One of the most important findings is that with regard to the WOMAC score there were small inter-group differences of unclear statistical significance. The difference in the parameters ‘postoperative WOMAC stiffness’ and ‘pre-to-postoperative gain in WOMAC total’ was p < 0.1 with power values of 28% and 53%. Hence, it may be assumed that differences in WOMAC scores between groups were present, but were not sufficiently detected by the statistics (statistical type 2 error, Hypothesis 1). Moreover, the revision-free survival rate was significantly poorer in patients with severe patellofemoral osteoarthritis when treated with TKA without patella resurfacing—poorer than in those with mild patellofemoral osteoarthritis (Hypothesis 2). On the basis of those findings the authors recommend that patients with preoperatively severe patellofemoral osteoarthritis (Iwano Stages 3 and 4) undergo patella resurfacing during TKA.

In an effort to compare our findings with previous research Cho et al. already investigated whether the preoperative severity of patellofemoral osteoarthritis affected the outcome of TKA without patella resurfacing [5]. The authors reported no significant differences between mild (Stage 0–1 Iwano) and moderate to severe (Stage 2–4 Iwano) cases according to WOMAC and Hospital for Special Surgery scores. From their findings the authors concluded that good results may be achieved with patella non-resurfacing, even in patients with severe patellofemoral osteoarthritis. The gross differences to the findings made in the current study may be explained by differences in the applied methods. First, the current study included far more patients (approx. 1200 vs. approx. 450). Second, the method of allocating patients to study groups according to their Iwano stage was not the same. Most importantly, Cho et al. analyzed postoperative patellofemoral tracking parameters, which may be regarded as surrogate parameters. Instead, the current study calculated the revision-free survival rate, which is regarded as a more robust outcome.

Also Feng et al. investigated the same topic and retrospectively analyzed the data from 167 patients [8]. Preoperative severity of patellofemoral osteoarthritis was again determined according to Iwano (Stage 0–1 was defined as mild and Stage 2–4 as moderate to severe). Several knee scores were applied. The authors determined no significant differences between the groups. The findings made by Feng et al. contradict the findings made in the current study. These differences may be attributed to the differences in numbers of patients and to the differences in the outcome parameters.

Schmidt et al. investigated preoperative severity of patellofemoral osteoarthritis in 193 individuals undergoing TKA without patella resurfacing according to Kellgren–Lawrence and according to the OARSI system [17]. The authors determined the Knee Society Score, global satisfaction, physical activity and the reoperation rate. The authors reported that patients with more severe preoperative osteoarthritis (lateral-sided) benefitted even more from the surgical procedure than did those with less severe osteoarthritis, as determined from the patient-reported outcomes. Reoperation rate was not affected by the stage of preoperative patellofemoral joint degeneration. Again, the differences to the current study are striking, but may be explained by the method differences. Schmidt et al. determined the location of the patellofemoral osteoarthritis in addition to the severity. Moreover, there were substantial differences in the methods regarding the patient-reported outcomes and the numbers of participants.

As stated above, the authors of the current study recommend patella resurfacing at least in patients with severe patellofemoral osteoarthritis. The authors are well aware that an additional surgical step during TKA may also bear additional risks. However, on the basis of the higher revision rates in the subgroup of patients with severe patellofemoral osteoarthritis those risks seem to be outweighed by the benefits. Clearly, thorough surgical training in how to perform patella resurfacing is indispensable. Typical errors like (a) introducing a too high overall patella thickness, (b) malalignment of the patella osteotomy, (c) implant overhang and (d) patella implant malrotation should be avoided.

The following potential limitations of the current study are acknowledged. First, the two sole outcome parameters that were available from the Arthroplasty Registry (WOMAC and revision-free survival) were susceptible not only to patellofemoral problems. In other words, although there was poorer revision-free survival in the group with more severe patellofemoral osteoarthritis, those revisions may not necessarily have been related to patellofemoral reasons.

An additional comparison of ‘reasons for revision’ did not reveal any differences between the two groups. Second, this was a retrospective study with the typical weaknesses associated with such studies: selection bias, information bias, inability to investigate parameters other than those previously collected during clinical routine, reliance on data collected by others etc. Third, although previously suggested [22], we did not succeed in collecting physical activity data and health-related quality-of-life data in conjunction with the knee-specific WOMAC data. Fourth, as mentioned above, our study must be regarded as underpowered with respect to the knee score outcome. However, no further patients would have been available with both complete WOMAC and x ray data. Thus, a-priori sample size calculation would not have solved that problem. Fifth, the retrospective approach based on an arthroplasty registry prevented us from determining how the non-resurfaced patellae were treated intra-operatively (circumpatellar electrocautery, osteophyte removal, neglected etc.). Regarding surgical technique, there was no standardization as the data originated from the state-run Arthroplasty Registry. It has to be assumed that a broad variety of surgical techniques was applied across the different hospitals.

It is regarded as a strength of our study that it employed the by far largest number of patients and the longest follow-up period (10 years). Moreover, the applied outcome parameters were well-established parameters and were regarded as having high data quality as they were derived from the state-run Arthroplasty Registry.

The study findings are deemed to have high clinical relevance. As meta-analyses of the last 10 years draw no clear general conclusion on the question whether or not to resurface patellae during TKA, the question may be addressed in a more individualized way. The findings made in the current study reveal that at least a small subgroup of patients (approx. 5–10%) is potentially better treated with additional patella resurfacing. This additional procedure brings an increase in implant costs and surgical risks, but appears to be beneficial and, therefore, justified in that subgroup of patients in the day-by-day clinical work.