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.
Competition, prices and quality in the market for physician consultations. The Journal of Industrial Economics [RePEc] Published 26th April 2016
In Australia you have to pay for a consultation with your GP, and GPs are free to set whatever price they like. Meanwhile, patients are free to choose to visit any GP they like. This means that there is a lot of scope for competition on price as well as quality. Australia’s Medicare reimburses a fixed rate (depending on type) for each consultation, and so patients only need to pay whatever the GP charges in excess of this. But it is also possible for GPs to use ‘bulk billing’ – whereby no additional co-payment above the rebate is charged – for specific groups such as children or old people. This study looks at whether competition influences the price and quality of consultations. Data are taken from the MABEL survey, which is a panel study of doctors. It includes information on consultation duration and fees and the number of people who aren’t charged a co-payment. Rural GPs are excluded from the analysis and the final sample includes 1966 GPs. The authors construct a measure of the level of competition according to the distance between GPs and their rivals, with a focus on the third-nearest GP practice. Quality is specified as consultation length (the longer, the better). A variety of models are run to predict price and quality and to examine the effects of competition, using an extension of the ‘Vickrey-Salop’ model. Results show that the greater the distance from competitors, the lower the proportion of patients who are not charged a co-payment and the higher the average price for those who are charged. This suggests that competition reduces prices. The magnitude of these effects is greater for people living in areas with a higher socio-economic status. There is also a positive effect on consultation duration, though this is small. One key implication of these findings is that as concentration in the market for GP consultations increases prices are likely to rise. Whether or not having fewer practices would also impact on quality is not clear from this study.
What determines the shape of an EQ-5D index distribution? Medical Decision Making [PubMed] Published 25th April 2016
If you’ve ever collected EQ-5D(-3L) data and looked at the distribution of index scores, you’ll probably have noticed that it looks a bit wobbly. Generally we see a cluster of people around the 0.3 mark, another cluster around 0.8, and then a stack of people at 1. It seems unlikely that this should be representative of the true distribution of health states for all conditions in which it’s observed. This new study looks at whether this odd distribution might be explained by the nature of the classification system or by the nature of the weights applied to the states. The most obvious culprit for the UK MVH values is the N3 term, which takes a disproportionately big cut for anyone with at least one level 3 response. Using a simulated data set, the authors show that if EQ-5D health states were randomly allocated (i.e. each state was equally likely) then a bimodal distribution would still be observed. The authors also looked at real data from the NHS PROMs programme and from a clinical trial. For both pre-surgery hip replacement patients and pre-surgery varicose vein patients, the usual 2 (or 3) cluster pattern can be observed. To the naked eye it looks as if the cut-off may be at 0.5, while a kmeans clustering procedure suggested that it may lie closer to 0.3. Looking at the dimension-level data for the hip replacement patients, it becomes clear that much of the difference can be explained by the usual activities and pain and discomfort dimensions, as there is little variation in the other dimensions to affect the scores. This is complicated by the N3 effect, but nevertheless it appears that the nature of the classification system can cause clustering in at least some conditions. So the authors go about ranking all of the EQ-5D states according to their index score, with the effect of removing any impact of the weighting on the distribution. Still, we observe the clustering. So it looks like six of one, half a dozen of the other. Clusters can result from the classification system, and then the difference between these clusters can be reinforced by the nature of the scoring algorithm. A condition-specific measure for the hip replacement patients does not show clustering and scores have a unimodal distribution. However, scores within the EQ-5D clusters are also unimodal and there is some evidence that the EQ-5D clustering may be genuine. What this really comes down to is the fact that we need to abandon our blinkered view that the EQ-5D is all about the index score. It’s important to bear in mind that the extent to which the classification system and/or the weighting might affect the distribution of index scores will differ between patient groups. Looking at the distribution of responses at the dimension level can tell us a lot about the nature of this statistical challenge.
On the estimation of the cost-effectiveness threshold: why, what, how? Value in Health Published 23rd April 2016
One day in history, NICE came up with a £20-30,000 threshold for the amount the NHS should be willing to pay for a QALY. I have no idea where this came from, or why it has remained the same since its conception (in theory, if not in practice). Few other countries specify a threshold as clearly as NICE, but some do adopt an indicative range. This study reviews the literature around the estimation of cost-effectiveness thresholds and the views and methods that have been used to support them; focussing on questions of ‘why’, ‘what’ and ‘how’. Why? Because the threshold approach is convenient and useful, and even if it is a bit of a fudge we can at least understand how decisions are made. What? Depending on who you ask, it’s either the social value of a QALY or it’s the opportunity cost associated with new investment, and what this really comes down to is whether or not you take the budget to be fixed. How? This is the main target of the review. 38 articles were identified and divided into demand-side (29) and supply-side (9) studies. Most of the former used willingness-to-pay surveys while a few inferred from the value of a statistical life. Values from the willingness to pay studies ranged from €1,000 to €5,000,000. Most of the problems associated with these derive from limitations in the QALY estimation process itself. The value of a statistical life studies turned up slightly higher values on average. Of the 9 studies looking at the supply-side, 4 looked at past investment decisions and 5 tried to estimate empirically the marginal cost of a QALY. Most of the studies looking at investment decisions observed too much inconsistency to actually come up with a threshold range. An Australian study inferred a threshold of €32,000 to €58,000. Estimating the marginal cost of a QALY is where the Claxton modern classic of £12,936 (€14,141) comes in. Other studies came up with similar estimates, though with some substantially higher for specific disease areas. So it looks as if most countries that talk about thresholds are more inclined to the social value of a QALY view, rather than the opportunity cost approach, because their thresholds are higher. Given institutional arrangements, this makes sense to me, as I’ve discussed before on this blog. But what it does imply is that some interventions that may be determined to be cost-effective by current standards might not be an appropriate use of resources within given budget constraints.