Trends in Comorbidities and Complications Using ICD-9 and ICD-10 in Total Hip and Knee Arthroplasties

Abstract

Background: The transition to the new ICD-10 (International Classification of Diseases, Tenth Revision) coding system in the U.S. poses challenges to the ability to consistently and accurately measure trends in comorbidities and complications. We examined the prevalence of comorbidities and postoperative medical complications before and after the transition from ICD-9 to ICD-10 among patients who underwent primary total hip or knee arthroplasty (THA or TKA). We hypothesized that the transition to ICD-10 codes was associated with discontinuity and slope change in comorbidities and medical complications. Methods: The Elixhauser comorbidities and medical complications were identified using the Premier Healthcare database from fiscal year (FY)2011 to FY2018. Using multivariable segmented regression models, we examined the changes in the levels and slopes after the transition from ICD-9 to ICD-10 coding. Odds ratios (ORs) of <1 and >1 indicate decreases and increases, respectively, in levels and slopes. Results: Overall, 2,006,581 patients who underwent primary THA or TKA were identified. The mean age was 65.9 ± 10.5 years, and the median length of the hospital stay was 2 days (interquartile range [IQR], 2 to 3 days). Of the comorbidities studied, congestive heart failure, hypertension, and obesity had a statistically significant but clinically small discontinuity after the transition from ICD-9 to ICD-10 coding. Of the complications, pneumonia (OR = 0.66, 95% confidence interval [CI] = 0.48 to 0.90), acute respiratory failure (OR = 1.88, 95% CI = 1.52 to 2.33), sepsis (OR = 2.54, 95% CI = 1.45 to 4.44), and urinary tract infection (OR = 1.79, 95% CI = 1.32 to 2.42) demonstrated statistically significant discontinuity. Alcohol abuse and paralysis had an increasing prevalence before the ICD transition, followed by a decreasing prevalence after the transition. In contrast, metastatic cancer, weight loss, and acquired immunodeficiency syndrome (AIDS) showed a decreasing prevalence before the ICD transition followed by an increasing prevalence after the transition. Generally, complications showed a decreasing prevalence over time. Conclusions: The discontinuities after the transition from ICD-9 to ICD-10 coding were relatively small for most comorbidities. Medical complications generally showed a decreasing trend over the quarters studied. These findings support caution when conducting joint replacement studies that rely on ICD coding and include the ICD coding transition period.

Administrative Data Use in National Registry Efforts: Blessing or Curse?

“…These findings were likely just the result of a changing definition for a complication, but the study demonstrated that the transition from the ICD-9 to the ICD-10 introduced a discontinuity in the data that can result in an artificial change in the reported prevalence of a given complication. These findings also have been documented in relation to the prevalence of comorbidities in patients who undergo TJA due to changes in the coding system 19 . Changes to administrative claims coding are not limited to major transitions, such as the switch from the ICD-9 to the ICD-10, because there are often changes to the definitions for CPT codes or the introduction of new CPT codes, which occur more frequently 21,22 .…”

“…Following the transition to the ICD-10, each specific complication has a separate code to designate the joint and laterality of the complication (e.g., ICD-10 T84.033: Mechanical loosening of internal left knee prosthetic joint). Although this transition brought much-needed specificity in TJA documentation, research has demonstrated that the transition to the ICD-10 significantly altered the reported prevalence of common surgical complications following TKA and total hip arthroplasty (THA) 19,20 . These findings were likely just the result of a changing definition for a complication, but the study demonstrated that the transition from the ICD-9 to the ICD-10 introduced a discontinuity in the data that can result in an artificial change in the reported prevalence of a given complication.…”

“…Just as health care continues to change and evolve, so does the administrative claims coding process for health-care delivery 18 . It is inevitable that periodic "improvements" will be made to administrative claims coding, but it is important to recognize that changes to the coding process can introduce discontinuities into a data set that must be considered by researchers 19,20 . If this pitfall is not avoided, then findings may be inappropriately attributed to an incorrect study variable.…”

The Rise of Medicare Advantage

“…Data from 2004 to 2020 were used to describe trends in insurance type. Only data from 2015 to 2020 were used for statistical analysis given the transition from ICD-9 to ICD-10 that occurred during 2015, which led to differences in the reported rates of comorbidities and complications between these 2 ICD systems 10,11 .…”

No difference in short-term readmissions following day-case vs. one overnight stay in patients having hip and knee arthroplasty: a nationwide register study of 51,042 procedures from 2010–2020

“…Time from discharge to readmission was reported as the number of days from the discharge date and the date of the first readmission. Readmissions registered with a diagnosis code previously associated with arthroplasty complications were also registered as complications [12,13] (Table 2, see Appendix). Readmissions less than 4 hours after discharge from the primary surgical admission were not considered readmissions.…”