Understanding patient flow forecasting
Patient flow forecasting is a fundamental tool in healthcare and pharmaceutical industries, enabling stakeholders to predict and plan for the treatment requirements of patients accurately.
This technique is particularly important in optimising resource allocation and making informed decisions, especially in the context of chronic diseases and rare conditions. Here, we will introduce the concept of patient flow forecasting, its methodology, common statistical models and methods, and how dedicated software is essential in mastering this complex forecasting technique.
Understanding patient flow forecasting
Patient flow forecasting is a methodological approach centred around monitoring the movement of patients as they progress through various stages of treatment and care. Unlike prevalence-based forecasting, which considers the total number of patients with a particular condition, patient flow forecasting relies on incident-based data. It involves tracking patients over time, offering insights into their journey through various lines of therapy or treatment regimens. This approach is particularly useful in chronic diseases and conditions that require progressive treatments.
Examples of such diseases include Alzheimer's, non-alcoholic steatohepatitis (NASH), cystic fibrosis, and other conditions with a chronic nature that necessitate transitions between different treatment options. Patient flow forecasting is also invaluable in managing rare or orphan diseases characterised by a limited patient population. In such cases, precision in tracking and predicting patient movements becomes vitally important, due to the small number of patients involved.
Applying patient flow forecasting methodology to data sources
Patient flow forecasting is based on the conversion of incident data, primarily data on new patient starts, into prevalent data, which represents the total number of patients on treatment. This conversion process is essential for understanding production demand and the number of units required to treat patients effectively. Several key concepts and data sources play pivotal roles in this methodology:
- Incidence data: Patient flow forecasting commences with incident data, which includes information on new patients entering treatment. This data is typically sourced from patient records, epidemiological databases, or primary research, providing the initial patient population.
- Persistency curves: These curves represent the average duration of therapy, offering insights into how long a patient remains on treatment. They play a critical role in converting incident data into prevalent data, estimating when patients might discontinue treatment or become eligible for progression. Data is often derived from clinical trials and scientific literature.
- Average length of therapy: This metric provides an understanding of the typical duration of treatment for patients, aiding in the estimation of when patients might stop receiving treatment or become ineligible for further therapy. Often this approach can be used when the more granular persistency curve data is not available or of poor quality.
- Treatment rates and diagnosis rates: These rates are essential for accurately delineating the patient landscape. They help filter the patient population, ensuring that only eligible patients are included in the forecast.
- Patient share data: Patient share data reveals how the patient population is distributed among different treatment options or products available in the market. Often, this data is obtained from comprehensive market tracking studies conducted by organisations like IQVIA.
- Uptake curves: Uptake curves are instrumental in modelling the launch of new treatments, patent expirations, and generic erosion. They provide insights into how a product gains market share and the time it takes to reach its peak.
- Conversion parameters: Data related to dosing, pricing, and other factors affecting treatment are crucial for patient flow forecasting and, specifically, the process of converting the total number of patients on treatment into units. Such data is typically obtained from clinical trials or internal sources.
A range of factors can significantly impact patient flow forecasting, including geographical variations. When selecting analogues for forecasting, factors such as healthcare infrastructure, accessibility to treatment facilities, and insurance coverage should be considered. It is essential to acknowledge that no two markets or regions are identical. While there is no perfect analogue, efforts should be made to match as many variables as possible to create accurate forecasts.
Statistical models commonly used in patient flow forecasting
Statistical models and time series methodologies are typically associated with sales-based or volume forecasting, rather than patient flow forecasting. Patient models are primarily driven by changes in the patient count within the total patient market. These changes can be influenced by factors such as demographic shifts, variations in diagnosis rates, or the introduction of new diagnostic tools and treatments.
While patient flow forecasting does involve some statistical modelling, its focus is less on trending sales and volume and more on understanding the evolution of the total patient market. This approach considers whether the number of available incident patients is growing or declining and identifies the reasons behind these trends. Statistical models commonly used in patient flow forecasting include:
Linear regression: These models are relatively simple and are used to establish relationships between variables. In patient flow forecasting, they can be employed to analyse factors influencing patient counts.
Vector autoregression (VAR): Useful for analysing time series data, making them relevant for certain aspects of patient flow forecasting, such as understanding the impact of treatment rate changes over time.
Risk analysis and Monte Carlo simulation: These specialist techniques are employed to assess and quantify the risk within a patient flow forecasting model. Given the complexity and numerous variables involved, risk analysis is crucial for understanding the potential uncertainties in forecasts.
The advantages of patient flow forecasting software
The utilisation of dedicated software in patient flow forecasting offers several distinct advantages:
- Efficiency: Specialised forecasting software streamlines the model-building process, reducing the time and effort required to create complex forecasts. It allows forecasters to focus more on analysis and less on manual data manipulation and coding.
- Transparency: Expertly built software provides transparency, storing formulas and algorithms in the background, allowing users to access and understand the calculations easily. This transparency is essential for validating forecasts and ensuring accuracy.
- Consistency: Forecasting software standardises models, ensuring consistency across different markets and indications. This enables easy consolidation of forecasts from various sources, improving the overall accuracy of predictions and enhancing collaboration across affiliate teams.
- Flexibility: Software offers flexibility in modelling various scenarios, making it easier to explore different assumptions and their potential impacts. This flexibility is crucial in a dynamic healthcare landscape.
- Treatment restrictions: Patient flow forecasting software can incorporate treatment restrictions, which are challenging to implement in manually built models. Understanding the eligibility of patients for different treatments is vital for accurate forecasts.
- In-built expertise: Users of specialist forecasting software will find essential modules built in, meaning that forecasters of all experience levels gain access to complex patient flow forecasting techniques.
Patient flow forecasting is a vital tool in healthcare and pharmaceutical industries, enabling precise predictions of patient movements through various stages of treatment. It relies on a combination of data sources, methodologies, and software tools to create accurate forecasts. Dedicated software solutions are at the forefront of this evolution, simplifying the forecasting process and making it accessible to a broader range of users, while sophisticated statistical models contribute to the understanding of patient count dynamics.
The future of patient flow forecasting
AI Integration is likely to play a significant role in patient flow forecasting, automating complex processes and improving prediction accuracy. Artificial Intelligence can also analyse vast datasets and identify patterns that might be overlooked by humans. Advances in AI and data analytics are also expected to lead to more accurate diagnosis tools and treatment options. This will potentially enhance patient outcomes and reduce the time patients spend on treatment.
We expect to see a hybrid approach to forecasting become more prevalent, combining the flexibility of Excel models with cloud-based analytics and reporting tools. This approach offers the best of both worlds, allowing for sophisticated modelling and efficient reporting.
Inevitably, patient flow forecasting within the pharmaceutical industry will continue to focus on treatment cost optimisation, helping healthcare providers allocate resources efficiently, leading to reduced overall healthcare costs, and benefitting both providers and patients.