Medicare bundled payments may not be effective for medical conditions
No association was found between hospital participation in five medical bundles included in Medicare's Bundled Payments for Care Improvement (BPCI) initiative and significant changes in several outcomes, including Medicare payments and length of stay, according to a recent study.
Researchers used data from Medicare claims from 2013 through 2015 to identify hospital admissions for congestive heart failure (CHF), pneumonia, chronic obstructive pulmonary disease (COPD), sepsis, and acute myocardial infarction (MI), the five most common BPCI medical conditions. Difference-in-difference analyses were used to examine potential changes in standardized Medicare payments for these conditions per episode of care, defined as the period of hospitalization plus 90 days postdischarge, at 492 BPCI hospitals and 898 matched control hospitals. The results were published July 19 by the New England Journal of Medicine.
One hundred twenty-five hospitals participated in BPCI for CHF, 105 participated for pneumonia, 101 participated for COPD, 88 participated for sepsis, and 73 participated for acute MI, with 62.6% of hospitals participating in more than one care bundle. The average Medicare payment per episode of care at baseline for all five conditions was $24,280 at BPCI hospitals and $23,901 at control hospitals. During the intervention period, these numbers decreased to $23,993 (difference, −286; P=0.41) and $23,503 (difference, −398; P=0.08), respectively. The difference in differences between the two hospital groups was $112 (P=0.79). Changes from baseline in clinical complexity, length of stay, ED use or readmission within 30 or 90 days postdischarge, or death within 30 or 90 days postadmission were also similar between intervention and control hospitals. The findings remained largely the same when the sample was limited to hospitals that did not drop out of BPCI.
The researchers noted that their findings differ from those of a previous study of BPCI for total joint replacement, which found an association between the program and lower overall Medicare payments. This could be because patients receiving joint replacement choose to undergo the procedure and are generally younger than those receiving hospital care for medical conditions, among other possible explanations, they wrote. They pointed out that because BPCI is a voluntary program and they looked at only five conditions, their results might not be generalizable to other settings. In addition, data on pricing, savings, and losses from the program were not available and follow-up time was limited, among other limitations.
Based on their findings, the researchers concluded that the five BPCI medical bundles included in their study did not appear to be associated with changes from baseline in total Medicare payments per episode, case complexity, length of stay, ED use, readmission, or mortality rates. “Bundling of services to encourage more efficient care has great face validity and enjoys bipartisan support,” the authors wrote. “For such bundling to work for medical conditions, however, more time, new care strategies and partnerships, or additional incentives may be required.”
Score may predict Pneumocystis pneumonia in ICU patients with hematologic malignancies and acute respiratory failure
A score based on a multivariable risk prediction model may help predict Pneumocystis pneumonia (PCP) in ICU patients with hematologic malignancies and acute respiratory failure, a recent study found.
Researchers in France examined a prospective multicenter cohort of ICU patients to identify factors associated with PCP (pneumonia caused by Pneumocystis jirovecii). The risk prediction model based on these factors was then tested in an independent prospective multicenter cohort. The study results were published July 11 by the American Journal of Respiratory and Critical Care Medicine.
Overall, the analysis included 1,330 patients, 1,092 in the derivation cohort and 238 in the validation cohort. Prevalence of PCP was 12.3% in the former and 6.3% in the latter, and time from ICU admission to PCP diagnosis was three days in both cohorts. Variables included in the model were age, lymphoproliferative disease, anti-Pneumocystis prophylaxis, number of days between onset of respiratory symptoms and ICU admission, shock, chest radiograph pattern, and pleural effusion. The median scores in the derivation and validation cohorts were 3.5 (range, −3.5 to 8.5) and 1.0 (range, −3.5 to 6.0), respectively.
Higher scores, and therefore higher risk, were associated with lymphoproliferative disease, no receipt of anti-Pneumocystis prophylaxis, over three days between onset of respiratory symptoms and ICU admission, and no alveolar patter on radiography, while lower scores were associated with age over 50 years, shock, and pleural effusion. Specificity for PCP was 88%, and the negative predictive value was 97%. Calibration and discrimination were good (area under the curve, 0.80 in the derivation cohort and 0.83 in the validation cohort).
The researchers noted that all of the included patients were admitted to the ICU for acute respiratory failure and that they did not examine how the score would perform in other patients and other settings, among additional limitations. They concluded that their score performed well in this patient population, however, and said that using it to predict PCP risk would probably decrease delays in appropriate therapy, adding that all of the variables included in the model are easily available at the bedside. “Studies are warranted to further assess the performance of the model and to determine whether its use improves the early diagnosis of [PCP], increases patient survival, and spares ICU resources,” the authors wrote.