Intelligent Health Blog

Refocusing Market Access for Difficult to diagnose diseases

Posted by Volv Global on Jun 26, 2019, 12:09:51 PM
Volv Global
Find me on:

By Christopher Rudolf, Founder and CEO at Volv Global

 

Redefining your strategy using machine learning with Volv inTrigue

In this series of conversations with Volv leaders, I have been talking with Dr Mike Tremblay, Volv’s Chief Scientific and Medical Officer, about what differentiates the company in terms of value delivery into markets for our customers. I hope that this is of interest to those of you working with difficult to diagnose diseases and rare diseases in your daily activities.

 

Volv’s 4 Market Access Pillars

 

Simply put, if a medicine is not available in a particular country, then patients in that country do not have access to it. ‘Market access’ is a term used to describe the process used by pharmaceutical companies to ensure that all patients who could benefit from a medicine can have access to it.

 

The market access process addresses the size of the treatment population and the uncertainty and risk as seen by payers, at the same time, to arrive at suitable pricing. By employing sophisticated machine learning and prediction models, it is possible to enable a new and potential disruptive approach to market access. We believe the new process is built on four pillars, which we will explain in turn:

 

1.     Precision identification of the treatment cohort

2.     Collaboration with tertiary centres and first prescribers

3.     Evidence-informed prevalence for pricing, reimbursement and dossier inclusion

4.     Improve quality of care itself with beneficial impact on patients.

Let’s look at each pillar in turn, as they apply to rare disease medicines.

1. Precision identification of the treatment cohort

 

Because rare diseases are by definition rare, clinicians are often unfamiliar with them. This lack of familiarity can lead to diagnostic errors, misdiagnosis and mistreatment, and it can cause the patient to be sent into the wrong treatment pathway. We suggest that using a prediction model to augment clinical reasoning has the potential not only to reduce diagnostic errors, but also to shorten the time to the correct diagnosis.

 

The time to correct diagnosis for a rare disease can be years, depending on the quality of the clinical diagnostic decision making. From a resource utilisation perspective, the evidence shows that some individuals will see six to 10 or more doctors to receive a correct diagnosis, and half of all individuals receive a least one incorrect diagnosis.

 

Figure 1: Time from 1st contact to correct diagnosis [from Molstar et al]

Figure 1: Time from 1st contact to correct diagnosis [from Molstar et al]

Do not using Clinical Coding

 

At Volv we also believe that it is important not to use ICPC or ICD10 codes to find patients, as in theory, they will have already been diagnosed! More challenging, is the idea to take the earliest possible signals by creating novel biomarkers that will find the patients as soon as possible.

 

A key feature of prediction models is the identification of predictive markers for a condition that are indicative, to a sufficient degree. These include biomarkers (e.g. phenotype/genotype), and cognitive and behavioural markers which need to be gathered from sparse data. 

Quick results and early validation of accuracy

 

The Volv prediction model has a particular strength to identify hidden, missed or misdiagnosed patients, who are in the wrong treatment pathway, from a retrospective observational analysis of clinical datasets. Using this approach enables a precision determination of the cohort of undiagnosed patients at risk of a specific condition.

A feature of the retrospective approach to patient cohort identification is the determination of a more precise prevalence.

 

Most rare diseases have estimated prevalence figures, normally determined solely from known treated patients, but this excludes missed or misdiagnosed patients, by definition. Figures so produced, which are erroneous, always under-report actual prevalence. With greater prevalence accuracy, it is also possible to construct a more evidence-informed incidence figure, which is often not known or not calculated for rare diseases.

2. Collaboration with tertiary centres and first prescribers

 

Normally, specialist medicines for rare diseases must be started by a clinician in a specialist treatment and referral centre or centres in a country; these are usually a tertiary centre at a university hospital. The treating clinicians usually treat or have knowledge of all the relevant patients in the country. These centres also have a wider role in clinical outreach and educational role to secondary centres and primary care.

 

Importantly, too, secondary and primary clinicians are unlikely to alter the specialist’s initial prescription, so they are critical in ensuring not just that the medicine is provided to the treatment population, but that it continues to be applied. 

3. Evidence-informed prevalence for pricing, reimbursement and dossier inclusion

 

Having got his far, we can see that machine learning may also have a disruptive impact on pricing and pricing discussions.

 

The ‘big tent’ approach that is envisioned uses real world data for robust treatment cohort membership criteria. This, coupled with a higher precision prevalence, directly addresses payer concerns, such as treatment uncertainty for the treatable patient population. Machine learning enables this uncertainty to be costed. Indeed, the focus on tertiary centres reduces the sales effort and channels investment normally required, and allows the pharmaceutical company instead to build collaborative working relationships for clinician and patient benefit to directly influence the non-ICER value for money analysis.

 

For market access, then, there is a more precise cohort model, and evidence-informed prevalence / incidence which can be used in reimbursement discussions, to assist payer determination of their costs in the present and in the future.

Wider applicability in the clinical setting

 

Additional benefits arising from the application of prediction modelling to a rare condition come in the form of a better understanding of the high value predictors of that condition. This new knowledge is available to augment clinical decision making by altering in what order symptoms are assessed for a diagnosis; the ‘stopping’ rule in the differential diagnosis (i.e. the criteria for the doctor settling on a diagnosis); and the development of ‘red flag’ decision tools to facilitate diagnostic insight by general practitioners and non-specialists. Dissemination of these capabilities from specialist centres enhances overall healthcare system performance and has the benefit of driving out system costs.

 

By centring market access on these few clinicians and centres, a lean market access strategy can be enabled which avoids costs that do not benefit the relevant treatable population and do not directly address the needs of these clinicians.

4. Improve quality of care itself with beneficial impact on patients

 

There is frequently little research on the personal and economic burden of most rare diseases, and there exist few studies of costs along the care pathway or evidence of the sources of delay and misdiagnosis in the care pathway. Indeed, the structure of the care pathway itself may be poorly understood and fail to identify key features of the patient experience of care, and hence key determinants of the standard of care.

We know, for instance, that many rare diseases have specific diagnostic difficulties which put patients at further risk. Examples include severe combined immune deficiency, Pompe's disease, Myelodysplastic syndrome and Fabry’s disease. [Blöß]

 

Doctors themselves are in need of specific support. While specialists largely depend on their own specialist networks and the quality of the published literature for the evidence base for their practice, primary care doctors depend on specialists. [Engel]

 

In determining system-level cost drivers for market access, pricing improvements in the standard of care alter the value for money equation. When rare diseases fail the usual ICER thresholds, the next level of analysis takes into account unmet needs and disease burden. However, few rare diseases have good whole system models, and the economic burden, in the absence of a full patient pathway analysis, is not often well documented. This approach facilitates that level of detail, by addressing these elements directly and reducing uncertainty. 

 

Volv Global redefines Market Access using AIRedefine your strategy

 

The result is a redefined market access strategy based on precision identification of the treatment cohort, detailed collaboration with tertiary centres and first prescribers, evidence-informed prevalence for pricing, reimbursement and dossier inclusion that then delivers improved quality of care itself with beneficial impact on patients.

At Volv, we have conceptualised and redefined market access as these pillars, using our machine learning and prediction capabilities as the pivotal point.

References

1.     Blöß S, Klemann C, Rother A-K, Mehmecke S, Schumacher U, Mücke U, et al. Diagnostic needs for rare diseases and shared prediagnostic phenomena: Results of a German-wide expert Delphi survey. Palau F, editor. PLOS ONE;12(2):e0172532.

2.     Molster C, Urwin D, Di Pietro L, Fookes M, Petrie D, van der Laan S, et al. Survey of healthcare experiences of Australian adults living with rare diseases. Orphanet Journal of Rare Diseases. 2016 Dec ;11(1).

3.     Engel P. et al. Physician and patient perceptions regarding physician training in rare diseases: the need for stronger educational initiatives for physicians. The Journal of Rare Disorders. 2013;1(2):1–15. 

 

 

Join Volv Global and the entire global AI healthcare community at Intelligent Health this September. 

 

Intelligent Health AI

11th-12th September 2019

Basel Congress Center, Switzerland

intelligenthealth.ai

 

Topics: AI, Medicine, MedTech