This issue of AHEHP poked some random vertebrae along the backbone of my career history, including digital health, screening, cost-effectiveness thresholds, and principles relating to decision modelling.
Starting with the zeitgeist, digital health.
This issue includes a report from a team involved in a NICE Medical Technology Guidance for a digital technology. Since my work on Sleepio, I’ve been absorbed in the difficulty of evaluating digital health technologies and providing access to them in the NHS. The technology in question for this study is myCOPD, which is an app-based technology to support patients’ self-management and pulmonary rehabilitation (among other functions).
The myCOPD assessment hit many of the big-ticket challenges for digital health evaluation, including reliance on limited evidence and ‘new’ types of evidence, big potential for cost-savings, and little or no risk to patients. The paper describes the assessment process, including the clinical and economic evidence reviews and the specifics of the company’s submission. Perhaps most interestingly, it reports on the changes that the External Assessment Group (EAG) made to the company’s costing model. The EAG demanded relatively few revisions. The main changes related to uptake, which was reduced from 100% to 46%. This is a recurring theme in NICE’s assessments of digital technologies, in a way that we don’t see for pharmaceuticals. Companies think lots of people will use their technology, while NICE isn’t so confident. The trouble with this is that forcing companies to cost at a lower uptake increases the price-per-patient at which they can market their technology, potentially making it less attractive to commissioners and thus NICE’s prophecy fulfils itself. Ultimately, myCOPD was recommended as a cost-saving technology.
Travelling back in time now to my PhD days, screening.
This issue includes an article on the design of screening programmes. The authors conducted a systematic review of colorectal cancer screening studies. They wanted to explore research relating to the variable inputs to the specification of a screening programme: screening intervals, age ranges, and test cut-offs. From 39 European cost-effectiveness studies, a few consistencies were found. Broader age ranges for eligibility and lower test cut-offs tend to be more cost-effective in colorectal cancer screening. Most importantly, annual screening is usually found to be more cost-effective than biennial. Despite this, and in contrast to North American recommendations, most screening programmes in Europe employ a biennial interval. Many researchers excluded annual screening as a comparator, and the authors speculate that this may be a result of analysts following the lead of clinical trialists, or that it may be an implicit acknowledgement of capacity constraints.
This issue includes two other systematic reviews. One considers decision models for anti-retroviral adherence interventions, finding that most interventions – such as smartphone reminders and counselling – tended to be cost-effective. Another review highlights the rising economic burden of non-communicable diseases in India.
Over to cost-effectiveness thresholds now, with a study from Germany.
Germany (notably? famously? stubbornly?) copes just fine without a cost-effectiveness threshold. Even researchers don’t seem very interested in the idea; there has been very little research into whether an implicit threshold decision rule might exist. This study goes some way towards filling that gap. The authors gathered information about technologies and their associated price negotiations from publicly available dossiers. Then they essentially estimated an incremental cost-effectiveness ratio for each technology and took an average to identify willingness to pay (WTP). Rather than focusing on identifying an overall threshold, WTP estimates were generated for specific clinical areas. Remember that Germany doesn’t make much use of QALYs. WTP was estimated in terms relevant to each clinical area and then linked to minimally important differences. The results are all over the place and difficult to interpret. For example, we see WTP estimates of €33,814 for a 1 percentage point reduction in HbA1c (for diabetes), and €10,971 for one life year gained (for cardiovascular disease). I’m not sure what we can do with such estimates. However, my main contention is that these figures do not represent willingness to pay. Unlike NICE et al., the German process operates with price negotiation built in. WTP is just one input here. Bargaining power is another. Market conditions probably matter too. The resulting price from a negotiation may not represent anything in particular when considered out of context, in which case an average across negotiations would be meaningless. The authors should have framed this as a study of shadow prices (or similar).
Finally, decision modelling.
There is one modelling study in this issue, which finds that CAR-T is not cost-effective for patients with diffuse large B-cell lymphoma in China. However, I was more interested in a novel kind of paper about modelling, which speaks to my interest in higher-level considerations. The authors report on the development and reception of a video series. The videos, which you can watch on YouTube, guide viewers through ideas relating to the role of social values in modelling (an underappreciated topic). 95 students were invited to complete a survey online after seeing the videos. The researchers’ intention was that the findings could inform the future use of the video series for educational purposes. In general, the students seemed to grasp the concepts quite well and enjoyed the videos. If you’re teaching modelling, you should browse through the results of the survey (despite their painful presentation) and consider how you might use this 18-minute video series.
Unfortunately, I can’t read everything. In this issue’s ‘also ran’ we have a commentary on health reform in Aotearoa New Zealand, estimates of the impact of COVID on EQ-5D-5L outcomes, Singaporean preferences for primary care services, and an evaluation of a payment scheme to promote biosimilar prescribing in France.