We are in a new world when it comes to access to and the use of data and evidence. Real world data and evidence takes us from structured studies to the routine delivery of healthcare, actual use of a medicine, and the patient’s actual health status.
What is knowing this worth and to whom?
Real world data is best understood in the context of decision making, or choices and how they are made and the consequences that flow from these decisions. To illustrate:
- Patients get the wrong treatment, i.e. they are misdiagnosed. This is a particular issue for patients with rare diseases which experience not just being treated for the wrong condition (i.e. they are in the wrong treatment pathway).
- Clinical reasoning may be flawed. The main issue here is medical misdiagnosis, and clinical reasoning itself (backward/forward driven reasoning), and the rules for diagnosis, guidelines, the order items are listed in the differential diagnosis, and behavioural heuristics that impact clinical reasoning. Medical errors are more associated with backward-driven reasoning, using the hypothetico-deductive method; while forward-driven begins with data, with fewer errors. Other reasoning concerns include: doctors are reluctant to make a rare disease diagnosis, called the zebra retreat; inappropriate referral and diagnosing a mimic and sending the patient off to the wrong specialist, not listening to the parents of ill children, and so on.
- The treatment is the problem. Even if the treatment is diagnostically correct, the success or failure of that treatment often depends on whether the patient is adherent. It also depends on whether there are adverse drug events which alter patient acceptance of the medicine. Some patients may be non-respondent to the treatment, too.
What does that mean?
Enabling much of this is the use of computational methods, and machine learning, which uses real world data to enable precision medicine, case finding, precision cohort identification and treatable populations.
Regulators currently rely on industry reporting for adverse drug event reporting. RWD could enable regulators to directly monitor the market in real time and identify AD events. This would alter the pharmacovigilance system. In addition, they could gather data on off-label use (for and against) to assess the validity of treatment claims.
RWE may speed regulatory approval as the studies are tightly focused, don’t make expansive product claims and benefits are easier to demonstrate, thereby reducing regulatory risk.
Reimbursement regulators, providers and payers benefit from the potential to improve the quality of care as delivered to patients. This is enhanced by the development of more sophisticated decision support tools built on e.g. computational approaches or embedded in electronic record systems. This includes, for example, ‘red flagging’ tools to improve differential diagnosis, identify mimics, and trigger appropriate clinical suspicion as well as ‘referral filters’ to address inappropriate referrals, and so on. All these improve the value for money equation, and importantly reduce treatment risk, which drives avoidable costs out of the system.
Pharmaceutical companies can use this type of data to inform their drug portfolio development process. This would bring some order to research and development to improve internal priority setting and assessment of research targets in particular to avoid research bias (the impact of behavioural heuristics in R&D decision making for instance). The impact on trials cannot be ignored, use of synthetic control arms, improve precision of trial cohorts to remove the 80% or more of individuals who are not selected for a trial and perhaps save 60% or more of trial costs, and predict trial outcomes.
The evidence base for dossier submissions can be evidence informed with respect to the size of the treatable population, and patient response to treatment, reducing payer risk which manifests itself in refusing to reimburse.
The table suggests just a few changes from current market access to data-driven RWE market access.
Needless to say, this alters the underlying assumptions of pharmacoeconomics, medicines pricing and positioning.
I’ve summarised just a few points in the table below, to distinguish between what today could be called “Push market access”, a sales driven approach to placement, to a “RWD/RWD market access” with reduced risk and improved opportunities for demonstrating product value.
| Stakeholder | ‘push’ Market Access | RWD/RWE Market Access |
|
|
|
|
| Patient | Risk of non-beneficial treatment | Precise patient treatment cohorts |
|
| Risk of mis-/missed diagnosis, medical error | Precision diagnosis with decision support tools |
|
|
|
|
| Clinician | Uncertainty of benefits of treatment and the ‘halo’ of uncertainty inherent in clinical decision making | Precision patient identification releases benefits through treatment targetting |
|
|
|
|
| Payers | Pay for uncertain benefits | Pay only for responders |
|
| Pay for treatment to non-responders | Precision medicine to demarcate treatable population |
|
| Pay for non-adherent treatments | Pay only for adherence, and risk reduction of non-adherence |
|
| Risk averse for uncertain treatable populations | Risk managed for an evidence treatable population |
|
|
|
|
| Pharma industry | Weak evidence for size of treatable population, with a “price per pill” | Precision patient cohorts defines treatable population with cohort pricing |
|
| Missing Phase 4 evidence | Good quality Phase 4 evidence |
|
| Risk of non-adherence, and non-responders | Reduce risk through precision case finding |
|
| Missed patients | Find the true treatable population |
|
| Drives costs into the healthcare system | Removes costs from the healthcare system |