Surgeons tackle readmission risk reduction

INDIANAPOLIS – Average 30-day readmission rates in a large national study varied widely by surgical specialty, ranging from 5% for general surgery patients to 12% for vascular surgery patients and 16% after hepato-pancreatic-biliary surgery.

This retrospective study was based upon American College of Surgeons National Surgical Quality Improvement Program (NSQIP) 2011 data on 240,125 patients discharged from 316 hospitals after these types of surgery. The results enabled investigators to identify the major risk factors for readmission. They used this information to generate a predictive model to identify patients at greatest risk with an eye toward introducing interventions to keep them out of the hospital, Dr. Timothy M. Pawlik said at the annual meeting of the American Surgical Association.

How effective such interventions will be in a surgical population is as yet unclear, he added.

The importance of this work lies in the growing emphasis health care payers are placing upon 30-day readmission as a quality-of-care indicator. In 2012, Medicare began cutting reimbursement by 1% to hospitals with above-average 30-day readmissions. Next year, this penalty is scheduled to increase to 3% under the Medicare Hospital Readmission Reduction Program. Prior research efforts to identify risk factors for readmission have focused chiefly on medical rather than surgical conditions, noted Dr. Pawlik, professor of surgery and chief of the division of surgical oncology at Johns Hopkins University, Baltimore.

The readmission risk predictive formula he and his coinvestigators developed is simple: American Society of Anesthesiologists physical status class + (length of stay/2). This number is rounded up to yield a risk score of 1-10. A patient with a readmission score of 1 has a 1% risk of readmission within 30 days of discharge. The risk climbs to 12% with a score of 5, 20% with a score of 8, and 40% with a score of 10.

"A score of 4 had an 8% readmission rate, with 77% sensitivity, 52% specificity, a low positive predictive value of 12%, but a high negative predictive value of 95%," the surgeon observed. "In essence, a score of 4 or higher was able to identify 80% of all readmissions, but it also included about half of all patients."

Because of a quirk in the NSQIP database – it records readmissions within 30 days of surgery instead of starting from discharge – the investigators had to limit the study population to surgery patients with a length of stay of 10 days or less and then apply statistical modeling. However, limiting the data set to patients with a maximum 10-day hospital stay only restricted the study cohort by 6%. Thus, the study results remain highly generalizable to U.S. surgery patients.

The area under the curve (AUC) of the receiver operating characteristic for the readmission risk formula was 0.70. Statisticians consider a test having an AUC of 0.50 to be worthless. An AUC of 1.0 would define a "perfect" test, while a test with an AUC in the 0.70-0.80 range is deemed of only "fair" accuracy.

The AUC for the readmission score varied considerably when applied to the various surgical subspecialties. For example, the AUC was 0.69 for general surgery patients but only 0.51 for thoracic surgery patients, 0.64 for vascular surgery patients, and 0.59 for colorectal surgery patients.

The fact that the predictive formula doesn’t perform any better than barely "fair" is testimony to the difficulty in identifying who will require readmission. That being said, the AUC for the readmission score after surgery compares favorably to published hospital readmission risk formulas developed for medical patients, which have similar and in many cases lower AUCs (JAMA 2011;306:1688-98), Dr. Pawlik observed.

At Johns Hopkins, the plan is to target surgery patients who have high readmission scores with interventions including more frequent follow-up phone calls and earlier scheduled postoperative clinic visits in an effort to keep them out of the hospital, he continued.

Discussant Dr. Keith D. Lillemoe was skeptical that this will result in reduced readmission rates. Indeed, he questioned whether the 30-day readmission rate is a legitimate quality measure for surgeons.

"I can look across the room and predict if a patient is going to come back in to the hospital. But what can we really do to keep such patients from being readmitted other than keeping them indeterminately long until we’ve passed that window? It seems like all the phone calls and early clinic visits in the world can’t stop the progression sometimes," reflected Dr. Lillemoe, professor and chairman of the department of surgery at Massachusetts General Hospital, Boston.

He noted with frustration that he had just stepped out of the lecture hall to arrange for readmission of a patient 1 week after discharge post pancreaticoduodenectomy.

"I made three phone calls myself to that guy trying to nurse him through his nausea and vomiting and other symptoms," the surgeon recalled.

Dr. Pawlik was sympathetic.

"It’s very hard to prevent readmissions. It’s a very complicated metric, and many would argue that it’s an inappropriate quality metric," he said. "Many things are outside our control before we even meet the patient, as are some things that occur in the hospital, and many things are beyond our control after discharge as far as where they live, their family structure, and their financial resources. In my opinion it’s a very problematic quality measure and I am not sure how we are going to tackle it."

He reported having no conflicts of interest.

[email protected]

INDIANAPOLIS – Average 30-day readmission rates in a large national study varied widely by surgical specialty, ranging from 5% for general surgery patients to 12% for vascular surgery patients and 16% after hepato-pancreatic-biliary surgery.

This retrospective study was based upon American College of Surgeons National Surgical Quality Improvement Program (NSQIP) 2011 data on 240,125 patients discharged from 316 hospitals after these types of surgery. The results enabled investigators to identify the major risk factors for readmission. They used this information to generate a predictive model to identify patients at greatest risk with an eye toward introducing interventions to keep them out of the hospital, Dr. Timothy M. Pawlik said at the annual meeting of the American Surgical Association.

How effective such interventions will be in a surgical population is as yet unclear, he added.

The importance of this work lies in the growing emphasis health care payers are placing upon 30-day readmission as a quality-of-care indicator. In 2012, Medicare began cutting reimbursement by 1% to hospitals with above-average 30-day readmissions. Next year, this penalty is scheduled to increase to 3% under the Medicare Hospital Readmission Reduction Program. Prior research efforts to identify risk factors for readmission have focused chiefly on medical rather than surgical conditions, noted Dr. Pawlik, professor of surgery and chief of the division of surgical oncology at Johns Hopkins University, Baltimore.

The readmission risk predictive formula he and his coinvestigators developed is simple: American Society of Anesthesiologists physical status class + (length of stay/2). This number is rounded up to yield a risk score of 1-10. A patient with a readmission score of 1 has a 1% risk of readmission within 30 days of discharge. The risk climbs to 12% with a score of 5, 20% with a score of 8, and 40% with a score of 10.

"A score of 4 had an 8% readmission rate, with 77% sensitivity, 52% specificity, a low positive predictive value of 12%, but a high negative predictive value of 95%," the surgeon observed. "In essence, a score of 4 or higher was able to identify 80% of all readmissions, but it also included about half of all patients."

Because of a quirk in the NSQIP database – it records readmissions within 30 days of surgery instead of starting from discharge – the investigators had to limit the study population to surgery patients with a length of stay of 10 days or less and then apply statistical modeling. However, limiting the data set to patients with a maximum 10-day hospital stay only restricted the study cohort by 6%. Thus, the study results remain highly generalizable to U.S. surgery patients.

The area under the curve (AUC) of the receiver operating characteristic for the readmission risk formula was 0.70. Statisticians consider a test having an AUC of 0.50 to be worthless. An AUC of 1.0 would define a "perfect" test, while a test with an AUC in the 0.70-0.80 range is deemed of only "fair" accuracy.

The AUC for the readmission score varied considerably when applied to the various surgical subspecialties. For example, the AUC was 0.69 for general surgery patients but only 0.51 for thoracic surgery patients, 0.64 for vascular surgery patients, and 0.59 for colorectal surgery patients.

The fact that the predictive formula doesn’t perform any better than barely "fair" is testimony to the difficulty in identifying who will require readmission. That being said, the AUC for the readmission score after surgery compares favorably to published hospital readmission risk formulas developed for medical patients, which have similar and in many cases lower AUCs (JAMA 2011;306:1688-98), Dr. Pawlik observed.

At Johns Hopkins, the plan is to target surgery patients who have high readmission scores with interventions including more frequent follow-up phone calls and earlier scheduled postoperative clinic visits in an effort to keep them out of the hospital, he continued.

Discussant Dr. Keith D. Lillemoe was skeptical that this will result in reduced readmission rates. Indeed, he questioned whether the 30-day readmission rate is a legitimate quality measure for surgeons.

"I can look across the room and predict if a patient is going to come back in to the hospital. But what can we really do to keep such patients from being readmitted other than keeping them indeterminately long until we’ve passed that window? It seems like all the phone calls and early clinic visits in the world can’t stop the progression sometimes," reflected Dr. Lillemoe, professor and chairman of the department of surgery at Massachusetts General Hospital, Boston.

He noted with frustration that he had just stepped out of the lecture hall to arrange for readmission of a patient 1 week after discharge post pancreaticoduodenectomy.

"I made three phone calls myself to that guy trying to nurse him through his nausea and vomiting and other symptoms," the surgeon recalled.

Dr. Pawlik was sympathetic.

"It’s very hard to prevent readmissions. It’s a very complicated metric, and many would argue that it’s an inappropriate quality metric," he said. "Many things are outside our control before we even meet the patient, as are some things that occur in the hospital, and many things are beyond our control after discharge as far as where they live, their family structure, and their financial resources. In my opinion it’s a very problematic quality measure and I am not sure how we are going to tackle it."

He reported having no conflicts of interest.

[email protected]

INDIANAPOLIS – Average 30-day readmission rates in a large national study varied widely by surgical specialty, ranging from 5% for general surgery patients to 12% for vascular surgery patients and 16% after hepato-pancreatic-biliary surgery.

This retrospective study was based upon American College of Surgeons National Surgical Quality Improvement Program (NSQIP) 2011 data on 240,125 patients discharged from 316 hospitals after these types of surgery. The results enabled investigators to identify the major risk factors for readmission. They used this information to generate a predictive model to identify patients at greatest risk with an eye toward introducing interventions to keep them out of the hospital, Dr. Timothy M. Pawlik said at the annual meeting of the American Surgical Association.

How effective such interventions will be in a surgical population is as yet unclear, he added.

The importance of this work lies in the growing emphasis health care payers are placing upon 30-day readmission as a quality-of-care indicator. In 2012, Medicare began cutting reimbursement by 1% to hospitals with above-average 30-day readmissions. Next year, this penalty is scheduled to increase to 3% under the Medicare Hospital Readmission Reduction Program. Prior research efforts to identify risk factors for readmission have focused chiefly on medical rather than surgical conditions, noted Dr. Pawlik, professor of surgery and chief of the division of surgical oncology at Johns Hopkins University, Baltimore.

The readmission risk predictive formula he and his coinvestigators developed is simple: American Society of Anesthesiologists physical status class + (length of stay/2). This number is rounded up to yield a risk score of 1-10. A patient with a readmission score of 1 has a 1% risk of readmission within 30 days of discharge. The risk climbs to 12% with a score of 5, 20% with a score of 8, and 40% with a score of 10.

"A score of 4 had an 8% readmission rate, with 77% sensitivity, 52% specificity, a low positive predictive value of 12%, but a high negative predictive value of 95%," the surgeon observed. "In essence, a score of 4 or higher was able to identify 80% of all readmissions, but it also included about half of all patients."

Because of a quirk in the NSQIP database – it records readmissions within 30 days of surgery instead of starting from discharge – the investigators had to limit the study population to surgery patients with a length of stay of 10 days or less and then apply statistical modeling. However, limiting the data set to patients with a maximum 10-day hospital stay only restricted the study cohort by 6%. Thus, the study results remain highly generalizable to U.S. surgery patients.

The area under the curve (AUC) of the receiver operating characteristic for the readmission risk formula was 0.70. Statisticians consider a test having an AUC of 0.50 to be worthless. An AUC of 1.0 would define a "perfect" test, while a test with an AUC in the 0.70-0.80 range is deemed of only "fair" accuracy.

The AUC for the readmission score varied considerably when applied to the various surgical subspecialties. For example, the AUC was 0.69 for general surgery patients but only 0.51 for thoracic surgery patients, 0.64 for vascular surgery patients, and 0.59 for colorectal surgery patients.

The fact that the predictive formula doesn’t perform any better than barely "fair" is testimony to the difficulty in identifying who will require readmission. That being said, the AUC for the readmission score after surgery compares favorably to published hospital readmission risk formulas developed for medical patients, which have similar and in many cases lower AUCs (JAMA 2011;306:1688-98), Dr. Pawlik observed.

At Johns Hopkins, the plan is to target surgery patients who have high readmission scores with interventions including more frequent follow-up phone calls and earlier scheduled postoperative clinic visits in an effort to keep them out of the hospital, he continued.

Discussant Dr. Keith D. Lillemoe was skeptical that this will result in reduced readmission rates. Indeed, he questioned whether the 30-day readmission rate is a legitimate quality measure for surgeons.

"I can look across the room and predict if a patient is going to come back in to the hospital. But what can we really do to keep such patients from being readmitted other than keeping them indeterminately long until we’ve passed that window? It seems like all the phone calls and early clinic visits in the world can’t stop the progression sometimes," reflected Dr. Lillemoe, professor and chairman of the department of surgery at Massachusetts General Hospital, Boston.

He noted with frustration that he had just stepped out of the lecture hall to arrange for readmission of a patient 1 week after discharge post pancreaticoduodenectomy.

"I made three phone calls myself to that guy trying to nurse him through his nausea and vomiting and other symptoms," the surgeon recalled.

Dr. Pawlik was sympathetic.

"It’s very hard to prevent readmissions. It’s a very complicated metric, and many would argue that it’s an inappropriate quality metric," he said. "Many things are outside our control before we even meet the patient, as are some things that occur in the hospital, and many things are beyond our control after discharge as far as where they live, their family structure, and their financial resources. In my opinion it’s a very problematic quality measure and I am not sure how we are going to tackle it."

He reported having no conflicts of interest.

[email protected]