The Unique Challenges and Opportunities of Forecasting for Orphan Diseases
An article written by David James, IPI – International Pharmaceutical Industry
Originally written for IPI – IPI International Pharmaceutical Industry
The term rare, or orphan, disease can encompass any one of more than 60001 conditions. Some, such as cystic fibrosis or Tourette’s syndrome are fairly well known, while others, including Acromegaly or Morquio syndrome are much less well recognised.
Historically, patients diagnosed with a disease classified as rare were denied access to effective medicines because prescription drug manufacturers could rarely make a profit from marketing drugs to such small groups. In 1983 however, the US passed its first major law on orphan drugs, known as the Orphan Drug Act, aimed at stimulating the production of medicines for rare diseases by offering pharmaceutical companies research grants, tax credits, fee waivers, and a seven-year market exclusivity for approved medications.
Following this, other pieces of legislation came in to force in the US, including the Rare Disease Act (2002), which established the Office of Rare Diseases at the National Institutes of Health (NIH) and legislation for single rare disease entities, such as the MD Care Act. Prior to this type of legislation, it was not financially viable to develop treatments for diseases with such a small target patient group.
Today, while the rare disease drug market has substantially improved, there is still some uncertainty around patient needs and numbers which makes it increasingly difficult for corporations to bring new drugs into the market. Consequently, forecasting at the very first stages of product development is imperative to making sure the product is successful; however, there are several factors that should be considered when forecasting in pharmaceutical markets, particularly in orphan or rare diseases.
Forecasting in any market involves looking ahead to explore what level of growth can be expected. The forecast also affects different business units within the organisation such as production, finance, operations, sales and marketing. Essentially, forecasting is about trying to predict the future based on the best information, utilising best practise forecasting principles with the best expertise in the organisation available.
The pharmaceutical market is also structured differently to most; it’s highly regulated, the purchaser isn’t the consumer, the supply chain is different, even the data used is different.
Generally, in consumer-driven markets there are hundreds of thousands, if not millions, of customers, so when looking at product forecasts, the forecaster doesn’t need to understand the individual, they need to know the whole group of customers and their general behaviours. Within the pharmaceutical industry, however, forecasters need to look at a specific customer, or type of customer, and then align that to what benefits their specific product is delivering. Understanding the individual customer is an even bigger consideration for orphan diseases, as the population is notably smaller so even one patient missed from the forecast could make a significant difference to the decision to invest in the product.
Forecasting in Orphan Diseases
The term orphan disease is very broad, covering many different diseases, each with different profiles, all falling into the rare/orphan category. The term orphan and rare are often used interchangeably when describing these types of diseases. There are also differing definitions of the criteria a disease must fulfil to be classified as rare. For example, in the USA a rare disease is classified as a condition which affects fewer than 200,000 people at any given time. In the EU, however, a disease is defined as rare when it affects fewer than 1 in 2000 patients.2 Based on these definitions, one rare disease may affect only a handful of patients whereas another could affect tens of thousands.
Measuring the number of rare diseases and the number of patients within them can also be challenging. Firstly, not all rare diseases are screened at birth so it could be months or years before diagnosis, if they are diagnosed at all. There is also a lack of scientific knowledge and quality information on some diseases which can contribute to the delay in diagnosis or eventual misdiagnosis. For example, adrenomyeloneuropathy (AMN), one of a group of genetically determined progressive disorders known as leukodystrophies that affect the brain, spinal cord, and peripheral nerves, is often misdiagnosed as multiple sclerosis (MS), according to the United Leukodystrophy Foundation.3
Secondly, rare disease patients are often scattered across multiple countries and not all countries report or record rare diseases, so forecasting for these types of diseases can pose a unique set of challenges as well as a unique set of opportunities.
With such a wide range of diseases falling into the orphan category some are complex to model from a methodology perspective, whilst others are simpler in nature. Having said this, one of the main challenges of forecasting orphan diseases is often data.
Due to the nature of pharmaceutical markets, the standard approach to forecasting the size of most diseases is to use epidemiologic data, however for orphan diseases there can be a challenge in finding accurate or consistent epidemiological data due to the small number of patients with the disease. A lack of accurate epidemiological data isn’t unique to orphan diseases; however, when this happens in other indications it’s possible to fall back on other data sources, such as prescription data or by running quantitative market research. This just isn’t an option for orphan diseases as secondary data sources aren’t available as often, there aren’t existing treatments available, and, if there are, the small number of prescriptions written make it unreliable. The same can be said for running any meaningful quantitative market research as there are so few treating physicians to interview. Another challenge is the lack of data consistency; if there is epidemiological data available it can vary within and across countries and regions.
The challenges of sizing a market in terms of patients extends to estimating the share a new drug may capture. Again, with more common diseases, this can be done with analogue data based on prescription audit data that can then be supported by quantitative market research.
Due to the lack of existing data, the World Health Organisation suggests that ongoing fundamental research into the disease process will result in the discovery of more targets for drug development for a specific rare disease. In particular, public funding of translational research, including proof of concept studies, might act as a catalyst to translate rare disease research into the development of new medicines. Making a disease easy to diagnose at an early stage will allow the development of prevention strategies that, even in the absence of an underlying treatment, can have a significant positive impact on a patient’s life.4
The effect of poor data can be compounded by a need for the forecasts to be extremely precise. By definition, the number of potential patients with an orphan disease will be small and therefore the cost of any treatment will need to be extremely high to cover the investment made, so variance of just a few patients can make a huge difference to a forecast in terms of go/no-go development decisions. For example, if it is estimated that 1000 people will be treated using a certain drug and the cost per patient is $1m, a company could be quickly down by millions of dollars if the forecast was off by just a few percentage points. This is one of the major challenges facing orphan disease forecasting; clients need very precise forecasts but forecasts in themselves are not a precise science.
Overcoming These Challenges
As with all forecasting, the way to overcome challenges is to be pragmatic. The forecasting adage of “It’s better to be roughly right than precisely wrong” is even more applicable for this type of market. Data for orphan diseases can be patchy so make what you have work harder, using data from one country as an analogue for another, for example. So, you can’t run a quantitative market research study; however, it’s worth remembering that the 10 KOLs you have probably represented a very high percentage of patient coverage so you can gain a lot of insight from a lot less.
Orphan diseases can be very complex in nature, particularly if they are progressive, where the patient’s disease changes over time. Progressive diseases are challenging to forecast particularly when a new treatment becomes available that changes the patient prognosis. The temptation when forecasting small, high value, complex diseases is to create a complex model to account for every variable that might make a difference. But this is precisely the opposite to how they should be approached; it really is essential to keeping the model as simple as it possibly can be. When the data doesn’t support the optimal methodology then compromise, or you will simply be applying estimates upon estimates without creating value.
The more a forecast is challenged, the more robust it will become, so as with any forecast it’s essential to gain input from different stakeholders. Most pharmaceutical companies are multinational businesses, operating in many countries and markets with key stakeholders in each. Running a bottom-up forecasting process will mean that market nuances such as differences in healthcare infrastructure or funding will be captured. Those on the ground may also identify alternative data sources not previously identified, therefore it is critical to ensure collaboration between multi-country stakeholders.
Finally, don’t create a single number forecast: if there’s variability due to lack of data then this needs to be shown either through sensitivity analysis or by creating a probability-based forecast. Any forecast, no matter what the market, is an estimate that will inevitably include inaccuracy. It’s a means of understanding both the risk and the potential of an investment. Therefore, it is essential that those who are basing their decision on the forecast understand both the range and the underlying drivers.
Orphan diseases can be very complex in nature yet have a huge impact on a patient’s life as well as providing a significant return on investment for a pharmaceutical company. Understanding the customer is imperative; however, very small patient pools mean there is often a lack of data available for analysis.
The availability and the consistency of this epidemiology data often mean alternative data sources are more relevant. There are often some really good sources of data that are sometimes unexplored as they come from surprising sources. There is an abundance of data in patient support groups or patient advocacy groups that can be utilised as well as KOLs, and as with forecasting in other markets, triangulating data sources where possible is often vital.
As with all forecasts, it is important to gain input from different stakeholders. When forecasting orphan diseases, the local market nuances are important; a bottom-up forecasting approach is likely to be more appropriate than a top-down, more standard approach. It’s also important that the expectations of stakeholders are managed from the outset. Familiarity with any weak data points means that the level of impact those weak spots are going to have on the outcome can be evaluated, thereby enabling forecasters to manage stakeholder expectations more effectively.
Finally, the objective for any forecast is to understand both the risk and the potential for an investment. Orphan diseases obviously bring challenges but there can be significant reward.