Tracking Quality, Risk and Cost: Three Things that Harm Value-based Payments
By Gene Rondenet and Lucy Zielinski
The Affordable Care Act in 2010 made a large shift from fee for service to fee for value (or pay for performance) and the trend has been increasing in recent years. According to Modern Healthcare Hospital Systems Survey, only about 15% of respondents (n=60) said they derived 10% or more of their net patient revenue in 2016 from risk-based contracts. Across the US, there is variability in the percentage of FFS vs FFV reimbursement—some provider organizations receive a majority FFS payments, while others are majority FFV, while most are somewhere in the middle, leaning more towards FFS. According to industry experts and CMS, the climb to FFV continues. As a result, provider organizations must prepare to understand the payor’s risk methodology and how to best represent the patient care provided through accurate documentation and coding. Additionally, organizations must realize the strategic value of clinical data, and how it can negatively impact or harm value-based payments. [QRC estimates the negative impact can be as high as 25%; $4M in payments—over $1M may be missed}
Risk models are technically complex, and they can be a very important driver of capitation rates. Data is collected from insurance claims and clinical systems (including EHRs and registries) and then aggregated and converted into a risk score. For example, the CMS HCC Risk Adjustment Model supports the Medicare Advantage program by adjusting payments based on several factors including beneficiary demographics, medical history and current conditions. Payment to the provider is based on this risk score and the risk level they have taken. The problem is that there is an inherent disconnect with the data that is required to accurately report a risk score and data that is captured by a physician. Physicians are not trained to be “data creators,” hence the data may not reflect the service or the complexity of the patient—or the system may not be set up to support accurate data capture. The challenge today is that meaningful data cannot be extracted, it may not exist in the medical record or HIT system or it is difficult to aggregate data from disparate systems.
{Examples: 1. Morbid obesity—physicians are reluctant to document the diagnosis because they don’t want to stigmatize the patient. This is a condition that is more complex to treat, and it does results in higher risk score and an increased capitation rate. Pre-ACA, once a patient is labelled with a diagnosis, coverage may be lost due to a pre-existing condition. 2-Proper codification of historical information—i.e., mastectomy. People are excluded from measures because codes are changing. In past procedures and illness. ICD-10 codes are not captured. There is no backup documentation or the information is captured by a bad historian. Issue—the date when it was done (i.e., colonoscopy—8 or 10 years…need more data) Ideal – subjectively trying to measure quality and using risk objectively. High risk patient, your standards on quality care change. I.e. diabetic – different measures on quality, i.e. blood sugar and BP. LDL under control.}
As payers continue to base providers’ payments on FFV methodologies or risk scores, it is pertinent that organizations review their data and understand risk. Three things that impact the risk accuracy rate, resulting in a harm to value based payments, are described below.
- Poor EHR set up.
With the EHR Incentive Program, many physicians and organizations were quick to take advantage of these dollars by selecting and implementing certified EHRs. However, many of these EHRs were not implemented with much thought to support the long-term capture of discrete clinical data elements. Healthcare organizations typically focused on just the financial implications of the data. For example, lab data with the LOINC codes and Diabetic A1cs are not documented appropriately to support the proper care of patients. - Inadequate training in comprehensive medical record documentation.
Providers have been trained to use EHRs, however, most have not received sufficient training on how to document accurately and comprehensively. EHR templates may have been implemented, however, may not support data capture. For example, a physician may use a template designed for a “Diabetic without complication(s)” that does not properly document the needed data for a patient with complications. This will adversely impact the risk score. Often, when data scientists review clinical data, ambiguous data can be detected and corrected. These sorts of retrospective reviews allow an organization to recoup missed payments, and more importantly improve data capture prospectively, resulting in accurate risk scores and payments. {Accurate risk scores lead to improved patient care and decreased medical costs.} - Deficiency in accurate diagnosis coding.
In FFS, payments are primarily made based on procedure or CPT-4 coding. In FFV, diagnosis (ICD-10) coding matters. The ICD-10 codes indicate the condition and any comorbidities of the patient, and factor into the risk score. For example, if a diabetic has complications and co-morbidity, the risk score is higher, resulting in higher payment. Not only will the provider be properly paid, but additionally the patient will also receive the appropriate level of care (i.e., screen for risk and falls, depression, etc.).
Provider organizations continue to be challenged with understanding the value of informatics; informatics are not regulated, nor widely understood. For example, physicians are charged with leading quality teams who make decisions about EHR templates, quality measures, and clinical documentation. These teams often have a clinical understanding of data; but do not have an understanding of how to capture and model the information in a way that is useable. Measure specifications supported by industry standard coding sets (i.e., CPT-4, ICD-10, LOINC, SNOMED, etc.) are complex. It really takes data scientists working with clinicians to bring clarity to the complexity of this data and these specifications.
Organizations, as they continue to inevitably move towards more risk-based payments, must assess the effectiveness of their data in support of these models. Some plans, such as Medicare Advantage, allow providers a 13-month prospective period, to review accuracy and submit corrected claims. This prospective review, may result in additional payments to the provider organizations, sometimes hundreds or even millions of dollars. They key is for organizations to track Quality, Risk and Cost accurately and frequently, capturing the right data – and reviewing it. In the end, the provider organization will be appropriately paid for managing the care of its population and the overall quality of patient care is improved.