Siu, Rex K.

Abstract Objectives To comprehensively evaluate the validity of International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) codes for both prevalent diagnoses and less common diseases, and to assess the performance of a large language model (LLM)-based system in validating these codes. Materials and Methods This retrospective study analyzed hospital admissions from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. We developed a validated LLM-based system using GPT-4o, refined through iterative prompt engineering, to assess ICD-10-CM code validity. We measured the positive predictive value (PPV) of ICD-10-CM codes, PPV of principal and secondary diagnoses, and the performance of an LLM-based system in code validation. Results Among 865 079 assigned codes, the PPV was 84.6% (95% CI, 84.5%-84.6%). Principal diagnoses had a PPV of 93.9% (95% CI, 93.7%-94.1%), while secondary diagnoses had a PPV of 83.8% (95% CI, 83.7%-83.9%). The LLM system demonstrated high performance in validating ICD codes, achieving 93.6% accuracy, 95.4% sensitivity, and 85.2% specificity. Among correctly assigned secondary diagnoses, the majority (67.9%) represented historical or baseline conditions, while 32.1% reflected active conditions that deviated from baseline status; 22.3% of these emerged after hospital admission. PPV decreases with later diagnosis positions, with the largest decline occurring between principal and secondary diagnoses. Discussion and Conclusion In this large-scale evaluation, ICD-10-CM codes exhibited generally high accuracy, though variability existed by position and condition type. A validated LLM system performed comparably to physician review and offers a scalable means to improve coding accuracy. These findings support the potential for integrating LLM-based auditing into routine workflows to strengthen the quality of administrative and research data.

ABSTRACT Objective To comprehensively evaluate the validity of ICD-10-CM codes for both prevalent diagnoses and less common diseases, and to assess the performance of a large language model (LLM)-based system in validating these codes. Materials and Methods This retrospective study analyzed hospital admissions from the Medical Information Mart for Intensive Care (MIMIC-IV) database. We developed a validated LLM-based system using GPT-4o, refined through iterative prompt engineering, to assess ICD-10-CM code validity. We measured the PPV of ICD-10-CM codes, PPV of principal and secondary diagnoses, and the performance of an LLM-based system in code validation. Results Among 865,079 assigned codes, the PPV was 84.6% (95% CI, 84.5%-84.6%). Principal diagnoses had a PPV of 93.9% (95% CI, 93.7%-94.1%), while secondary diagnoses had a PPV of 83.8% (95% CI, 83.7%-83.9%). The LLM system demonstrated high performance in validating ICD codes, achieving 93.6% accuracy, 95.4% sensitivity and 85.2% specificity. Among correctly assigned secondary diagnoses, the majority (67.9%) represented historical or baseline conditions, while 32.1% reflected active conditions that deviated from baseline status; 22.3% of these emerged after hospital admission. PPV decreases with later diagnosis positions, with the largest decline occurring between principal and secondary diagnoses. Discussion and Conclusion In this large-scale evaluation, ICD-10-CM codes exhibited generally high accuracy, though variability existed by position and condition type. A validated LLM system performed comparably to physician review and offers a scalable means to improve coding accuracy. These findings support the potential for integrating LLM-based auditing into routine workflows to strengthen the quality of administrative and research data.

ABSTRACT Treatment options for patients with inflammatory bowel disease are constantly evolving; however, medication-refractory disease remains an issue. Pediatric case series show the potential benefit of sirolimus therapy in refractory Crohn's disease (CD); however, limited data exist in adult patients. As such, we retrospectively identified and report clinical outcomes for 4 patients prescribed sirolimus for treatment of refractory CD. Despite a median sirolimus therapy duration of 524 days and some therapeutic benefits, all patients discontinued therapy due to adverse effects. Our findings suggest that while sirolimus may have clinical utility, its role may be limited by treatment-derived adverse effects.

Abstract Introduction Crohn’s disease (CD) with short bowel syndrome (SBS) can present as chronic intestinal failure (CIF) often requiring nutritional support. Teduglutide is a treatment option for these patients. We investigated clinical outcomes of CD-CIF patients with SBS treated with teduglutide. Methods Adults with CD-CIF and SBS who received teduglutide were identified at a tertiary care academic center between 2012 and 2023. Data was collected retrospectively. Primary outcome measured was reduction in parenteral support (PS) by ≥20% volume, with PS defined as utilization of parenteral nutrition (PN) or intravenous fluids (IVF). Several secondary outcomes included immunosuppressive medication changes, subjective symptom improvement, and stool output. Results We identified 32 patients with CD-CIF and SBS receiving teduglutide. Comparing clinical outcomes before and after teduglutide, 26 of 32 patients achieved the primary outcome of ≥20% PS reduction. A decrease was seen in patients requiring PN + IVF, with corresponding increases in patients requiring PN only and IVF only. Among all 3 groups, a total of 23 patients received PN prior to teduglutide, which decreased to 14 following teduglutide. Weekly PN volume reduced from 7.00 to 3.55 L and weekly frequency decreased from 7.00 to 3.00 instances (P < .01). Reductions in weekly volume and frequency were observed among all patients receiving IVF support (25 vs 15). Secondary outcomes showed improvement in patient reported subjective symptoms (84.4%), stool output (90.6%), patients meeting criteria for diarrhea/high ostomy output (27 vs 14), and use of unique antidiarrheal medications (3.0 vs 2.0). Conclusions This retrospective case series demonstrated improved clinical outcomes in patients with CD-CIF and SBS treated with teduglutide resulting in decreased PS requirements, antidiarrheal medications requirement, and stool output without significant effects on immunosuppressive therapy.