Osteoarthritis (OA) is a multifactorial disease with genetic and environmental determinants with the weight of causes forming a continuum between the extremes of predominantly genetic or predominantly environmental.1 Meniscus lesions are associated with a high OA risk.2 3 In a manner similar to knee OA in general, the risk is influenced by the presence of nodal OA of the hand (a marker of the presence of heredity for generalised OA), by obesity, and by sex.4^–6 MRI based studies show that meniscus lesions, traditionally associated with “secondary” OA, are much more common in “primary” knee OA than was previously thought, in support of an important role of such lesions also in what was previously termed “primary OA”.7^–10 We have thus proposed that OA after meniscectomy is not markedly different from “primary OA” and that meniscectomised subjects provide a useful model for studies on OA development and treatment.11

Little is known about the natural course of symptoms and function in knee OA patients. In a recent literature review of nine articles of varying methodological quality, there was conflicting evidence for the first 3 years of follow-up and limited evidence for worsening of pain and function in follow-up longer than 3 years.12 To our knowledge, no previous publications are available on long-term longitudinal development of symptoms and self-reported and objectively-assessed function in middle-aged subjects with or at high risk of knee OA. Such information would improve our limited knowledge of the natural course of symptoms and function in early-stage knee OA, and after meniscectomy. Further, identification of prognostic factors of worse outcome would help identify patients likely to benefit from preventive interventions.

The primary aim of this study was to describe self-reported outcomes and objectively measured physical function over time in a cohort of mid-life subjects with or at high risk of knee OA. Secondary aims were to determine prognostic factors of a more rapid worsening and to determine if measures of self-reported and objectively assessed physical function can be used interchangeably. We studied a middle-aged cohort of previously meniscectomised patients and compared the outcome with that of a population-based age and sex-matched reference sample.

METHODS

Subjects

The subjects were identified through the surgical code system, following an isolated meniscectomy at Lund University Hospital in 1973, 1978, 1983, 1984, or 1985. In 1973 and 1978 >95% were total meniscectomies, while in 1983–1985 >95% were partial meniscectomies. As described,5 6 patients with a diagnosis other than meniscectomy affecting the knee, meniscectomy in both knee compartments, or radiographs indicating knee OA at time of surgery were excluded. Out of 859 originally identified subjects, 403 were excluded due to the prespecified exclusion criteria and 127 declined participation.5 6 The remaining eligible 329 subjects, meniscectomised at age 13–68 years, were examined twice, at 15–22 years and 19–32 years after the surgery. The patients were invited to a first assessment in 1994 (meniscectomised in 1973), 1995 (meniscectomised in 1978), or in 2000 (meniscectomised in 1983–85). The second assessment for all subjects was performed in 2004. The observation window relevant for this report thus ranged from 4–10 years for different patients. In this report, data from questionnaires at both follow-ups, functional performance tests at both follow-ups, and radiographs of knees at first follow-up are included.

Reference subjects

In total, 68 individuals with no previous knee surgery and no known meniscal or cruciate ligament injury, constituted the original reference group.3 5 They were identified using the National Population records and matched for sex, birth year, and zip code with the cohort meniscectomised in 1973. Age, sex, and body mass index (BMI) were similar to the total cohort of meniscectomised participants. The reference subjects were invited to a first assessment in 1996 and exclusion criteria were the same as for meniscectomised subjects. Knee pain was not an exclusion criterion for the reference group. The second assessment was performed in 2004.

All subjects signed an informed consent form and the study was approved by the Ethics Committee at the Medical Faculty of Lund University.

Self-reported outcomes

The Knee injury and Osteoarthritis Outcome Score (KOOS) is a valid and reliable patient-relevant questionnaire for patients with knee injury and knee OA.13^–15 The KOOS is an extension of the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC),16 comprising five different subscales; pain, other symptoms, activities in daily living, function in sport and recreation, and knee-related quality of life. Each subscale is scored separately from 0–100 on a worst to best scale. A difference of 10 units is considered to represent a clinically important difference.17 The questionnaire we used and a user guide can be found online at http://www.koos.nu.

Tests of physical function

Two tests were used for evaluation of lower extremity function: maximum length of one-leg hop for distance and maximum number of knee-bendings performed in 30 s. These tests were selected out of 10 tests since they had low floor effects (9% and 3%, respectively), good reliability, and fulfilled the criteria of discriminative validity in middle-aged subjects with knee injury and knee OA.18

One-leg hop for distance is a test commonly used for evaluation of functional performance in patients with anterior cruciate ligament deficiency.19 The test mimics sporting activities and demands muscle explosivity, balance and functional stability of the knee.20 The maximum length of one-leg hop correlates strongly with isokinetic knee extensor strength in middle-aged meniscectomised patients (r = 0.8)21 and moderately with isokinetic knee extensor strength in patients with anterior cruciate ligament reconstruction (r = 0.62).22 The test–retest reliability in the current study group was excellent (ICC = 0.93). Inter-rater reliability has not been assessed.

The testing procedure was as follows. The subject stood on one leg, hands behind their back, and was asked to hop as long as possible, landing and balancing on the same leg long enough for the examiner to determine the distance measured by a tape measure fixed to the floor. The distance was measured (in cm) from toe in the starting position to heel in the landing position. The best of three hops was recorded, and if the subject improved more than 10 cm between the second and third hop an additional hop was performed.18

Maximum number of knee-bendings in 30 s was originally used in long-term follow-up of meniscectomised subjects where a worse result was seen compared to matched controls.23 A good result requires the ability to alternate rapidly between concentric and eccentric work of the extensor muscles over the hip and knee joints, an ability that is impaired in patients with knee OA, especially in knee joint positions of 15 to 30 degrees of flexion.24 The test–retest reliability in the current study group was excellent (ICC = 0.92).18 Inter-rater reliability for maximum number of knee-bendings performed in 30 s was evaluated in the first 50 subjects during second assessment and found to be excellent (ICC = 0.96).

The testing procedure was as follows. The patient stood aligned with the long axis of the foot to a straight line and toes placed on a perpendicular line; fingertip support for balance was provided by the examiner. The subject was then asked to bend the knee, without bending forward from the hip, until he/she no longer could see the line along the toes (about 30 degrees of knee flexion). The angle of knee flexion was continuously controlled by the physical therapist and the number of knee-bendings performed in 30 s was noted. Maximum number of knee-bendings in 30 s was always performed prior to one-leg hop for distance and recorded by either of two physical therapists who had undergone familiarisation and training sessions.

Potential risk factors for worsening over time

Age, sex, self-reported BMI, and radiographic knee OA were considered as potential risk factors for worsening in self-reported and objectively assessed outcomes from first to second assessment.

Radiographic knee OA

At first assessment, radiographic knee OA was determined in both knees by standing anteroposterior tibiofemoral radiographs guided by fluoroscopy (knee in 15 degrees of flexion) and skyline views of the patellofemoral joint (knee in 50 degrees of flexion), read by one examiner (blinded to clinical details and outcome) using the atlas of Osteoarthritis Research Society International.5 25 The subjects were classified as having radiographic OA if findings approximated a grade equal to or more than 2 on the Kellgren and Lawrence scale.25 26 The radiographic procedure and results from this cohort was reported in detail.26 27

Statistical analysis

Baseline values were non-normally distributed and thus the non-parametric Mann–Whitney test was used for comparisons between groups at baseline. Changes over time were approximately normally distributed and the analysis of covariance (ANCOVA) regression model was used for analysis. In all models, predictors were adjusted for each other, follow-up time in years (second assessment–first assessment) and value at first assessment of the dependent variable. Follow-up time in years adjusted for “cohort effect” since the cohorts had different follow-up times. For comparison, follow-up time in years was substituted with cohort as categorical data (73 cohort, 78 cohort and 83–85 cohort). The results were similar and we chose to adjust for follow-up time in years. For physical function tests, the meniscectomised subjects contributed with the meniscectomised leg only. In the reference group, the correlation between test result of right and left leg was high (Pearson r>0.90), so we chose to use the mean value of both legs for analyses. To study tibiofemoral OA as a predictor of worsening, tibiofemoral radiographic status of the meniscectomised knee was used in the regression models. For the reference group, we used tibiofemoral radiographic status in either knee.

To explore the hypothesis that radiographic OA in more knee compartments is associated with worsening in the measured outcomes, an ordinal knee OA score was constructed by counting how many compartments (tibiofemoral or patellofemoral) from either knee had OA (range 1–4, the medial and lateral tibiofemoral compartments were counted together as one compartment). Only meniscectomised subjects with tibiofemoral OA of the operated knee were included.

Spearman’s Rho was used to correlate the self-reported and objectively assessed outcomes. All tests were two-tailed and we considered a p value ⩽0.05 statistically significant (SPSS for Windows, V.13.0; SPSS Inc., Chicago, Illinois, USA).

RESULTS

Subject characteristics

The analyses were restricted to the 259 meniscectomised patients and 50 reference subjects for whom longitudinal KOOS data was available, ie, 79% and 74% respectively of the meniscectomised subjects and reference subjects examined at first assessment. Subjects lost to second assessment (n = 70) were older (58.9 vs 52.7 years, p<0.001), reported significantly worse activites of daily living (ADL) function (83 vs 89, p = 0.017), and performed on average three fewer knee-bendings per 30 s (27 vs 24, p = 0.030). There was no significant difference in the prevalence of radiographic knee OA in the subjects available for the second follow-up (47%) and the subjects lost to follow-up (55%), p = 0.162.

Longitudinal objectively assessed physical function was available for 67% of the included subjects, 173 patients (38 (22%) women) and 47 reference subjects (10 (21%) women). The group for which KOOS and objective physical function was available did not differ from the group where only KOOS data was available with regard to subject characteristics or self-reported KOOS data (p>0.14).

The mean (SD) age of the meniscectomised subjects was 52.6 (10.4) years at first assessment and 59.4 (10.3) years at second assessment, and 81% were men. The reference subjects had a mean age of 53.4 (9.8) years at the first assessment and 62.1 (10.1) years at second assessment and 78% were men. Radiographic tibiofemoral OA was present in 47% of the operated knees of meniscectomised subjects and in 8% and 10% of the right and left knees of reference subjects respectively at first assessment; p<0.001 (table 1).

Self-reported and objectively assessed outcomes at first assessment and over time

At first assessment, meniscectomised patients reported statistically and clinically significantly worse self-reported KOOS scores compared to the reference group (table 2). Meniscectomised patients also performed a significantly lower number of knee-bendings in 30 s. Between first and second assessment, meniscectomised patients worsened significantly in all outcomes, both self-reported and objectively assessed, while the reference group worsened in objectively assessed tests only (table 2). When restricting the analyses to men only or to those without OA, similar results were found (data not shown). When comparing the worsening over time in the meniscectomised group to the reference group, we found the meniscectomised patients to worsen more, significantly so with regard to KOOS pain and one leg hop (table 2).

Predictors of change in self-reported outcomes and objective function

In the meniscectomised group, being a woman or having radiographic tibiofemoral OA was associated with greater worsening in most self-reported outcomes and in objective function over the 4–10 year follow-up time (table 3). Older age and higher BMI was associated with worsening in objectively assessed function, but not in self-reported outcomes (table 3).

Due to the low number of control subjects (n = 50) predictors of change are not reported for this group.

The association between self-reported and objectively assessed physical function

To determine if self-reported and objectively assessed physical function represent similar constructs, the association of self-reported ADL function with maximum number of knee-bendings/s was analysed, and self-reported sport and recreation function with one-leg hop, since these combinations of measures theoretically would measure similar constructs: less demanding function related to activities of daily living and more demanding function related to sports and recreation. The correlation between self-reported and objective function was weak at first follow-up and even weaker for the longitudinal analyses. At first follow-up, the correlation between self-reported ADL function and maximum number of knee-bendings in 30 s was r_S = 0.26 (p = 0.001) and for self-reported Sport and Recreation Function and one-leg hop r_S = 0.14 (p = 0.08). The correlations of the corresponding longitudinal changes were 0.09 (p = 0.2) and 0.08 (p = 0.3), respectively.

DISCUSSION

This longitudinal study of early stages of OA development in a well described cohort with previous meniscectomy adds novel data on change in knee pain and self-reported as well as objective physical function of the lower extremity. In our middle-aged sample, self-reported outcomes and objective physical function were impaired compared with a reference population, and worsened during the 4–10 year observation window, similar to older age cohorts with knee pain and or OA.28^–30 Due to the limited sample size, all impairment differences were not significant. The null hypothesis cannot be ruled out in the non-significant cases.

Meniscectomised women are at increased risk of worsening

Few studies have addressed self-reported outcomes and objectively assessed physical functioning in middle-aged individuals. One cross-sectional study showed that middle-aged women have compromised physical functioning.31 What was not shown previously is that middle-aged women deteriorated at a higher rate than men, not only with regard to self-reported and objectively assessed physical function, but also with regard to symptoms from the knee and knee-relevant quality of life. These changes may precede and contribute to development of OA in this group. Our present findings suggest that middle-aged women are an important target group for exercise interventions to forestall future worsening of physical functioning.31 This is consistent with women in the population aged 55–74 evaluated with the KOOS reporting worse knee-related pain, other symptoms and physical function related to activities of daily living, compared to age-matched men.32 The small number of women in the reference group of the present study did not allow any conclusions regarding an interaction of previous knee injury and gender on worsening in knee-related outcomes.

Radiographic OA predicted worsening

We have shown a degenerative meniscal lesion to be associated with an increased prevalence of radiographic OA and worse self-reported outcome, compared with a traumatic meniscus lesion.26 Because of the strong association of type of meniscal tear with radiographic OA, we choose to use radiographic OA, which was determined at first assessment, as a predictor of change in self-reported and objectively assessed outcomes instead of type of tear determined at surgery, on average 18 years prior to first assessment.

In population-based samples, radiographic knee OA and self-reported pain and function are poorly correlated.33 In patients, the correlation is higher, most probably reflecting the fact that joint-related pain is the most common reason for patients to seek medical care and the most common indication for doctors to perform a radiographic examination of the knee. The present study cohort was examined about two decades after meniscus surgery and would be expected to perform in between population-based and hospital-based samples. In support of this finding, in a cross-sectional analysis of a subset of the current study sample, severe joint space narrowing—but not less severe joint space narrowing or the presence of osteophytes—was associated with worse self-reported outcomes 19 years after surgery.23 The role of OA grade as predictor of longitudinal change in self-reported and observed physical function is less well documented.12 The present study shows that radiographic tibiofemoral OA in the index knee equivalent to Kellgren and Lawrence grade 2 or higher predicts worse self-reported pain, knee-related quality of life, and self-reported and objectively assessed physical function after about 7 years of follow-up. Moreover, a higher number of OA-affected knee compartments was associated with a greater worsening in self-reported outcomes and objectively assessed knee function. These predictors are clinically relevant in that they help to identify patients at an earlier age than previously reported and where interventions may be more successful.

Methodological aspects of measurement of function

The present report shows that validated measures of self-reported and objective function used in middle-aged subjects with knee OA represent separate entities and cannot be used interchangeably or be assumed to measure similar concepts. This is in line with previous findings in elderly.34^–37 In elderly with knee OA, tests mimicking activities of daily living are frequently used for assessment of objective function. When applied in this middle-aged cohort however, tests typically used in elderly populations were associated with high ceiling effects. As an example, the 11 tests included in the Index of Muscle Function validated for RA and OA38 had ceiling effects of 60–90% in the current study population,18 indicating these tests having poor validity in this middle-aged sample. We thus choose to use more demanding tests with low floor and ceiling effects proven valid for the current study group.18

The presence of knee pain could explain some of the differences in performance between meniscectomised subjects and the reference group. When analysing cases and controls together we found that change in pain (measured by the KOOS) tended to affect the change in performance of knee-bendings and one-leg hop. When analysing cases and controls separately, change in pain still tended to affect performance in the reference group but not in the cases. We interpret this finding such that other factors, including meniscectomy, affect change in functional performance more over time than change in pain does.

To fully evaluate function, self-reported and objective function needs to be assessed. This conclusion is also supported by our observation that higher age and BMI were related to a worse decline in objective function only, and not in self-reported function.

Strengths and limitations

This study is categorised as a high quality study applying the criteria of methodological quality used in a recent literature review of the course of pain and functional status in hip and knee OA.12 A limitation is the lack of consideration of psychosocial factors as predictors of decline. Another limitation of our study is missing performance outcome data, the major reason being logistic difficulties. The proportions of subjects lost to follow-up were similar in the meniscectomised group and in the control group. The subjects lost to follow-up were older and performed worse on the knee-bending test at first assessment. To impute a low result for those who were missing would likely result in an overestimate in the change, whereas imputing the last observed value carrying forward would underestimate the change. Since the analyses were controlled for age and performance at first assessment, and since the extent of radiographic knee OA was comparable between the studied group and the excluded subjects, this was unlikely to affect the results.

Can improvements in muscle strength and function prevent OA?

In our middle-aged sample, self-reported outcomes and objective physical function worsened over 4–10 years, similar to older age cohorts with knee pain and or OA.28^–30 We used previously meniscectomised patients as a model for studies on OA development and treatment. Indeed, 18 years after meniscectomy 53% had developed radiographic knee OA. Decreased muscle strength and muscle function was found to precede development of radiographic OA or self-reported OA.39^–41 Intervention studies suggest that exercise improves cartilage quality in meniscectomised patients,42 and slows radiographic progression of existing OA.43 Restoration of muscle function may be beneficial in those at risk for OA or with early-stage OA. For subjects with prior knee injury and high risk of developing knee OA, we suggest that physical function in the lower extremity should be evaluated, and if declining, exercise to restore function be initiated.

CONCLUSIONS

Meniscectomised subjects have impaired knee-related outcomes at 18 years after surgery. Worsening over time was greater compared with reference subjects, especially so in women and subjects with radiographic OA. Self-reported and objectively assessed knee-related physical function represents separate entities that should be evaluated separately. The present findings improve our ability to identify middle-aged individuals at high risk for worsening in knee-related outcomes known to be associated with development and further progression of OA. Rehabilitative efforts may be warranted in middle-aged meniscectomised patients, especially in women and those who have developed radiographic OA.