In October, hospitals’ fledgling ICD-10 systems and processes will be put to their first major test with the addition of more than 6,130 new codes.
In October, hospitals’ fledgling ICD-10 systems and processes will be put to their first major test with the addition of more than 6,130 new codes. Without sufficient bandwidth to manage and verify appropriate utilization of the expanded code set, risk adjustment, quality measures reporting, and reimbursement levels may all be impacted, as well as revenue cycles that are likely still bouncing back from the initial post-ICD-10 hit. Not to mention the impact on data integrity and, subsequently, the effectiveness of the analytics needed for population health reporting. But the code set expansion isn’t the only data-driven challenge on the horizon — MACRA and MIPS will both require accurate analytics for successful participation.
Recently, Dr. Brian Levy, vice president, Global Clinical Operations with Wolters Kluwer, took time to discuss the role of analytics in effective ICD-10 management, MIPS and MACRA participation and quality reporting, and how hospitals can ensure that they are ready for the enhanced analytics these changes require.
Levy has 20 years of experience in medical informatics, with particular expertise in development of terminologies and clinical content, and the use of the Internet by patients and physicians to improve care delivery. He leads a team of physicians, nurses, PhDs, and professional medical coders who maintain the terminologies, mappings, and other content in the Language Engine. Levy is also a practicing Board Certified Internist and acts as a hospitalist at local hospitals who received his medical school training at the University of Michigan. His numerous articles and presentations in the healthcare informatics field detail his experience and expertise.
Q: What role does data normalization play in analytics?
Levy: Data normalization is the first step to achieving your analytics initiative. In order to accurately and efficiently report analytics, you need to be sure that your data is up to date, accurate, and reliable. Analytics often relies on data collected from multiple sources. Data normalization is required to coalesce these data from their disparate systems into unified codes. For example, finding all patients who have had an HbA1c test requires all the disparate lab codes for HbA1c be normalized.
Q: What kind of implications will occur if you have the wrong ICD-10 codes or your ICD-10 codes are not updated after the October 1, 2016 deadline?
Levy: With the wrong ICD-10 code or if a code is still documented as unspecified, the DRG shift can fluctuate dramatically, leaving you with potentially lower reimbursement than anticipated. For example, a physician can document Major depressive disorder, single episode, unspecified (ICD-10:F32.9), or if that same physician documented with more specificity to better reflect the patient’s actual condition to Major depressive disorder, single episode, severe with psychotic features (ICD-10: F32.3), the DRG shifts from 881 to 885 causing about $2,000 increase in reimbursement. So a 14-site hospital system could potentially lose $3 million in revenue with just this simple change in specificity. Getting more granular on diagnosis often not only impacts DRG weights, but also affects HCC, E&M codes, and more.
Treatment guidelines and other decision support are going to vary based on the severity of the diagnosis. Using Type 2 Diabetes as an example, a doctor is going to have a much different plan for a patient with Type 2 w/o complications (e11.9), but if there is a way to be guided to navigate clinical refinements or attributes of the diabetes, the patient may actually have something a lot more complicated like Type 2 Diabetes w/renal complications. Thus, more appropriate treatment guidelines can be presented to the clinician.
It is simply good practice for provider groups to make sure doctors are documenting correctly. Not just for money and outcomes, but for analytics and to ensure their own care guidelines are being followed. Payers are increasingly denying unspecified codes.
Q: How will accurate analytics affect other sections of an organization?
Levy: Without normalized data, you will greatly impact your billing workflow, you won’t be able to accurately report on quality measures, and potentially you may not be able to comply with MACRA/MIPS requirements.
When it comes to billing, for instance, with unspecified codes, provider workflow diminishes. Clinician and coder satisfaction decreases because they are constantly having to have a clinician specify codes that they have already submitted — wastes of time, money, and resources — and decreases in productivity when documentation isn’t accurate at the point of care with the required specificity.
Quality measure reporting requires data such as lab tests, medications, problems and diagnosis, and procedures. The quality measures themselves are composed of a pre-defined list (Value Set) of codes that can be used to trigger the quality measure, or may even be a set of codes that exclude one or more triggering measures. Ensuring all the patients that may be reportable are found is critical for quality measure reporting. Missing reportable patients may result in inaccurate reporting and result in financial impacts, inaccurate HEDIS Star Ratings, and other public ratings based on quality measure statistics. Normalized data allows physicians to find the right patients that meet the criteria for quality measure reporting.
The MIPS program will rely on scores assigned to providers based on their performance in four areas: quality, resource use, clinical practice improvement, and meaningful use of EHR technology. Accuracy of data is a critical component to all four of these areas. For example, interoperability will be a major focus for the updated Meaningful Use aspects under MIPS — and normalized data will be a critical component to actually achieving interoperability.
Q: Do billing codes reflect the problem list data from the EMR?
Levy: Traditionally, clinicians maintained a patient problem list independent from the billing codes submitted for reimbursement. So, often the problem list more accurately described the true patient details, while the billing data may have included unspecified codes or other less detailed codes chosen more for convenience rather than trying to capture the patient’s true conditions. In the EHR era, there can now be more synergies between the problem list and billing code data. For example, a problem can be dragged from the problem list over to the diagnosis list — converting more specific SNOMED codes into the best ICD-10-CM codes for billing. As billing data more accurately reflects patients’ true conditions, then analytics, quality measure reporting, data warehouse storage, HIEs, etc. will be improved through cleaner, more normalized data.
Q: What steps should be taken to prepare for MIPS and its financial implications?
Levy: MIPS solidifies the importance of collecting accurate and more complete data for reporting, achieving true interoperability of clinical information, and enhancing patient engagement.
Patient information is scattered across disparate systems, all of which is useful for effective provider decision-making and quality reporting. Bringing in out-of-network data requires a data normalization strategy and enterprise-wide information governance. Doing this will ensure clinical caregivers have necessary information at the point of care and that data is useful for analytics and quality reporting.
For instance, reporting under MIPS will require lab, drug, procedure, and diagnosis codes. As local lab results are reported, they may need to be normalized to the right standard (LOINC) required for the reporting. Problem list codes in SNOMED may need to be converted to ICD-10 codes for reporting, and drug codes converted to RxNorm.
To touch on the financial implications, documentation must be correct and as specific as possible to fully capitalize on reimbursement under MIPS. If clinicians do not properly document patient encounters, EHR and analytics reports will miss identifying patients that belong to population cohorts required for reporting.
For instance, if the right sinusitis codes or the right asthma codes are not selected, these patients will be missed when data is aggregated for quality measure reporting, leading to inaccurate results and potential payment reductions. Most commonly, EHRs are connecting clinicians to non-specified codes that result in lower reimbursements and less accurate analytics.
Health Language offers an EHR plug in called Provider Friendly Terminology that allows providers to document conditions, diagnoses, and procedures in language that aligns with their everyday language while coding to ICD-10 and SNOMED on the back-end, with clinical guides to avoid selecting unspecified codes. Time is of the essence as the first performance period for MIPS starts January 1, 2017. Terminology management that promotes optimal clinical documentation, interoperability, and patient engagement is now a priority.