The United States spends more on health care than do 13 otherhigh-income countries, according to The Commonwealth Fund’s “U.S.Health Care from a Global Perspective” report.

It is estimated that nearly 20 percent of that spend goes towaste, including overtreatment, lack of care coordination, andfraud.^[1]

Minimizing wasteful medical spending —spending that could be cut without negatively affecting access tocare, quality of care, or health outcomes — represents asignificant opportunity to reduce not just direct medical costs butcosts associated with insurance premiums, which have beenincreasing between 3 and 13 percent per year since 2000 andoutpacing inflation and earning increases (which usually hoverbetween 2 and 4 percent).^{[1, 2]}

Doing so also has the potential to improve quality in ways thatbenefit employers, health plans and patients alike.

Predictive analytics — using data and algorithms to predict aparticular event — provides a tool to reduce wasteful spending insuch areas as unnecessary tests or procedures, inappropriate placeof service, and wrong diagnosis or care.

Unnecessary tests or procedures

A commonly cited example of wasteful spending is prostate cancer testing. TheU.S. Preventive Services Task Force gives the PSA test (a bloodtest) a “D” grade, noting that the test often produces falsepositives and exposes the patient to the subsequent risk and costof unnecessary treatment.^[3]

Further, as the website Choosingwisely.org explains, usingimaging scans (such as CT, PET or bone scans) on patients who havebeen deemed to have early-stage prostate cancer that will notlikely spread, exposes the patient to the risk of radiation, highcosts, and false positives that could lead to additional stress andthe cost and risk of unneeded treatment.

Put simply, the risks outweigh the potential benefits forotherwise healthy men. Predictive analytics can help reduce thiswasteful spending by proactively reaching out to men withinformation about the risks and benefits of such tests and lettingthem know to question their doctor’s recommendation for prostatecancer tests.

Inappropriate place of service

An illustrative example is the common ordering of an imagingtest. The out-of-pocket price for astandard chest x-ray, CT scan, or ultrasound can vary by hundredsof dollars, depending on where the imaging is done. Often thestress, confusion or simple lack of knowledge can lead to suchtests being performed at unnecessary expensive facilities.

Take, for example, a doctor who orders an MRI test for apatient. The patient then schedules the MRI at the local hospital,even though the same screening could have been done at a nearbystandalone clinic for a fraction of the price. In some of thesecases, the individual may not have the comparative informationreadily available when making a care decision; and anurgent-seeming situation precludes the patient from doingresearch.

In other cases patients simply don’t know that price differencesexist from facility to facility (or how great those differences canbe), and therefore don’t choose to shop around. Predictiveanalytics can be used to proactively intercept individuals who mayrequire an MRI based on risk scoring models and generate awarenessof the disparities in price in order to promote shopping aroundwhen they actually end up in the market for this service.

Delivering information proactively can save the patient frommaking a potentially very costly choice that ultimately does notimprove their quality of care or health outcomes.

Wrong diagnosis or care

Incorrect diagnoses and/or inappropriate treatment plans accountfor another portion of wasteful medical spending. For example, arecent update on the Spine Patient Outcomes Research Trial revealedthat eight-year outcomes were not significantly different betweensurgical and nonsurgical treatment for spinal stenosispatients.^[4]

Despite this evidence and regardless of the fact that the costsof surgery and lost productivity due to absences at work aresignificantly greater than those for non-surgical treatments, therates of spinal stenosis surgery have risen considerably over thepast decades.

This source of wasteful spending may be more common than manythink. According to a 2012 survey, 37 percent of medical diagnosesare wrong, and 75 percent of treatment plans require acorrection.^[5]

In these types of examples, predictive analytics can help targetpatients who have received particular diagnoses and may be headingdown a path of treatment for which the outcomes are questionableand equip them with resources (such as information about freeexpert medical-opinion services) that can help ensure they have thecorrect diagnoses and treatment plans before moving forward.

Put simply big data (and the predictive analytic applicationsthat use them) helps faster identification high-risk patients;better intervention; and better follow-through fromHIPAA-compliant, data-driven monitoring.^[6]

A recent Forrester research study shows that predictive analytictechnology is on the minds of human resources professionals.

In a survey of 100 such professionals, 62 percent said theywould like to use predictive analytics to provide the right care toeach employee in order to optimize benefits utilization; halfreported that they see value in using this type of technology todirect employees to the right provider.^[7]

Predictive analytics gives HR professionals real-time insight,allowing them to connect with the right people at the right time(i.e. before they make a potentially costly or harmfulchoice).

Key recommendations

The Forrester research study concluded that hyper-personalizedmessaging can help employees use the rightbenefits, in the right way, and at the right time —and big data and predictive analytics are needed to create andexecute these messages. Organizations wishing to adopt thesetechnologies should look at these five key capabilities.

1. The ability to engineer predictive analytics into existingbenefits hubs or implement a single-source portal.

2. The ability to ingest a myriad of data, including demographicsdata; clinical data, such as medical and pharmacy claims and datafrom labs, biometrics, and wearables; encounter/program utilizationdata; environmental data, such as weather, pollen and air quality;contextual data, such as location or device; and self-reported datathat provides as close to a real-time picture of the employee’sprioritized needs and opportunities as possible.

3. Predictive analytics to build personalized benefits profiles foreach employee and covered spouse that “learns” about the individualwith each interaction with benefits and during specific lifeevents.

4. Hyper-personalized push messages to employees and covereddependents when they need it.

5. The ability to monitor the impact of messages on benefitutilization and costs.

Utilizing technologies with these capabilities promises to be amore reliable way to reduce wasteful health care spending, thussaving money for patients and payers and reducing the risk, harm,and cost of unneeded or ineffective care.

Sources:
^[1] Health Care Costs: A primer. The Henry J. KaiserFamily Foundation, 2012.
^[2] Squires, D., & Anderson, C. (2015). U.S. HealthCare From a Global Perspective: Spending, Use of Services, Prices,and Health in 13 Countries. Issue brief (CommonwealthFund), 15, 1-15.
^[3]http://www.uspreventiveservicestaskforce.org/Page/Document/RecommendationStatementFinal/prostate-cancer-screening

|

^[4] Weinstein, James N. DO, MS et al. Long-TermOutcomes of Lumbar Spinal Stenosis: Eight-Year Results of the SpinePatient Outcomes Research Trial (SPORT). Spine,January 2015
^[5] https://www.bestdoctors.com/contact-us/faqs.
^[6] Groves, P., Kayyali, B., Knott, D., & VanKuiken, S. (2013). The ‘big data’ revolution inhealthcare. McKinseyQuarterly, 2.
^[7] ″Employees Don't Know What Benefits They AreMissing, But Predictive Analytics Can Help″ Forrester Consulting,2016.