Since June 30, 2011, all Medicare FFS claims are analyzed using predictive modeling before they are paid. Section 4241 of the Small Business Jobs Act of 2010 mandated the Centers for Medicare and Medicaid Services (CMS) implement a predictive analytics system to analyze Medicare claims to detect patterns associated with high risk of fraudulent activity. Based on the analysis, profiles of providers, networks, billing patterns, and beneficiary utilization have been developed. These profiles have been used to create risk scores which are used to alert CMS of aberrant behavior.
Home health claims contain a significant amount of information to provide CMS with a picture of the patient’s current condition, the services provided, and length of services. The claims include Diagnoses, which should support the medical necessity of the ordered services. The Statement Covers Period (From/Through date) and the Admission Date, which indicate how long each patient has been actively on service, can be used to determine if an agency has a high recertification rate.
The HIPPS code submitted on the claim reveals if the episode is an early episode (the first or second episode of home health care without a 60-day break in services) or a late episode (the third or later episode without a 60-day break in services). Claims submitted with a new Admission Date but a late episode indicate the agency is admitting a patient who has already received two episodes of home health care within the past 60 days from another agency or that they are readmitting the patient to their own home health agency within 60 days of discharge. Both of these situations occur in the normal course of home health care, but if they occur frequently could indicate fraudulent behavior.
The HIPPS score also includes the clinical, functional, and therapy utilization needs of the patient based on the completed OASIS assessment. Typically this value should correspond with the intensity of services provided. If agencies have a pattern of submitting claims with high therapy utilization and HIPPS scores indicating low clinical and functional needs, it can be indicative of a higher risk of fraudulent activity.
However, the claim is only a snapshot and not a detailed portrait; only the complete clinical record can support or reject an assumption of inappropriate billing. Therefore, CMS does not deny claims solely based on the alerts, but uses the data to assist in claims review activity and to enhance its database for future activities.
Forms of predictive analysis have been used for decades to manage patients at high risk for hospitalization, falls, etc. Predictive modeling risk adjustments are used to compare outcomes based on OASIS assessments. History has taught agencies to be proactive in approaching the results of the analysis and to develop a plan of action based on the findings.
Agencies have taken action based on predictive modeling when developing programs for their patients. Falls and hospitalization prevention programs, resource management plans, disease management programs, and other programs based on the predictive characteristics of the patients they serve have long been a part of home health care. Risk-adjusted outcomes from their OASIS assessments are used to create performance improvement plans to address less-than-optimal outcomes.
Now agencies must apply the same approach to the analysis of their claims and the services they provide. They should be aware of their numbers: what is the agency’s recertification rate, average number of visits per patient episode, average length of stay, average number of therapy visits per patient episode, and the correlation between diagnoses, functional scores, and visit intensity. Based on these findings, agencies should prioritize their clinical record audits-choose the cases in which the patient has been on service for multiple episodes, been transferred from another agency and/or has high therapy utilization. They must evaluate if the documentation clearly identifies the patient’s need for service and supports the skilled need and medical necessity of the services provided as reported in the OASIS assessment.
Failure of documentation to do so can usually be traced to one of three occurrences: First, the OASIS was not accurately completed to reflect the patient’s true functional needs. This typically occurs when the functional scores are determined by a nurse instead of a therapist. Agencies who identify this problem will need to either retrain the nurses or get more input from the therapists. Second, the documentation is poorly written and does not reflect the true condition of the patient or the skill level of the services provided. Training in proper documentation techniques is the only solution to this problem. Third, there can be a true disconnect between the patient’s needs and the services provided. Agencies who identify this problem will need to work on care planning and oversight.
CMS believes that ultimately, predictive modeling will benefit honest providers by minimizing the disruption to providers with low risk scores. In most instances, agencies that are aware of the information being provided on their claims and conduct their own internal reviews should not be seriously affected by the use of predictive modeling. However, as with any profiling method, there will be some good providers caught in the risk pool. While unfortunately, this will add cost for copying records and slow cash flow for the episodes selected, it should not result in denial of claims, and agencies should be spared additional scrutiny when proven correct.
About the Author
Bonny Kohr, RN, CHCE, HCS-D, is the manager of clinical services for FR &R Healthcare Consulting, Inc. She is a registered nurse, certified homecare coding specialist, and certified homecare and hospice executive. Bonny has worked 23 years in home health care, beginning as a field staff nurse, then as a clinical director, and finally as the chief operating officer.
Contact the Author
To comment on this article please go to email@example.com