Thesis Thursday: Matthew Quaife

On the third Thursday of every month, we speak to a recent graduate about their thesis and their studies. This month’s guest is Dr Matthew Quaife who has a PhD from the London School of Hygiene and Tropical Medicine. If you would like to suggest a candidate for an upcoming Thesis Thursday, get in touch.

Title
Using stated preferences to estimate the impact and cost-effectiveness of new HIV prevention products in South Africa
Supervisors
Fern Terris-Prestholt, Peter Vickerman
Repository link
http://researchonline.lshtm.ac.uk/4646708

Stated preferences for what?

Our main study looked at preferences for new HIV prevention products in South Africa – estimating the uptake and cost-effectiveness of multi-purpose prevention products, which protect against HIV, pregnancy and STIs. You’ll notice that condoms do this, so why even bother? Condom use needs both partners to agree (for the duration of a given activity) and, whilst female partners tend to prefer condom-protected sex, there is lots of evidence that male partners – who also have greater bargaining power in many contexts – do not.

Oral pre-exposure prophylaxis (PrEP), microbicide gels, and vaginal rings are new products which prevent HIV infection. More importantly, they are female-initiated and can generally be used without a male partner’s knowledge. But trials and demonstration projects among women at high risk of HIV in sub-Saharan Africa have shown low levels of uptake and adherence. We used a DCE to inform the development of attractive and usable profiles for these products, and also estimate how much additional demand – and therefore protection – would be gained from adding contraceptive or STI-protective attributes.

We also elicited the stated preferences of female sex workers for client risk, condom use, and payments for sex. Sex workers can earn more for risky unprotected sex, and we used a repeated DCE to predict risk compensation (i.e. how much condom use would change) if they were to use HIV prevention products.

What did you find most influenced people’s preferences in your research?

Unsurprisingly for products, HIV protection was most important to people, followed by STI and then pregnancy protection. But digging below these averages with a latent class analysis, we found some interesting variation within female respondents: over a third were not concerned with HIV protection at all, instead strongly caring about pregnancy and STI protection. Worryingly, these were more likely to be respondents from high-incidence adolescent and sex worker groups. The remainder of the sample overwhelmingly chose based on HIV protection.

In the second sex worker DCE, we found that using a new HIV prevention product made condoms become less important and price more important. We predict that the price premium for unprotected sex would reduce by two thirds, and the amount of condomless sex would double. This is an interesting labour market/economic finding, but – if true – also has real public health implications. Since economic changes mean sex workers move from multi-purpose condoms to single-purpose products which need high levels of adherence, we thought this would be interesting to model.

How did you use information about people’s preferences to inform estimates of cost-effectiveness?

In two ways. First, we used simple uptake predictions from DCEs to parameterise an HIV transmission model, allowing for condom substitution uptake to vary by condom users and non-users (it was double in the latter). We were also able to model the potential uptake of multipurpose products which don’t exist yet – e.g. a pill protecting from HIV and pregnancy. We predict that this combination, in particular, would double uptake among high-risk young women.

Second, we predicted risk compensation among sex workers who chose new products instead of condoms. We were also able to calculate the price elasticity of supply of unprotected sex, which we built into a dynamic transmission model as a determinant of behaviour.

Can discrete choice experiments accurately predict the kinds of behaviours that you were looking at?

To be honest, when I started the PhD I was really sceptical – and I still am to an extent. But two things make me think DCEs can be useful in predicting behaviours.

First is the data. We published a meta-analysis of how well DCEs predict real-world health choices at an individual level. We only found six studies with individual-level data, but these showed DCEs predict with an 88% sensitivity but just a 34% specificity. If a DCE says you’ll do something, you more than likely will – which is important for modelling heterogeneity in uptake. We desperately need more studies following up DCE participants making real-world choices.

Second is the lack of alternative inputs. Where products are new and potential users are inexperienced, modellers pick an uptake number/range and hope for the best. Where we don’t know efficacy, we may assume that uptake and efficacy are linearly related – but they may not be (e.g. if proportionately more people use a 95% effective product than a 45% effective one). Instead, we might assume uptake and efficacy are independent, but that might sound even less realistic. I think that DCEs can tell us something about these behaviours that are useful for the parameters and structures of models, even if they are not perfect predictors.

Your tread the waters of infectious disease modelling in your research – was the incorporation of economic factors a challenge?

It was pretty tricky, though not as challenging as building the simple dynamic transmission model as a first exposure to R. In general, behaviours are pretty crudely modelled in transmission models, largely due to assumptions like random mixing and other population-level dynamics. We made a simple mechanistic model of sex work based on the supply elasticities estimated in the DCE, and ran a few scenarios, each time estimating the impact of prevention products. We simulated the price of unprotected sex falling and quantity rising as above, but also overlaid a few behavioural rules (e.g. Camerer’s constant income hypothesis) to simulate behavioural responses to a fall in overall income. Finally, we thought about competition between product users and non-users, and how much the latter may be affected by the market behaviours of the former. Look out for the paper at Bristol HESG!

How would you like to see research build on your work to improve HIV prevention?

I did a public engagement event last year based on one statistic: if you are a 16-year old girl living in Durban, you have an 80% lifetime risk of acquiring HIV. I find it unbelievable that, in 2018, when millions have been spent on HIV prevention and we have a range of interventions that can prevent HIV, incidence among some groups is still so dramatically and persistently high.

I think research has a really important role in understanding how people want to protect themselves from HIV, STIs, and pregnancy. In addition to highlighting the populations where interventions will be most cost-effective, we show that variation in preferences drives impact. I hope we can keep banging the drum to make attractive and effective options available to those at high risk.

Chris Sampson’s journal round-up for 2nd April 2018

Every Monday our authors provide a round-up of some of the most recently published peer reviewed articles from the field. We don’t cover everything, or even what’s most important – just a few papers that have interested the author. Visit our Resources page for links to more journals or follow the HealthEconBot. If you’d like to write one of our weekly journal round-ups, get in touch.

Quality-adjusted life-years without constant proportionality. Value in Health Published 27th March 2018

The assumption of constant proportional trade-offs (CPTO) is at the heart of everything we do with QALYs. It assumes that duration has no impact on the value of a given health state, and so the value of a health state is constant regardless of its duration. This assumption has been repeatedly demonstrated to fail. This study looks for a non-constant alternative, which hasn’t been done before. The authors consider a quality-adjusted lifespan and four functional forms for the relationship between time and the value of life: constant, discount, logarithmic, and power. These relationships were tested in an online survey with more than 5,000 people, which involved the completion of 30-40 time trade-off pairs based on the EQ-5D-5L. Respondents traded off health states of varying severities and durations. Initially, a saturated model (making no assumptions about functional form) was estimated. This demonstrated that the marginal value of lifespan is decreasing. The authors provide a set of values attached to different health states at different durations. Then, the econometric model is adjusted to suit a power model, with the power estimated for duration expressed in days, weeks, months, or years. The power value for time is 0.415, but different expressions of time could introduce bias; time expressed in days (power=0.403) loses value faster than time expressed in years (power=0.654). There are also some anomalies that arise from the data that don’t fit the power function. For example, a single day of moderate problems can be worse than death, whereas 7 days or more is not. Using ‘power QALYs’ could be the future. But the big remaining question is whether decisionmakers ought to respond to people’s time preferences in this way.

A systematic review of studies comparing the measurement properties of the three-level and five-level versions of the EQ-5D. PharmacoEconomics [PubMed] Published 23rd March 2018

The debate about the EQ-5D-5L continues (on Twitter, at least). Conveniently, this paper addresses a concern held by some people – that we don’t understand the implications of using the 5L descriptive system. The authors systematically review papers comparing the measurement properties of the 3L and 5L, written in English or German. The review ended up including 24 studies. The measurement properties that were considered by the authors were: i) distributional properties, ii) informativity, iii) inconsistencies, iv) responsiveness, and v) test-retest reliability. The last property involves consideration of index values. Each study was also quality-assessed, with all being considered of good to excellent quality. The studies covered numerous countries and different respondent groups, with sample sizes from the tens to the thousands. For most measurement properties, the findings for the 3L and 5L were very similar. Floor effects were generally below 5% and tended to be slightly reduced for the 5L. In some cases, the 5L was associated with major reductions in the proportion of people responding as 11111 – a well-recognised ceiling effect associated with the 3L. Just over half of the studies reported on informativity using Shannon’s H’ and Shannon’s J’. The 5L provided consistently better results. Only three studies looked at responsiveness, with two slightly favouring the 5L and one favouring the 3L. The latter could be explained by the use of the 3L-5L crosswalk, which is inherently less responsive because it is a crosswalk. The overarching message is consistency. Business as usual. This is important because it means that the 3L and 5L descriptive systems provide comparable results (which is the basis for the argument I recently made that they are measuring the same thing). In some respects, this could be disappointing for 5L proponents because it suggests that the 5L descriptive system is not a lot better than the 3L. But it is a little better. This study demonstrates that there are still uncertainties about the differences between 3L and 5L assessments of health-related quality of life. More comparative studies, of the kind included in this review, should be conducted so that we can better understand the differences in results that are likely to arise now that we have moved (relatively assuredly) towards using the 5L instead of the 3L.

Preference-based measures to obtain health state utility values for use in economic evaluations with child-based populations: a review and UK-based focus group assessment of patient and parent choices. Quality of Life Research [PubMed] Published 21st March 2018

Calculating QALYs for kids continues to be a challenge. One of the challenges is the choice of which preference-based measure to use. Part of the problem here is that the EuroQol group – on which we rely for measuring adult health preferences – has been a bit slow. There’s the EQ-5D-Y, which has been around for a while, but it wasn’t developed with any serious thought about what kids value and there still isn’t a value set for the UK. So, if we use anything, we use a variety of measures. In this study, the authors review the use of generic preference-based measures. 45 papers are identified, including 5 different measures: HUI2, HUI3, CHU-9D, EQ-5D-Y, and AQOL-6D. No prizes for guessing that the EQ-5D (adult version) was the most commonly used measure for child-based populations. Unfortunately, the review is a bit of a disappointment. And I’m not just saying that because at least one study on which I’ve worked isn’t cited. The search strategy is likely to miss many (perhaps most) trial-based economic evaluations with children, for which cost-utility analyses don’t usually get a lot of airtime. It’s hard to see how a review of this kind is useful if it isn’t comprehensive. But the goal of the paper isn’t just to summarise the use of measures to date. The focus is on understanding when researchers should use self- or proxy-response, and when a parent-child dyad might be most useful. The literature review can’t do much to guide that question except to assert that the identified studies tended to use parent–proxy respondents. But the study also reports on some focus groups, which are potentially more useful. These were conducted as part of a wider study relating to the design of an RCT. In five focus groups, participants were presented with the EQ-5D-Y and the CHU-9D. It isn’t clear why these two measures were selected. The focus groups included parents and some children over the age of 11. Unfortunately, there’s no real (qualitative) analysis conducted, so the findings are limited. Parents expressed concern about a lack of sensitivity. Naturally, they thought that they knew best and should be the respondents. Of the young people reviewing the measures themselves, the EQ-5D-Y was perceived as more straightforward in referring to tangible experiences, whereas the CHU-9D’s severity levels were seen as more representative. Older adolescents tended to prefer the CHU-9D. The youths weren’t so sure of themselves as the adults and, though they expressed concern about their parents not understanding how they feel, they were generally neutral to who ought to respond. The older kids wanted to speak for themselves. The paper provides a good overview of the different measures, which could be useful for researchers planning data collection for child health utility measurement. But due to the limitations of the review and the lack of analysis of the focus groups, the paper isn’t able to provide any real guidance.

Credits

 

Bad reasons not to use the EQ-5D-5L

We’ve seen a few editorials and commentaries popping up about the EQ-5D-5L recently, in Health Economics, PharmacoEconomics, and PharmacoEconomics again. All of these articles have – to varying extents – acknowledged the need for NICE to exercise caution in the adoption of the EQ-5D-5L. I don’t get it. I see no good reason not to use the EQ-5D-5L.

If you’re not familiar with the story of the EQ-5D-5L in England, read any of the linked articles, or see an OHE blog post summarising the tale. The important part of the story is that NICE has effectively recommended the use of the EQ-5D-5L descriptive system (the questionnaire), but not the new EQ-5D-5L value set for England. Of the new editorials and commentaries, Devlin et al are vaguely pro-5L, Round is vaguely anti-5L, and Brazier et al are vaguely on the fence. NICE has manoeuvred itself into a situation where it has to make a binary decision. 5L, or no 5L (which means sticking with the old EQ-5D-3L value set). Yet nobody seems keen to lay down their view on what NICE ought to decide. Maybe there’s a fear of being proven wrong.

So, herewith a list of reasons for exercising caution in the adoption of the EQ-5D-5L, which are either explicitly or implicitly cited by recent commentators, and why they shouldn’t determine NICE’s decision. The EQ-5D-5L value set for England should be recommended without hesitation.

We don’t know if the descriptive system is valid

Round argues that while the 3L has been validated in many populations, the 5L has not. Diabetes, dementia, deafness and depression are presented as cases where the 3L has been validated but the 5L has not. But the same goes for the reverse. There are plenty of situations in which the 3L has been shown to be problematic and the 5L has not. It’s simply a matter of time. This argument should only hold sway if we expect there to be more situations in which the 5L lacks validity, or if those violations are in some way more serious. I see no evidence of that. In fact, we see measurement properties improved with the 5L compared with the 3L. Devlin et al put the argument to bed in highlighting the growing body of evidence demonstrating that the 5L descriptive system is better than the 3L descriptive system in a variety of ways, without any real evidence that there are downsides to the descriptive expansion. And this – the comparison of the 3L and the 5L – is the correct comparison to be making, because the use of the 3L represents current practice. More fundamentally, it’s hard to imagine how the 5L descriptive system could be less valid than the 3L descriptive system. That there are only a limited number of validation studies using the 5L is only a problem if we can hypothesise reasons for the 5L to lack validity where the 3L held it. I can’t think of any. And anyway, NICE is apparently satisfied with the descriptive system; it’s the value set they’re worried about.

We don’t know if the preference elicitation methods are valid for states worse than dead

This argument is made by Brazier et al. The value set for England uses lead time TTO, which is a relatively new (and therefore less-tested) method. The problem is that we don’t know if any methods for valuing states worse than dead are valid because valuing states worse than dead makes no real sense. Save for pulling out a Ouija board, or perhaps holding a gun to someone’s head, we can never find out what is the most valid approach to valuing states worse than dead. And anyway, this argument fails on the same basis as the previous one: where is the evidence to suggest that the MVH approach to valuing states worse than dead (for the EQ-5D-3L) holds more validity than lead time TTO?

We don’t know if the EQ-VT was valid

As discussed by Brazier et al, it looks like there may have been some problems in the administration of the EuroQol valuation protocol (the EQ-VT) for the EQ-5D-5L value set. As a result, some of the data look a bit questionable, including large spikes in the distribution of values at 1.0, 0.5, 0.0, and -1.0. Certainly, this justifies further investigation. But it shouldn’t stall adoption of the 5L value set unless this constitutes a greater concern than the distributional characteristics of the 3L, and that’s not an argument I see anybody making. Perhaps there should have been more piloting of the EQ-VT, but that should (in itself) have no bearing on the decision of whether to use the 3L value set or the 5L value set. If the question is whether we expect the EQ-VT protocol to provide a more accurate estimation of health preferences than the MVH protocol – and it should be – then as far as I can tell there is no real basis for preferring the MVH protocol.

We don’t know if the value set (for England) is valid

Devlin et al state that, with respect to whether differences in the value sets represent improvements, “Until the external validation of the England 5L value set concludes, the jury is still out.” I’m not sure that’s true. I don’t know what the external validation is going to involve, but it’s hard to imagine a punctual piece of work that could demonstrate the ‘betterness’ of the 5L value set compared with the 3L value set. Yes, a validation exercise could tell us whether the value set is replicable. But unless validation of the comparator (i.e. the 3L value set) is also attempted and judged on the same basis, it won’t be at all informative to NICE’s decision. Devlin et al state that there is a governmental requirement to validate the 5L value set for England. But beyond checking the researchers’ sums, it’s difficult to understand what that could even mean. Given that nobody seems to have defined ‘validity’ in this context, this is a very dodgy basis for determining adoption or non-adoption of the 5L.

5L-based evaluations will be different to 3L-based evaluations

Well, yes. Otherwise, what would be the point? Brazier et al present this as a justification for a ‘pause’ for an independent review of the 5L value set. The authors present the potential shift in priority from life-improving treatments to life-extending treatments as a key reason for a pause. But this is clearly a circular argument. Pausing to look at the differences will only bring those (and perhaps new) differences into view (though notably at a slower rate than if the 5L was more widely adopted). And then what? We pause for longer? Round also mentions this point as a justification for further research. This highlights a misunderstanding of what it means for NICE to be consistent. NICE has no responsibility to make decisions in 2018 precisely as it would have in 2008. That would be foolish and ignorant of methodological and contextual developments. What NICE needs to provide is consistency in the present – precisely what is precluded by the current semi-adoption of the EQ-5D-5L.

5L data won’t be comparable to 3L data

Round mentions this. But why does it matter? This is nothing compared to the trickery that goes on in economic modelling. The whole point of modelling is to do the best we can with the data we’ve got. If we have to compare an intervention for which outcomes are measured in 3L values with an intervention for which outcomes are measured in 5L values, then so be it. That is not a problem. It is only a problem if manufacturers strategically use 3L or 5L values according to whichever provides the best results. And you know what facilitates that? A pause, where nobody really knows what is going on and NICE has essentially said that the use of both 3L and 5L descriptive systems is acceptable. If you think mapping from 5L to 3L values is preferable to consistently using the 5L values then, well, I can’t reason with you, because mapping is never anything but a fudge (albeit a useful one).

There are problems with the 3L, so we shouldn’t adopt the 5L

There’s little to say on this point beyond asserting that we mustn’t let perfect be the enemy of the good. Show me what else you’ve got that could be more readily and justifiably introduced to replace the 3L. Round suggests that shifting from the 3L to the 5L is no different to shifting from the 3L to an entirely different measure, such as the SF-6D. That’s wrong. There’s a good reason that NICE should consider the 5L as the natural successor to the 3L. And that’s because it is. This is exactly what it was designed to be: a methodological improvement on the same conceptual footing. The key point here is that the 3L and 5L contain the same domains. They’re trying to capture health-related quality of life in a consistent way; they refer to the same evaluative space. Shifting to the SF-6D (for example) would be a conceptual shift, whereas shifting to the 5L from the 3L is nothing but a methodological shift (with the added benefit of more up-to-date preference data).

To sum up

Round suggests that the pause is because of “an unexpected set of results” arising from the valuation exercise. That may be true in part. But I think it’s more likely the fault of dodgy public sector deals with the likes of Richard Branson and a consequently algorithm-fearing government. I totally agree with Round that, if NICE is considering a new outcome measure, they shouldn’t just be considering the 5L. But given that right now they are only considering the 5L, and that the decision is explicitly whether or not to adopt the 5L, there are no reasons not to do so.

The new value set is only a step change because we spent the last 25 years idling. Should we really just wait for NICE to assess the value set, accept it, and then return to our see-no-evil position for the next 25 years? No! The value set should be continually reviewed and redeveloped as methods improve and societal preferences evolve. The best available value set for England (and Wales) should be regularly considered by NICE as part of a review of the reference case. A special ‘pause’ for the new 5L value set will only serve to reinforce the longevity of compromised value sets in the future.

Yes, the EQ-5D-3L and its associated value set for the UK has been brilliantly useful over the years, but it now has a successor that – as far as we can tell – is better in many ways and at least as good in the rest. As a public body, NICE is conservative by nature. But researchers needn’t be.

Credits