“Pain elsewhere” and the impact of knee pain in older people
To investigate how often knee pain is accompanied by pain elsewhere and to determine whether the presence of “pain elsewhere” than the knee influences either knee-related disability or the impact of knee pain on health and psychological status.
A survey was mailed to 8,995 individuals (age ≥50 years) registered with 3 general practices in North Staffordshire in the UK. Patients recorded pain on a manikin and completed the Short-Form 36 health survey, the Hospital Anxiety and Depression Scale, and the Western Ontario and McMaster Universities Osteoarthritis Index. Responders were categorized as having no pain, knee pain with or without pain elsewhere, or other pain (no knee pain, but pain elsewhere). Those with knee pain or other pain were subdivided by the extent of pain elsewhere.
The adjusted survey response rate was 70%. Fifty-seven percent of responders with knee pain had pain in at least 2 other joint areas. Pain elsewhere was associated with lower physical function and with anxiety and depression both in the knee-pain group and in the other-pain group (associations were stronger in the knee-pain group). Knee pain and disability were less severe in those with knee pain alone than in those with knee pain and pain elsewhere, even after adjusting for age, sex, obesity, laterality of pain, and depression.
Most people with knee pain have multiple joint site pain. The importance of this to clinicians and researchers is that the severity of knee pain and related disability is worse in the presence of pain elsewhere. This finding has implications for the management and treatment of older people with knee pain.
Patients who present with a regional musculoskeletal pain can be separated into those who have the regional pain only and those whose regional pain is accompanied by pain in other parts of the body. Natvig et al have argued that isolated back pain should be considered as a separate syndrome from back pain presented in the presence of other reported pain symptoms (1). A rationale for this is that one of the strongest predictors of the outcome of low back pain is the number of other areas that are painful (2).
Joint pain is the most common syndrome of regional pain in older people, but it is mostly considered in the context of a local pathologic condition (osteoarthritis [OA]) in the joint. However, there is evidence that the development of OA, as evidenced by radiographic changes, is weakly linked to joint pain severity and disability in the general population of older adults (3). Pain occurring elsewhere than in the particular joint under scrutiny might increase the impact of the regional syndrome on a patient's health status and daily activity, as seems to be the case with back pain. Such generalized pain might arise either as a result of OA in other joints or from other causes of chronic or widespread musculoskeletal pain.
To investigate this further, we studied the example of knee pain, a common site for joint pain in the older age group and for radiographic OA. In a sample of older people, we investigated 1) how often knee pain is accompanied by pain elsewhere, and 2) whether the presence of “pain elsewhere” than the knee influences either knee-related disability or the impact of knee pain on general health and psychological status.
PATIENTS AND METHODS
We undertook a cross-sectional survey of all 8,995 individuals age ≥50 years who were registered with 3 general practices in North Staffordshire (4). These practices are members of the North Staffordshire GP Research Network. More than 98% of the general population of the UK is registered with a general practice (5), and therefore the practice registers form a convenient frame for identifying a representative sample of a local population. Local health service and Keele University research governance procedures were followed. The North Staffordshire Local Research Ethics Committee approved the project.
The questionnaire included a blank body manikin. Responders were asked to shade on the manikin any pain or ache that had lasted for ≥1 day in the last month. Transparent templates were applied to the manikin to categorize responders according to the presence or absence of pain in the knee (“knee pain”) and in the following regions (“pain elsewhere” than the knee): neck, hand, lower back, hip, and foot and ankle. The template scoring method has been shown to be reliable (6).
Three groups of survey responders were defined on the basis of the following manikin definitions: group 1, no pain (no knee pain, no pain elsewhere); group 2, knee pain (knee pain with or without pain elsewhere); group 3, other pain (no knee pain, pain elsewhere). Groups 2 and 3 were then categorized into the following subgroups based on the number of “pain elsewhere” regions shaded on the manikin: group 2, subgroup 1 (knee pain only); group 2, subgroup 2 (knee pain plus 1 pain elsewhere); group 2, subgroup 3 (knee pain plus ≥2 pains elsewhere); group 3, subgroup 1 (pain elsewhere than the knee in 1 region); group 3, subgroup 2 (pain elsewhere than the knee in 2 regions); group 3, subgroup 3 (pain elsewhere than the knee in ≥3 regions).
The Short-Form 36 health survey (SF-36) (7) was used to determine the influence of pain elsewhere on general health status. The Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) (8) was chosen to measure pain and disability specifically related to the knee. Since obesity and anxiety and depression are linked with widespread pain, we included measures of these to explore their influence on any link between number of pain areas and impact of knee problems. We used the Hospital Anxiety and Depression Scale (HADS) (9) to measure anxiety and depression, and we used self-reported weight and height to calculate body mass index (BMI). Obesity was defined as BMI >30 kg/m2.
Item responses in the SF-36 were summed and transformed to provide scores varying from 0 (worst health) to 100 (best health) for each of the 8 domains. Scores were calculated when more than half of the items on a scale were completed. Means and SDs were calculated for each SF-36 domain within each pain subgroup. Differences in domain scores between the 3 groups were assessed using analysis of variance (ANOVA), adjusting for age, sex, and obesity, followed by Tukey's honest significant difference post hoc multiple comparison test. ANOVA (again adjusting for age, sex, and obesity) was used to compare the “knee pain plus ≥2 pains elsewhere” subgroup with the “pain elsewhere than the knee in ≥3 regions” subgroup.
HADS data were summed to obtain an anxiety score and a depression score. Those scoring above the highest tertile were considered to be “more” anxious and “more” depressed. Differences between the pain subgroups were assessed using multiple logistic regression with age, sex, and obesity also included in the model.
Item responses for the WOMAC were summed to produce subscale scores (pain 0–20, stiffness 0–8, physical function 0–68). Higher scores indicate worse knee-related pain and disability. Guidelines for missing data were followed (8). For each subscale (pain, stiffness, and restricted physical function), responders who scored “severe” or “extreme” on at least 1 item on that subscale were defined as “severe” (10).
Within the knee-pain group, associations between pains elsewhere and the presence of severe pain, difficulty with physical function, or stiffness as indicated by the WOMAC were estimated, again using multiple logistic regression and with age, sex, obesity, laterality of pain, and depression included as independent variables in the models (entered concurrently). Laterality of pain was defined as pain in both knees (bilateral) or pain in just one knee (unilateral). Since depression and anxiety scores were highly correlated (Pearson's r = 0.65), only depression was included as a potential confounder. Statistical analysis was performed using SPSS 12.0 for Windows (SPSS, Chicago, IL).
Of those surveyed, 6,177 (70% adjusted) returned a questionnaire having completed the manikin and having answered the screening question on the presence of knee pain in the last 12 months. Of these subjects, 813 were excluded since they reported knee pain in the last year but not on the manikin (which was used to identify body pain in the last month only). The mean ± SD age of the remaining 5,364 subjects was 65 ± 10.0 years; 55% of subjects were women and 99.6% were white UK citizens or white Europeans.
Table 1 shows the demographics of pain in survey responders. Slightly more than one-third (n = 1,909) were in the no-pain group, 41% were in the knee-pain group (n = 2,210), and 23% were in the other-pain group (n = 1,245). Proportionally more women than men and more younger responders than older responders reported multiple pains in the knee-pain group (both P < 0.001 by chi-square test for trend); this was less apparent in the other-pain group (P = 0.50 for age; P = 0.10 for sex).
Table 1. Age, sex, anxiety, depression, and physical function by pain status*
|Women||2,959 (55)||983 (51)||226 (49)||274 (55)||782 (62)||328 (54)||241 (56)||125 (60)|
|Age, years|| || || || || || || || |
| 50–64||2,746 (51)||921 (48)||215 (47)||242 (49)||690 (55)||339 (56)||221 (52)||118 (57)|
| 65–74||1,592 (30)||582 (30)||136 (30)||165 (33)||355 (28)||157 (26)||138 (32)||59 (29)|
| ≥75||1,026 (19)||406 (21)||106 (23)||89 (18)||212 (17)||114 (19)||69 (16)||30 (14)|
|Anxious‡||1,489 (28)||337 (18)||102 (23)||155 (32)||530 (43)||163 (27)||119 (28)||83 (41)|
|Depressed‡||1,340 (25)||293 (16)||82 (19)||140 (29)||518 (42)||120 (20)||117 (28)||70 (34)|
|Physical function, mean ± SD SF-36 score§||–||76.8 ± 26.8||59.3 ± 28.6||53.4 ± 29.7||40.8 ± 29.4||69.8 ± 27.9||63.0 ± 29.4||53.1 ± 28.7|
Physical function restriction was worse in the knee-pain group than in the other-pain group, even after adjustment for age, sex, and obesity (mean difference in SF-36 scores 17.5, 95% confidence interval [95% CI] 15.3–19.9, P < 0.001), and worse in the other-pain group than in the no-pain group (mean difference 12.3, 95% CI 9.9–14.7, P < 0.001). There was an increase in physical function limitation (lower mean SF-36 scores) with an increase in the number of pains elsewhere, both in the knee-pain group and in the other-pain group (Table 1). When we compared the two subgroups with at least 3 pain regions (knee pain plus ≥2 pains elsewhere versus pain elsewhere than the knee in ≥3 regions), the knee-pain subgroup had worse physical function restriction after adjustment for age, sex, and obesity (mean difference 11.1, 95% CI 6.8–15.3, P < 0.001). There were similar but less marked differences in the other SF-36 dimensions (data not shown).
There were increasingly strong associations between pains elsewhere and anxiety and depression within both pain groups as the number of pain regions increased. Responders with knee pain plus ≥2 pains in areas other than the knee had 4 times the odds of being depressed as those reporting no pain at all (odds ratio [OR] 4.0, 95% CI 3.3–4.8), but they also had a higher likelihood of being depressed than those with pain elsewhere than the knee in ≥3 regions (OR 1.4, 95% CI 1.0–1.9).
The other analyses were restricted to the knee-pain group. Table 2 summarizes the crude scores and associations between knee pain, restricted physical function, and stiffness severity and the cumulative number of regional pains elsewhere identified from the manikin. Mean knee-specific pain, physical function, and stiffness scores on the WOMAC were lower (i.e., indicating less severe knee pain and less knee-related disability and stiffness) in those with isolated knee pain than in those with pains elsewhere. For each additional pain elsewhere, the severity of each problem increased incrementally, and there was a corresponding pattern in the ORs summarizing the associations.
Table 2. Severity of knee problems in patients with knee pain, by number of pains elsewhere than the knee*
|Knee alone (457)||5.9 ± 3.9||115 (26)||Referent||18.6 ± 14.6||166 (37)||Referent||2.6 ± 1.9||72 (16)||Referent|
|Knee plus 1 (496)||6.5 ± 4.0||146 (30)||1.2 (0.9–1.7)||21.5 ± 15.1||215 (44)||1.4 (1.1–1.8)||2.8 ± 1.9||96 (20)||1.3 (0.9–1.8)|
|Knee plus 2 (470)||6.8 ± 4.5||151 (33)||1.4 (1.0–1.8)||22.7 ± 15.6||204 (44)||1.4 (1.0–1.8)||3.0 ± 2.0||102 (22)||1.5 (1.1–2.1)|
|Knee plus 3 (374)||7.9 ± 4.1||138 (38)||1.7 (1.3–2.4)||27.4 ± 15.3||211 (57)||2.3 (1.7–3.1)||3.5 ± 1.9||110 (30)||2.3 (1.6–3.2)|
|Knee plus 4 (241)||9.1 ± 4.5||117 (50)||2.8 (2.0–3.9)||32.0 ± 16.1||156 (66)||3.3 (2.4–4.6)||4.0 ± 1.8||95 (40)||3.5 (2.4–5.0)|
|Knee plus 5 (172)||9.7 ± 4.5||93 (56)||3.6 (2.5–5.2)||35.2 ± 15.8||123 (74)||4.9 (3.3–7.3)||4.4 ± 1.9||84 (51)||5.5 (3.7–8.1)|
Table 3 further illustrates the associations between number of pains elsewhere in those with knee pain (categorized as having no pain elsewhere, 1 pain elsewhere, or ≥2 pains elsewhere) and severity of pain, restricted function, and stiffness. The association between severity and multiple pains elsewhere was independent of age, sex, obesity, and laterality and was not explained by further adjustment for the presence of depression.
Table 3. Associations of pain elsewhere with severe WOMAC knee-specific pain, restricted physical function, and stiffness in those with knee pain*
|Alone||115 (26)||Referent||Referent||166 (37)||Referent||Referent||72 (16)||Referent||Referent|
|Plus 1 pain elsewhere||146 (30)||1.2 (0.9–1.6)||1.1 (0.8–1.5)||215 (44)||1.4 (1.1–1.9)||1.3 (0.9–1.7)||96 (20)||1.2 (0.8–1.7)||1.0 (0.7–1.5)|
|Plus ≥2 pains elsewhere||499 (41)||1.8 (1.4–2.4)||1.5 (1.1–2.0)||694 (56)||2.2 (1.7–2.8)||1.7 (1.3–2.3)||391 (32)||2.1 (1.6–2.8)||1.7 (1.2–2.3)|
This study adds to previous studies in 3 ways. First, we have shown that knee pain does not tend to occur in isolation. Second, we have shown that the presence of pain in regions other than the knee is associated with poorer general and psychological health in those with knee pain. A small number of studies have previously illustrated an additional impact of widespread pain on general disability status. In the study by Natvig et al (1), responders with lower back pain as part of widespread pain had worse functional ability than those with lower back pain alone. In a smaller Swedish study (11), higher pain intensity, longer duration of pain, and greater disability (measured through work absence) was found in those with chronic pain at multiple sites compared with those with localized neck–shoulder pain. Thomas et al found that, compared with adults with back pain and without widespread pain, adults with both back pain and widespread pain had a greater risk of persisting disabling back pain at 12 months (2).
Third, our results further suggest that there is a link between the extent of pain experienced in the body and the impact of a particular regional pain such as knee pain. We have shown that knee-specific pain and disability are actually worse in the presence of pains elsewhere than the knee, even after accounting for poorer psychological health. This has implications for randomized controlled trials and other studies using the WOMAC as an outcome measure, since investigators in such trials and studies might need to consider the presence of widespread pain when interpreting their results.
Our findings are in contrast with the model of knee pain in older people that attributes the effects of knee pain directly to the underlying pathologic development of OA in the joint. One explanation for this is that multiple-site pain is a different syndrome from single-site pain (1). Single-site pain may more obviously be related to local factors, while other influences on pain experience may determine the impact of more widespread pain, including an additional effect on local sites, given the strong associations between number of joint sites and psychological and general health status.
An alternative explanation is that widespread pain in older people represents generalized OA, which is the underlying pathologic process determining the added impact of widespread pain. However, this cannot explain the effect of widespread pain on knee-related pain and disability. There remains something about the existence of other pain that inflates the effect of the knee problem and that does not seem to be explained by psychological distress or severity of the knee problem. One candidate for this is abnormal central sensitization to pain (12).
Another explanation for the additional impact of widespread pain on joint-specific pain is the presence of comorbidity, which we have not measured in this study. This effect should, however, be limited in the current study since we asked survey responders to think specifically about their knee pain when answering the survey questions about disability.
Response rates to mailed surveys in the range of 70–85% are viewed as very good (13). Since our survey achieved a good response and completion rate (70%), the effects of nonresponse bias should be limited, although there were slight differences between responders and nonresponders in terms of age, sex, and location of general practice. However, nonresponse is unlikely to have had a major effect on the comparison between those with different levels and patterns of pain. Another possible bias in this study is that people with multiple joint site pain may have a general tendency to report multiple problems in response to a questionnaire (“reporting effects”). One such effect could be represented by a propensity to report more severe pain irrespective of joint site.
Age is clearly an important potential confounder of the associations observed in our study because of the sharp rise in reported joint symptoms with age. However, all analyses were adjusted for age within a group of subjects who were all age ≥50 years.
One limitation of this study is its cross-sectional nature, which means we have investigated pain and disability at one point in time. Based on mid-2000 population estimates (14), the demographic profile of the sample was similar to that of North Staffordshire and England and Wales except for ethnicity; 99.6% were white UK citizens or white Europeans, reflecting the ethnicity of our local population.
The practical importance of our findings is that the presence and extent of pain in other sites may be an important determinant of outcome in patients who present with knee pain, just as it appears to be in those who present with back pain. Only a minority of those with knee pain in our study had isolated knee pain, so pains in multiple sites are a common experience of older people with joint pain in the general population. This was a cross-sectional study, but the implication is that both general and knee-specific outcomes, and both physical and mental state, will depend on addressing the management of the generalized pain, since knee pain, already disabling on its own (and a recognized public health concern), is worse in the context of widespread pain.
However, there may be implications in the other direction, in that adequate management of the local joint problem might have beneficial effects on the generalized experience and perception of pain. The outcome of total knee replacement might, for example, be to reduce the frequency and impact of pains elsewhere in the body, if the regional source of pain from the knee turns out to be an important driver for the maintenance of the generalized pain. This has indeed been suggested by results of studies of hip replacement by Kosek and Ordeberg (15).