A generalized covariate-adjusted top-scoring pair algorithm with applications to diabetic kidney disease stage classification in the Chronic Renal Insufficiency Cohort (CRIC) Study.

BACKGROUND: The growing amount of high dimensional biomolecular data has spawned new statistical and computational models for risk prediction and disease classification. Yet, many of these methods do not yield biologically interpretable models, despite offering high classification accuracy. An exception, the top-scoring pair (TSP) algorithm derives parameter-free, biologically interpretable single pair decision rules that are accurate and robust in disease classification. However, standard TSP methods do not accommodate covariates that could heavily influence feature selection for the top-scoring pair. Herein, we propose a covariate-adjusted TSP method, which uses residuals from a regression of features on the covariates for identifying top scoring pairs. We conduct simulations and a data application to investigate our method, and compare it to existing classifiers, LASSO and random forests. RESULTS: Our simulations found that features that were highly correlated with clinical variables had high likelihood of being selected as top scoring pairs in the standard TSP setting. However, through residualization, our covariate-adjusted TSP was able to identify new top scoring pairs, that were largely uncorrelated with clinical variables. In the data application, using patients with diabetes (n = 977) selected for metabolomic profiling in the Chronic Renal Insufficiency Cohort (CRIC) study, the standard TSP algorithm identified (valine-betaine, dimethyl-arg) as the top-scoring metabolite pair for classifying diabetic kidney disease (DKD) severity, whereas the covariate-adjusted TSP method identified the pair (pipazethate, octaethylene glycol) as top-scoring. Valine-betaine and dimethyl-arg had, respectively, ≥ 0.4 absolute correlation with urine albumin and serum creatinine, known prognosticators of DKD. Thus without covariate-adjustment the top-scoring pair largely reflected known markers of disease severity, whereas covariate-adjusted TSP uncovered features liberated from confounding, and identified independent prognostic markers of DKD severity. Furthermore, TSP-based methods achieved competitive classification accuracy in DKD to LASSO and random forests, while providing more parsimonious models. CONCLUSIONS: We extended TSP-based methods to account for covariates, via a simple, easy to implement residualizing process. Our covariate-adjusted TSP method identified metabolite features, uncorrelated from clinical covariates, that discriminate DKD severity stage based on the relative ordering between two features, and thus provide insights into future studies on the order reversals in early vs advanced disease states.

Adverse Safety Event Characteristics and Predictive Factors in Hospital Encounters for Patients with Chronic Kidney Disease.

BACKGROUND: Adverse safety events (ASE) during hospitalization may contribute to renal decline or poor outcomes. Understanding factors contributing to ASE in chronic kidney disease (CKD) is limited. The objective is to compare differences and determine predictors of renal pertinent ASE in discharges for CKD. METHOD: A cross-sectional analysis of the National Inpatient Sample, Healthcare Cost and Utilization Project, Agency for Healthcare Research and Quality 2012 data. The study included adults age ≥18 years with discharge diagnosis for CKD stages 1-4, excluding cancer of the kidney and renal pelvis, renal transplant, end-stage renal disease. Predictors included study sample characteristics, including patient demographics, comorbidity, and hospitalization-related variables. Outcomes assessed included distribution of ASE (angioedema, confusion, muscle weakness or cramps, lower extremity edema (LEE), falls, hypoglycemia, nausea-vomiting-diarrhea (NVD), and skin rash), mean total charge per hospital event, and length-of-stay. The analytical approach used descriptive statistics (means and proportions) and bivariate analysis to compare differences (ASE versus none). Predictors of ASE were explored using multivariate logistic regression. RESULTS: 10.3% of inpatient discharges for CKD showed an ASE. Mean charges (USD 48,072 vs. 46,996), days length-of-stay (6.8 vs. 5.7), number of diagnosis on record (6.8 vs. 5.7), geographical region (Midwest, and West), and type of hospital (rural) were significantly associated with ASE. Most common ASEs were confusion (18%), LEE (21.3%), and NVD (50.7%). Odds of ASE increased for age, female gender, rural hospitals, geographical region, and diagnosis for anemia, coagulopathies, depression, fluid and electrolyte disorders, neurological disorders, psychoses, and weight loss. CONCLUSIONS: We identified key factors that increase the risk of ASE in patients with CKD. Opportunities exist to reduce ASE in CKD.

Telemedicine to Promote Patient Safety: Use of Phone-Based Interactive Voice-Response System to Reduce Adverse Safety Events in Pre-dialysis CKD.

CKD patients have several features conferring on them a high risk of adverse safety events, which are defined as incidents with unintended harm related to processes of care or medications. These characteristics include impaired kidney function, polypharmacy, and frequent health system encounters. The consequences of such events in CKD can include new or prolonged hospitalization, accelerated kidney function loss, acute kidney injury, ESRD, and death. Health information technology administered via telemedicine presents opportunities for CKD patients to remotely communicate safety-related findings to providers for the purpose of improving their care. However, many CKD patients have limitations that hinder their use of telemedicine and access to the broad capabilities of health information technology. In this review, we summarize previous assessments of the pre-dialysis CKD populations' proficiency in using telemedicine modalities and describe the use of interactive voice-response system to gauge the safety phenotype of the CKD patient. We discuss the potential for expanded interactive voice-response system use in CKD to address the safety threats inherent to this population.

Patterns of NSAIDs Use and Their Association with Other Analgesic Use in CKD.

BACKGROUND AND OBJECTIVES: Avoiding nonsteroidal anti-inflammatory drugs is important for safe CKD care. This study examined nonsteroidal anti-inflammatory drug use patterns and their association with other analgesic use in CKD. DESIGN, SETTING, PARTICIPANTS, & MEASUREMENTS: The Chronic Renal Insufficiency Cohort Study is an observational cohort study that enrolled 3939 adults ages 21-74 years old with CKD between 2003 and 2008 using age-based eGFR inclusion criteria. Annual visits between June of 2003 and December of 2011 were organized into 15,917 visit-pairs (with an antecedent and subsequent visit) for 3872 participants with medication information. Demographics, kidney function, and clinical factors were ascertained along with report of nonsteroidal anti-inflammatory drug or other analgesic use in the prior 30 days. RESULTS: In our study, 24% of participants reported nonsteroidal anti-inflammatory drug use at baseline or at least one follow-up study visit. Having a 10 ml/min per 1.73 m2 higher eGFR level at an antecedent visit was associated with higher odds of starting nonsteroidal anti-inflammatory drugs at a subsequent visit (odds ratio, 1.44; 95% confidence interval, 1.34 to 1.56). Seeing a nephrologist at the antecedent visit was associated with lower odds of starting or staying on nonsteroidal anti-inflammatory drugs at a subsequent visit (odds ratio, 0.70; 95% confidence interval, 0.56 to 0.87 and odds ratio, 0.61; 95% confidence interval, 0.46 to 0.81, respectively). Starting and stopping nonsteroidal anti-inflammatory drugs were both associated with higher odds of increasing the number of other analgesics (odds ratio, 1.52; 95% confidence interval, 1.25 to 1.85 and odds ratio, 1.78; 95% confidence interval, 1.39 to 2.28, respectively) and higher odds of increasing the number of opioid analgesics specifically (odds ratio, 1.92; 95% confidence interval, 1.48 to 2.48 and odds ratio, 1.46; 95% confidence interval, 1.04 to 2.03, respectively). CONCLUSIONS: Nonsteroidal anti-inflammatory drug use is common among patients with CKD but less so among those with worse kidney function or those who see a nephrologist. Initiation or discontinuation of nonsteroidal anti-inflammatory drugs is often associated with supplementation with or replacement by, respectively, other analgesics, including opioids, which introduces possible drug-related problems when taking these alternative analgesics.

Nephrotoxic Medication Exposure in U.S. Adults with Predialysis Chronic Kidney Disease: Health Services Utilization and Cost Outcomes.

BACKGROUND: Nephrotoxic medication exposure increases risks for acute kidney injury, permanent renal function loss, and costly preventable adverse drug events. Exposure to medications associated with inducing acute tubular nephritis or tubular toxicity versus nonexposure among those with predialysis renal disease-a population vulnerable to increased risk of kidney injury-may affect health services utilization and cost outcomes. Few studies quantify nephrotoxic medication exposure in chronic kidney disease (CKD) and associated costs. OBJECTIVE: To examine exposure to medications associated with inducing acute tubular nephritis or tubular toxicity versus nonexposure and the effect on health services utilization and cost outcomes in a nationally representative sample of adults with predialysis CKD. METHODS: This retrospective study used Medical Expenditure Panel Survey (MEPS) household component longitudinal files (years 2006-2012; panels 11-16). Participants included 809 MEPS respondents aged > 18 years with predialysis CKD, after excluding those participants with cancer, kidney stone, renal dialysis, or transplant procedures (approximately 14.7 million U.S. noninstitutionalized individuals). Two groups were created to evaluate the main measures: (1) participants prescribed 1 or more medications associated with risk of acute tubular nephritis and/or tubular toxicity (termed "nephrotoxic exposure") and (2) participants with nonexposure. Medications cited in published literature as associated with tubular kidney damage were used. Multivariable regression models assessed the pattern of nephrotoxic medication exposure and its effect on health services utilization and expenses. RESULTS: Nephrotoxic medication exposure occurred in 72% of adult MEPS respondents. Of those, 47.2% and 52.8% were prescribed 1 and at least 2 nephrotoxic medications, respectively. Coexistent chronic conditions included hypertension (72.3%), diabetes (49.5%), coronary heart disease (33%), arthritis (23.6%), and chronic obstructive pulmonary disease (17.6%). Eligible MEPS respondents aged ≥ 65 years, from the U.S. South region, and with Charlson Comorbidity Index (CCI) score > 0 were 75% (vs. aged 18-45 years), 83% (vs. Northeast), and 72%-96% (vs. CCI = 0) more likely to be exposed to nephrotoxic medications. Uninsured participants showed 55% less likelihood of nephrotoxic exposure, compared with privately insured participants. Higher utilization was shown in the nephrotoxic medication exposure group (vs. nonexposure): prescription fills (52.8 vs. 26.8, P < 0.001), emergency department visits (56.2 vs. 29.3 per 1,000 patient months, P < 0.001), and hospitalization (51.8 vs. 23.4 per 1,000 patient months, P < 0.001). Unadjusted all-cause expenses were greater for the following categories: medical ($119,935 vs. $11,462, P < 0.001), prescription drug ($4,828 vs. $2,816, P < 0.001), and total health expenses ($24,663 vs. $14,277, P < 0.001). Adjusted all-cause expenses were greater for total (29.7% greater, P = 0.003), prescription medications (56.6% greater, P < 0.001), and medical (23.4% greater, P = 0.036), but there were no differences in predialysis CKD-related utilization and expenses. CONCLUSIONS: Increased vigilance is needed when prescribing nephrotoxic medications in predialysis CKD, particularly in patients with comorbid conditions and the elderly. Nephrotoxic medication exposure in predialysis CKD has the potential for increased health services utilization and cost outcomes. DISCLOSURES: There was no grant or intramural funding for this research. The authors have no conflicts of interest, financial or otherwise, to disclose. Study concept and design were primarily contributed by Davis-Ajami, along with Fink and Wu. Davis-Ajami took the lead in data collection, along with Wu, and data interpretation was performed by David-Ajami, Wu, and Fink. All authors participated in manuscript preparation and revision.