Today, errors in complex claims lead to substantial levels of health plan overpayment. Detecting and fixing those errors, however, requires a high level of manual effort, which is hard to believe considering these clinical claim reviews can require the review of hundreds, if not thousands, of pages of documentation. The manual aspect of this review process makes it difficult to identify errors and even more difficult to identify trends at scale.
Why Claim Errors Occur
The rules governing claim coding are notoriously complex and difficult to understand, resulting in a high number of errors and overpayments. Centers for Medicare & Medicaid Services (CMS) data suggests a Medicare error rate of 7.38%, many of them due to upcoding (using billing codes that reflect a more severe illness than actually existed or a more expensive treatment than was provided) to the tune of $1.5 billion in improper payments. Additionally, mistakes may occur if billing departments make errors when interpreting physician notes.
Some common causes for claims errors include:
Incomplete Data: The claim does not have complete information about the patient or the claim (e.g. date of care)
Wrong Data: The claim contains mistakes in patient data (misspelled patient name, inaccurate policy number, etc.) or provider data (provider address, etc.)
Eligibility: The patient is not eligible for the billed service or a prior authorization was not obtained
Late Filing: The claim exceeds the deadline for filing
Coding Error: The claim contains an obsolete, incorrect, or missing code
Insufficient Evidence: The medical record documentation does not support the code used (fraud and abuse)
Whether through accident or intent, claim errors are a source of billions of dollars in excess health plan spending. These errors not only impact the healthcare system broadly, but can have a direct impact on patient care, potentially leading to clinically unnecessary interventions that could be harmful or create an inaccurate picture of patient health.
The Top 12 Claim Errors
According to data normalized across health plans covered by Machinify Audit, a collection of information from 52 million patients, a variety of clinical codes are being used subjectively, representing high levels of coding errors, fraud, and abuse. With a lack of standardized criteria for identifying specific codes, there’s often a disconnect between providers and health plans, allowing providers to take advantage and force incorrect payments. We’ve identified the top 12 or, as we’re calling them, the Dirty Dozen.
The Dirty Dozen
Sepsis: We’re seeing rising rates of admissions that include a sepsis ICD-10 diagnostic code in the Medicare population. This growth is disproportionate to the growth of the Medicare population, suggesting the need for closer inspection. While timely treatment remains critical to maintaining the current decline in mortality associated with sepsis, an increased use of Diagnosis Related Group (DRG) codes assigned to sepsis, along with the CMS weights assigned to those DRGs, has resulted in higher costs. Our own analysis of claims data across health plans suggests that not all sepsis claims meet the clinical, objective requirements for sepsis diagnosis set by CMS and Medicare.
Acute Kidney Injury (AKI): The claim does not meet the clinical criteria for acute kidney injury or may have been upcoded from a similar condition (e.g. acute renal failure). For example, AKI is often upcoded when patients are just dehydrated. The temporary increase in creatinine seen with dehydration can be identified correctly with advanced AI isolates the correct lab values and performs the calculation indicating the absence of AKI.
Chronic Obstructive Pulmonary Disease (COPD) or COPD exacerbation: The claim may be incorrectly coded without proof of disease or with symptoms that do not correlate to COPD exacerbation codes. For COPD in particular, it’s not just about money. The Dept of Justice has sanctioned hospitals for performing unnecessary procedures and treatments on patients with a range of pulmonary conditions including COPD. including invasive procedures like blood draws or administering antibiotics and oxygen, even when they were not clinically warranted.
Congestive Heart Failure (CHF): Miscoding CHF may lead to $993 million in overpayments per year. Often, there is no one lab test or x-ray that validates whether a patient had an exacerbation of CHF. It requires an assessment of multiple data points that don’t fit neatly into a formula.
Diabetic Ketoacidosis: If the patient does not meet the clinical criteria for this higher level code, then it is often coded based upon patient symptoms without clinical testing.
Malnutrition: ADHS study determined that hospitals overbill by $1 billion by incorrectly assigning malnutrition diagnosis codes not supported by the medical record, which reflected less severe malnutrition or no malnutrition at all.
Pneumonia: A pneumonia code or more complex pneumonia code is used for a less severe set of symptoms (e.g. fever and cough) without further diagnostic testing or associated with a length of stay that was not indicative of complex pneumonia.
Acute Blood Loss Anemia: This code is often incorrectly used when chronic blood loss anemia is the appropriate choice or the documentation is insufficient to distinguish between anemia types.
Urinary Tract Infection (UTI): Currently, a UTI is the most common hospital-acquired infection, but not always reflected in claims data or incorrectly attributed as present on admission.
Encephalopathy: The number of codes for encephalopathy and incorrect attribution of altered mental status as encephalopathy often lead to incorrect use of codes without support from the medical record.
Myocardial Infarction: The patient may have received monitoring or pain medication, not meeting the diagnostic criteria for myocardial infarction or following specific interventions for myocardial infarction (e.g. heart cath or bypass).
Experimental, Investigational or Unproven (EIU): The claim includes a drug, surgery, or device that is not yet accepted by the medical community.
How to Reduce Errors and Ensure Payment Integrity
Payment integrity (PI) programs are integral to finding these errors, but often these programs are highly manual, generating a low hit rate (the rate that claims are correctly selected for audit).
Often, health plans turn to purely outsourcing their payment integrity programs to solve or catch these problems, but outsourcing makes the programs more difficult to manage. This is particularly true when multiple vendors are brought in to assist. Plus, the post-pay claim correction process, or recouping incorrect payments that have already been paid out, is, no surprise, a source of significant provider abrasion.
To improve the payment integrity process, health plans should increase claim identification accuracy, reduce the number of vendors needed for reviews, and help drive case auto-validation to reduce provider abrasion. To do so, health plans need:
A Policy Source of Truth: Policies covering thousands of procedures and diagnosis codes should be standardized and codified to eliminate subjective human judgment and biases and support more consistent, accurate, and timely decision making with clear, defensible rationales for all clinical and coding decisions.
Autonomous Claim Review: Leverage software that has a deep understanding of patient care trajectory and knows the language of medical coding to identify claim coding with higher hit rates.
A Human-in-the-Loop Approach: Treat AI as an assistant. It can scan medical records and identify critical elements for a human clinical reviewer to validate or invalidate. Trained clinical reviewers are still in the driver’s seat, AI is just riding shotgun.
An Insights Feedback Loop: Ensure that post-pay audit insights are incorporated into pre-pay decisioning. For policy review, help pinpoint areas for policy improvements to better clarify and standardize best practices in diagnosis or treatment of these issues.
Improving Audit Efficiency & Accuracy with AI
Machinify Audit has been proven to reduce medical record review time by 50% in large scale production studies, helping increase the volume of reviews health plans are able to review through a build-operate-transfer (BOT) model that combines elements of insourcing and outsourcing.
Our AI-powered capabilities examine the claim against the medical record with great accuracy, presenting a summarized finding of documentation to human reviewers to help remove the subjectivity in determinations. In combination with Machinify's deep understanding of the care journey of typical patients, we identify anomalous claim coding with high accuracy, seeing hit rates over of 60%.
For decades, the healthcare industry has considered medical records a source of truth. However, Machinify has uncovered a trove of misleading information in medical records, both intentional and unintentional. Understanding and identifying these elements is key to finding the truth about the care delivered.
To learn more about how Machinify can help address your payment challenges and to see Machinify in action, schedule a demo.