Predicting unplanned readmission due to cardiovascular disease in hospitalized patients with cancer: a machine learning approach.

Cardiovascular disease (CVD) in cancer patients can affect the risk of unplanned readmissions, which have been reported to be costly and associated with worse mortality and prognosis. We aimed to demonstrate the feasibility of using machine learning techniques in predicting the risk of unplanned 180-day readmission attributable to CVD among hospitalized cancer patients using the 2017-2018 Nationwide Readmissions Database. We included hospitalized cancer patients, and the outcome was unplanned hospital readmission due to any CVD within 180 days after discharge. CVD included atrial fibrillation, coronary artery disease, heart failure, stroke, peripheral artery disease, cardiomegaly, and cardiomyopathy. Decision tree (DT), random forest, extreme gradient boost (XGBoost), and AdaBoost were implemented. Accuracy, precision, recall, F2 score, and receiver operating characteristic curve (AUC) were used to assess the model's performance. Among 358,629 hospitalized patients with cancer, 5.86% (n = 21,021) experienced unplanned readmission due to any CVD. The three ensemble algorithms outperformed the DT, with the XGBoost displaying the best performance. We found length of stay, age, and cancer surgery were important predictors of CVD-related unplanned hospitalization in cancer patients. Machine learning models can predict the risk of unplanned readmission due to CVD among hospitalized cancer patients.

Prevalence of primary and secondary hypertension among hospitalized patients with cancer in the United States.

BACKGROUND: Hypertension is the most common comorbidity in patients with cancer. We aimed to estimate the prevalence of hypertension by demographic characteristics and cancer type among hospitalized patients with cancer. METHODS: Hospitalized cancer patients were included using 2016-2018 National Inpatient Sample data. The independent variable was the presence of hypertension, which was further classified as primary, secondary, and other hypertension. Patient characteristics were grouped by age, sex, race/ethnicity, and the 12 most common cancer types. Multinomial logistic regression was used. RESULTS: Among 638,670 hospitalized patients with cancer, 56.8% had hypertension. The predicted percentages of having any hypertension were higher with age, male gender, and black race. The predicted percentages of any hypertension were the highest in kidney cancer patients across all age and race/ethnicity groups. Uterine cancer was associated with the highest percentages of primary hypertension, followed by kidney cancer. Leukemia was associated with the highest percentages of secondary hypertension, followed by non-Hodgkin lymphoma. DISCUSSION: Kidney cancer patients had the highest predicted percentage of hypertension overall, while uterine cancer and leukemia had the highest percentages of primary and secondary hypertension, respectively. This study provides evidence for identifying cancer patients who need more attention for the prevention and management of hypertension.

Food Insecurity Among Medicare Beneficiaries With Type 2 Diabetes: Implications for Diabetes Care and Education Specialists.

PURPOSE: The purpose of the study was to examine factors associated with food insecurity among Medicare beneficiaries with type 2 diabetes. METHODS: The 2019 Medicare Current Beneficiary Survey Public Use File of beneficiaries ≥65 years old with type 2 diabetes (n = 1343) was analyzed. A binary variable was created to represent food insecurity (1 = food insecurity, 0 = without food insecurity) with ≥2 affirmative responses adapting an established algorithm of the United States Department of Agriculture food insecurity questionnaire. A survey-weighted logistic model was performed to examine factors (ie, sociodemographic characteristics, health status, and insurance coverage) associated with food insecurity. RESULTS: Approximately 11.6% of study Medicare beneficiaries with type 2 diabetes reported food insecurity. Non-Hispanic Black beneficiaries were more likely to report food insecurity than non-Hispanic White beneficiaries. Beneficiaries with incomes < $25 000 were more likely to report food insecurity than those with higher incomes. Beneficiaries enrolled in Medicare Advantage programs (vs traditional Medicare), having Medicare-Medicaid dual eligibility (vs nondual), and living with instrumental activities of daily living or activities of daily living limitations (vs without) were also more likely to report food insecurity than their respective counterparts. CONCLUSIONS: Sociodemographic disparities in food insecurity were observed among Medicare beneficiaries with type 2 diabetes. Implementation of screening protocols, interventions related to social determinants of health, and the diabetes care continuum may mitigate the prevalence of food insecurity in this demographic.

Cost-related medication nonadherence among Medicare beneficiaries with cardiovascular disease risk factors: The role of comprehension of the Medicare programme and its prescription drug benefits.

RATIONALE, AIMS, AND OBJECTIVES: This study aims to investigate how reported comprehension of the Medicare programme and its prescription drug benefits is associated with cost-related medication nonadherence (CRN) among Medicare beneficiaries with cardiovascular disease (CVD) risk factors. METHODS: This cross-sectional study used the 2017 Medicare Current Beneficiary Survey Public Use File data and included Medicare beneficiaries aged ≥65 years who reported having at least one CVD risk factor (i.e., hypertension, hyperlipidemia, diabetes, smoking and obesity) (n = 2821). A survey-weighted logistic model was used to examine associations between perceived difficulty of understanding the Medicare programme and its prescription drug benefits and CRN, controlling for beneficiaries' demographic (e.g., age) and clinical characteristics (e.g, comorbidities). This study further analyzed five subgroups based on the type of CVD risk factors involved. RESULTS: Among Medicare beneficiaries with CVD risk factors, 14.4% reported CRN. Medicare beneficiaries with CVD risk factors who reported difficulty understanding the overall Medicare programme and its prescription drug benefits were more likely to report CRN, compared to those who reported easy understanding of the overall Medicare programme (OR = 1.50; 95% CI = 1.11-2.04; p = 0.009) and its prescription drug benefits (OR = 2.01; 95% CI = 1.52-2.66; p < 0.001). Similar results were obtained for the subgroups with obesity, hypertension or hyperlipidemia. CONCLUSIONS: Perceived difficulty of understanding the Medicare Programme and its prescription drug benefits is associated with CRN among Medicare beneficiaries with CVD risk factors, especially those with obesity, hypertension or hyperlipidemia. Monitoring and enhancing Medicare beneficiaries' overall understanding of the Medicare programme may reduce CRN.

A systematic review of economic evaluations of tyrosine kinase inhibitors for non-small cell lung cancer (NSCLC).

INTRODUCTION: Although tyrosine kinase inhibitors (TKIs) have improved the efficacy of treatment for non-small cell lung cancer (NSCLC), the accessibility of TKIs is limited due to high costs. Despite the critical role of the cost-effectiveness of TKIs on decision-making, no systematic reviews have compared the cost-effectiveness of comparable TKIs. Therefore, we systemically reviewed the economic evaluation studies on various TKIs for NSCLC. AREAS COVERED: We searched PubMed and the Cochran Library to identify the published economic evaluation studies of TKIs in NSCLC patients that were published by January 2022. All of the included studies (n = 38) evaluated the cost-effectiveness of epidermal growth factor receptor (EGFR)-TKIs (n = 29) or anaplastic lymphocyte kinase (ALK)-TKIs (n = 9). The cost-effectiveness results were reported as the incremental cost-effectiveness ratio per quality-adjusted life-year, except for three studies. EXPERT OPINION: We found that the economic evaluation studies of the first and second generation of EGFR-TKIs and ALK-TKIs varied by the country and study settings, such as comparator and input parameters. In 12 studies, osimertinib (EGFR-TKI) was not cost-effective compared to other first/second EGFR-TKIs, regardless of the study settings. More evidence can be provided about cost-effectiveness of the third-generation TKIs in future research.

Prescription drug coverage satisfaction and medication nonadherence among Medicare beneficiaries with cancer.

BACKGROUND: Medication nonadherence among older patients with cancer can have profound health consequences. This study examines the association between prescription drug coverage satisfaction and medication nonadherence among Medicare beneficiaries with cancer. METHODS: We analyzed the 2017 Medicare Current Beneficiary Survey Public Use File of beneficiaries aged ≥65 years with reported non-skin cancer (n = 806). Beneficiaries were considered to have medication nonadherence if they reported: skipping doses, taking smaller doses than prescribed, or delaying or not filling a prescription because of cost. A survey-weighted logistic model, adjusted for covariates, was conducted to examine the association between prescription drug coverage satisfaction and medication nonadherence. RESULTS: Of study beneficiaries with cancer, 14.7% reported medication nonadherence. Higher proportions of beneficiaries with medication nonadherence were dissatisfied with the amount paid for medications (33.2% vs. 11.0%, p < 0.001) and the medications included on formulary (29.5% vs 5.2%, p < 0.001). In the adjusted analysis, the risk for medication nonadherence was higher among those who were dissatisfied with the amount paid for medications (OR = 2.22; p = 0.050) and the medications included on formulary (OR = 5.03; p = 0.005). CONCLUSIONS: Strategic mitigation of these barriers is essential to improving health outcomes in this at-risk population.

Health-related quality of life among elderly breast cancer patients treated with adjuvant endocrine therapy: a U.S Medicare population-based study.

OBJECTIVES: The evidence regarding the impact of individual adjuvant endocrine therapies (AET) on health-related quality of life (HRQoL) is limited. We aimed to assess the association between the type of AET and HRQoL and to examine the relationship between HRQoL and one-year mortality among women with breast cancer in the USA. METHODS: This retrospective cross-sectional study used the 2006-2017 Surveillance, Epidemiology, and End Results (SEER)-Medicare Health Outcomes Survey database to identify older women with early-stage hormone receptor-positive breast cancer. Multivariate linear regressions were used to assess the association between types of AET (anastrozole, letrozole, exemestane, and tamoxifen) and HRQoL scores (physical component summary (PCS) and mental component summary (MCS)). Multivariate logistic regressions were used to predict the impact of PCS and MCS on one-year mortality. RESULTS: Out of 3537 older women with breast cancer, anastrozole was the most commonly prescribed (n = 1945, 55.0%). Regarding PCS, there was no significant difference between the four AET agents. Higher MCS scores, which indicate better HRQoL, were reported in patients treated with anastrozole (vs. letrozole [ß = 1.26, p = 0.007] and exemestane [ß = 2.62, p = 0.005) and tamoxifen (vs. letrozole [ß = 1.49, p = 0.010] and exemestane [ß = 2.85, p = 0.004]). Lower PCS and MCS scores were associated with higher one-year mortality, regardless of type of AET initiated, except for tamoxifen in MCS. CONCLUSION: Although there was no significant difference in physical HRQoL scores between AET agents, anastrozole and tamoxifen were associated with better mental HRQoL scores.

Maladaptive Fall Risk Appraisal and Falling in Community-Dwelling Adults Aged 60 and Older: Implications for Screening.

OBJECTIVES: We aimed to examine the characteristics of maladaptive fall risk appraisal (FRA), discrepancies between physical and perceived-fall risk, and their associations with falls. METHODS: Fall risk appraisal was determined using the full-tandem stand test as an objective measure and the Fall Efficacy Scale-International as a subjective measure, and 433 adults aged ≥60 years from Thailand were classified into four groups: irrational (low physical/high perceived risk), incongruent (high physical/low perceived risk), congruent (high physical/high perceived risk) and rational (low physical/low perceived risk) FRAs. RESULTS: Only 20.8% of adults aged ≥60 years had rational FRA. The rest of the participants had either irrational (57.3%) or incongruent (2.3%), or congruent (19.6%) FRAs. Approximately 74% of those with congruent FRA reported experiencing a fall last year, followed by incongruent (60%), irrational (41.1%), and rational FRAs (27.8%, p < .001). After covariates adjustment, participants with congruent FRA were 3.06 times more likely of falling than those with rational FRA (p = .011). CONCLUSIONS: Maladaptive FRA is highly prevalent among adults aged ≥60. Identifying maladaptive FRA is essential for ensuring that adults aged ≥60 receiving early treatment associated with falls. CLINICAL IMPLICATIONS: Preventing a transition from rational to irrational, incongruent, and congruent fall risk appraisals is vital to prevent falls and mitigate this problematic health condition.

Hospital outcomes of male breast cancer in the United States.

OBJECTIVES: Hospital utilization and costs of female breast cancer have been well documented. However, evidence focusing on male breast cancer is scarce, despite the different clinical characteristics between female and male breast cancer. We aim to estimate hospital length of stay (LOS) and costs associated with male breast cancer in the United States. STUDY DESIGN: Retrospective observational study. METHODS: We analyzed the 2012-2016 Health Care Utilization Project National Inpatient Sample of 416 hospitalization events of male patients with breast cancer. Patients who had breast cancer diagnoses were selected based on the primary International Classification of Disease, Ninth Revision or Tenth Revision, Clinical Modification codes. A negative binomial regression and a generalized linear model with a gamma distribution and log-link function were conducted to estimate the LOS and hospital costs after controlling for sociodemographics, clinical characteristics (eg, metastatic status, Elixhauser Comorbidity Index [ECI] score), and hospital characteristics. RESULTS: On average, male patients with breast cancer stayed for 2.42 days and expensed $9059 per hospital visit. Patients with metastatic status had longer LOS (5.39 vs 3.24 days; P = .005) and higher hospital costs ($11,185 vs $8547; P = .03) than those without. Patients with an ECI score of 3 or more showed longer LOS (4.05 vs 2.68 days; P = .003) and higher hospital costs ($10,043 vs $7022; P < .001) than those with an ECI score of 0. CONCLUSIONS: LOS and hospital costs for male patients with breast cancer were associated with metastatic status and comorbidities. This information can be used to assess the health care resources needed to treat male breast cancer.

Iterative guided machine learning-assisted systematic literature reviews: a diabetes case study.

BACKGROUND: Systematic Reviews (SR), studies of studies, use a formal process to evaluate the quality of scientific literature and determine ensuing effectiveness from qualifying articles to establish consensus findings around a hypothesis. Their value is increasing as the conduct and publication of research and evaluation has expanded and the process of identifying key insights becomes more time consuming. Text analytics and machine learning (ML) techniques may help overcome this problem of scale while still maintaining the level of rigor expected of SRs. METHODS: In this article, we discuss an approach that uses existing examples of SRs to build and test a method for assisting the SR title and abstract pre-screening by reducing the initial pool of potential articles down to articles that meet inclusion criteria. Our approach differs from previous approaches to using ML as a SR tool in that it incorporates ML configurations guided by previously conducted SRs, and human confirmation on ML predictions of relevant articles during multiple iterative reviews on smaller tranches of citations. We applied the tailored method to a new SR review effort to validate performance. RESULTS: The case study test of the approach proved a sensitivity (recall) in finding relevant articles during down selection that may rival many traditional processes and show ability to overcome most type II errors. The study achieved a sensitivity of 99.5% (213 out of 214) of total relevant articles while only conducting a human review of 31% of total articles available for review. CONCLUSIONS: We believe this iterative method can help overcome bias in initial ML model training by having humans reinforce ML models with new and relevant information, and is an applied step towards transfer learning for ML in SR.

Cost-effectiveness of Interventions to Manage Diabetes: Has the Evidence Changed Since 2008?

OBJECTIVE: To synthesize updated evidence on the cost-effectiveness (CE) of interventions to manage diabetes, its complications, and comorbidities. RESEARCH DESIGN AND METHODS: We conducted a systematic literature review of studies from high-income countries evaluating the CE of diabetes management interventions recommended by the American Diabetes Association (ADA) and published in English between June 2008 and July 2017. We also incorporated studies from a previous CE review from the period 1985-2008. We classified the interventions based on their strength of evidence (strong, supportive, or uncertain) and levels of CE: cost-saving (more health benefit at a lower cost), very cost-effective (≤$25,000 per life year gained [LYG] or quality-adjusted life year [QALY]), cost-effective ($25,001-$50,000 per LYG or QALY), marginally cost-effective ($50,001-$100,000 per LYG or QALY), or not cost-effective (>$100,000 per LYG or QALY). Costs were measured in 2017 U.S. dollars. RESULTS: Seventy-three new studies met our inclusion criteria. These were combined with 49 studies from the previous review to yield 122 studies over the period 1985-2017. A large majority of the ADA-recommended interventions remain cost-effective. Specifically, we found strong evidence that the following ADA-recommended interventions are cost-saving or very cost-effective: In the cost-saving category are 1) ACE inhibitor (ACEI)/angiotensin receptor blocker (ARB) therapy for intensive hypertension management compared with standard hypertension management, 2) ACEI/ARB therapy to prevent chronic kidney disease and/or end-stage renal disease in people with albuminuria compared with no ACEI/ARB therapy, 3) comprehensive foot care and patient education to prevent and treat foot ulcers among those at moderate/high risk of developing foot ulcers, 4) telemedicine for diabetic retinopathy screening compared with office screening, and 5) bariatric surgery compared with no surgery for individuals with type 2 diabetes (T2D) and obesity (BMI ≥30 kg/m2). In the very cost-effective category are 1) intensive glycemic management (targeting A1C <7%) compared with conventional glycemic management (targeting an A1C level of 8-10%) for individuals with newly diagnosed T2D, 2) multicomponent interventions (involving behavior change/education and pharmacological therapy targeting hyperglycemia, hypertension, dyslipidemia, microalbuminuria, nephropathy/retinopathy, secondary prevention of cardiovascular disease with aspirin) compared with usual care, 3) statin therapy compared with no statin therapy for individuals with T2D and history of cardiovascular disease, 4) diabetes self-management education and support compared with usual care, 5) T2D screening every 3 years starting at age 45 years compared with no screening, 6) integrated, patient-centered care compared with usual care, 7) smoking cessation compared with no smoking cessation, 8) daily aspirin use as primary prevention for cardiovascular complications compared with usual care, 9) self-monitoring of blood glucose three times per day compared with once per day among those using insulin, 10) intensive glycemic management compared with conventional insulin therapy for T2D among adults aged ≥50 years, and 11) collaborative care for depression compared with usual care. CONCLUSIONS: Complementing professional treatment recommendations, our systematic review provides an updated understanding of the potential value of interventions to manage diabetes and its complications and can assist clinicians and payers in prioritizing interventions and health care resources.

Health Shocks and Initiation of Use of Preventive Services Among Older Adults.

This article examines whether adverse changes to health or functioning serve as an impetus to begin using preventive services among older individuals with a history of non-use. Using data from the 1998-2008 Health and Retirement Study, the use of mammograms, pap smears, prostate cancer screenings, cholesterol checks, and flu shots is examined among 2,975 self-reported non-users of such services. Older women who experience a health shock are 1.86, 1.50, 1.79, and 1.46 times more likely to begin getting mammograms, pap smears, cholesterol checks, and flu shots, respectively. Older men who experience a health shock are 2.24, 2.72, and 1.64 times more likely to begin getting prostate cancer screenings, cholesterol checks, and flu shots, respectively. All of these results are statistically significant. Thus, older adults often improve their health behaviors after experiencing an adverse health event.

Effects of Medicare Coverage of a "Welcome-to-Medicare" Visit on Use of Preventive Services among New Medicare Enrollees.

In January 2005, Medicare began covering a one-time initial preventive physical examination (IPPE), also called a "Welcome-to-Medicare" visit, during a beneficiary's first 6 months under Part B. This paper examines the effects of offering Medicare IPPE coverage on the use of mammograms, breast self-exams, Pap smears, prostate cancer screenings, cholesterol screenings, and flu vaccines among beneficiaries new to Part B. We adopt a difference-in-difference estimator and estimate a set of multivariate logit models to quantify the effects of introducing Medicare IPPE coverage on the use of preventive services. Models are estimated separately for men and women. Data for the analysis come from the 1996-2008 Health and Retirement Study. Among both men and women, having coverage for a one-time IPPE under Medicare had no effects on the utilization of any of the preventive services listed above. In this study, we find that offering coverage for a one-time IPPE under Medicare was insufficient to spur greater use of preventive services among new Medicare beneficiaries. These findings are important and suggest that policy-makers may need to consider other approaches to increase the use of recommended preventive services.