Inaccurate patient flow analysis can create severe consequences even for the most diligent pharmaceutical companies.
Risks include vastly overestimating or underestimating the value or opportunity for a particular product which may ultimately impact on patients’ ability to access potentially life-saving treatments.
So, how can pharma companies minimize these risks and actively deliver products which support better patient services? Mastering patient flow analysis can be challenging due to the stringent data-gathering and complex modeling required.
Here, our forecasting experts, Andrew Ward and Kris Barker discuss three key considerations that can help companies produce more accurate patient-based product forecasts.
1. Accurate data is a critical challenge for successfully predicting patient flow
Gathering sufficient and good quality patient data is a critical challenge for pharmaceutical forecasters. The standard approach for understanding patient flow is to conduct market research. This often takes the form of interviewing doctors about their patient journeys, yet this method is highly reliant on anecdotal evidence and human recall. Not only does this introduce inaccuracies and inconsistencies, this type of market research also reduces the patient journey to just the period when they are actively seeing their physician. Doctors also tend to overestimate their importance in the treatment paradigm of a patient, thus, market research that focuses on physicians tends to leave significant data gaps in the overall patient journey.
An alternative approach is to use existing patient record data gathered by third-party research firms and government entities. While this may be straightforward in countries like the US, this type of data may not be available or be patchy in developing nations. In Europe, GDPR can be a hindrance to obtaining sufficient quantities of quality patient data. This is especially true for pharmaceutical companies who are targeting underserved or rare diseases where there is a chronic lack of data.
Overcoming these data gaps and the tendency towards physician bias can be tricky. A recommended approach is to draw on a variety of data sources to build a rich, multi-layered understanding of patient flow. For example, secondary data from patient records can be combined with patient interviews and audit data.
2. Apply the right methodology and variables to model accurately
Obtaining accurate and sufficient data is the first critical step in building a successful model for patient flow analysis, but just as important is how the data is used. The ideal methodology for this type of model would look at sub-patient groups using key variables such as progression rates. It is important to remember, though, that there is a trade-off between data and methodology; where data is not available for particular variables or patient groups it can be tempting to rely on estimates which can erode the accuracy of the forecast.
There is also often a judgement call to be made in terms of the complexity of the methodology chosen. This begins with understanding the core objectives, the product life cycle, and selecting the right variables.
For example, patient flow can be an appropriate tool when working with a product in the later line of treatment for a progressive disease where there are various treatment options. In this scenario, it can help to eliminate value overestimation by providing a comprehensive understanding of the opportunity for this new product. Yet in other scenarios, such as an early-stage product assessment, there may be little justification for investing in such a complex methodology. Where there is not a big investment decision to be made, it usually makes sense to stick to a simpler, lower-budget approach.
3. Complexity of models and process discourages stakeholders from investing time and resources
Constructing an accurate patient flow analysis model is challenging due to the sheer range of variables that need to be included and the consequences of getting it wrong. The complexity required from such models alone can be a stumbling block, discouraging stakeholders and teams from investing the necessary time and resources in the first place.
Complex models that actually deliver a comprehensive understanding of patient flow often require significant time investment to accurately interpret the results. This can pose challenges for decision-makers who are not familiar with the model and the underlying data.
To tackle this, stakeholders should be given a clear narrative overview of the methodology and the data to empower them to take action on the results, even when they don’t align with previous expectations. Having a product champion on the team who fully understands the model is another option for pharmaceutical companies who are heavily reliant on consultants.
Convincing stakeholders of the necessity of a complex and comprehensive data-gathering phase is another key challenge within pharmaceutical firms. Investing in extensive market research and database access can be a hard sell without having a clear idea of the eventual findings. Stakeholders also need to understand the importance of keeping the data and the model up to date. Ideally, this should be done every 2-3 years to keep pace with market developments.
One final consideration here is the importance of back testing the model using historical data. Accurate results based on this historic data will prove that the chosen methodology and the variables being used are suitable for the core objective.
Final Takeaways
- Complex models can be off-putting for stakeholders. Creating a digestible narrative of the data and methodology used can help secure the required investment and support for better data-gathering and modelling.
- Gathering sufficient and appropriate data is a key challenge for achieving forecast accuracy. Overcoming data gaps and physician bias can be achieved by tapping into numerous data sources, including secondary data from patient records, patient interviews, and audit data.
- Methodology and selecting the right variables are the final keys to successful patient flow analysis. The complexity of the model ultimately needs to align with the degree of investment involved.
