A simple risk predictor may identify patients at higher risk for an unplanned readmission within 30 days of transcatheter aortic valve replacement (TAVR), a new study suggests.
The Centers for Medicare & Medicaid Services (CMS) “has their own all-cause readmission tools, hospitals have their own EPIC readmission risk scores, but I don’t really think we use them specifically for TAVI patients right now,” lead study author, Sahil Khera, MD, MPH, Massachusetts General Hospital, Harvard Medical School, Boston, told theheart.org | Medscape Cardiology. “They’re generic readmission tools.”
“But this one I would want to use for my patients to understand their risk of coming back and how intensive my transition of care should be,” he said. “It’s something geared towards TAVI patients.”
The investigators recently reported that 15% to 20% of patients are readmitted within 30 days of TAVR, driven by noncardiac causes in more than 60% of cases. Though several risk models and apps are available to assess TAVR in-hospital mortality risk, this is the first tool designed to predict these readmissions.
To fill the data gap, investigators examined data on 39,305 patients in the Nationwide Readmissions Database (NRD) who underwent TAVR between January 2013 and August 2015. Of these, 6380 patients (16.2%) were readmitted within 30 days, Khera reported at the Society for Cardiovascular Angiography and Interventions (SCAI) 2018 Scientific Sessions.
A nomogram was developed based on parameters identified through univariable and logistic regression analyses, with model calibration performed with bootstrapping. Internal validation was done with K-fold cross-validation.
The final risk model includes nine variables scored on a 350-point scale: anemia, atrial fibrillation, chronic liver disease, chronic lung disease, chronic kidney disease, end-stage renal disease on dialysis, length of stay 5 days or more, acute kidney injury, and discharge disposition.
“The beauty of this tool is, because we made it using administrative datasets, it’s very easy to use,” Khera said. “You can pull out any chart from your patient’s record and will find these variables. You don’t need to do any complex GFR [glomerular filtration rate] calculations, any STS [Society of Thoracic Surgeons] calculations. So case managers, administrators, nurses, anyone can calculate this.”
For example, an 82-year-old woman with chronic obstructive pulmonary disease, atrial fibrillation, and chronic kidney disease who developed acute kidney injury after TAVR and required 6 days of hospitalization would receive 182 points and thus have more than a 25% risk for 30-day readmission.
A score of more than 212 points predicts a greater than 30% risk for 30-day readmission (C-statistic, 0.63).
The C-statistic mirrors those for the acute myocardial infarction, heart failure, and pneumonia measures CMS currently uses (0.61 – 0.63), noted Khera. “All the administrative datasets usually have C-statistics that are on the lower end because they lack more granular data,” he said. “It is also possible that social determinants that are not routinely captured in administrative datasets influence readmissions.”
Indeed, the investigators acknowledge they lacked data on STS score, drug therapy, echocardiographic variables, valve type, and paravalvular leak. In addition, they intentionally excluded vascular and bleeding complications from the model because the vast number of ICD-9 codes for these events would have made it difficult for users to easily incorporate, Khera said.
On the other hand, he observed that the model includes chronic liver disease, which is an important predictor of readmission that falls out of many surgical scores, including the STS cardiac surgery risk score.
“Clearly the C-statistic is not very strong but it is strong enough to help us make an association, to give us a signal that this patient is sick and may require additional attention,” Khera said.
Commenting for theheart.org | Medscape Cardiology, Chandan Devireddy, MD, from Emory University School of Medicine, Atlanta, Georgia, who was not involved in the study, said, “I think having a procedural score such as this is a good start; I don’t know that this is a finished product.”
He noted that bleeding complications can have a large impact on outcomes and said using other databases, such as the STS/American College of Cardiology TVT registry, may be useful to refine the model.
“Nevertheless, including periprocedural and perihospitalization data does serve a purpose in helping to stratify patients for additional resources they may need,” Devireddy said. “But I would ask the question, ‘Is the use of the scale best-suited to identify patients before, during the procedure, or both?’ and maybe it’s both.”
He observed that today’s TAVR patients are vastly different from the high-risk candidates screened in the early days of TAVR. “But you can still have a patient who is extraordinarily healthy and for whatever reason experiences challenges with the procedure that takes them on a very different trajectory than what they may expect postprocedure,” he added. “In that case, this would be very helpful.”
Khera said they are working on validating the algorithm in their institution and creating an Apple app but stressed it should not be used to deny a patient TAVR and can only be calculated after the procedure, when discharge planning starts.
“If the case manager is doing the discharge planning and knows the patient has a higher risk score, perhaps they will want to talk to the primary care provider or the nursing care facility, or provide home nursing care, or use telemonitoring,” he said. “What we want to promote is better transition of care.”
“In the end, we want to give these patients a better quality of life. Because whenever you have a patient that’s coming in again and again, you’re taking away the quality of life.”
Khera has disclosed no relevant financial relationships. Devireddy reported serving on a scientific advisory board for Medtronic and receiving travel and logistical funding for meetings from Edwards Lifesciences.
Follow Patrice Wendling on Twitter: @pwendl. For more from theheart.org, follow us on Twitter and Facebook.
Tidak ada komentar:
Posting Komentar