AAOS Now

Published 1/1/2014
|
Thomas K. Fehring, MD; David C. Ayers, MD; Patricia D. Franklin, MD, MPH, MBA; the FORCE-TJR investigators

Improving 30-day Readmission Models for TJR

FORCE-TJR and AAHKS collaboration results in addition of clinical risk factors

The latest findings from a collaboration between the American Association of Hip & Knee Surgeons (AAHKS) and FORCE-TJR (Function and Outcomes Research for Comparative Effectiveness in Total Joint Replacement) were presented at the 2013 AAHKS Annual Meeting in Dallas. The AAHKS and FORCE-TJR had been working together to advocate for the addition of clinical data to the risk adjustment models for total knee replacement (TKR) and total hip replacement (THR) used by the Centers for Medicare & Medicaid Services (CMS).

These models were developed by CMS to enable the public reporting of readmissions and complications after total joint replacement on Medicare’s Hospital Compare website. AAHKS and FORCE-TJR leaders met with CMS to advocate for the addition of clinical data to administrative data, and presented their work illustrating that the addition of clinical risk factors (from the FORCE-TJR database) to the CMS model improves the risk adjustment model significantly.

This improvement in the CMS model would enable a fairer public reporting process that would benefit patients, surgeons, and hospitals. CMS and the AAHKS have agreed to work together on improved models containing clinical risk factors.

Public reporting
In 2009, the National Quality Forum (NQF) published complication rates following primary THR and TKR surgeries. Hospital complication rates varied from 2.2 percent to 8.9 percent, with a median rate of 4.2 percent. The NQF concluded that “variability is a signal of differences in quality of care received.”

In September 2012, CMS issued preliminary hospital-specific reports of “risk-adjusted” 30-day readmission and total complication rates after TJR. This year, CMS will publicly disseminate hospital-specific TJR outcome reports and will compare risk-adjusted 30- and 90-day outcomes.

Although the steps taken by the CMS are important, the risk-adjustment models used raise two major concerns. First, the models are based solely on administrative billing data and comorbidity codes based on the International Classification of Diseases, 9th Edition (ICD-9) and lack surgeon- and patient-reported health information. Surgeon-reported clinical risk factors are central to outcome measurement systems for high-volume surgical procedures, as noted in the Society for Thoracic Surgeons’ (STS) coronary artery bypass graft (CABG) outcome reports. The second limitation is that CMS reports are based solely on patients 65 years or older. However, more than 40 percent of all TJR surgeries in the United States are performed in patients younger than age 65. Therefore, the CMS models would be unable to assess the key risks in this younger, rapidly growing cohort, and the public data would reflect care for only half of the TJR patients in the hospitals surveyed.

This raised the concern that risk factors, such as patients with severe joint deformity, would not be captured, and outcomes might appear poor due to risks that are not included in the models. If outcome measures do not fully adjust for all patient-risk factors, patients with the greatest risks may be at a disadvantage because surgeons would be hesitant to operate on them. Finally, clinically refined risk-adjustment models would be needed to fairly compare hospitals, surgeons, and patient outcomes.

Improving the model
The AAHKS and FORCE-TJR collaborated on analyses to determine whether the addition of clinically refined risk measures to the existing CMS model would improve the risk-adjustment process. FORCE-TJR is a federally funded, national cohort of 126 surgeons in 22 states, enrolling all TJR patients in a comprehensive outcomes registry. Data include patient-reported outcomes, clinical measures, and implant status.

Clinical measures (from the FORCE-TJR database) proven to be associated with post-TJR readmission were combined with the 2011 CMS administrative data. Among the clinical measures added were the following:

  • smoking status
  • body mass index (BMI)
  • musculoskeletal comorbidities such as pain in other joints
  • patient-reported function prior to TJR surgery

In early analyses, the combination of clinical and administrative data significantly improved the risk prediction models. For example, the readmission rate increased as BMI increased; this fact might be missed in the CMS model because the administrative data only record BMIs greater than 40. In addition, the rate of readmission within 30 days was 3.7 times higher for current smokers compared to nonsmokers.

As the 2012 CMS data become available, FORCE-TJR will repeat and further validate these analyses. After validation of the key clinical risk factors, work with CMS will continue to move toward a more refined risk-adjustment methodology that results in fair outcome comparisons.

Lessons from CABG
The STS cardiac outcome system has monitored risk-adjusted mortality for decades. Although CABG models predict mortality rates and CMS models for TJR outcomes focus on 30-day readmission and complication rates, lessons from the CABG outcome system have been extremely valuable.

The CABG model uses both administrative and clinician-reported data. Statisticians found that the addition of two clinical measures (actual ejection fraction or creatinine level) improved the risk-adjustment model beyond administrative data, and the combination of both achieved the full predictive power of the model.

Next steps
To summarize, the addition of surgeon- and patient-reported preoperative risk factors to the CMS model significantly improved the 30-day post-TJR readmission model. FORCE-TJR will repeat these analyses with 2012 data to validate the model. In addition, future collaborations with the research committees of The Hip Society, The Knee Society, and AAOS will determine whether other clinical measures—such as joint-specific physical exam measures or radiographic assessments of joint alignment and deformity—will further improve risk models.

As patients, referring physicians, surgeons, and policy makers move toward using risk-adjusted post-TJR outcome data to improve patient care, it is critical to develop brief, yet comprehensive, risk-adjustment models to discriminate between variations in outcomes due to suboptimal quality of care and those due to pre-existing patient risk factors.

Thomas K. Fehring, MD, is the president of the AAHKS and co-director of the Hip and Knee Center at OrthoCarolina; David C. Ayers, MD, is professor of orthopedics and physical rehabilitation at the University of Massachusetts Medical School; Patricia D. Franklin, MD, MPH, MBA, is a professor of orthopedics and physical rehabilitation at the University of Massachusetts Medical School.

Disclosure information: Dr. Fehring—DePuy, a Johnson & Johnson Company; AAHKS; Hip Society; Knee Society. Dr. Ayers—Journal of Bone & Joint Surgery; American Orthopaedic Association. Dr. Franklin— Wolters Kluwer Health - Lippincott Williams & Wilkins.