We found no significant change over 2 years in the average patient-relevant outcome scores for this study group of individuals who had undergone meniscectomy about 15 years earlier, even though the group was highly enriched in early-stage and incipient radiographic knee OA. However, we found substantial change in the self-report for individual subjects over the same time period.
The generally worse KOOS scores for the individuals with radiographic knee OA, compared to those without, are consistent with earlier reports. Thus, the Baltimore Longitudinal Study of Aging reported that patients with a Kellgren-Lawrence score of 1 were almost twice as likely to report ever having knee joint pain compared with those who had a score of zero. The strength of the association increased with increasing Kellgren and Lawrence score .
Similarly, there was in meniscectomized individuals evidence for a graded increase in pain and functional limitations with increasing severity of radiographic signs of OA . However, a discrepancy between knee pain and radiographic features of knee OA has also been noted, both cross-sectionally and longitudinally [3, 24, 25]. Depression and lack of muscle strength have been shown to better explain pain than radiographic findings [26–28].
Individual vs. group analysis
Few reports have explored OA symptom variation on an individual level [2–4]. A detailed comparison of our results with earlier reports is difficult, since they were conducted before validated and patient-relevant OA disease-specific measurement tools had been widely introduced. The "Bristol OA 500" were patients with advanced radiographic knee OA and a mean age of 65 years recruited from a hospital based rheumatology clinic.
In contrast, the mean age of the present study cohort was 50 years, with 2/5 having mild-moderate radiographic OA, and another 1/5 incipient radiographic changes. Further, the cohort reported on here was recruited from a group of individuals that had undergone isolated meniscectomy 15 years earlier, but independent of their subsequent symptom level or disease history. The mean scores of our study group were relatively good and not representative of subjects with advanced OA seeking medical care.
The rationale for investigating this particular cohort at this time after surgery was its enrichment in early-stage knee OA, and that it consequently may represent a study group suitable for future pharmacological disease-modifying intervention. We assessed our patients at an interval of 2 years; this period of time being suggested as a minimum for clinical trials of disease modification in OA to detect both structural and symptom change .
It could be that the findings reported here are valid only for post-injury, secondary OA, or for this particular cohort. However, the criteria and delimitations for posttraumatic OA compared to primary OA have recently been shown to be much less clear than thought [13, 14, 30], and meniscal pathology is common also in primary, garden-variety, knee OA . Tibiofemoral OA was observed in 53 out of 133 patients who were underwent radiographic examination. Isolated patellofemoral OA was rare and, since it did not affect the final results, was not taken into account. A further argument favoring the general applicability of the present results is the concordance of our findings with other longitudinal studies on OA [2–5, 32].
We applied the criteria for minimal perceptible clinical improvement (MPCI) obtained for the WOMAC; since KOOS contains the WOMAC items and is similar in format. The KOOS subscale ADL is equivalent to the WOMAC subscale Function, while new items have been added to the KOOS subscales Pain and Symptoms. The dimensions assessed by the KOOS subscales Sport and Recreation Function and knee-related Quality of Life are not assessed by the WOMAC. The MPCI for the WOMAC is in the range of 8 to 12 points on a 0–100 scale .
This threshold coincides with the change observed in KOOS scores between 3 and 6 months postoperatively when assessing rehabilitation following reconstruction of the anterior cruciate ligament and concurs with the OARSI definition of moderate improvement in the knee pain assessment for clinical trials in OA [18, 22]. However, the OARSI responder criteria were designed for the evaluation of the patient's response to oral NSAID and intra-articular treatment and may differ for other interventions.
It may be argued that the subject-related changes observed in this study represent inherent instrument instability. However, validation studies of KOOS support the reproducibility and stability of the KOOS instrument [17–19]. Test-retest data on the KOOS subscale pain obtained from 75 patients about to undergo knee arthroscopy  was used to determine the number of subjects improving, deteriorating or not changing over an average period of 5 days.
The proportion of subjects changing over 5 days was approximately half of that changing over 2 years in the present study, in further support that the variation in the present study cannot be explained solely by instrument noise (data not shown). A 'frame shift' in the priorities of the individual patient may occur during long term studies. However, we suggest that a significant frame shift is unlikely to have occurred over this 2 year study period of a cohort with a mean age of 50 years.
Significant change of KOOS scores over time were noted in 1/3 of the cohort studied. About half of those who changed clinically improved. This was true in particular for patients with lower (worse) baseline scores. It is thus possible that the lower proportion of 'changers' among those with better baseline scores may have been, at least in part, due to a ceiling effect.