Category Archives: Decisions

Managerial control of medicines cost drivers

It is not unreasonable to have concerns about the cost of medicines.

Drug costs are usually influenced by government policies on pricing and reimbursement of medicines themselves. These range from simple discount seeking to more complex approaches such as conditional approvals, and value-based pricing (perhaps a subject for another posting). These can achieve a measure of drug cost control, but may also distort the market of medicines themselves.

For instance, tendering for generic medicines can sometimes lead to unacceptable consequences, such as unexpected product substitution by suppliers, patient and clinician confusion as medicines change appearance, and complications in medicines management or pharmacists. And a ‘winner take all’ award of contract can mean that the losers exit the market, removing a source of price competition and choice for consumers and governments. This is an unintended but avoidable consequence of using this crude procurement instrument.

Regulations and health technology assessment together are challenging to free pricing of medicines, but it is unsurprising that medicines should be subject to some assessment of efficacy and performance in the real world, and not just on the results of clinical trial evidence on a highly selected study population. HTA has also thrown into the spotlight the logic by which drug prices are established by the pharmaceutical industry. This scrutiny is not a bad thing as it highlights the methodologies used, whether they accurately produce a price reflecting the value of the medicine as used. Separately, the cost of the research to produce the medicine is a factor, and one should not be surprised that the prices of successful drugs should try to recoup the costs of all the failed drug research, even if those costs could be seen as the price of the risk of doing business for the industry.

Apart from these approaches to drug cost control, there are opportunities to reduce costs within the healthcare system itself.

Achieving improved cost control, value for money and improved health outcomes are consequence of better management of medicines procurement, patient adherence, dispensing and waste reduction and reduction in variations in prescribing practices.

These are processes and organisational interventions designed to enable improved professional practice through hospital formulary controls and best practice in medicines logistics. These enable the ability to reduce prescribing variance, strengthen quality systems and improve patient acceptability whilst strengthening the foundations of professional practice.

The following “logic map” shows how this works:

A central feature of any high-performing healthcare system or organisation includes best-practice in medicines use and clinical management.

As all aspects of healthcare are under varying degrees of financial stress, cost controls and appropriate use of medicines are a legitimate focus of scrutiny to achieve the highest standards of clinical practice and safe patient care.

Failure to achieve clinical and managerial control of the use of medicines across the patient treatment pathway may arise from:

  • misuse of medicines (failure to prescribe when appropriate, prescribing when not appropriate, prescribing the wrong medicine, failure to reconcile medicines use across clinical hand-offs
  • “clinical inertia” and failure to manage patients to goal (e.g. management of diabetes, and hypertension post aMI) [see for example: O’Connor PJ, Sperl-Hillen JM, Johnson PE, Rush WA, Blitz WAR, Clinical inertia and outpatient medical errors, in Henriksen K, Battles JB, Marks ES et al, editors, Advances in Patient Safety: From Research to Implementation Vol 2: Concepts and Methodology), Agency for Healthcare Research and Quality, 2005]
  • failure to use or follow best-practice and rational prescribing guidance
  • lack of synchronisation between the use of medicines (demand) and procurement (supply), with an impact on inventory management and
  • loss of cost control of the medicines budget.

The essential challenge is ensuring that the healthcare system and its constituent parts are fit for purpose to address and avoid these failures or at least minimise their negative impact.

Medicines costs are the fastest growing area of expenditure and comprise a major constituent of patient treatment and recovery.

The cost of drug mortality was described in 1995 [Johnson JA, Bootman JL. Drug-related morbidity and mortality; a cost of illness model. Arch Int Med. 1995;155:1949/56] showing the cost of drug mortality and morbidity in the USA and costed the impact at $76.6 billion per year (greater than the cost of diabetes).

The study was repeated five years later [Ernst FR, Grizzle A, Drug-related morbidity and mortality: updating the cost of illness model, J Am Pharm Assoc. 2001;41(2)] and the costs had doubled.

And costs and use have continued to rise since then.

Evidence from a variety of jurisdictions suggests that drugs within the total cost of illness can be substantial, for instance:

  • Atrial fibrillation: drugs accounted for 20% of expenditure [Wolowacz SE, Samuel M, Brennan VK, Jasso-Mosqueda J-G, Van Gelder IC, The cost of illness of atrial fibrillation: a systematic review of the recent literature, EP Eurospace (2011)13 (10):1375-1385]
  • Pulmonary arterial hypertension: drugs accounted for 15% in a US study [Kirson NY, et al, Pulmonary arterial hypertension (PAH): direct costs of illness in the US privately insured population, Chest, 2010; 138.]

There are upward pressures that increase costs, downward pressures that decrease costs and pressures that influence costs in either direction; the diagram illustrates a few:

Many of the drivers can be addressed through a combination of professional staff development, better use of information, particularly within decision-support systems to support guidelines and prescribing compliance, and organisational interventions.

An interventional strategy to manage medicines cost drivers involves a structured review of central drivers of drug cost and use within existing national or organisational priorities.

The range of possible solutions fall across of spectrum of interventions and any or all of these are good starting points:

  1. development of drug use policies
  2. development of clinical policies, guidelines, and clinical decision-support algorithms
  3. drug-use evaluation studies
  4. clinical and medical audit
  5. cost-benefit studies
  6. professional development
  7. procurement effectiveness performance review
  8. patient treatment pathway analysis
  9. analysis of waste reduction opportunities
  10. management/organisational improvements to support appropriate behaviours.

To start involves assessing the current state of these aspects, and determine any gaps with national or organisational policy, or evidence-informed best practice. As a proxy measure of the necessary changes, measurement of this gap becomes the focus, and requires evidence of current practice against the desired goal. In many cases, where systems are weak or poorly performing a comprehensive root-and-branch review may be needed, with a corresponding impact on existing managerial, organisational and professional practice.

All healthcare systems and organisations are different and whilst it is difficult to precisely quantify the outcomes in advance, organisations undertaking a sustained process of medicines review and optimisation should be able to release more than 10% of existing drug expenditure and possibly more.

In organisations with a less-well developed clinical pharmacy, where medicines information systems are not well developed, and where clinical guidance is not proceduralised, greater savings are likely, perhaps to 25% or more, reflecting the possibility that the lack of information conceals upward drivers of costs, masks inefficient medicines management or evidence of misuse and waste.

In the longer run, healthcare organisations will need to ensure sustainability of any medicines optimisation review, by ensuring strong organisational structures, practices and behaviours. Development of these frameworks is an important by-product of medicines optimisation interventions, with a corresponding improvement in medicines safety.

9 Tribes of the Internet and their health interests

Discussions on health literacy are increasing as healthcare providers, clinicians, payers and patients consider what this means for healthcare. Having been involved in launching the world’s first digital interactive health channel in the UK in 2000, one thing I learned is not to assume that everyone is alike or has common interests.

Healthcare systems are poor doing what retailers take for granted, namely the segmentation of their users. When we did the health channel, we worked with a simple framework drawing on work by the California HealthCare Foundation, in their report “Health E-People”. This gave us a workable model of the different types of users and their different needs, and that in developing content and services for them through the Channel, we needed to be mindful of this. More recent work by the Pew Internet Project has identified the “9 Tribes of the Internet”, to reveal how different people interact and use technology. Of course, segmentation can be quite elaborate, but at this stage we need a scaffold to guide our further understanding.

The main assumption we need to make about technology is how it will be used by people and thereby how this informs the adoption/diffusion process. Health and social care are traditionally “high touch” activities, given the way that knowledge has been organised, who knows it and how it is used. This, however, is being challenged by technologies that embody what traditionally has been found in the brains of specialist clinicians — what I call ‘cognologies’.

Increasingly we are seeing technological innovations that can embody both that knowledge (in decision algorithms for instance) and in skill (in robotic devices, vision systems for instance). Will people accept a shift toward high technology care at the expense of its traditional focus on care by humans? Is that an aesthetic preference (we like it) or might people come to prefer “lower touch” technologically-enabled services if it is reliable, and on-demand?

As we think about this, I suggest the following as some thoughts for policy makers and care providers:

  1. Eventually, the individual will have to own, in some form, their own health record if much of the desired changes in patient behaviours are to be realised. This will lead to patients having a new understanding of information about themselves, and as such this information will need to be clear without mediation or interpretation by others. Patients will, therefore, become involved in decisions about what to do with their information, and with whom it is shared and used; for instance, use in databases whether in commercial or public organisations that will be accountable to the patient for the use of that information. The patient, as what I call the ‘auditor of one’ will come to take a keener interest in the accuracy of the health record and be less tolerant of mistakes or inaccuracies, as is the case in other areas (e.g. banking, credit scoring).
  2. Not everyone will be digitally enabled in the way technology pundits fantasise about. This is not a digital divide and is not evidence of social exclusion, but is a personal choice of people to lead their lives as they wish in a pluralistic society; it may be that in the end, we all end up as digital natives over time, but some will still be hold-outs, or ‘islands in the net. The key implication here is that service providers will need to move in some cases very slowly to adopt technologies with some types of people. In time, perhaps people may adopt low-level access and interactivity, but for some people technological interactivity will remain at best an option not a preference within an evolving technological ecosystem. It remains to be seen whether this will continue to be the case; evidence from other technologies suggests not, that in time, technologies are broadly universally accepted, but not necessarily used in the same ways by everyone.
  3. The assessment of benefits of technologies in the traditional health technology assessment [HTA] model will need to pay much greater attention to the segment of the population likely to be involved and the social context of that group, taking account of distinct patterns of use and preferences. This challenges the current paradigm used within HTA communities. The conclusion that one-size-fits-all HTA assessment will increasingly prove inadequate. This means that designing and implementing technologies will need to be far more flexible when it comes to the structure of service delivery as the adoption/diffusion process itself will come to determine the socio-economic benefits. Consider that few today would subject the telephone to an impact assessment – it is now part of our expectations, and we should not be surprised if the same thing happens to evolving technologies in healthcare focused on the use by consumers and patients.
  4. The tribes model suggests that not everyone will necessarily buy into the technology revolution. For many people, they work in care precisely because they want to have personal contact with people, and not through intermediating technologies. Since many patients also would have that preference, organisations may need to structure services and staffing to ensure the right mix of people to service the right publics. This will challenge approaches to the organisational design of service providers, in the main suggesting more pluralism in variety, scale and function.
  5. Patient compliance, concordance, adherence may become more dependent on the features of the technologies, their design and ease of use, than on the willingness of the patient to follow a particular care regime. Patients are deliberately non-adherent for many good reasons (some of which reflect fundamental flaws from the medicine itself, its delivery system, or side-effects). Accidental non-adherence is another matter obviously. Helping people understand their limitations in using and working with technologies as matter of personal preferences will become very important, which increases the focus on personalisation.

It is common for health and social care systems, especially where the state is the main source of funding, to tend toward omnibus systems of service delivery, which has difficulty dealing with individual service preferences. Whether it is fully appreciated, such systems favour professional and provider interests and depend on proxy interpretations of patient preference. It would be a mistake to assume a similar approach with technologies. Instead, we should be encouraging approaches which are sensitive to the preferences and usage patterns of individuals. In this way, too, we may actually see services being offered that people will value and use.

The 9 Tribes in Health

Background

Pew Internet Project identified the “9 Tribes of the Internet” in a report in 2009 [http://www.pewinternet.org/2009/06/10/the-nine-tribes-of-the-internet/], to ascertain how different people interact and use technology. The California HealthCare Foundation, in its “Health E-People” [http://www.pewinternet.org/2009/06/10/the-nine-tribes-of-the-internet/], identified three broadly defined populations: the well with an interest in health, the newly diagnosed, and those with long-term or chronic health conditions.

The Pew research was instructive in thinking about how people might deal with a more technologically enabled health and social care system. I have sketched out some relationships in the table which gives an overview of the sort of considerations that are likely relevant and important.

NOTE: This was first written in 2010, and updated in 2019.

Payer decision making

The relevance of value in establishing the positioning of medicines is the new normal for pharmaceutical marketing. Pharmaceutical companies have customers who are highly constrained by whether healthcare system funding is sustainable long term. Remember, payers think epidemiologically and in multiple years of costed care so industry needs to assess how that can be understood for product value. The pharmaceutical industry is constrained by its ability to generate revenues from medicines sales to cover the costs of research and development.

These two collide in the decision making process to adopt, or not, a medicine. The payers broadly have to balance the sustainability of their budgets with a potentially innovative medicine that will improve care outcomes. The pharmaceutical companies have to construct the value case to demonstrate these care outcomes. That probably means at least two things among many;

  1. Stop pricing drugs by the pill or pack, and start pricing valued outcomes for a defined set of patients over a number of treatment years, and
  2. Forget about trying to ‘time’ the market for product launch. The right time is set by payer budget cycles and their drug investment and disinvestment decisions. And, oh yes, the evidence.

By the way, my approach does differ from the journey model of Ed Schoonveld in important respects, by identifying the structured, and gated, decision processes involved; that why medicines aren’t sold, but bought.

Let’s first look at the colliding priorities. The diagram shows that payers are concerned with the value of a medicine in minimising treatment risk for the treated population. A company is seeking the value of the medicine by maximising the size of the treatment population that they believe benefits. As you grow the treatable population beyond the evidence, risk rises; for payers, reducing that risk is addressed through evidence.

No alt text provided for this image

This is a collision of notions of ‘uncertainty’ in decision making and folks on the industry side should be used to requests for more evidence and novel access arrangements such as conditional reimbursement with evidence generation, and so on. As in any model of competing interests seeking a common price, the intersection of these two notions of uncertainty is defined by a price at which both parties will agree the price pays for the uncertainty it quantifies (i.e. it quantifies uncertainty in a certain way). The intersect quantifies risk, and sets the size of the treatment population that can benefit for that price.

The resulting curve may be thought of the ‘community effectiveness curve‘ depicting the optimal balancing of risk for the treatment community and a proxy for price agreement along that curve. This, by the way, is a better way to identify price corridors for people who still think that way.

This structured process is what this article is about.

Here is the gated decision process for payer decision making. While payers may not formally see themselves going through this in a linear way, they are thinking these thoughts, in this order.

No alt text provided for this image
Gated Payer Decision Making for Market Entry of New Medicines

From the payer perspective, information needs to be specific to the decision gate and having the wrong information at the wrong time (e.g. the right information at the wrong gate) will just frustrate folks and probably irritate decision makers.

The diagram is read left to right, and a ‘yes’ answer to a question is needed in order to move through the gate. Getting a ‘no’ means the information supplied failed to make the case.

The following is a quick tour of the underlying logic. By the way, I call this a gated process as there are criteria for satisfying the conditions for passing through the gate; it is, I believe, unhelpful to decision making to characterise them as hurdles, as this suggests they are imposed to make life difficult. They are, actually, simply the structure of decision making.

Looking at this from a behavioural perspective, i.e. psychology informing decision making, each gate means this:

  • To get through the first gate, the payer is confronted with existing treatment options and asks why do I need another, or why change? Unfamiliarity may also be at work, with novel treatment benefits that lack comparators. Evidence of unmet need might be helpful along with good epidemiology to demonstrate the possibility of better outcomes.
  • Satisfied that a new therapy may be warranted, there is the question of risk and benefit compared to current treatment. While a new therapy might be indicated (yours?), the associated risk may be unacceptable compared to not using it. The benefits really do have to hold under increased uncertainty for a payer to agree to increased treatment risk. I suggest this is where discussion of standards of care begin to be quantified, having been introduced at the first gate. Payers often are not as aware as they should be on the current standards of care evidence in misdiagnosis, medical error and patient dissatisfaction.
  • Then having agreed that this uncertainty and its associated risk are acceptable, we are confronted with the cost and efficacy issue. Now we are beginning to price that risk. Good analysis of the costs of care and mis-care are useful, again because payers are not often aware of whole system costs (i.e. the costs of a treatment pathway) either because they are using using a fee schedule linked to DRG type classification or haven’t proofed their capitation models.
  • Success in pricing that risk moves to the question of the medicine in the context of total treatment costs and whether the treatment costs themselves for the patient population can be managed or will the scaling of the costs overwhelm the system for this treatment population versus all other options. Companies may see themselves as just suppliers of medicines for a price, and not a partner in the total system. But understanding the cost drivers along the whole treatment pathway, not just the costs a new medicine may drive, becomes an important element in final value pricing. If you have a medicine that reduces associated costs, or avoids certain costs (think the Triple Aim, here), then the determinants of value are much clearer. It may be that a biomarker is a value-add from one perspective but only if it reduces medical error and misdiagnosis, without increasing costs, so precision patient identification becomes important. If you’ve got this far, though, you’ll have already shown you can demarcate the treatment population, including the responder subset with a degree of precision.
  • Finally, the payer thinks about the future and whether there will be new medicines coming along that might address the same treatment population, alter risk differently, improve outcomes, avoid costs, with better patient adherence, and so on. Given, broadly, a medicine is alone in its treatment class for months, rather than years, payers may choose to delay decision making or consider options you’ve ignored that may trade off future costs and present priorities. This may be where a payer will be thinking disinvestment or product substitution and the determinants of that are critical in this final phase. Here’s a scenario: Why might a particular medicine not be a preferred medicine on a hospital formulary? The answer is simple: don’t have production problems where supply cannot be guaranteed. The lesson is that this is where the long game gets played out.

For those of you who read Kahneman’s “Thinking Fast and Slow”, or similar, there are decisional heuristics at work here. And across that gated process, you are contending not just with highly structured evidence informed quantitative information, but also how humans can be influenced by how human’s think they think. This has a raft of factors such as confirmation bias, hyperbolic discounting, choice overload, loss aversion, endowment effect, anchoring, mental accounting and social proof. It will pay to be attentive to when you present what information and the frame of mind decision makers are in. The reason this is important is that that regulators and payers in different countries, hospitals or regions can make different decisions from the same evidence, so something else is going on.

And so, a comment on pricing. To short-circuit this challenging gated process, it is common simply to cut the price, i.e. discount. Discounting is a quick win trick that only works if payers are trying to reduce present costs, which they all are. However, payers with their eye on the future are more likely to be interested in pricing arrangements that address uncertainty over time and so will be amendable to arrangements such as coverage with evidence development or outcomes guarantee. If they are focused on whole system issues, they will be interested in care pathway (cohort/whole system) pricing for instance. If, though, the future costs are a priority, think about capitation arrangements, or simple price/volume but be mindful that this last is like selling products door-to-door in the 1950’s.

I happen to think care pathway pricing of carefully demarcated patient populations with costs taken over say 5 years is a better pricing model for both parties. Value can be demonstrated on both sides along with evidence of such things as improved adherence (to reduce waste by non-responders) or diagnostic decision support aids to address misdiagnosis and sources of medical error or reduce time to the correct diagnosis, in the case of rare diseases for instance.

This article is designed to emphasise product value determination under conditions of uncertainty to arrive at a sustainable long-term relationship.

Decision Making

Good decision-making depend on understanding what goals the decisions serve, and what information and processes we will use to make that decision. It is not unknown for decisions to be made in haste, or with poor understanding of the context, or miscontrue external influences.

An example is mismatching the evidence generation process, used to provide clinical research evidence of a product’s value to payers, with the stage of decision-making by the payer. Payer decision-making is a gated, binary process; that is to say, it has a linear structure of stages, and at each stage (gate), it is a yes/no decision. Failing at a gate (getting a no) means the product goes no further. What this process looks like is important. Not understanding it means that the wrong evidence is provided at the wrong time, betraying a lack of understanding of how information is used.

Clinical workflow is also a decision-making process so we need to know how clinical decisions are made.  The same applies to how patients make decisions around their use, or not, of their medicines (called adherence).

We also know about the risks of groupthink, and its impact on the quality of decision-making. Methods designed to challenge thinking are important to ensure that the right problems and issues are being addressed. In policy making environments, this involves critical issues as policy making processes usually lead to legislative processes, and the use of instruments (laws, penalties, etc.) to enforce the law: getting this wrong can lead to policy gaming, or non-compliance. Commercially, this can mean that the decisions reflect personal preferences or influences, rather than an evidence-informed assessment of what the real commercial options are. Tools such as Devil’s Advocacy, Red Team are well-established approached used to ensure high quality decision-making.