OA in Men	Symptomatic Radiographic Knee OA in Men
Lawrence,53 1966	>35	35-44 45-54 55-64 65+	4.0 13.1 40.0 49.1	2.8 5.4 21.8 28.6	5.6 8.2 28.2 26.4	2.5 4.1 9.7 14.3
Felson,54 1987	63-94	<70 70-79 ≥80	25.1 36.2 52.6	7.6 13.0 15.8	30.4 30.7 32.6	6.2 7.8 5.4
Anderson,55 1988	35-74	35-44 45-54 55-64 65-74	1.2 3.6 7.5 20.3		1.2 2.2 5.1 9.0
van Saase,19 1989	>45	45-49 50-54 55-59 60-64 65-69 70-74 75-79 ≥80	12.7 16.1 14.0 24.2 33.3 40.2 40.2 54.6		7.7 11.2 11.8 23.0 13.1 24.7 22.0 22.2

In a cohort of men of age 60 years and above in the Baltimore Longitudinal Study of Aging (BLSA), the incidence was highest at the DIP joints and increased with age in all hand joints.⁵² Oliveria et al. reported an age- and sex-standardized incidence rate for symptomatic, radiographic hand OA of 100 per 100,000 person-years (95% CI 86-115).⁴⁸

Prevalence

Studies of the prevalence of knee OA are summarized in Table 1-3.¹⁹^,⁵³^,⁵⁴^,⁵⁵ The prevalence of radiographic knee OA rises in women from 1% to 4% in those 24 to 45 years of age to 53% to 55% in those of age 80 years and older. In men, the prevalence rises from 1% to 6% in those 45 years and younger to 22% to 33% in those 80 years and

older. Other studies report a prevalence of 12% (Chingford, women 45 to 64 years),⁵⁶ 3.6% (Michigan Bone Health Study, women 24 to 45 years),⁵⁷ and 29% (Rotterdam, >55 years).⁵⁸ In the Beijing Osteoarthritis study, prevalence of radiographic OA in Chinese men rose from 10% at 60 to 64 years to 45.7% at ages over 80 years, similar to findings in Framingham men.⁵⁹ Rates in Beijing women in these age groups were 39.6% and 59.1% respectively, about 40% higher than what was found in Framingham women, applying the same case definitions and radiographic methods.

TABLE 1-4 PREVALENCE OF HIP OSTEOARTHRITIS

Study	Age Range of Participants	Age Subset	Radiographic Hip OA in Women	Symptomatic Radiographic Hip OA in Women	Radiographic Hip OA in Men	Symptomatic Radiographic Hip OA in Men
Lawrence,53 1966	>55		6.2	3.4	16.5	5.5
Maurer,60 (1979)	55-74		2.8	0.7	3.5	0.7
Danielsson,21 1984	40-89	40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85-89	0.4 0.4 0.4 1.2 1.6 0.8 5.2 4.7 5.0 10.0		* 0.4 0.8 1.2 1.6 2.8 2.4 6.4 11.5 5.6
van Saase,19 1989	>45	45-49 50-54 55-59 60-64 65-69 70-74 75-79 80+	2.6 2.0 2.6 3.8 10.9 14.8 14.5 26.0		2.8 2.2 5.9 10.1 11.2 4.7 10.2 11.1
*no cases found

TABLE 1-5 PREVALENCE OF RADIOGRAPHIC HAND OSTEOARTHRITIS

Study	Age Range of Participants	Age Subset	DIP OA in Women	PIP OA in Women	MCP OA in Women	CMC-1 OA in Women	DIP OA in Men	PIP OA in Men	MCP OA in Men	CMC-1 OA in Men
Lawrence,53 1966	>15	15-24 25-34 35-44 45-54 55-64 65+	* * 7.7 22.9 54.8 80.4	* * * 5.9 28.7 49.1	* 0.6 2.2 6.5 25.2 43.8		* 1.8 5.5 22.5 47.1 65.6	* * 2.4 5.5 10.6 32.3	* 1.8 * 12.0 17.3 43.0
van Saase,19 1989	>20	20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80+	1.0 0.5 0.3 4.2 8.9 22.0 41.6 55.5 68.9 76.0 74.7 73.5 72.7	* * 0.3 1.0 1.4 3.6 9.7 20.5 29.6 31.1 35.2 44.4 48.1	0.7 1.0 2.9 4.2 8.2 15.5 22.5 29.2 45.6 54.6 56.6 63.2 55.8	0.3 * * 1.5 5.8 10.9 16.4 24.5 34.5 42.1 46.7 53.0 57.1	0.3 0.3 2.2 2.3 8.6 14.1 24.0 40.5 48.9 51.7 58.8 64.4 48.1	0.3 0.3 0.3 1.0 1.0 3.0 5.4 12.3 11.8 18.1 20.0 32.2 18.5	0.7 1.8 3.7 4.4 9.6 9.7 16.7 29.1 40.5 40.5 50.6 45.8 37.0	* * 0.9 1.0 3.5 4.4 11.5 15.5 20.8 18.1 23.5 42.4 25.9
*no cases found

Studies of the prevalence of hip OA are summarized in Table 1-4. ¹⁹^,²¹^,⁵³^,⁶⁰ An increase in hip OA with age is seen in both genders, especially in women. Other studies report a prevalence of 3% in women and 3.2% in men (NHANES I, 55 to 74 years),⁶¹ and15.9% in women and 14.1% in men (Rotterdam, >55 years). ⁵⁸ Hoaglund et al. found only five cases of hip OA (graded K/L 3-4) in 500 Hong Kong southern Chinese hospital patients.⁶² A subsequent study similarly found a lower prevalence of hip OA in Hong Kong Chinese men.⁶³ Prevalence of hip OA in men and women was substantially lower in Beijing men and women than in U.S. cohorts assessed using the same methods.⁶⁴ For radiographic hip OA, prevalence ratios were 0.07 (Chinese women to white women in the SOF), 0.22 (Chinese women to women in the NHANES I), and 0.19 (Chinese men to white men in the NHANES I).

Though reported differences may relate to study methodology, the prevalence of hip OA may be lower in Jamaicans,⁶⁵ Asian Indians,⁶⁶ and Nigerians⁶⁷ than in European populations.

Two studies of the prevalence of hand OA are summarized in Table 1-5.⁵³^,⁶⁹ The prevalence of radiographic and also of symptomatic, radiographic hand OA at all sites rises with age, and is greater in women. Plato and Norris provide age-specific prevalence rates for individual hand joints for men in the BLSA, and summarize several previous U.S. studies.⁶⁸ Other studies report a prevalence of 14% in the DIP and 16% in CMC-1 (Chingford, women 45 to 64 years);⁵⁶ 3% in DIP joints, 1.0% in PIP, 0.7% in MCP, 0 in CMC-1, 0.5% in IP-1 (Michigan Bone Health Study, women 24 to 45 years);⁵⁷ 35% in DIP joints in women, and 24% in men (Hong Kong southern Chinese, >54 years);⁶² and 30%, for symptomatic hand OA in women, and 29% in men (National Household Education Surveys [NHES] 25 to 74 years).⁶⁹

Risk Factors for Incident Osteoarthritis

Individual studies have traditionally sought to identify risk factors for incident disease or OA progression but not both, in large part due to the cost and logistics of powering both outcomes. The current era is witness to the development and initiation of large-scale studies that will have power to look at incidence, progression, and disability within the same study: the Rotterdam study MOST (Multicenter Osteoarthritis Study), and the OAI (Osteoarthritis Initiative).

A candidate’s risk factor’s effect on each outcome—incidence and progression—should be specifically and separately examined. It is widely believed that the risk factor profiles for each of these key outcomes may overlap but are not identical. Also, the magnitude of the effect of a given risk factor may differ according to the stage of OA disease present in a given joint, i.e., prior to definite OA (the stratum for study of incident OA), or after OA is definitely present (the stratum for progression). As is the case in these studies, it is ideal that the examination of effect on incident and progressive OA occurs within the same study. Otherwise, if the effect on incidence differed from that on progression when these outcomes were examined in separate studies, it would remain possible that the difference was linked to methodologic differences between studies.⁷⁰ An additional key design element of MOST and the OAI reflects evolution in views of the basic OA condition that is of highest priority to study in terms of potential intervention and/or prevention strategy development: the cohorts of each of these studies includes individuals with symptomatic radiographic knee OA or those at higher (than the general population) risk to develop it.⁷⁰

To identify risk factors for incident OA, longitudinal studies, which allow determination of the relationship of a risk factor at baseline to new disease development over time, are of course optimal, but are more expensive to perform. Cross-sectional studies of the relationship between exposure to a given factor and risk of having disease have been more common. OA development is often attributed to a joint-specific local mechanical environment within a systemic milieu, leading to categorization of risk factors as either systemic or local. However, certain risk factors like age, a systemic factor that may act in part by altering the mechanical environment, illustrate that categorization may oversimplify what are complex risk factor effects. Unless otherwise specified, the following studies have focused on a radiographic definition of OA (i.e., K/L ≥2).

Body Weight

In Framingham participants of median age 37 years, weight predicted the presence of knee OA 36 years later.⁷¹ The age-adjusted relative risk for knee OA in the heaviest quintile of baseline weight versus the lightest three quintiles was 2.07 (95% CI 1.67–2.55) for women and 1.51 for men. Results were unaffected by adjustment for serum uric acid level and physical activity. Weight change in Framingham women affected the risk for developing knee OA.⁷² A decrease in BMI of 2 units over the previous 10 years decreased the odds of knee OA (OR 0.46, 95% CI 0.24–0.86). Analyses were adjusted for age, baseline BMI, knee injury, smoking, job physical labor, habitual physical activity, and educational level.

A subsequent Framingham study (mean subject age 70.5 years) in subjects free of disease at baseline confirmed that higher BMI increased the risk of OA (OR 1.6/5 unit increase, 95% CI 1.2–2.2) and weight change was directly related to risk of OA (OR 1.4/10 lb change in weight).⁷³ These findings were present in women; per the authors, the absence of relationship in men may reflect a gender difference or the small number of incident cases in men.

In a longitudinal study of the Chingford population (women, mean age 54 years), belonging to the top BMI tertile was associated with an increased risk of knee OA (OR
2.38, 95% CI 1.29–4.39), adjusting for hysterectomy, estrogen replacement therapy, physical activity, knee pain, and social class.⁷⁴ In Chingford participants with unilateral knee OA, 46% in the top BMI tertile developed OA in the uninvolved knee over 2 years versus 10% in the lowest tertile.¹⁵

Obesity was more strongly associated with bilateral (OR 6.6, 95% CI 4.71–9.18) than unilateral OA (OR 3.4 in right knees), adjusting for injury, age, and gender in 45 to 74 year old NHANES I participants.¹² There is little evidence of a metabolic link between body weight and knee OA. With one exception,⁷⁵ population-based studies have not revealed an independent relationship of a metabolic correlate of obesity (e.g., serum lipids, glucose or glucose tolerance test, body fat distribution, and blood pressure) with knee OA.⁷⁶^,⁷⁷^,⁷⁸^,⁷⁹

In both the Framingham and Chingford populations, while BMI was linked to all patterns of knee OA (tibiofemoral, patellofemoral, and mixed), odds ratios were highest for mixed involvement.¹⁷^,⁸⁰

The possibility remains, albeit small, that knee symptoms preceding OA lead to lower levels of activity which contribute to obesity, and that other factors cause OA. However, analysis of NHANES I data⁵⁵ revealed no evidence that the association between BMI and knee OA is stronger for those with knee symptoms. In Framingham women, the association between weight and knee OA was stronger in those without symptoms; in men, the association appeared to be stronger in those with symptomatic disease.⁷¹

In contrast to the knee, a more modest association between body weight and hip OA has been described. In NHANES I, neither obesity nor fat distribution was associated with hip OA.⁶¹ However, being overweight was more closely associated with bilateral (OR 2.0, 95% CI 0.97- 4.15) than unilateral hip OA (OR 0.54, 95% CI 0.26- 1.16), adjusting for gender, age, race, and education. In the Zoetermeer study, obesity was linked to OA in the right but not the left hip in men, and was not associated with hip OA in women⁸¹ In another study involving farmers, the risk of hip OA was highest in the tallest and heaviest members of the sample though the association with weight, height, or BMI did not achieve significance.⁸² In an early report from the Rotterdam study it was stated that being overweight increased the risk of incident knee OA but not incident hip OA.⁸³

Support for a link between obesity and hand OA comes from one longitudinal study⁵¹ and some cross-sectional studies⁷⁶^,⁷⁹^,⁸¹ but not others.⁵⁷^,⁸⁴^,⁸⁵ In the Tecumseh study, mean age- and smoking-adjusted baseline weight was higher among those who developed hand OA than among those who remained free of disease.⁵¹ Blood pressure, cholesterol, uric acid, and glucose were not linked to the development of hand OA. A cross-sectional relationship was detected in men and women in the Zoetermeer study,⁸¹ men and women in NHES and NHANES I for combined hand/foot OA,⁷⁶ and in men only of the Goteborg population,⁷⁹ but not in men⁸⁴ or women⁸⁵ in the BLSA. In the Michigan Bone Health Study, no association was detected between the presence of hand OA and BMI; relationship between BMI and hand radiographic scores did not persist after adjusting for age and bone mineral density (BMD).⁵⁷

Cross-sectional studies examining the relationship between specific hand joint groups and weight have had conflicting results. In the Zoetermeer survey, obesity was associated with DIP and PIP OA but not with CMC OA,⁸¹ while in the Chingford population, BMI was associated with CMC OA but not with DIP or PIP OA.⁸⁶

Age

Aged cartilage has altered chondrocyte function and material properties and responds differently to cytokines and growth factors. In addition, joint-protective neural and mechanical factors may become impaired with age, such as proprioception, varus-valgus laxity,⁸⁷ and muscle strength.⁸⁸

In a longitudinal study of the Chingford population (women, mean age 54 years), belonging to the highest of three age groups was associated with an increased risk of knee OA (OR 2.41, 95% CI 1.11–5.24), adjusting for hysterectomy, estrogen replacement therapy, smoking, physical activity, pain, social class, height, and weight.⁷⁴ Knee osteophyte development increased by 20% per 5-year age increase. The magnitude of risk associated with aging appears to decrease as older ages are reached. Age did not affect the risk of knee OA in a longitudinal Framingham study in which the mean subject age at baseline was 70.5 years.⁷³ Several cross-sectional studies have demonstrated a higher prevalence of knee OA with increasing age, including those of Lawrence et al.,⁵³ the Framingham study,⁵⁴ NHANES I,¹²^,⁵⁵ and the Zoetermeer survey.¹⁹

Hip OA is more prevalent at older ages. A relationship between age and hip OA is supported by two Scandinavian studies,²¹^,²³ the Zoetermeer survey,¹⁹ and NHANES I.⁶¹ In the NHANES I data, age increased the risk of hip OA (OR 2.38 for ages 70 to 74 years versus 55 to 59 years, 95% CI 1.15–4.92), adjusting for gender, race, marital status, education, and family income.⁶¹

Age is closely associated with the development of hand OA¹¹ as shown in the reports of Lawrence,⁵³ the BLSA,⁵²^,⁸⁴ the Zoetermeer survey,¹⁹ and the Michigan Bone Health Study.⁵⁷ In a longitudinal BLSA study, Kallman et al. found that age increased the risk of OA for almost every radiographic feature in every hand joint group.⁵² In pre- and perimenopausal women, age was more strongly linked to hand than knee OA.⁵⁷

Only gold members can continue reading. Log In or Register to continue