Development of a mortality score to assess risk of adverse drug reactions among hospitalized patients with moderate to severe chronic kidney disease

External validation of a 2-year all-cause mortality prediction tool developed using machine learning in patients with stage 4-5 chronic kidney disease

BACKGROUND

Chronic kidney disease (CKD) is associated with increased mortality. Individual mortality prediction could be of interest to improve individual clinical outcomes. Using an independent regional dataset, the aim of the present study was to externally validate the recently published 2-year all-cause mortality prediction tool developed using machine learning.

METHODS

A validation dataset of stage 4 or 5 CKD outpatients was used. External validation performance of the prediction tool at the optimal cutoff-point was assessed by the area under the receiver operating characteristic curve (AUC-ROC), accuracy, sensitivity, and specificity. A survival analysis was then performed using the Kaplan-Meier method.

RESULTS

Data of 527 outpatients with stage 4 or 5 CKD were analyzed. During the 2 years of follow-up, 91 patients died and 436 survived. Compared to the learning dataset, patients in the validation dataset were significantly younger, and the ratio of deceased patients in the validation dataset was significantly lower. The performance of the prediction tool at the optimal cutoff-point was: AUC-ROC = 0.72, accuracy = 63.6%, sensitivity = 72.5%, and specificity = 61.7%. The survival curves of the predicted survived and the predicted deceased groups were significantly different (p < 0.001).

CONCLUSION

The 2-year all-cause mortality prediction tool for patients with stage 4 or 5 CKD showed satisfactory discriminatory capacity with emphasis on sensitivity. The proposed prediction tool appears to be of clinical interest for further development.

A Random Forest Algorithm for Assessing Risk Factors Associated With Chronic Kidney Disease: Observational Study

BACKGROUND

The prevalence and mortality rate of chronic kidney disease (CKD) are increasing year by year, and it has become a global public health issue. The economic burden caused by CKD is increasing at a rate of 1% per year. CKD is highly prevalent and its treatment cost is high but unfortunately remains unknown. Therefore, early detection and intervention are vital means to mitigate the treatment burden on patients and decrease disease progression.

OBJECTIVE

In this study, we investigated the advantages of using the random forest (RF) algorithm for assessing risk factors associated with CKD.

METHODS

We included 40,686 people with complete screening records who underwent screening between January 1, 2015, and December 22, 2020, in Jing'an District, Shanghai, China. We grouped the participants into those with and those without CKD by staging based on the glomerular filtration rate staging and grouping based on albuminuria. Using a logistic regression model, we determined the relationship between CKD and risk factors. The RF machine learning algorithm was used to score the predictive variables and rank them based on their importance to construct a prediction model.

RESULTS

The logistic regression model revealed that gender, older age, obesity, abnormal index estimated glomerular filtration rate, retirement status, and participation in urban employee medical insurance were significantly associated with the risk of CKD. On RF algorithm-based screening, the top 4 factors influencing CKD were age, albuminuria, working status, and urinary albumin-creatinine ratio. The RF model predicted an area under the receiver operating characteristic curve of 93.15%.

CONCLUSIONS

Our findings reveal that the RF algorithm has significant predictive value for assessing risk factors associated with CKD and allows the screening of individuals with risk factors. This has crucial implications for early intervention and prevention of CKD.

Development of Screening Tools to Predict Medication-Related Problems Across the Continuum of Emergency Department Care: A Prospective, Multicenter Study

Background: Medication-related problems (MRPs) occur across the continuum of emergency department (ED) care: they may contribute to ED presentation, occur in the ED/short-stay unit (SSU), at hospital admission, or shortly after discharge to the community. This project aimed to determine predictors for MRPs across the continuum of ED care and incorporate these into screening tools (one for use at ED presentation and one at ED/SSU discharge), to identify patients at greatest risk, who could be targeted by ED pharmacists.

Methods: A prospective, observational, multicenter study was undertaken in nine EDs, between July 2016 and August 2017. Blocks of ten consecutive adult patients presenting at pre-specified times were identified. Within 1 week of ED discharge, a pharmacist interviewed patients and undertook a medical record review to determine a medication history, patient understanding of treatment, risk factors for MRPs and to manage the MRPs. Logistic regression was undertaken to determine predictor variables. Multivariable regression beta coefficients were used to develop a scoring system for the two screening tools.

Results: Of 1,238 patients meeting all inclusion criteria, 904 were recruited. Characteristics predicting MRPs related to ED presentation were: patient self-administers regular medications (OR = 7.95, 95%CI = 3.79–16.65), carer assists with medication administration (OR = 15.46, 95%CI = 6.52–36.67), or health-professional administers (OR = 5.01, 95%CI = 1.77–14.19); medication-related ED presentation (OR = 9.95, 95%CI = 4.92–20.10); age ≥80 years (OR = 3.63, 95%CI = 1.96–6.71), or age 65–79 years (OR = 2.01, 95%CI = 1.17–3.46); potential medication adherence issue (OR = 2.27, 95%CI = 1.38–3.73); medical specialist seen in past 6-months (OR = 2.02, 95%CI = 1.42–2.85); pharmaceutical benefit/pension/concession cardholder (OR = 1.89, 95%CI = 1.28–2.78); inpatient in previous 4-weeks (OR = 1.60, 95%CI = 1.02–2.52); being male (OR = 1.48, 95%CI = 1.05–2.10); and difficulties reading labels (OR = 0.63, 95%CI = 0.40–0.99). Characteristics predicting MRPs related to ED discharge were: potential medication adherence issue (OR = 6.80, 95%CI = 3.97–11.64); stay in ED > 8 h (OR = 3.23, 95%CI = 1.47–7.78); difficulties reading labels (OR = 2.33, 95%CI = 1.30–4.16); and medication regimen changed in ED (OR = 3.91, 95%CI = 2.43–6.30). For ED presentation, the model had a C-statistic of 0.84 (95% CI 0.81–0.86) (sensitivity = 80%, specificity = 70%). For ED discharge, the model had a C-statistic of 0.78 (95% CI 0.73–0.83) (sensitivity = 82%, specificity = 57%).

Conclusion: Predictors of MRPs are readily available at the bedside and may be used to screen for patients at greatest risk upon ED presentation and upon ED/SSU discharge to the community. These screening tools now require external validation and implementation studies to evaluate the impact of using such tools on patient care outcomes.

Drug-Related Problems in Hospitalised Patients with Chronic Kidney Disease: A Systematic Review

INTRODUCTION

Globally, chronic kidney disease (CKD) is one of the leading causes of mortality. Impaired renal function makes CKD patients vulnerable to drug-related problems (DRPs).

AIM

The aim of this systematic review was to investigate the prevalence and nature of DRPs among hospital in-patients with CKD.

METHODS

A systematic review of the literature was conducted using Medline, EMBASE, PsycINFO, Web of Science (Core Collection), CINAHL plus (EBSCO), Cochrane Library (Wiley), Scopus (ELSEVIER) and PubMed (U.S.NLM) from index inception to January 2020. Studies investigating DRPs in hospitalised CKD patients published in the English language were included. Two independent reviewers extracted the data and undertook quality assessment using the Joanna Briggs Institute (JBI) tool.

RESULTS

A total of 2895 unique titles were identified; with 20 meeting the inclusion criteria. DRPs prevalence in CKD was reported between 12 and 87%. The most common DRPs included ineffective treatment, inappropriate drug choice and dosing problems. Antibiotics, H2-antihistamines and oral antidiabetics (metformin) were common drug classes involved in DRPs. Factors associated with DRPs included severity of CKD, the number of medications taken, age, length of hospital stay, and gender.

CONCLUSION

This systematic review provides evidence that DRPs are a frequent occurrence and burden for hospitalised patients with stage 1-4 CKD. Heterogeneity in study design, case detection and definitions are common, and future studies should use clearer definitions and study designs. Protocol Registration: PROSPERO: CRD42018096364.

Prediction of mortality among patients with chronic kidney disease: A systematic review

BACKGROUND

Chronic kidney disease (CKD) is a common medical condition that is increasing in prevalence. Existing published evidence has revealed through regression analyses that several clinical characteristics are associated with mortality in CKD patients. However, the predictive accuracies of these risk factors for mortality have not been clearly demonstrated.

AIM

To demonstrate the accuracy of mortality predictive factors in CKD patients by utilizing the area under the receiver operating characteristic (ROC) curve (AUC) analysis.

METHODS

We searched Ovid MEDLINE, EMBASE, and the Cochrane Library for eligible articles through January 2021. Studies were included based on the following criteria: (1) Study nature was observational or conference abstract; (2) Study populations involved patients with non-transplant CKD at any CKD stage severity; and (3) Predictive factors for mortality were presented with AUC analysis and its associated 95% confidence interval (CI). AUC of 0.70-0.79 is considered acceptable, 0.80-0.89 is considered excellent, and more than 0.90 is considered outstanding.

RESULTS

Of 1759 citations, a total of 18 studies (n = 14579) were included in this systematic review. Eight hundred thirty two patients had non-dialysis CKD, and 13747 patients had dialysis-dependent CKD (2160 patients on hemodialysis, 370 patients on peritoneal dialysis, and 11217 patients on non-differentiated dialysis modality). Of 24 mortality predictive factors, none were deemed outstanding for mortality prediction. A total of seven predictive factors [N-terminal pro-brain natriuretic peptide (NT-proBNP), BNP, soluble urokinase plasminogen activator receptor (suPAR), augmentation index, left atrial reservoir strain, C-reactive protein, and systolic pulmonary artery pressure] were identified as excellent. Seventeen predictive factors were in the acceptable range, which we classified into the following subgroups: predictors for the non-dialysis population, echocardiographic factors, comorbidities, and miscellaneous.

CONCLUSION

Several factors were found to predict mortality in CKD patients. Echocardiography is an important tool for mortality prognostication in CKD patients by evaluating left atrial reservoir strain, systolic pulmonary artery pressure, diastolic function, and left ventricular mass index.