OARSI: Model helps predict readmission after joint replacement

SEATTLE – Adding clinical factors to a commonly used risk model improves by about 20% the prediction of which patients will be readmitted after undergoing total joint replacement surgery, new data show.

The new, combined model employs diagnostic codes and a variety of clinical factors, such as preoperative patient-reported measures – emotional status assessed with the mental component score of the 12-item Short Form Health Survey, moderate to severe pain in other weight-bearing joints (the contralateral joint, the hips or knees, and low back), and history of smoking – and the Charlson comorbidity index.

Investigators led by Dr. Patricia D. Franklin, a professor in the department of orthopedics and physical rehabilitation, University of Massachusetts, Worcester, set out to improve on the Centers for Medicare & Medicaid (CMS) risk prediction model for 30-day readmission. Rates of readmission using the model are publicly reported as an indicator of the quality of care at hospitals.

“Analyses are described as risk adjusted, implying that the remaining outcome variation is due to variation in quality of care,” she told attendees of the World Congress on Osteoarthritis. “The CMS risk adjustment model, while expertly developed, was limited to the billing and diagnosis data that were available to them and for patients over 65 years of age.”

The researchers tested the model using data from FORCE-TJR (Function and Outcomes Research for Comparative Effectiveness in Total Joint Replacement and Quality Improvement), the first national total joint replacement registry. Analyses were based on 2,560 patients aged 65 years or older who underwent total knee or hip replacement during 2011 and 2012.

Overall, 4.7% of the patients were readmitted within 30 days. About one-third of the readmissions were for implant-related issues, while the rest were due to complications of a major surgery in older patients with comorbidities, according to Dr. Franklin, who disclosed that she has received investigator-initiated research awards from Zimmer and Biomet.

In multivariate analysis, discrimination of patients having a readmission with the combined model was superior to that with the model based on diagnostic codes alone (c-statistic, 0.75 vs. 0.64), she reported at the meeting sponsored by the Osteoarthritis Research Society International.

“We believe that in the combined risk-adjustment model … the key predictors were indeed older age and gender and that list of ICD codes. However, the addition of medical and musculoskeletal comorbidities – in particular, moderate and severe pain in other weight-bearing joints – was important to the success,” Dr. Franklin commented, noting that the findings will be validated using 2013 registry data.

“We have been advising total joint replacement surgeons to collect these musculoskeletal factors in a systematic way so that they can be included in these future analyses. We believe it would be wise to integrate clinician-defined and -reported risk factors before public reporting,” she concluded.

Session attendee Dr. Jeffrey N. Katz, a professor at Harvard Medical School and codirector of the Brigham Spine Center at Brigham and Women’s Hospital, Boston, wondered whether pain in the other knee is a proxy for the severity and systemic nature of the osteoarthritis in these patients.

Identifying the real risk factors at play will be important, Dr. Katz said, adding, “I think we see this a lot in risk models. ... and addressing the proxy may not address the underlying issue.”

“I’m just very afraid of the law of unintended consequences” of risk models potentially being used to deny some patients surgery, commented session comoderator Dr. Nigel Arden, professor of rheumatology at the University of Oxford, England.

“A lot of the risk factors you identified were not obviously reversible,” Dr. Arden commented. “So are these people going to get prehab, or would you operate but do lots of extensive postoperative rehab? Or is this going to be a patient decision aid?”

The main aims of the research were to better ensure that the model was truly reflecting surgical care and to provide surgeons with information for planning and counseling patients, according to Dr. Franklin.

“I think patients should be informed what their risk is. … Average readmission rates or mortality rates don’t inform individuals,” she added. “So our goal actually is to begin to have some strategies to parse out subgroups of patients at greater or lesser risk for complications, and for benefits from the surgery.”

SEATTLE – Adding clinical factors to a commonly used risk model improves by about 20% the prediction of which patients will be readmitted after undergoing total joint replacement surgery, new data show.

The new, combined model employs diagnostic codes and a variety of clinical factors, such as preoperative patient-reported measures – emotional status assessed with the mental component score of the 12-item Short Form Health Survey, moderate to severe pain in other weight-bearing joints (the contralateral joint, the hips or knees, and low back), and history of smoking – and the Charlson comorbidity index.

Investigators led by Dr. Patricia D. Franklin, a professor in the department of orthopedics and physical rehabilitation, University of Massachusetts, Worcester, set out to improve on the Centers for Medicare & Medicaid (CMS) risk prediction model for 30-day readmission. Rates of readmission using the model are publicly reported as an indicator of the quality of care at hospitals.

“Analyses are described as risk adjusted, implying that the remaining outcome variation is due to variation in quality of care,” she told attendees of the World Congress on Osteoarthritis. “The CMS risk adjustment model, while expertly developed, was limited to the billing and diagnosis data that were available to them and for patients over 65 years of age.”

The researchers tested the model using data from FORCE-TJR (Function and Outcomes Research for Comparative Effectiveness in Total Joint Replacement and Quality Improvement), the first national total joint replacement registry. Analyses were based on 2,560 patients aged 65 years or older who underwent total knee or hip replacement during 2011 and 2012.

Overall, 4.7% of the patients were readmitted within 30 days. About one-third of the readmissions were for implant-related issues, while the rest were due to complications of a major surgery in older patients with comorbidities, according to Dr. Franklin, who disclosed that she has received investigator-initiated research awards from Zimmer and Biomet.

In multivariate analysis, discrimination of patients having a readmission with the combined model was superior to that with the model based on diagnostic codes alone (c-statistic, 0.75 vs. 0.64), she reported at the meeting sponsored by the Osteoarthritis Research Society International.

“We believe that in the combined risk-adjustment model … the key predictors were indeed older age and gender and that list of ICD codes. However, the addition of medical and musculoskeletal comorbidities – in particular, moderate and severe pain in other weight-bearing joints – was important to the success,” Dr. Franklin commented, noting that the findings will be validated using 2013 registry data.

“We have been advising total joint replacement surgeons to collect these musculoskeletal factors in a systematic way so that they can be included in these future analyses. We believe it would be wise to integrate clinician-defined and -reported risk factors before public reporting,” she concluded.

Session attendee Dr. Jeffrey N. Katz, a professor at Harvard Medical School and codirector of the Brigham Spine Center at Brigham and Women’s Hospital, Boston, wondered whether pain in the other knee is a proxy for the severity and systemic nature of the osteoarthritis in these patients.

Identifying the real risk factors at play will be important, Dr. Katz said, adding, “I think we see this a lot in risk models. ... and addressing the proxy may not address the underlying issue.”

“I’m just very afraid of the law of unintended consequences” of risk models potentially being used to deny some patients surgery, commented session comoderator Dr. Nigel Arden, professor of rheumatology at the University of Oxford, England.

“A lot of the risk factors you identified were not obviously reversible,” Dr. Arden commented. “So are these people going to get prehab, or would you operate but do lots of extensive postoperative rehab? Or is this going to be a patient decision aid?”

The main aims of the research were to better ensure that the model was truly reflecting surgical care and to provide surgeons with information for planning and counseling patients, according to Dr. Franklin.

“I think patients should be informed what their risk is. … Average readmission rates or mortality rates don’t inform individuals,” she added. “So our goal actually is to begin to have some strategies to parse out subgroups of patients at greater or lesser risk for complications, and for benefits from the surgery.”

SEATTLE – Adding clinical factors to a commonly used risk model improves by about 20% the prediction of which patients will be readmitted after undergoing total joint replacement surgery, new data show.

The new, combined model employs diagnostic codes and a variety of clinical factors, such as preoperative patient-reported measures – emotional status assessed with the mental component score of the 12-item Short Form Health Survey, moderate to severe pain in other weight-bearing joints (the contralateral joint, the hips or knees, and low back), and history of smoking – and the Charlson comorbidity index.

Investigators led by Dr. Patricia D. Franklin, a professor in the department of orthopedics and physical rehabilitation, University of Massachusetts, Worcester, set out to improve on the Centers for Medicare & Medicaid (CMS) risk prediction model for 30-day readmission. Rates of readmission using the model are publicly reported as an indicator of the quality of care at hospitals.

“Analyses are described as risk adjusted, implying that the remaining outcome variation is due to variation in quality of care,” she told attendees of the World Congress on Osteoarthritis. “The CMS risk adjustment model, while expertly developed, was limited to the billing and diagnosis data that were available to them and for patients over 65 years of age.”

The researchers tested the model using data from FORCE-TJR (Function and Outcomes Research for Comparative Effectiveness in Total Joint Replacement and Quality Improvement), the first national total joint replacement registry. Analyses were based on 2,560 patients aged 65 years or older who underwent total knee or hip replacement during 2011 and 2012.

Overall, 4.7% of the patients were readmitted within 30 days. About one-third of the readmissions were for implant-related issues, while the rest were due to complications of a major surgery in older patients with comorbidities, according to Dr. Franklin, who disclosed that she has received investigator-initiated research awards from Zimmer and Biomet.

In multivariate analysis, discrimination of patients having a readmission with the combined model was superior to that with the model based on diagnostic codes alone (c-statistic, 0.75 vs. 0.64), she reported at the meeting sponsored by the Osteoarthritis Research Society International.

“We believe that in the combined risk-adjustment model … the key predictors were indeed older age and gender and that list of ICD codes. However, the addition of medical and musculoskeletal comorbidities – in particular, moderate and severe pain in other weight-bearing joints – was important to the success,” Dr. Franklin commented, noting that the findings will be validated using 2013 registry data.

“We have been advising total joint replacement surgeons to collect these musculoskeletal factors in a systematic way so that they can be included in these future analyses. We believe it would be wise to integrate clinician-defined and -reported risk factors before public reporting,” she concluded.

Session attendee Dr. Jeffrey N. Katz, a professor at Harvard Medical School and codirector of the Brigham Spine Center at Brigham and Women’s Hospital, Boston, wondered whether pain in the other knee is a proxy for the severity and systemic nature of the osteoarthritis in these patients.

Identifying the real risk factors at play will be important, Dr. Katz said, adding, “I think we see this a lot in risk models. ... and addressing the proxy may not address the underlying issue.”

“I’m just very afraid of the law of unintended consequences” of risk models potentially being used to deny some patients surgery, commented session comoderator Dr. Nigel Arden, professor of rheumatology at the University of Oxford, England.

“A lot of the risk factors you identified were not obviously reversible,” Dr. Arden commented. “So are these people going to get prehab, or would you operate but do lots of extensive postoperative rehab? Or is this going to be a patient decision aid?”

The main aims of the research were to better ensure that the model was truly reflecting surgical care and to provide surgeons with information for planning and counseling patients, according to Dr. Franklin.

“I think patients should be informed what their risk is. … Average readmission rates or mortality rates don’t inform individuals,” she added. “So our goal actually is to begin to have some strategies to parse out subgroups of patients at greater or lesser risk for complications, and for benefits from the surgery.”