This study compared the clinimetric properties of four tools to assess the condition of knee or hip arthroplasty patients during the acute recovery phase after surgery. The instruments demonstrated the generally negative impact of surgery on function, on both patient-perceived and actual functional performance. Results also showed that function usually did not return to preoperative values before the 6-week interval. Similar recovery patterns have been demonstrated in other studies [28–31]. Improving function shortly after surgery can help prevent deconditioning, increase patient safety and decrease the need for external resources after hospital discharge. The results of the present study highlight the need of a valid outcome measure to assess function shortly after surgery, in order to evaluate the efficacy of perioperative interventions aimed at improving short-term function. Also, the anticipated impact of surgery on function should be discussed pre and postoperatively with the patient, as it has been shown that patient satisfaction can decrease if patients experience slower recovery than what they expected [32]. When comparing knee with hip arthroplasty, patients undergoing knee arthroplasty generally tended to have poorer outcomes both pre and postoperatively. This is especially the case for pain, where recovery was significantly less for knee compared to hip patients. Pain could explain why poorer function was also obtained for knee patients. Poorer outcomes for knee compared to hip arthroplasty have been demonstrated in other studies [28, 33]. The amount of change in outcomes was however similar for both groups, although recovery (especially for the TUG, ILAS and PQRS function) was more in the first 2 weeks for knee patients, and more in the period from 2 to 6 weeks for hip patients. This could be explained by the poorer preoperative outcomes for knee patients, allowing quicker recovery after surgery to preoperative outcome levels in this patient subgroup.

Although surgery had a considerable impact on function, other dimensions assessed by PROMs such as cognitive, emotional and social factors were much less affected. Postoperatively, function improved considerably while cognitive, emotional and support factors were stable for the majority of patients. These dimensions therefore appear less critical as an outcome measure with patients undergoing arthroplasty during the acute recovery period. Other studies have demonstrated the limited importance of psychosocial dimensions in the acute recovery phase of arthroplasty patients [34, 35]. Arthroplasty is generally planned well in advance, with generally predictable results when compared to other types of surgeries, such as unplanned cardiac surgery. It is therefore possible that arthroplasty is generally less emotionally involving than other types of surgeries, limiting the usefulness of psychosocial outcome measures in the acute recovery phase.

When comparing performance measures with PROMs, results show that the amount of change was similar for both categories of tools in the days after surgery. However, the relationships between performance measures and PROMs were moderate at best. PROMs were related to patient perceived improvement in the days following surgery, whereas performance measures were related to patient perceived improvement only after 6 weeks. It thus appears that patient perception of function and actual functional performance are somewhat different dimensions in the days after surgery, but more related after several weeks. In the first days after surgery, it is probably difficult for patients to determine their level of function since they have performed limited functional tasks. Also, perception of function is probably influenced by other factors present shortly after surgery, such as pain, nausea/dizziness and level of alertness. When the effects of anaesthesia subside and patients gradually increase their levels of activity in the weeks after surgery, it is probably easier for patients to ascertain their level of function. It could therefore be important to assess both perceived function and functional performance in the days after surgery, but not necessarily after 6 weeks. Another factor that could explain this discrepancy could be previous arthroplasties, where perceptions of function for patients having undergone previous arthroplasties could be different from patients undergoing an arthroplasty for the first time, based on past experience and clearer expectations. Other studies have demonstrated the discrepancies between perceived function and functional performance after arthroplasty [3–5, 36, 37]. Studies are however needed to understand these discrepancies and assess their impact, since the consequences of patients over or underestimating their perceived function is not known. When there is a discrepancy, it could be argued that the focus should be functional performance, since it objectively demonstrates what the patient is capable of doing. Conversely, it could also be argued that patient perception is the most important, since it reflects what the patient is going to try at home. A clinical objective after the functional performance assessment could be to educate the patient, in order to reduce this discrepancy. Research is however needed to determine how to manage these discrepancies.

Both performance measures (ILAS and TUG) were highly related in the days after surgery, but less in the weeks after. This reduced relationship could be explained by the ceiling effect of the ILAS in the weeks after surgery. The ILAS changed more in the days after surgery, while the TUG changed more in the weeks after. These results seem to demonstrate that the ILAS assesses more basic function than the TUG, which is consistent with the tasks assessed by both tools. The ILAS assesses the patient’s capacity to do basic tasks in isolation, while the TUG is a timed combination of tasks. Although there were slight result differences between these tools, they were quite similar in the way they performed. The TUG however takes much less time, space and resources than the ILAS. The TUG can also be used over a wider period since it did not show a ceiling effect as the ILAS did. An MIC of 43.4 s over 6 weeks was also obtained for the TUG for hip patients, while none was obtained for the ILAS. Thus, the TUG appears superior to the ILAS to assess function in the first 6 weeks following arthroplasty. Several studies have also demonstrated the validity of TUG in arthroplasty patients, while studies are less frequent for the ILAS [6]. Although specificity was high for the TUG MIC, sensitivity was moderate. Thus, patients below the MIC TUG cut-off could be misclassified as not having improved when, in fact, they did improve. Conversely, patients above the cut-off can be considered improved with more certainty. High specificity combined with moderate sensitivity has been found in another study aimed at identifying the TUG MIC in patients with hip osteoarthritis [21].

Contrary to other studies that demonstrated the lack of responsiveness of PROMs in the acute recovery phase [3–5, 36, 37], the personal status subscale of the RHDS was shown to be responsive according to SRMs. MICs were also obtained for the global and personal status subscale of the RHDS for knee patients. These results are probably attributable to the fact that the RHDS was specifically developed for the postoperative recovery phase, where other PROMs studied in arthroplasty patients were originally developed for osteoarthritis patients and assess more advanced tasks. The RHDS can however only be used during hospitalization, which is generally brief in arthroplasty, limiting its usefulness as an outcome measure. As for the PQRS, it appears more appropriate in the longer term since it assesses the patient’s capacity to return to preoperative status. There thus appears to be a need for a postoperative recovery PROM developed specifically for arthroplasty patients, since there are currently none.

The study results are limited to the context in which they were collected, i.e. a large Canadian university-affiliated hospital performing a significant number of arthroplasties. Care should be taken when transposing the present results to other settings. This study should therefore be replicated in other contexts. The rate of ineligible patients is also relatively high, although this can mostly be explained by out-of-town patients not available for follow-ups. Subgroup analyses comparing knee and hip arthroplasties were also not possible due to the limited sample size. Finally, outcome measure results could have been influenced by the presence of osteoarthritis in other joints, but this was not assessed.