Reimbursing health care providers is more complex that it might seem, and as a result, payments are often inaccurate. When health insurers overpay for care, premiums rise and consumers pay more – that’s why overpayment recovery programs are an important investment in more affordable care. However, these programs don’t always deliver the ROI plan leaders are looking for. One area where an upgrade is in order is data mining. SmartBrief spoke with SCIO Health Analytics’ Lalithya Yerramilli to learn more.
Where does data mining fit into health plans’ overpayment recovery programs?
When building a program as complex as overpayment recovery, there will never be a perfect one-size-fits-all solution that ensures system-wide payment integrity. Instead, to maximize recoveries, health plans are wise to employ a variety of tactics, each designed with separate and distinct (but complementary) purposes in mind. Complementary payment integrity tactics include:
- High-dollar claim reviews
- Fraud and abuse detection
- Coordination of benefits
- Subrogation
- Claim editing
- Data mining (prepay and postpay)
- Post-payment medical record reviews
- Prepayment medical record reviews
- Provider behavior modification
- Special investigations units
That said, due to high ROI, relatively low provider abrasion and the ability to identify systemic issues that can be fixed internally, we see data mining as one of the most important tools health plans have at their disposal for correcting billing and payment errors. However, like any other tool, it needs to be maintained in order to stay effective.
What is longitudinal data mining, and how does it contrast with traditional data mining?
In a recent webinar, we categorized data mining as a “trustworthy tool ripe for innovation” — meaning that while just about every health plan has some form of data mining in place, most were implemented years ago and have since grown stale. Updating current programs with what we call “longitudinal data mining” is an excellent way to incorporate the latest innovations and maximize outcomes. In traditional data mining, claims are reviewed one by one. In contrast, a longitudinal approach uses past claims behavior to inform the present analysis and identify data-mining trends. This wider view across claims catches rare error patterns that otherwise would remain hidden if relying on traditional queries.
What types of patterns emerge when you take a longer-term view across claims, and what might follow-up analysis show?
The truth is you never know what will appear. Because by its very nature longitudinal data mining uncovers rarer issues, often we find patterns by the same provider or at the same point in the adjudication process that have gone unnoticed for years. One especially useful application of longitudinal data mining is what we call “care transition analysis.” This application uncovers overpayments that can occur as members move between sites of care. The ability to link and track claims from site to site also allows health plans to uncover questionable coding and billing patterns such as inappropriate referrals or when hospitals and skilled nursing facilities both bill for the same rehabilitative care.
Does this process enable plans to address certain root causes of overpayment, reducing overpayments altogether?
Yes, because longitudinal data mining uncovers previously unknown patterns. Health plans can use this information along with benchmark trends to identify and fix issues that have been uncovered. For many plans, this is a pragmatic first steps toward reducing waste and preventing the majority of overpayments from occurring in the first place. Longitudinal data mining can also be applied as a prepayment solution, which would prevent payment errors, if not the initial billing error.
What benefits can health plans expect from adopting this technique?
Health plans that elect to use longitudinal data mining will benefit from increased recoveries without any increase in provider abrasion. Because data mining is an established tactic that does not require medical record requests, adopting this approach simply enhances an already-accepted tactic for payment integrity. If anyone is interested in learning more about longitudinal data mining, I would encourage them to check out our recent white paper on the topic, and also to explore the importance of custom content development as part of a robust data mining program.
Lalithya Yerramilli has 15 years of experience in analytics in insurance, health care, and life sciences industries working with customer info-base, transactional, physician level, patient level, claims and longitudinal datasets. She enjoys understanding the complexities of a business process and translating that information to develop predictive models that provide insightful solutions. Her experience has been in consumer, provider, and fraud, waste and abuse analytics using Bayesian statistics, cohort analytics, data modeling, factor analysis, clustering, CHAID, multivariate testing and machine learning algorithms. Learn more about SCIO Health Analytics’ smarter approach to overpayment identification and other health plan challenges at sciohealthanalytics.com.