Medical Risk Scores Research Articles

Predicting Hospital Readmissions After Total Shoulder Arthroplasty Within a Bundled Payment Cohort.

Given the rising demand for shoulder arthroplasty, understanding risk factors associated with unplanned hospital readmission is imperative. No study to date has examined the influence of patient and hospital-specific factors as a predictive model for 90-day readmissions within a bundled payment cohort after primary shoulder arthroplasty. The purpose of this study was to determine predictive factors for 90-day readmissions after primary shoulder arthroplasty within a bundled payment cohort. After obtaining IRB approval, a retrospective review of a consecutive series of Medicare patients undergoing primary shoulder arthroplasty from 2014 to 2020 at a single academic institution was conducted. Patient demographic data, surgical variables, medical comorbidity profiles, medical risk scores, and social risk scores were collected. Postoperative variables included length of hospital stay, discharge location, and 90-day readmissions. Multivariate analysis was conducted to determine the independent risk factors of 90-day readmission. Overall, 3.6% of primary shoulder arthroplasty patients (127/3,523) were readmitted within 90 days. Readmitted patients had a longer hospital course (1.75 versus 1.45 P = 0.006), higher comorbidity profile (4.64 versus 4.24 P = 0.001), higher social risk score (7.96 versus 6.9 P = 0.008), and higher medical risk score (10.1 versus 6.96 P < 0.001) and were more likely to require a home health aide or be discharged to an inpatient rehab facility or skilled nursing facility ( P = 0.002). Following multivariate analysis, an elevated medical risk score was associated with an increased risk of readmission (odds ratio = 1.05, P < 0.001). This study demonstrates medical risk scores to be an independent risk factor of increased risk of 90-day hospital readmissions after primary shoulder arthroplasty within a bundled payment patient population. Additional incorporation of medical risk scores may be a beneficial adjunct in preoperative prediction for readmission and the potentially higher episode-of-care costs. Level III, retrospective cohort.

Predicting with Proxies: Transfer Learning in High Dimension

Predictive analytics is increasingly used to guide decision making in many applications. However, in practice, we often have limited data on the true predictive task of interest and must instead rely on more abundant data on a closely related proxy predictive task. For example, e-commerce platforms use abundant customer click data (proxy) to make product recommendations rather than the relatively sparse customer purchase data (true outcome of interest); alternatively, hospitals often rely on medical risk scores trained on a different patient population (proxy) rather than their own patient population (true cohort of interest) to assign interventions. Yet, not accounting for the bias in the proxy can lead to suboptimal decisions. Using real data sets, we find that this bias can often be captured by a sparse function of the features. Thus, we propose a novel two-step estimator that uses techniques from high-dimensional statistics to efficiently combine a large amount of proxy data and a small amount of true data. We prove upper bounds on the error of our proposed estimator and lower bounds on several heuristics used by data scientists; in particular, our proposed estimator can achieve the same accuracy with exponentially less true data (in the number of features d). Finally, we demonstrate the effectiveness of our approach on e-commerce and healthcare data sets; in both cases, we achieve significantly better predictive accuracy as well as managerial insights into the nature of the bias in the proxy data. This paper was accepted by George Shanthikumar, big data and analytics.

Data Mining Derived Insights into the Regional Character of Medical Risk Scores

Data Mining Derived Insights into the Regional Character of Medical Risk Scores

Comparison of Perception of Pregnancy Risk of Nulliparous Women of Advanced Maternal Age and Younger Age

Over the last 3 decades, the proportion of women who have delayed childbearing into their mid 30s and early 40s has been increasing. Because advanced maternal age (AMA) is associated with several adverse maternal, fetal, and neonatal outcomes, these pregnancies are considered to be "high risk." Research indicates that pregnancy risk perception is an important factor in pregnant women's health care use and decision making during pregnancy. The objectives of this study were to compare risk perception in pregnant women of AMA (aged 35 years or older) with that of younger women and to explore the relationship between perception of pregnancy risk and selected variables. A sample of 159 nulliparous pregnant women (105 aged 20-29 years and 54 aged 35 years or older) was recruited from a variety of settings in Winnipeg, Manitoba, Canada. Women were asked to complete questionnaires to assess perception of pregnancy risk, risk knowledge, pregnancy-related anxiety, perceived control, health status, and medical risk. Women of AMA had higher education levels, were more likely to work during pregnancy, and had higher medical risk scores than younger women. Women of AMA perceived higher pregnancy risk for both themselves and their fetuses than did younger women. They rated their risks of cesarean birth, dying during pregnancy, preterm birth, and having a newborn with a birth defect or one needing admission to a neonatal intensive care unit higher than those of younger women. There were no significant differences between the 2 age groups in pregnancy-related anxiety, knowledge of risk, perceived control, and health status. Women of AMA have a higher perception of pregnancy risk than younger women, regardless of their medical risk. This evidence suggests that incorporating discussions of pregnancy risk into prenatal care visits may assist pregnant women of AMA to make more informed choices, reduce anxiety, and avoid unnecessary interventions.

Correlations between early stress experiences and later behavior regulation in premature born toddlers with low and very low birth weight at the age of two years

Introduction: In a period of rapide brain development, preterm born children have to deal with a great amount of stressful and painful experiences after birth. Studies using animal models have demonstrated the association between adverse experiences and later behavioral problems. By now there is only few evidence showing a direct link between early stressful experiences of the premature baby and the well known developmental and behavioral problems in childhood. The present study examines the correlation between early stress experiences and later behavioral regulation in preterm born children at the age of two, controlling for medical risk and cognitive development. Methods: Data of N=80 preterm born children at the age of two years (birth weight >1500g: N=50; birth weight <1500g: N=30) were collected. The amount of stressful (e.g. stimulation after bradycardia) and painful (e.g. veine puncture) procedures and medical risk were assessed by analyzing charts of the hospital stay following birth. In a laboratory session of 90 minutes, four tasks of inhibitory control (adapted by Kochanska et al., 2000) and the Bayley II (mental scale, MDI), as a measure of cognitive development, were administered. Results and conclusion: Data collection is already finished; data analysis is in progress. It is hypothesized that a greater amount of early stress experiences is linked with lower abilities in behavioral regulation at the age of two years, even when controlling for medical risk and cognitive development. Concerning early stress experiences (qualified for duration of hospital stay), preliminary analysis (N=25) reveals a moderate positive relationship with the two medical risk scores (SNAP II PE: ρ=0.42, p=0.03 and NNBRS: ρ=0.43, p=0.02), as well as a marginally significant negative relationship with the MDI (ρ=–0.37, p=0.07). Results of the whole data set are discussed with regard to the heightened vulnerability of low birth weight preterm born children and the meaning of behavioral control as a basis of successful learning.

Predicting Motor Outcome at Preschool Age for Infants Tested at 7, 30, 60, and 90 Days After Term Age Using the Test of Infant Motor Performance

Accurate and diagnostic measures are central to early identification and intervention with infants who are at risk for developmental delays or disabilities. The purpose of this study was to examine (1) the ability of infants' Test of Infant Motor Performance (TIMP) scores at 7, 30, 60 and 90 days after term age to predict motor development at preschool age and (2) the contribution of the home environment and medical risk to the prediction. Sixty-one children from an original cohort of 90 infants who were assessed weekly with the TIMP, between 34 weeks gestational age and 4 months after term age, participated in this follow-up study. The Peabody Developmental Motor Scales, 2nd edition (PDMS-2), were administered to the children at the mean age of 57 months (SD=4.8 months). The quality and quantity of the home environment also were assessed at this age using the Early Childhood Home Observation for Measurement of the Environment (EC-HOME). Pearson product moment correlation coefficients, multiple regression, sensitivity and specificity, and positive and negative predictive values were used to assess the relationship among the TIMP, HOME, medical risk, and PDMS-2 scores. The correlation coefficients between the TIMP and PDMS-2 scores were statistically significant for all ages except at 7 days. The highest correlation coefficient was at 90 days (r=.69, P=.001). The TIMP scores at 30, 60, and 90 days after term; medical risk scores; and EC-HOME scores explained 24%, 23%, and 52% of the variance in the PDMS-2 scores, respectively. The TIMP score at 90 days after term was the most significant contributor to the prediction. The TIMP cutoff score of -0.5 standard deviation below the mean correctly classified 80%, 79%, and 87% of the children using a cutoff score of -2 standard deviations on the PDMS-2 at 30, 60, and 90 days, respectively. The results compare favorably with those of developmental tests administered to infants at 6 months of age or older. These findings underscore the need for age-specific test values and developmental surveillance of infants before making referrals.