Advanced Analytics Methodology: Describe, Diagnose, Predict & Prescribe
The spiraling cost of healthcare faced by our nation has reached more than $2 trillion and its inflationary pressures remain thus far unabated, outpacing all other categories of spending increase by many multiples. Within this vast number, healthcare waste and inefficiencies add up to more than 30% of each dollar spent. This portion of healthcare spending not only results in no therapeutic value to consumers; at times, damage is actually done to consumers through medical errors or consequences of overtreatment.
For decades, stakeholders have tackled this national emergency by using time consuming tactics that ultimately only affect 3-5% of the cost. The rationale behind this strategy has been that the adjustments were easy to make and provided some temporary relief, the proverbial low-hanging fruit which would partially offset the precipitous increases in healthcare spending. While these methods were a way to kick the can a bit further down the road, these strategies never provided the level of relief stakeholders sought.
Today’s healthcare market turmoil makes it all the more important for more innovative methods to be utilized in meeting these challenges. With the advent of better business intelligence technologies and the Big Data Movement, the time is right for a strategic approach that brings to bear 4th Generation analytic tools applied to population health management. These tools go beyond historic data warehouse solutions in speed and power. In addition, 4th Generation tools facilitate the use of more elegant predictive and prescriptive data strategies that allow users of these solutions to identify more meaningful opportunities and quickly ascertain the best courses of action. Every manager of a group’s health needs a customized real solution that fits the members of their specific group. Managing an ailment with the right treatment by the right provider at the right time is the “Holy Grail” that these 4th Generation tools can uncover and the best way to effectively tackle the 30% waste and inefficiency in the current system.
To maximize the value derived from these tools, stakeholders must go beyond the more basic ways of looking at a population and instead utilize the 4-step process of Advanced Analytics Methodology. These 4 steps – descriptive, diagnostic, predictive, and prescriptive analytics – can be seen as stair steps towards the goal of better understanding a population and how to optimally identify opportunities for more efficient and effective utilization of healthcare. This more complete and comprehensive strategy allows stakeholders to focus on the 30% – not just the 5% – and is a game changer for those innovative enough to seize this opportunity.
The first two steps are the ones most commonly undertaken when stakeholder try to understand their population. Descriptive analysis looks at the characteristics of the population and focuses on the past. Data is historic and is used to describe the population. This information is valuable for reporting and record keeping but really just scratches the analytic surface. Next, diagnostic analysis begins to explore drivers of cost, use, and other factors of healthcare utilization. More value is found here, but it still only provides very rudimentary guidance as to how to impact these drivers of cost.
Let’s look at an example of what traditional analytics tend to miss and what stakeholders can now largely avoid through more advanced business intelligence methods.
A large manufacturer found that their mammography rate was 80%. Their carrier reports put them above the benchmark and touted this screening rate as a great accomplishment for the organization. Seeing this as a win, the firm’s carrier had never taken them beyond the most basic descriptive statistics. This same employer had 6 cases of breast cancer diagnosed at stage 4 by the end of the year which they found surprising given their 80% mammography rate. This company had a predominantly male employee population, but all 6 advance breast cancer cases came from the 15% female employees. Upon closer examination, while spouses of male employees had a greater than 90% screening rate, female employees had a mammography rate of only 10%. None of these 6 affected women had had their recommended screening in the past several years. Even diligently applied diagnostic statistics might have found this need for intervention and education about preventive care, but imagine if predictive and prescriptive analytics had been in play. This and many other risks could be identified sooner, providing opportunities for action and often, avoidance of illness, expense, and disruption of lives.
The power of 4th Generation Analytics shows its value as one reaches the final two steps, predictive and prescriptive analytics. Through predictive analytics, the probability of various outcomes can start to be assessed. This allows strategy to flip from taking a reactive stance to adapting a more proactive position. Plan design, supportive programs, and education can all focus on what is expected or what needs to be changed making for more accurate and efficacious population management. Finally, prescriptive analytics are an area largely untouched by previous data strategies. Prescriptive analytics take all that was assessed, analyzed, and anticipated in the previous 3 steps and applies not just opinion but the body of population health science to recommend the best actions to take based on the user’s focus. In this way, stakeholders have the opportunity to take the lead.
Weighing out what choice is best when facing limited time and resources is often essential to avoiding waste and maximizing the value of healthcare dollars. In this example, an employer was steered toward a less valuable program without a complete analytic view of the situation.
A real estate firm was paying for an asthma disease management (DM) program at the advice of their carrier, because reports showed a spike in ER cost by asthmatics. The program was represented to the company as a success as ER costs did reduce significantly after the program was implemented. However, when population health analysts studied this company’s ER use over the last 3 years, it was found that a single asthmatic ER patient with complex comorbid conditions generated the majority of the costs used to justify an asthma DM program. This same organization has a significant prevalence of diabetics that are poorly managed with very low level of compliance with standards of care. Using 4th Generation Analytics, it was seen retrospectively and could have been seen prospectively that the resources spent on asthma would have been better utilized to engage the diabetic population where the greater need was. They could have implemented a more meaningful intervention by better managing this much larger group of patients who needed eye exams, blood tests, and assessment for neuropathies, potentially saving this organization over $400,000 in short-term savings. The complex asthmatic patient would have been better served by a more personalized care management approach that addressed his multiple conditions rather than just his asthma, still saving both the patient and employer money through more coordinated care but without offering a program that ended up going largely unused by the company.
These systematic and scientific steps would benefit all those trying to improve the efficiency and quality of care and would avoid the unexpected outcomes many stakeholders are left to managing. Without bringing advanced analytics and the vast knowledge available from population health research, mistakes and inefficient use of the system will continue unchecked. Companies with the power of these population health analytic tools will have the edge in providing sustainable, high quality health solutions to their populations as well as fostering a healthier and more productive workforce, a critical two-way win for our nation’s health.
About the author
Rabih Suki, MPH, PhD, Chief Strategy Officer of Dynamic Health Strategies
- PhD—Bio-Statistics, and Minor in Epidemiology and Operational Research, The University of Texas School of Public Health, Houston Health Science Center
- MPH International and Family Health—The University of Texas School of Public Health, Houston Health Science Center