Both total hip and total knee replacements are known to be highly cost-effective; so why are there so many people waiting in severe pain for this treatment in the UK? Around the world, there's an increasing demand for healthcare interventions to be economically viable and cost-effective. To assess this, a ‘quality-adjusted life years’ (QALY) is often calculated based on the EuroQol-5D (EQ-5D) score, as well as population demographics such as life expectancy. In this interview, Mr Andrew Duckworth speaks to Edinburgh colleagues Ms Chloe Scott and Professor Colin Howie and discusses their study entitled ‘‘Worse than death’ and waiting for joint replacement’, which was published in the August 2019 edition of The Bone & Joint Journal.

The aim of your study was to look at patients awaiting hip and knee arthroplasty and categorise their health state in particular category known ‘worse than death’. Can you give us a brief background to the paper?

Over many years orthopaedics has looked carefully at the outcome of treatments that it has undertaken. We've developed very sophisticated scores, such as the Oxford score so that we can detect differences in the way that we operate. More recently, across the country, and particularly in our department, we noticed that more and more patients were complaining about the time that we kept them waiting for surgery. We thought it'd be interesting to see if the waiting time had made any difference to their state.

The EQ-5D is a general score and it's used across all specialties in medicine. It was developed in the latter half of the last century, and it covers five domains; mobility, self-care, ability to perform unusual tasks, pain and anxiety and depression. In orthopaedics we've largely ignored it because there are only three scores that you can give of those five domains; it perhaps didn't give us enough granularity to make decisions.

However, like many people, we have discovered that this was more informative than we had thought in the past. What happens in the EQ-5D is you don't just add up the score for each domain. Each domain is given a number one to three, and each state for each individual patient is allocated one of 243 states. The difference between the EQ-5D and other scores is that those 243 states were given to members of the public in each country in Europe and it was up to them to assess how long they’d like to live when in that particular state. Each state was allocated a number and the numbers go from -1 to 1, with 0 being a state on the borderline between what the public would regard as acceptable or unacceptable and anything less than 0 is a state worse than death.

Since then, they've taken these numbers and looked at the cost utility, i.e. how much healthcare improvement you give the patient over time by your treatment. You allocate the costs of giving the treatment and the cost to society of not giving the treatment; that's how a QALY is worked out.

In relation to hip and knee arthroplasty, what do we already know about cost-effectiveness in terms of the QALY in particular?

For a treatment to be considered cost-effective, NICE has determined that it should cost less than £20,000 per QALY gained. For cancer therapies this figure is increased to less than £100,000 per QALY gained. In addition to being highly clinically effective, both total hip and total knee replacements are known to be highly cost-effective too. Work from our centre by Paul Jenkins and colleagues published in The Bone & Joint Journal in 2013, shows that for total hip replacement, the cost of the QALY is £1371, and for total knee replacement, it's £2101.

Can you describe the type of prosthesis we use in our standard sort of postoperative protocol?

Our centres are a university teaching hospital, and we perform about 800 total hips and 800 total knees per year in addition to revision arthroplasty. Our standard is to use cemented implants for both procedures. For a total hip replacement, we use cemented Exeter stem with a cemented contemporary cup, normally implanted by a posterior approach. For a total knee replacement, we use the triathlon cruciate-retaining TKA inserted using a measured resection technique.

Postoperatively, both total hips and total knees are allowed to wait there or from the first postoperative morning. They start initially with a frame and progressed to sticks and they normally go home on day two to four when they can do stairs independently with their sticks. They receive daily physiotherapy while in hospital. In addition to occupational therapy and foot, after discharge they're reviewed at six to eight weeks by our team of arthroplasty practitioners and patient-reported outcome measures (PROMs) are routinely collected.

Please can you give us a bit more detail regarding the patient questionnaire and how the waiting list and time data was obtained and calculated.

PROM questionnaires include the EQ-5D as we've already mentioned, as well as the appropriate Oxford score, either hip or knee, and a series of detailed comorbidity questions, which cover 12 specific conditions, including heart failure, a MI, stroke, peripheral arterial disease, CAPD, diabetes, connected tissue disorders and inflammatory arthropathies, and kidney and liver disease.

Using the UK scoring system, the possible scores for the EQ-5D index range from -0.6 to one, 1 defined as full health;0 is death and negative scores are, therefore, ‘worse than death’. The questionnaire also included Oxford knee score and comorbidity scores. These were applied to patients two to three weeks prior to their surgery in a pre-assessment clinic where they were completed independently. The same scores were posted out to patients at a year. In addition to these PROMs, demographic data, including the Scottish Index of Multiple Deprivation and body mass index were collected from all the patients.

How was the waiting list and time data collected?

The paper isn't primarily about waiting lists, but we thought including this as a variable was important in the analysis, because we wondered whether waiting undue amounts of time for joint replacement would affect your list and death status. The length of time from the decision to operate to the actual surgery was provided by our waiting list office.

Can you give us a brief overview of the analysis of the PROMs and how the receiver operating characteristic (ROC) curve analysis was used to identify the thresholds of both the Oxford hip and knee scores associated with the state worse than death category.

Total hip and total knee patients were separately analysed using univariate analysis to identify significant associations with their preoperative worse than death status. As our sample sizes were large (2073 hips and 2168 knees), a number of variables were statistically significant at the 10% level or less; a multivariate analysis was performed on all of these variables (multivariate binary, logistic regression analysis performed using SPSS). Oxford scores, as we'll go on to find, were found to be independently associated with worse than death status for both hips and knees, this was further investigated using ROC curve analysis. This a method you use to determine the sensitivity and specificity of a test and can be used, therefore, to determine a threshold value or a cut-point of a continuous variable. Like the Oxford hip score, Oxford knee score is associated with a dichotomous outcome, like worse than death status.

Moving onto the findings of the study; how many patients prior to joint replacement were defined as having a worse than death status?

Prior to total hip replacement, 91 patients (19%) were defined as being in a health state worse than death, with a negative EQ-5D score. Of these worse than death patients, 99% reported extreme levels of pain. Following total hip arthroplasty, the median EQ-5D index for the whole cohort improved from 0.36 to 0.79. The number of patients worse than death reduced from 19% to 2%. It's important to note that these patients weren’t the same patients; the ones that were worse than death postoperatively weren't necessarily preoperatively, so it doesn't seem to be a function of personality.

The multivariate analysis determined that for hips, having a poor joint specific hip function, measured using the Oxford hip score was an independent predictor of worse than death status in addition to the presence of CAPD. None of the other demographic variables and none of the other comorbidities were significant in the multivariate analysis. The ROC analysis was then associated with a highly significant area under the curve. We identified a threshold Oxford hip score value of 14.5 as being associated with worse than death status with fairly high sensitivity of 80% and a specificity of 75%. The key message here is that worse than death status was significantly determined by the hip specific function rather than other comorbidities and general health.

The second key message is that those patients who were worse than death preoperatively achieved significantly worse outcomes at one-year with Oxford hip scores, which were a mean of 7.7 points worse than those who weren't worse than death preoperatively. This is important because this value exceeds the minimal clinically important difference for this score, which is 5. It was also associated with worse patient satisfaction rates. Patients worse than death preoperatively were satisfied 85% of the time versus 92% when not worse than death.

Let’s move on to the knee replacements, what did you find regarding these outcomes we've just discussed and how did that compare to the total hip replacements?

The picture was very similar for knees though it wasn’t quite as bad; 12% of patients were worse than death preoperatively, all of whom again reported extreme levels pain. Following the replacement, the median EQ-5D index improved from 0.59 to 0.76 and the number of patients worse than death reduced from 12% to 3%. Again, the multivariate analysis determined the Oxford knee score was an independent association with worse than death status. This time though, peripheral arterial disease was the only comorbidity associated with this status. Again, the ROC analysis provided adequate sensitivity and specificity and suggested a cut-point of 17.5.

In terms of the outcomes at a year, the pattern was similar. Patients worse than death preoperatively achieved worse one-year outcomes with a mean Oxford knee score, which was 8.2 points less than those not worse than death and satisfaction rates of 73% versus 84. Again, that exceeds the MCID.

What did you find with regards to the waiting list status of both?

Because it was over a relatively short period of time, it doesn't fully achieve statistical significance, but there's absolutely no doubt that the patients who wait longer have a higher worse than death score. The interesting thing is that the patients who were in a state of worse than death were often seen earlier, perhaps because GPs are good at picking out patients who are suffering.

What are the implications of the study?

19% of patients with joint disease of the hip and 12% within the knee are defined as having a state worse than death using the EQ-5D score. The strengths of this study are without doubt. There's a large number of patients from a robust, prospective database, and obviously excellent and robust analysis has been performed throughout.

What do you feel are any potential limitations of the data?

First of all, we have to accept that the state worse than death is a state that was given by members of the public who didn't have any diseases whatsoever, but it was allocated by over 3500 people, in an independent review, which had nothing to do with arthritis of the hip.

It has strengths, but it also has a weakness; if you ask the people who are in a state worse than death, only 15% of them would actually like to be dead. However, they do see their life as miserable and it's loss of function and pain that are the two predictors of it.

The other major thing at this point is the difference between the preoperative score and the postoperative score. Almost everybody gets better and loses the pain within two days of surgery, which is a huge difference.

You emphasise in the discussion the importance of interpreting results in view of how the EQ-5D score is calculated. Can you expand on that?

As we have already discussed, although the EQ-5D is widely used, it's not necessarily widely understood. We've talked about the 243 potential health states that it defines and the fact that this definition is based on nearly 3500 members of the UK population valuing these states and deciding that they wouldn't be able to endure living in them. A criticism of the paper and of the worse than death status is that it doesn't apply to the individual patients. We haven't asked the individual patients ‘do you feel that you're in a state worse than death’? I think it's significant that it's a value judgment that has been placed on these health things by the general population, many of whom will ultimately go on to develop degenerative joint diseases. I don't think you can ignore the fact that the public obviously consider these conditions to be serious and unpleasant to put up with. In the context of having a treatment that works reliably and is highly cost-effective, I think it's difficult to argue that arthroplasty shouldn't be delivered.

How does this compare to other medical conditions?

Because QALYs are used to calculate the cost-effectiveness of all medical interventions EQ-5D indexes are available for a number of chronic health conditions. If we consider our mean preoperative EQ-5D index for end-stage degenerative joint disease of the hip, which was 0.30 and of the knee, which was 0.43, these are notably worse than those reported in the literature for a number of chronic health conditions, which include type two diabetes, which is 0.78, heart failure 0.64, COPD 0.52, asthma 0.42, the list goes on.

I think this is especially important to appreciate in the context of resource management in the health service, where hip and knee arthroplasty are often some of the first procedures to be reduced or rationed in order to save money. They'd been designated as procedures of limited clinical value by number of clinical commissioning groups down south. What this paper shows is that doing this isn't a benign act and leaving patients in pain with end-stage hip or knee degenerative joint disease has significant effects, both on their health-related quality of life and on the outcome that they can expect to achieve when they are treated for joint replacement surgery.

What do you feel is the next step particularly when considering that increasingly referrals to secondary care for joint replacement are being monitored by various criteria, such as body mass index and PROMs?

First of all, patients with degenerative joint disease have a worse health status than many of the conditions that are not limited to access to healthcare and where the cost benefit is very much poorer than that. The joint replacement is a major question for those commissioning healthcare. The other thing for orthopaedic surgeons is to realise that we do something which is hugely beneficial for our patients; the outcome is almost universally successful. We take somebody who is ill and we almost instantly make them better.

It's cruel to create a healthcare system, which claims to have universal access, free at the point of delivery but actually limits access to the patients who would benefit most. For one of the most common surgical procedures carried out in the health service! No matter how it's been done to reduce instant costs, within a year, it has an increased cost to society. And more importantly, to the patients, future voters. They sit in the community in a state worse than death.

If you’d like to read the full paper you can do so here. You can listen to the podcast version of this interview here.