Reducing Healthcare Costs by Managing Clinical Variation - mHealth to the Rescue?
The headlines continue to beat out the message that in the United States we spend too much on healthcare, in aggregate and in relation to what value is received. Value has been alternatively defined as Quality over Cost, or in some cases (Quality and Service) over Cost. These value statements have been vetted and discussed at length in numerous presentations and seminars. The constant is cost, and there is a lot of attention being paid to what makes cost high relative to “quality” as well as in absolute terms. Will the emerging digital health and electronic health innovations, collectively "mobile health" or mHealth, be the answer?
Let's look at costs first. Allow me to summarize the many healthcare cost drivers into two categories, social and systemic.
Social, or community based variables include water quality, immunization rates, unemployment, and percentage of population over age 65. Income level and education levels within a community also reflect on how people will engage the system, as income and preferences drive personal habits. In addition to these variables with publicly available data sources, there are quantifiable personal lifestyle events that include but are not exclusive to amount of sleep, stress levels, quantity and quality of food consumed, amount of exercise/activity, how much time is spent with family or in community-related events (e.g. volunteer time). Taken together, these public and personal data define social determinants of health.
Systemic cost drivers are related to process efficiency and effectiveness in care delivery. Variation in process and clinical results is a persisting issue, despite being discussed for what feels like decades.
A quick review of the 2014 CMS data for Medicare inpatient payments illustrates this persisting issue, at least at a high level. I compiled these illustrations to highlight the cost impact that clinical (and other causes of) variation create. The analysis is not meant to be relied upon other than to gauge the relative differences and to gain insights as to why these differences exist. In order for this analysis to be relied upon for any other purpose would require significant additional and clinically-based analysis.
In the first illustration (Figure 1), we see that the Aggregate Cost of Inpatient Payments per capita differs significantly by geographic area. I highlighted a few states in the circle, and then, drilling down to just look at California, the differences for a single diagnostic code are broken out. The per capita cost differences within a diagnostic code within the geographic area is significant. The two examples given here are for pneumonia and cardiac arrhythmia. One more layer down, I look at a couple of specific hospitals in similar geographic markets with credible patient volumes. No matter which way you slice it, and no matter what diagnostic code selected, the analysis highlights a significant range of inpatient payments that cannot be explained by differences in charge masters/revenue cycle management, or patient mix. My conclusion: Clinical variation remains a significant issue.
Figure 1: Drilling into Per Capita Inpatient Payment Differences
One more chart. I have found in my work for various clients that there are multiple “distributions” of patients within a given diagnostic cohort. Generally, there are patients that incur a high number of low dollar claims, then there are patients that incur a lesser number of intermediate valued claims. The third group is less numerous still, but generally incur a wide range of expensive treatments as they are usually poly-chronic, and require the most intensive care and case management. The fourth tier, or fourth modal distribution, is typically catastrophic. I like to break down any cost data into these groupings whenever I can to gain better insights into the data “noise” that in fact may be systemic “noise”, e.g. clinical variation. Ideally this analysis would be done on a longitudinal and total cost of care basis, that incorporates all claims and associated non-facility based expenses incurred to regain health. I consider the illustration here to be directional, and indicative of the nature of the issue.
In this chart I illustrate significant differences between the seven hospital systems with the greatest volume of patients receiving care only for the cardiology-related diagnosis captured in the range of DRG codes 200- 400. There is a large patient population flowing through these diagnostic codes, so the differences in the lines illustrates tremendous economic impact. How would these hospitals manage this variation, this financial risk, if they were to contract on a value-based care basis with revenue at risk and a function of care provided and costs incurred? Will an analysis like this help drive the determination of evidence-based guidelines? Will evidence-based care guidelines and clinical pathways be identified and managed to? What social determinants would complicate patient compliance or success within those guidelines?
Figure 2: Hospitals Face Cost Risk Under Value Based Contracts
The cost issue is a complicated one, and this discussion just scratches the surface. Healthcare cost is complex because there are social determinant influences as well as systemic influences. Therefore, to impact cost we need to address physician behavior and associated clinical variation as well as understand and incorporate social determinants of health.
Within social determinants of health, digital health and mobile health (mHealth or eHealth) technology enables better information collection and knowledge sharing. This in turn enables better clinical decision making and teaming, realigns physician decision making with consumer preferences, and creates a dynamic environment that can be used to modify individual behaviors and habits. This conclusion about mHealth is supported by the HIMSS/PCHA 2016 survey. The survey noted that mHealth and related connected health technologies (such as telehealth) are important components to manage risk (clinical variation) and support value-based care strategies that are intended to address physician alignment and cost related concerns.
The success of mHealth or any other technology depends on following a defined strategy and purposeful implementation. Resources, both money and people, are too limited, such that a scatter-shot approach is wasteful and time consuming. We need improved population health and community health, improved efficiency, and improved outcomes. These are difficult challenges. To be successful, we will need to look not on the near-term and obvious, resulting in uncoordinated efforts and internal conflicts, but beyond, to understand an overall goal, and strive for what perhaps has not been considered before.
Winning emerging technology platforms will incorporate this comprehensive approach up front, with interface and data management from consumer engagement all the way through supportive decision making and clinical services delivery. This will increase their value to potential clients who are already struggling to adopt changes to address clinical variations and other systemic inefficiencies. Unfortunately, the cultural and systemic challenges are currently not well understood by many technology innovators.
mHealth could be the answer but there is still a lot of effort required of all parties.
Comments and opinions shared are always welcome. Follow the conversation on Twitter @ShiftHealthcare.