Warning! OUTDATED BROWSER DETECTED!   Please update your browser immediately for a better experience on this website. Learn More
21/xsl/MobileMenu.xsltmobileNave880e1541/WorkArea//http://www.rsna.org/RSNANewsDetailWireframe.aspx?pageid=15319&id=22608&ekfxmen_noscript=1&ekfxmensel=falsefalsetruetruetruefalsefalse10-18.0.0.0730truefalse
  •  
     
  • Stochastic Modeling Helps Predict Radiology’s Financial Future

    Understanding financial analytics can help radiologists make better business decisions and gauge outcomes. By Richard Dargan


    August 1, 2017

    While the field of advanced financial analytics may be foreign territory for many radiologists, experts in the field say they should make it their business to be educated on applying these tools in their everyday practices.

    Hospital administrators must balance many factors and interests when making business decisions. Yet many radiologists do not have the background or training to communicate with hospital administrators on a business level, according to the authors of the recent Radiology article, “Financial Forecasting and Stochastic Modeling: Predicting the Impact of Business Decisions.”

    Radiologists who want a voice in this process not only need a working knowledge of financial concepts and terms, but an understanding of analytics — including the modeling of future financial scenarios. Such models are relatively inexpensive and easy to implement, experts say.

    “Most radiologists don’t have formalized business training,” said co-author Geoffrey D. Rubin, MD, MBA, the George Barth Geller Professor in the Department of Radiology, Duke University School of Medicine in Durham, NC. “But because financial considerations are key drivers of healthcare decisions, radiologists — and all physicians — should understand and participate in the process.”

    Business decisions play an important role in the professional lives of radiologists, influencing where and how they practice, the tools available for image acquisition and interpretation, and ultimately their job satisfaction. Decisions on capital outlays like equipment are subject to financial modeling, an analytical approach that uses existing numbers to generate projections of future earnings. The vast majority of models rely on a deterministic approach, which defines inputs as fixed values based upon historical precedent or the best judgment of experts.

    “The problem with deterministic models is that even though these fixed value inputs could be highly principled and based on the informed opinions of bright people, at the end of the day they produce a single number,” said Dr. Rubin, a member of the RSNA Public Information Advisors Network. “They can’t accommodate for the uncertainty of projections into the future and therefore run the risk of being wrong.”

    A better approach to modeling, according to the review authors, is the stochastic — derived from the Greek word for “capable of guessing” — method. Stochastic modeling accommodates for uncertainty by showing many possible future states and outcomes.

    “The real world has uncertainty,” said review co-author Bhavik N. Patel, MD, MBA, from Duke University School of Medicine. “The stochastic approach gives you a wide range of probability, including the best-case and worst-case scenarios.”

    By returning a range of values, a stochastic model can provide a much better idea of the risks behind a decision.

    “Healthcare organizations are faced with limited resources,” Dr. Rubin said. “When they are attempting to decide how to allocate funds in the most effective manner, this type of financial modeling is critical.”

    Stochastic Modeling Offers Range of Financial Scenarios

    The authors presented a scenario in which a radiology practice compares the annual profit from interpreting emergency CT scans at a remote hospital through fixed, variable or self-billing payment options. The deterministic model uses a single value that corresponds to the most commonly encountered or “expected” values. The results greatly overestimate the value of the fixed payment option, while underestimating the value of the variable and the self-billing options.

    Rather than taking a single estimate for key variables, the stochastic model accommodates the spectrum of exams performed over the course of a year and create a distribution to accommodate for those factors. For example, if a minimum profit of $30,000 was required, the stochastic model shows that the variable contract would achieve this return 95 percent of the time, while the self-billing option would achieve it only 84 percent of the time. However, when the model forecasts the difference between the returns of these two models, it reveals that the self-billing option provides superior returns to the variable contract in 70 percent of simulations.

    “Ultimately the correct choice between these two options would depend upon the priorities and risk tolerance of the group,” the authors noted, “but it should be clear that stochastic models provide substantially greater information than the base case deterministic models.”

    Despite its strengths, stochastic modeling is rarely used in healthcare, Dr. Rubin said, partly due to misconceptions over its complexity.

    In fact, stochastic modeling is easy to introduce, requiring only a personal computer, spreadsheet software and a spreadsheet-based predictive modeling application. Risk analysis software allows the user to add stochastic assumptions to spreadsheets and generate Monte Carlo simulations, which produce distributions of possible outcome values. The more Monte Carlo simulations are run, the closer the approximation is to the true distribution. The forecasted output distribution is used to assess the riskiness of the situation.

    “This software is not esoteric or expensive; it’s readily available and a small investment will open the door to more effective financial modeling within the radiology environment,” Dr. Rubin said.

    Financial Modeling is the Future of Healthcare

    Independent of the financial model, radiologists also need to have a strategic approach that incorporates robust financial analytics. For instance, a radiologist may see a new iterative reconstruction algorithm that promises to significantly reduce CT radiation dose without a change in image quality as a major boost for patient safety, but an administrator may balk at the expensive purchase price and the need for a dedicated server.

    “These financial tools allow you to speak the language and frame things in a way that makes sense to the person at the other end of the table,” said Dr. Patel. “If we show that radiation dose reduction can lead to fewer capital equipment costs, then the administrator might be inclined to say, ‘I don’t understand the science behind it, but I do understand the numbers.’”

    The authors caution that financial modeling has limitations. The models are not representative of immutable facts, Dr. Rubin said, as they are subject to the veracity and accuracy of inputs. In addition, the mathematical relationships built into models are subject to potential biases.

    But even with these limitations, financial modeling and, more broadly, an understanding of financial terms and concepts are increasingly important as the healthcare industry continues to undergo seismic changes.

    “The goal here is to give radiologists a tool kit that will help them approach administrators and ask probing questions,” Dr. Rubin said.

    “Future physician-leaders have to become familiar with these terms,” Dr. Patel added. “There are a number of courses out there and an abundance of literature, and once you get your feet wet, the intimidation factor goes away quickly.”




    Rubin
    Rubin

    Bhavik
    Bhavik

    Radiology
    Probability distributions of profit forecasts from 10,000 simulations for each of the three contract options. The base case values calculated (left) are indicated by stars at the top of the graph, illustrating the extent with which the base case model misrepresents the probability distributions of the stochastically forecast results. The fixed contract has substantially lower profit than the other two options. While the distribution for the self-billing contract appears to have the highest mean value, it also has a much wider dispersion, indicating substantially greater uncertainty associated with its forecast when compared with the slightly lower return, but more predictable variable contract. (Radiology 2017;283;2;342-358) ©RSNA 2017. All rights reserved. Printed with permission.

    Web Extras




    To:
    From:
    Subject:
    Comment:
    Link:
      
  •  
    comments powered by Disqus