By Brian Heacox, Decision Point Healthcare Solutions
Measuring provider performance is complex. The analytics required to produce valid and reliable metrics is challenging. Using those metrics to make a tangible impact on outcomes is even harder.
To date, healthcare organizations have focused most of their provider-focused analytics on cost and utilization, quality and operations-based metrics, and have delivered those profiles of physician performance to providers attempting to highlight opportunities for improvement. Though the analytics driving initiatives have solid underpinnings, efforts to improve outcomes have been sporadic and hard to measure.
In order to measure true provider performance and deliver actionable feedback, we, as analysts, need to fundamentally shift in the way we analyze provider performance and realign the expectations and responsibilities for performance improvement.
Backdrop Of Traditional Provider Performance Analytics
The concept of provider performance measurement has been around for a long time, and there are numerous technologies and tools available to help us measure cost, utilization and quality more accurately. Episode classifications, population risk groupers, type of service categorizations, compliance and adherence guidelines, for example, all help us more accurately understand the risk, utilization and compliance of a patient panel. They also can help us more appropriately compare the cost and utilization of one provider to another by adjusting for the mix of episodes or clinical severity of patients in the panel. These technologies are critical tools in the evaluation of providers.
Though, even with great tools, answering the question “what’s causing the provider’s sub-par performance?” cannot easily be done. We’re conditioned to look for the root cause because we want to succinctly communicate the root cause to the provider so that they can take direct action. This type of traditional data analysis, however, often results in many false positives. We’ve found that with this method many “lower performing” providers are flagged as “low” because of a very small number of “one-time” or outlier events like an expensive prescription fill for a rare condition.
Physicians generally want to deliver high-quality, cost-effective care to their patients. Instances of fraud, waste and abuse are extremely rare and are relatively easy to find with traditional tools. A true physician evaluation tool must evaluate a physician’s decisions against his available alternatives and provided recommendations based on the alternatives available to him.
Plans also have a difficult time displaying performance data to physicians in an effective way. Cost and utilization indices comparing performance to a risk- or case-mix-adjusted average do not provide enough depth or insight to motivate a provider to “change behavior.” Worse, it often leads the provider to question whether the data and methodologies are correct.
Social Determinants Of Health, Health Literacy And Health Engagement
The concepts of social determinants, health literacy and health engagement are directly linked and can provide crucial analytic value both separately and together when analyzing provider behavior.
For example, two members with an almost identical utilization and disease history can experience vastly different outcomes because of their social determinants of health. The probability of an admission (and readmission) for a 73-year old diabetic man with a comorbidity and a recent ER visit is 25-30 percent higher if the man lives alone, is low income, and has limited access to transportation compared to a man without those socio-economic barriers. Social determinants have a significant impact on utilization patterns, adherence/compliance and outcomes
Likewise, health engagement and health literacy play a role in utilization and outcomes. A member that is engaged with their doctor, has established a strong relationship with their care team and has taken an active role in self-care is 20 percent-50 percent less likely to be readmitted compared to a member who is less engaged.
The combination of derived, imputed values from health plan administrative data combined with third-party, member-specific consumer data can provide these social and engagement insights. Not applying these data points to the analytic exercise makes provider performance assessments indefensible.
Member Access, Provider Communications & Satisfaction
A member’s perception of their access, satisfaction and experience with their providers drives utilization behavior. A member’s perception could be the result of actual access and/or satisfaction issues, such as difficulty in making appointments because of an over-burdened doctor; or it could be a result of low member expectations, such as perception of difficulty in making appointments even through the member may not have actually attempted to make an appointment in several years. Regardless of the reason, perception impacts outcomes.
It’s crucial to not only understand each provider’s ratings across access, satisfaction and experience measures, but also to understand the difference behind real issues and perceived issues.
The healthcare industry’s CAHPS (Consumer Assessment of Healthcare Providers and Systems) survey is a healthcare consumer satisfaction survey that’s designed to gauge an individual’s perception of healthcare, access to care and specific areas of satisfaction. Clearly high CAHPS ratings start with having a high-quality healthcare organization with excellent access to care and meaningful processes in place to promote satisfaction. However, even with all this, CAHPS is heavily dependent on the demographic, utilization and disease profile of the population.
Because CAHPS is a government-sponsored anonymous survey it is not possible to link the survey responder to individual providers or to each individual’s disease, utilization, engagement and socio-economic profile. A “mock” or “diagnostic” CAHPS survey using the same (or similar) survey questions, can help uncover provider- and member-specific issues by linking individual responses to their providers and to their clinical, utilization, and consumer profile. Since these mock surveys are typically administered to a portion of the entire population and yield a 25 percent-60 percent response rate (depending on the channel used for the survey), plans can use machine learning techniques to extrapolate the results of the survey to all members in the plan. In short, machine learning helps identify members that exhibit similar behavior to members responding negatively to the CAHPS survey.
Using these methodologies, identifying high-volume providers and provider groups that are consistently high risk for predicted CAHPS can help target specific providers that may be impacting CAHPS scores. These providers and groups also have higher voluntary disenrollment rates, lower HEDIS rates, and higher risk adjusted utilization rates.
A Collaborative Platform For Continuous Improvement
A collaborative provider analytics platform that can measure performance by differentiating members not only by their disease and utilization history, but also by their engagement patterns, social determinants of health, access and satisfaction, delivers insights that can help the provider, the plan and the member to work together to improve outcomes.
This type of platform enables each stakeholder to view the population that is within their purview and promotes comparisons to benchmarks that are adjusted for provider engagement and social determinants. For example, how can a PCP be responsible for utilization behavior of clinically-complex members that they haven’t even seen? How can a specialists’ performance be compared to a benchmark that doesn’t adjust for the significant socio-economic factors that his patients’ experience?
A true provider analytics system shifts the methodical underpinnings and promotes performance improvement by focusing on:
About The Author
Brian Heacox is the product manager at Decision Point Healthcare Solutions.