Clinical validation is a hot topic across the healthcare spectrum.

With the expansion of public and private payer audits and the trend of increased claim denials clinical validation has a growing impact on healthcare organizations nationwide. In fact, in a recently published survey by the Association of Clinical Documentation Integrity Specialists (ACDIS), participants indicated that up to 46 percent of their organization’s denials fall in the clinical validation bucket. ¹

So, what exactly is clinical validation? Clinical validation means that the diagnoses documented in a patient’s record must be substantiated by clinical criteria generally accepted by the medical community. It has implications for diagnostic documentation, coding, claims submission, audits and denials, recovery audit processes, regulatory compliance, and sanctions. ²  

Addressing Clinical Validation Denials

Because of the increasing importance of clinical validation, clinical documentation improvement (CDI) and coding programs have started to incorporate clinical validation review, either at the point of care or at the time of coding, to increase accuracy and reduce post-claim denials.

Unfortunately, for most healthcare providers, identifying cases that fall into the at-risk bucket is difficult and time-consuming. It is typically a manual process and requires a secondary review to determine whether the provider documentation of a definitive diagnosis is supported by the clinical facts in the record. Accurately detecting the absence of key clinical evidence can necessitate a painstaking review of the patient record and involve advanced skills requiring a strong understanding of clinical pathology.

Applying NLP to Clinical Validation

Natural language processing (NLP) plays an essential role in improving physician and hospital coding, quality measures, utilization management, and CDI. Now artificial intelligence (AI) and NLP can be applied to clinical validation to positively impact the integrity of medical record documentation.

Clinical validation is exceptionally challenging because you need to identify gaps in the patient’s story. Unlike NLP for coding, which recognizes what is clearly stated by the physician in the record, NLP for clinical validation looks deeper into the supporting evidence—including laboratory results, vital signs, and key findings—to understand what the physician didn’t say. This requires a deep understanding of the evidence within records, models that effectively capture and weight the evidence, and reasoning that connects the evidence to specific diagnoses and procedures.

Revolutionizing CDI with Automated Case Finding

NLP-driven automated case finding proactively recognizes when the clinical picture is not fully consistent with the coded diagnosis. It does so by identifying the clinical evidence documented within the medical record related to specific conditions and then compares that evidence to definitive documentation. For example, in a patient record where the physician documented acute respiratory failure, the NLP will look for a range of clinical evidence, including pulse oximetry, respiratory rate, blood gas results, symptoms such as shortness of breath and stridor, and treatment with supplemental oxygen. All of that evidence paints a clinical picture of the underlying disease.

NLP and automated case finding can assist CDI professionals with the challenging task of clinical review. By precisely identifying records with documentation weaknesses, automated case finding lets CDI specialists zero in on the most critical opportunities. By presenting CDI experts with the right charts at the right time, automated case finding, combined with dynamic workflow, dramatically decreases the time required to conduct an initial review, reduces the need to add additional staff, increases accuracy and helps prevent downstream denials.  

Modeling Complex Processes

Clinical validation is an excellent example of what is possible when NLP and AI are applied to model a complex process. This modeling requires extensive data from which to learn and an expansive knowledge of medical specialty guidelines. By effectively integrating CDI case finding and NLP-driven coding, all powered by the same clinical NLP technology, healthcare providers can successfully apply AI to increase accuracy and reduce clinical-validation-based denials.

To learn more about clinical validation and artificial intelligence check out our podcast.

References

1 “2020 CDI Week Industry Overview Survey,” ACDIS, September 13, 2020.

2. “What Is Clinical Validation,” Coding Corner, ACP Hospitalist, December 2016.