Forecasting in Pharmaceutical Industry (Patient-Level) – Part 1

Forecasting in the Pharmaceutical Industry: A Patient-Centric Approach

Accurate forecasting is critical in the pharmaceutical industry, especially when preparing to launch a new drug. Forecasts not only shape important strategic decisions but also ensure companies can meet patient demand while managing costs and inventory. Getting the forecast wrong can lead to product shortages that negatively impact patients or excess inventory that hurts the bottom line.

While there are many approaches to pharmaceutical forecasting, centering the process around patients provides a logical framework to project drug performance. In this post, we‘ll walk through the key steps of a best-in-class patient-based forecasting methodology for new product launches. Future posts will address forecasting for in-market products and those with raw material supply constraints.

Determining the Target Patient Population

The first step in patient-based forecasting is to define and quantify the target patient population for a new drug. This starts with assessing the total population in a given market and then applying a series of filters:

• Prevalence rate: What percentage of the population has the condition the drug treats? Applying this to total population yields the prevalent population.

• Diagnosis rate: What percentage of the prevalent population is properly diagnosed? Some patients may have a disease but not be aware of it. Undiagnosed patients are unlikely to seek or receive treatment.

• Treatment rate: What percentage of the diagnosed population currently receives treatment? Not all diagnosed patients will be prescribed drug therapy.

Depending on the disease area and market, additional filters may be relevant:

• Symptomatic rate: For conditions where not all patients exhibit symptoms, the symptomatic rate is the percentage of diagnosed patients who are actually showing symptoms that require treatment. Asymptomatic patients may not need pharmacological intervention.

• Healthcare access: Particularly in developing markets, the percentage of the population with reliable access to healthcare services can be an important filter. Patients without access to a doctor are unlikely to be prescribed a new medicine.

• Affordability: For conditions disproportionately affecting low-income populations, the affordability of a new branded medicine can be a limiting factor. Only patients with the means to pay for a drug are viable candidates for treatment.

By sequentially applying relevant filters to the total population, the target patient population can be estimated. For example:

• Total population: 100,000,000
• Prevalence rate: 5% → 5,000,000 prevalent population
• Diagnosis rate: 70% → 3,500,000 diagnosed population
• Treatment rate: 80% → 2,800,000 treated population
• Symptomatic rate: 60% → 1,680,000 symptomatic population
• Access rate: 90% → 1,512,000 accessible population
• Affordability rate: 70% → 1,058,400 realistically treatable population

So by this calculation, the realistically treatable population is just over 1% of the total population after all the filters are applied. Small changes in any of the filter assumptions could have a big impact on the ultimate patient potential.

Another consideration at this stage is patient segmentation. The target patient population does not have to be treated as one homogeneous group. It may make sense to divide patients into meaningful segments if different assumptions apply. Examples could include:

• Pediatric vs. adult populations if dosing and treatment practices differ
• Mild vs. moderate vs. severe disease if progression impacts likelihood of treatment
• Affluent vs. disadvantaged socioeconomic status if access and affordability vary
• Urban vs. rural geographies if healthcare infrastructure is inconsistent

Each patient segment can then be forecasted with a unique set of assumptions to more precisely model market realities.

Factoring Comorbidities and Concomitant Medications

There are two important dynamics to consider in patient-based forecasting:

• Comorbidity: This is when a single patient has multiple disease conditions. If the new drug can treat more than one of a patient‘s illnesses, they may be counted in the patient potential multiple times.

• Concomitancy: This is when a patient takes multiple medications to treat the same condition. Concomitancy increases the overall volume of drugs utilized per patient.

For example, imagine a new diabetes drug that can also treat comorbid hypertension. If there are 1,000,000 diabetes patients and 500,000 hypertension patients, but 30% of each group has both conditions, the total addressable patients are:

• Diabetes-only: 700,000
• Hypertension-only: 200,000
• Diabetes + hypertension: 300,000
• Total: 1,200,000

So while there are only 1,000,000 unique patients, the drug can be utilized to treat a total of 1,200,000 indications. This can have a meaningful impact on the total opportunity.

Concomitancy comes into play if some patients will use the new drug in combination with other diabetes and/or anti-hypertensive medicines. If we assume 20% concomitancy in each of the patient groups above, the total utilization potential is:

• Diabetes-only: 700,000 x 1.2 = 840,000
• Hypertension-only: 200,000 x 1.2 = 240,000
• Diabetes + hypertension: 300,000 x 1.2 x 1.2 = 432,000
• Total: 1,512,000

Failing to account for comorbidities and concomitancy can lead forecasts to underestimate the true drug utilization opportunity.

Calculating Patient Share

Once the target population is established, the next step is to project what share of those patients will receive the new drug. This involves estimating two key parameters:

• Peak market share: The highest share the new drug is expected to achieve at its peak
• Time to peak share: How many months or years it will take from launch to reach that peak share

A simple but non-transparent approach is to enter expected shares by time period without any underlying logic. A more robust method ties share projections to anticipated market events and data.

Peak share will be shaped by factors like:

• Number of competitors, both current and expected future entrants
• Relative efficacy, safety, convenience vs. competitors based on clinical data
• Prioritized profiles and preferences of prescribing physicians
• Affordability and access vs. other treatment options

Time to peak is driven by the launch trajectory which is influenced by:

• Maturity of the category and speed of adoption of innovative brands
• Size and scale of planned launch promotional activities by the company
• Expected competitive responses and counter-detailing
• Accessibility and reimbursement of the new drug at launch and over time

The most common approach is to categorize the expected launch trajectory as rapid, medium or slow and apply a standardized time to peak share accordingly, such as:

• Rapid uptake: 50% of peak in year 1, 80% in year 2, 100% in year 3
• Medium uptake: 30% of peak in year 1, 60% in year 2, 85% in year 3, 100% in year 4
• Slow uptake: 15% of peak in year 1, 35% in year 2, 60% in year 3, 80% in year 4, 100% in year 5

For example, if a new drug is expected to achieve a 25% peak share and is projected to have a medium launch uptake trajectory, the share forecast would be:

• Year 1: 25% x 30% = 7.5% share
• Year 2: 25% x 60% = 15% share
• Year 3: 25% x 85% = 21.3% share
• Year 4: 25% x 100% = 25% share
• Year 5+: 25% share

More sophisticated models may use historical uptake analogues specific to a therapeutic area to plot expected share over time rather than simplified trajectories.

Translating Patient Share to Sales

The final step is converting patient share into financial forecasts. This is done by estimating the number of units (doses, cycles, etc.) of the new drug each patient utilizes on average and then multiplying total patients by share and unit utilization.

For example, let‘s say the new diabetes and hypertension drug from the prior example is an oral pill taken twice daily. At peak 25% share of the 1,512,000 patient utilization potential, that equates to:

• 1,512,000 patients x 25% share = 378,000 treated patients
• 378,000 treated patients x 2 pills per day x 365 days = 275,940,000 pills

If we assume a 10% gross-to-net discount/rebate and a net price per pill of $2, the peak year forecast would be:

• 275,940,000 pills x $2 net price per pill = $551,880,000 net sales

To complete the annual forecast, repeat this calculation with the expected patient share for each year:

• Year 1 at 7.5% share = 41,391,000 pills or $82,782,000 net sales
• Year 2 at 15% share = 82,782,000 pills or $165,564,000 net sales
• Year 3 at 21.3% share = 117,207,720 pills or $234,415,440 net sales
• Year 4 at 25% share = 275,940,000 pills or $551,880,000 net sales
• Year 5+ at 25% share = 275,940,000 pills or $551,880,000 net sales

There are many other considerations in real-world forecasting models:

• Not all demand is equal – forecasts should distinguish between paid/reimbursed demand and clinical trial, compassionate use, or free drug utilization which generate no revenue

• Compliance and persistency are important – many chronic patients don‘t take drugs as prescribed so average utilization may be lower

• Pricing and access change over time – introduction of new competitors, loss of patent exclusivity and changes to reimbursement can materially impact price and volume

• Geographic mix is critical – parallel trade between EU countries due to reference pricing and widely varying affordability worldwide mean regional assumptions are required

Despite the inherent complexity, grounding pharmaceutical forecasts in a patient-based framework provides a solid foundation to pressure-test assumptions and build sound projections to inform decision making.

In the next post, we‘ll examine how forecasting approaches differ for in-market products compared to new launches. We‘ll also discuss how to adapt forecasting techniques for drugs with constrained raw material supply.

Until then, always remember the most important stakeholder in this process – the patient. Robust forecasting helps ensure new medicines reach those who need them most.