Guidance assessment consultation document for HTG10887 Digital platforms to support preparation before and rehabilitation after primary elective hip or knee replacement surgery: early-use assessment

Evidence relevant to the decision problem was identified for 13 out of 16 technologies, as summarised in table 2.

The primary outcomes were:

• health-related quality of life

• pain and joint-specific function

• meaningful function and participation in daily living

• confidence in recovery and self-management

• mobility

• functional performance

• face-to-face clinical review.
  
  Evidence was identified for all primary outcomes except for escalation to face-to-face clinical review.

The committee acknowledged the range of study designs outlined in the evidence base and the conclusions of the EAG. The EAG's conclusions were that, while there was no clear and consistent benefit of digital platforms on any primary outcomes, the evidence indicated that some digital platforms may be equal or superior to current practice. It felt that RCT evidence provided the most methodologically robust findings and that other study designs did not provide enough good quality evidence to justify an evidence generation recommendation. In particular, the committee did not think that evidence from single-arm studies was useful for decision making. This is because it did not think that it was possible to determine if the changes in clinical outcomes were a result of the digital platform used. The committee agreed that digital platforms with RCT evidence for either total hip replacement or total knee replacement could be used for further evidence generation in the joint in which the RCT was done, provided they were cost effective under pessimistic assumptions. It stated that comparative studies, ideally RCTs for digital platforms without this evidence, with a comparator that reflects current practice in the NHS are a priority for further research.

The committee noted that generalisability of the evidence to current NHS practice is a key limitation. This is because of the difficulty interpreting the comparative evidence because some studies had comparators that were standard care in the NHS, or similar, but most were compared with in-person physiotherapy. Clinical experts confirmed that access to face-to-face physiotherapy after leaving hospital is variable. They estimated that around 50% of people with a total knee replacement and 90% of people with a total hip replacement receive information to self-manage their recovery with no face-to-face physiotherapist support. The committee acknowledged that the higher intensity of physiotherapy used in many comparator arms, compared with standard care in the NHS, meant that the effects of digital platforms have the potential to be replicated if used in the NHS. But it stated that well-conducted comparative studies done in an NHS setting should be prioritised for further evidence generation.

The committee noted that the same digital platform would be offered before and after surgery. It acknowledged that many studies used digital platforms either before or after surgery, but not both before and after. The committee was unable to determine how studies on digital platforms used only before or after surgery would translate into the NHS. It said that this was because the degree to which the before and after surgery elements of the platforms contributed to potential benefits was uncertain. It concluded that evidence generation needs to capture the use of digital platforms both before and after surgery to replicate how the platforms would most likely be used in the NHS.

The committee noted that most of the evidence was on total knee replacement, with limited evidence on partial knee replacement and total hip replacement. Most of the evidence came from studies including mixed indications, with no evidence on hip resurfacing. Clinical experts estimated that around 10% to 20% of knee replacement procedures are partial, and that around 1% to 2% of hip replacement procedures are resurfacing procedures. They noted that the care pathways for all hip and knee replacement procedures are similar. But they said that the populations, recovery times and recovery expectations differ between each. The committee agreed that evidence for total knee replacement could be transferred to partial knee replacements and that evidence for total hip replacements could be transferred to hip resurfacing procedures. But it concluded that evidence for total knee replacement was not transferable to total hip replacement. So, the committee split the recommendations by hip and knee replacement, making recommendations for use with evidence generation based on the RCT evidence available per joint.

The committee discussed the risks of using digital platforms. It was concerned by the lack of adverse-event data because most studies did not report this outcome. Experts stated that the main types of complications after surgery are:

• acute surgical complications and medical complications, which happen in 1% to 2% of cases

• escalating rehabilitation needs after surgery, which is more common.
  
  Although digital platforms have red-flag mechanisms to direct people to seek further clinical support, the committee was concerned that digital platforms could reduce contact with healthcare professionals, which could risk complications being missed. The committee agreed that data on adverse events was a priority for evidence generation.

The committee acknowledged that there were 15 ongoing studies for 11 digital platforms, with the EAG considering 7 of them being likely to address key evidence gaps. Of the ongoing studies, there are 7 RCTs. Four RCTs are investigating digital platforms that had no RCTs identified in the evidence review (Good Boost, Joint Academy, moveUP and Slider). Five RCTs are being done in the UK, or include UK sites (BPMpathway, Good Boost, Joint Academy, Slider and Sword Thrive). The committee considered this when making its recommendations and agreed that ongoing research could address its concerns regarding the clinical and cost effectiveness of some digital platforms.

The EAG developed a decision-tree economic model for total hip and knee replacements with a 1-year time horizon. It compared digital platforms with standard care in the NHS for a 'generic' digital platform in the base case as well as the optimistic and pessimistic scenarios for each digital platform. The optimistic scenario used the claimed benefits of the technology for each input when available, or the base case values if there was no evidence. It assumed a reduction in length of hospital stay and a gain in quality-adjusted life years (QALYs). The pessimistic scenario assumed that no reduction in length of stay or QALY gains would occur with digital platforms. The committee acknowledged from the EAG assessment that all digital platforms were cost effective for both total hip and total knee replacement at a willingness to pay threshold of £20,000 per QALY gained in the optimistic scenario. Under pessimistic assumptions, BPMpathway, ForPatientApp, Good Boost, GoWellHealth, moveUP, mymobility, myrecovery and Physitrack remained cost effective for both total knee and total hip replacements. PreActiv remained cost effective for total knee replacement, but not total hip replacement.

The committee was cautious about the assumptions used to inform the model, specifically the:

• non-platform-specific use of clinical parameters

• weak evidence when platform-specific evidence was used in addition to the use of expert assumptions.
  
  The committee was uncertain whether the inputs were robust enough to support decision making based on cost effectiveness. But the committee acknowledged that it is an early model and that evidence generation is needed to provide robust outcome measures for each technology to inform future modelling. Because the model inputs were highly uncertain, the committee used the most conservative scenario in its decision making. It concluded that digital platforms that had RCT evidence for the joint, and were cost effective under pessimistic assumptions, can be recommended for use while further evidence is generated.

The committee discussed the plausibility of the QALY input used in the model. The model incorporated a QALY gain of 0.035 with digital platforms based on a non-randomised comparative UK study of 1,160 people having total knee replacement. The committee noted that although the evidence was the most generalisable to the NHS, the non-randomised and retrospective study design and large amounts of missing data introduced uncertainty into the robustness of the QALY gain. The committee was also concerned that the health-related quality-of-life outcomes were variable between identified studies, with some non-UK randomised trials finding no health-related quality-of-life benefit with digital platforms. The committee acknowledged that non-UK evidence was limited by the generalisability of the comparator. It was not confident in the robustness of the QALY estimates and whether the digital platforms would result in any QALY gain. As a result, it preferred the pessimistic scenario which omitted QALY gains with digital platforms. It concluded that evidence generation is needed on health-related quality of life from comparative studies done in the NHS.

The committee discussed the plausibility of the 0.5-day reduction in length of stay used in the model base case and optimistic scenario. It heard that the estimate was based on several non-randomised studies, which generally reported small reductions in length of stay, and clinical expert opinion. The committee acknowledged that length of stay is a key driver in the model but that a sensitivity analysis showed that platforms could still be cost effective if there was no reduction in length of stay in the base case. The committee also noted the Getting It Right First Time Orthopaedic Elective Surgery guide to delivering ambulatory perioperative care for joint replacement surgery (2023) which states that 'a stay of 0- or 1-night is the default expected'. The committee agreed that this might affect the potential benefit of digital platforms on length of stay in the future. Overall, the committee was not confident that using digital platforms would result in any reduction in length of stay. So, it preferred the pessimistic scenario which omitted length of stay reductions with digital platforms.

The committee understood that patient engagement and adherence to digital platforms was not included in the model because of inadequate evidence. The committee was concerned that not modelling adherence could impact the cost effectiveness of the technologies, especially because the limited evidence available suggested low compliance with digital platforms. But, the committee acknowledged that this is an early model and that adherence is a concern across all digital technologies. The committee concluded that patient engagement and adherence is a priority for evidence generation.