Use of Japanese big data from electronic medical records to investigate risk factors and identify their high-risk combinations for linezolid-induced thrombocytopenia.

PURPOSE: Thrombocytopenia is a major event associated with linezolid (LZD) therapy. Factors affecting LZD-induced thrombocytopenia (LIT) have been reported in previous studies. However, several issues pertaining to LIT have not yet been clarified. In the present study, we used Japanese big data to investigate associated factors and their high-risk combinations that influence LIT. METHODS: Patients administered LZD between May 2006 and October 2020 were included in this study. LIT was defined as either a 30% or more reduction from the baseline platelets or platelet values of < 100,000/µL. We evaluated factors affecting LIT and combinations of factors that alter LIT risk according to a decision tree (DT) analysis, a typical machine learning method. RESULTS: We successfully enrolled 1399 patients and LIT occurred in 44.7% of the patients (n = 626). We classified the laboratory data on renal function, LZD duration, age, and body weight (BW) into smaller categories. The results of multivariate analysis showed that prolonged LZD therapy, BW < 45 kg, estimated glomerular filtration rate (eGFR) < 30 mL/min/1.73 m2, and dialysis were risk factors for LIT. The DT analysis revealed that the highest risk was a combination of LZD duration ≥ 14 days and eGFR < 30 mL/min/1.73 m2. CONCLUSIONS: The present study extracted four risk factors and identified high-risk combinations for LIT. Patients with these risk factors should be closely monitored.

Risk Factor Analysis of Vancomycin-Induced Nephrotoxicity in Paediatric Patients Aged 0-1 Year Using Japanese Medical Database.

Vancomycin (VCM)-induced nephrotoxicity (VIN) is a major side effect in paediatric patients. However, most studies are limited to patients aged 0-18 years. We evaluated the risk factors of VIN in patients aged 0-1 year using Japanese electronic medical record database. We used RWD database which was contained electronic medical records and claims data of approximately 20 million people from 160 medical institutions. We targeted hospitalized patients who were administered VCM between June 2000 and December 2020. VIN was defined by two criteria: Criterion 1 was an increase in serum creatinine (Scr) ≥ 0.5 mg/dL or 50% during VCM treatment period compared to the Scr baseline; and criterion 2 was an increase in Scr ≥50% within seven days or Scr ≥0.3 mg/dL within two days during VCM treatment. The risk factors of VIN were evaluated using multivariate logistic regression analysis. We analysed 446 patients; patients with VIN in Criteria 1 and 2 were 33 and 58, respectively. In Criterion 1, multivariate logistic regression analysis identified four independent factors with p-value <0.05 (VCM concentration ≥20 mg/L, amphotericin B (AMPH-B), piperacillin-tazobactam (TAZ/PIPC), and vasopressor drugs). In Criterion 2, multivariate logistic regression analysis identified concomitant use of vasopressor drugs with p-value <0.05. Therefore, concomitant use of vasopressor drugs was suggested to affect the risk of VIN in patients aged 0-1 year. The findings may help in developing estimation models to assess the risk of VIN in paediatric patients.

Factors affecting creatine phosphokinase elevation during daptomycin therapy using a combination of machine learning and conventional methods.

AIMS: Musculoskeletal toxicity is a typical side effect of daptomycin (DAP). However, the risk factors have not been well established. Here, we aimed to identify independent factors affecting DAP-induced musculoskeletal toxicity using a combination of machine learning and conventional statistical methods. METHODS: A population-based, retrospective, observational cohort study was conducted using the Japanese electronic medical record database. Patients who received DAP between October 2011 and December 2020 were enrolled. Two definitions of musculoskeletal toxicity were employed: (1) elevation of creatine phosphokinase (CPK) value more than twice from baseline and >200 IU/L, and (2) >1000 IU/L. First, multiple logistic regression analyses (a conventional statistical method) were performed to identify independent factors affecting CPK elevation. Then, decision tree analyses, a machine learning method, were performed to detect combinations of factors that change CPK elevation risk. RESULTS: Of the 2970 patients who received DAP, 706 were included. Elevation of CPK values >200 IU/L and >1000 IU/L occurred in 83 (11.8%) and 17 (2.41%) patients, respectively. In multiple logistic regression analysis, baseline CPK value and concomitant use of hydrophobic statins were commonly extracted as independent factors affecting each CPK elevation, but concomitant use of hydrophilic statins was not. In decision tree analysis, patients who received hydrophobic statins and had high baseline CPK values were classified into the high-risk group. CONCLUSION: Our novel approach revealed new risk factors for CPK elevation. Our findings suggest that high-risk patients require frequent CPK monitoring.

Using Japanese big data to investigate novel factors and their high-risk combinations that affect vancomycin-induced nephrotoxicity.

AIMS: Several factors related to vancomycin-induced nephrotoxicity (VIN) have not yet been clarified. In the present study, we used Japanese big data to investigate novel factors and their high-risk combinations that influence VIN. METHODS: We employed a large Japanese electronic medical record database and included patients who had been administered intravenous vancomycin between June 2000 and December 2020. VIN was defined as an increase in serum creatinine ≥0.5 mg/dL or 1.5-fold higher than the baseline. The outcomes were: (1) factors affecting VIN that were identified using multiple logistic regression analysis, and (2) combinations of factors that affect the risk of VIN according to a decision tree analysis, which is a typical machine learning method. RESULTS: Of the 7306 patients that were enrolled, VIN occurred in 14.2% of them (1035). A multivariate analysis extracted 22 variables as independent factors. Concomitant ramelteon use (odds ratio 0.701, 95% confidence interval 0.512-0.959), ward pharmacy service (0.741, 0.638-0.861), duration of VCM < 7 days (0.748, 0.623-0.899) and trough concentrations 10-15 mg/L (0.668, 0.556-0.802) reduce the risk of VIN. Meanwhile, concomitant piperacillin-tazobactam use (2.056, 1.754-2.409) and piperacillin use (2.868, 1.298-6.338) increase the risk. The decision tree analysis showed that a combination of vancomycin trough concentrations ≥20 mg/L and concomitant piperacillin-tazobactam use was associated with the highest risk. CONCLUSIONS: We revealed that the concomitant ramelteon use and ward pharmacy service may decrease the risk of VIN, while the concomitant use of not only piperacillin-tazobactam but also piperacillin may increase the risk.

Machine Learning-Based Model for Estimating Vancomycin Maintenance Dose to Target the Area under the Concentration Curve of 400-600 mg·h/L in Japanese Patients.

In therapeutic drug monitoring of vancomycin (VCM), the area under the concentration-time curve (AUC) is related to clinical efficacy and toxicity. Determining the maintenance for patient is necessary since VCM concentrations are affected by factors such as renal function. We constructed a machine learning-based model to estimate the maintenance dose to target an AUC of 400-600 mg⋅h/L in each combination of patient's factors. This retrospective observational study was conducted at two hospitals. Patients who received VCM intravenously with measured trough and another point (e.g., peak) concentrations within the November 2011 to March 2019 period were enrolled. We extracted the factors that affect VCM concentration and constructed a decision tree model using a classification and regression tree algorithm. Of the 1380 patients, 822 were included. Training data were split up to four times and included 24 subgroups. The average corrected VCM daily doses ranged 17.6-59.4 mg/kg. Estimated glomerular filtration rate, age, and body mass index were selected as predictive variables that affected the recommended daily dose. In the validation data, our model had slightly higher proportions of AUC of 400-600 mg⋅h/L than other nomograms. However, our model was based only on limited patients. Thus, further clinical studies are needed to develop a general-purpose model in the future. We successfully constructed a model that recommends VCM maintenance daily doses with AUC of 400-600 mg⋅h/L for each combination of independent variables. Our model has the potential for application as a simple decision-making tool for medical staff.

Urinary Protein-to-creatinine Ratios Predict Recurrence in Pediatric and Young Adult Cases of Minimal Change Nephrotic Syndrome.

High-dose steroids are required for the treatment of minimal change nephrotic syndrome (MCNS), especially for episodes of recurrence. Predicting and avoiding recurrence can help reduce the steroid dose, but prediction is currently difficult. We herein examined whether changes in laboratory data, especially the urinary protein- to-creatinine ratio (UTP/UCr), can predict clinical recurrence. We also assessed differences in clinical features between children and young adults. We included 36 patients with MCNS; for each case, we retrospectively studied laboratory data during stable remission and pre-recurrence, with the "stable" period defined as all but the 6 weeks before recurrence, and pre-recurrence defined as the 4±2 weeks before recurrence. UTP/UCr, serum albumin, etc. were measured every 5 years during stable periods. We divided patients into cohorts by age at recurrence, < 15 years and ≥ 15 years, and compared stable and pre-recurrence values for the two groups. UTP/UCr values during stable periods tended to be higher in younger patients. UTP/UCr and serum albumin showed statistically significant changes during pre-recurrence periods, but only in those aged ≥ 15 years. Thus, clinical features of recurrence differed depending on age. Signs of recurrence can be confirmed via UTP/UCr or serum albumin several weeks before recurrence in patients ≥ 15 years.

Probiotic Prescription Status of Pediatric Patients with Otitis Media Receiving Oral Amoxicillin or Amoxicillin/Clavulanate from April 2016 to March 2017 Using a Japanese Health Insurance Claims Database.

Antibiotic-associated diarrhea (AAD) is a typical side effect of antibiotic treatment, especially in children. Amoxicillin (AMPC) and amoxicillin/clavulanate (AMPC/CVA) are associated with high risk of AAD; however, these antibiotics are important in the pediatric field. Recent research suggests that probiotics prevent pediatric AAD, including that caused by AMPC and AMPC/CVA. Indeed, guidelines for acute otitis media in children recommend the concomitant use of probiotics. However, the prescription status of probiotics for pediatric patients with otitis media receiving oral AMPC and AMPC/CVA remains unknown. We therefore conducted a survey to clarify the current status of these prescriptions and, in particular, to identify specific populations with a low proportion of probiotic prescriptions. Pediatric patients (≤15 years of age) newly prescribed oral AMPC or AMPC/CVA for otitis media between April 2016 and March 2017 were identified from a Japanese health insurance claims database. Eligible patients were divided into the AMPC (1303 patients) and AMPC/CVA (424 patients) groups, in which 659 (50.6%) and 293 (69.1%) patients were prescribed probiotics, respectively. Of the patients receiving probiotic prescriptions in the AMPC and AMPC/CVA groups, 632 (95.9%) and 286 (97.6%) patients received antibiotic-resistant probiotic prescriptions, respectively. When classified by the prescribing clinical department and patient age, the proportions of probiotic prescriptions in Internal Medicine and Pediatrics departments were lower than those in the Otorhinolaryngology department regardless of age. These results indicate the probability of insufficient probiotic prescriptions for pediatric patients with otitis media. Solving this issue may lead to the provision of safer antimicrobial therapy.

Implementation Status of Liver Function Tests for Monitoring Benzbromarone-Induced Hepatotoxicity: An Epidemiological Survey Using the Japanese Claims Database.

A major adverse effect of benzbromarone is hepatotoxicity. Therefore, periodic liver function tests are required at least for the first 6 months of benzbromarone administration. However, it is not clear whether the relevant blood tests are implemented appropriately. Here, we performed a cross-sectional survey of the implementation status of liver function tests in patients who were newly prescribed benzbromarone, using the Japanese large claims database. Male patients who were newly prescribed benzbromarone from January 2010 to December 2016 were included. We targeted patients who continued benzbromarone during the observation period (up to 180 d from the start of administration). The primary endpoint was the proportion of patients in whom periodic liver function tests were implemented. A periodic liver function test was defined as one or more liver function tests performed during both 1-90 and 91-180 d of initial benzbromarone administration. We labeled the tests as a "periodic test" or "non-periodic test" based on whether periodic liver function tests were performed or not, respectively. Furthermore, factors influencing non-periodic test were analyzed. Periodic testing was implemented only in 28.7% of patients. Additionally, factors such as number of hospital beds ≤19 (compared to 100-199 beds) and duration of the first prescription of benzbromarone were associated with non-periodic testing. Our study revealed that periodic liver function tests are not performed sufficiently in Japan. Thus, clinicians prescribing benzbromarone should be educated about the test. Our blood-test-based approach should be applied to other drugs and countries in future research.

Individualized concept for the treatment of autosomal recessive polycystic kidney disease with end-stage renal disease.

Management of children with autosomal recessive polycystic kidney disease (ARPKD) who develop end-stage renal disease (ESRD) remains challenging because of concomitant liver disease. Patients with recurrent cholangitis are candidates for liver-kidney transplantation, while the treatment for patients with splenomegaly and pancytopenia due to portal hypertension is controversial. Herein, we report 7 children who were treated using an individualized treatment strategy stratified by liver disease. Two patients with recurrent cholangitis underwent sequential liver-kidney transplantation, while 4 patients with splenomegaly and pancytopenia but without recurrent cholangitis underwent splenectomy followed by isolated kidney transplantation. The remaining patient, who did not have cholangitis and pancytopenia, underwent isolated kidney transplantation. Blood cell counts were normalized after splenectomy was performed at the median age of 8.7 (range, 7.4-11.7) years. Kidney transplantation was performed at the median age of 8.8 (range, 1.9-14.7) years in all patients. Overwhelming post-splenectomy infections and cholangitis did not occur during the median follow-up period of 6.3 (range, 1.0-13.2) years. The estimated glomerular filtration rate at the last follow-up was 53 (range, 35-107) mL/min/1.73 m2 . No graft loss occurred. Our individualized treatment strategy stratified by recurrent cholangitis and pancytopenia can be a feasible strategy for children with ARPKD who develop ESRD and warrants further evaluation.

Nonsteroidal anti-inflammatory drugs use in patients with chronic kidney disease are often prescribed from different clinicians than those who diagnosed them.

PURPOSE: When prescribing nonsteroidal anti-inflammatory drugs (NSAIDs) for chronic kidney disease (CKD), patients' pathology and concomitant medications should be considered. In our pharmaceutical experience, NSAIDs are often prescribed by departments that are different from those that diagnosed CKD. That is, NSAIDs may be prescribed for patients without the advice of the clinicians who diagnosed them. In this study, we aimed to elucidate how frequently such cases occur. METHODS: We used the large health insurance claims database constructed by JMDC Inc., Tokyo. We evaluated the proportions of CKD diagnosis and NSAID prescription by different clinical departments and institutions. RESULTS: A total of 224 014 out-patients were included in the analysis; they were divided into CKD (n = 1501) and non-CKD groups (n = 222 513). The internal medicine departments diagnosed CKD most frequently (74.8% of the patients) and surgical departments rarely diagnosed CKD. However, the proportion of prescribed NSAIDs was high in other departments, especially surgical departments. In the CKD group, 50.4% of the patients received CKD diagnosis and NSAID prescription from different clinical departments; 72.8% of the patients received a diagnosis and prescription from different medical institutions. CONCLUSION: Our study revealed that NSAIDs are often prescribed to patients with CKD from different clinicians than those who diagnosed them.

Prescription of Colchicine with Other Dangerous Concomitant Medications: A Nation-Wide Survey Using the Japanese Claims Database.

The anti-inflammatory agent colchicine may cause toxic effects such as rhabdomyolysis, pancytopenia, and acute respiratory distress syndrome in cases of overdose and when patients have renal or liver impairment. As colchicine is a substrate for CYP3A4 and P-glycoprotein (P-gp), drug-drug interactions are important factors that cause fatal colchicine-related side effects. Thus, we conducted a nation-wide survey to determine the status of inappropriate colchicine prescriptions in Japan. Patients prescribed the regular use of colchicine from April 2014 to March 2017 were identified using the Japanese large health insurance claims database. As the primary endpoint, we evaluated the concomitant prescription proportions of strong CYP3A4 and/or P-gp inhibitors classified as "contraindications for co-administration" with colchicine in patients with renal or liver impairment. We defined these cases as "inappropriate colchicine prescriptions." Additionally, factors affecting inappropriate colchicine prescriptions were analyzed. Among the 3302 enrolled patients, 43 (1.30%) were inappropriately prescribed colchicine. Of these 43 patients, 11 had baseline renal and/or liver impairment. By multiple regression analysis, the primary diseases "gout" and "Behçet's disease" were extracted as independent factors for inappropriate colchicine prescriptions with odds ratios of 0.40 (95% confidence interval: 0.19-0.84) and 4.93 (95% confidence interval: 2.12-11.5), respectively. We found that approximately 1% of patients had important colchicine interactions. Particularly, Behçet's disease was a risk factor for inappropriate prescriptions, with approximately 25% of patients showing renal and/or liver impairment (classified as "contraindications for co-administration"). These findings may be useful for medical professionals who prescribe colchicine therapy.

A New Algorithm Optimized for Initial Dose Settings of Vancomycin Using Machine Learning.

This study aimed to construct an optimal algorithm for initial dose settings of vancomycin (VCM) using machine learning (ML) with decision tree (DT) analysis. Patients who were administered intravenous VCM and underwent therapeutic drug monitoring (TDM) at the Hokkaido University Hospital were enrolled. The study period was November 2011 to March 2019. In total, 654 patients were included in the study. Patients were divided into two groups, training (patients who received VCM from November 2011 to December 2017; n = 496) and testing (patients who received VCM from January 2018 to March 2019; n = 158) groups. For the training group, DT analysis of the classification and regression tree algorithm was performed to construct an algorithm (called DT algorithm) for the initial dose settings of VCM. For the testing group, the rates of attaining the VCM therapeutic range (trough value = 10-15 and 10-20 mg/L) with the DT algorithm and three conventional dose-setting methods were compared for model evaluation. The DT algorithm was constructed to be used for patients with estimated glomerular filtration rate ≥50 mL/min and body weight ≥40 kg. As a result, the recommended daily doses ranged from 20.0 to 58.1 mg/kg. In model evaluation, the DT algorithm obtained the highest rates of attaining the VCM therapeutic range compared to conventional dose-setting methods. Therefore, our DT algorithm can be applied to clinical practice. In addition, ML is useful for setting drug doses.

Comparison of interactions between warfarin and cephalosporins with and without the N-methyl-thio-tetrazole side chain.

Cephalosporins with an N-methyl-thio-tetrazole (NMTT) side chain interact with warfarin by reducing the production of blood clotting factors. However, cephalosporins without the NMTT side chain also enhance the effects of warfarin. Thus, we aimed to compare the effects of warfarin modified by cephalosporins with and without the NMTT side chain, using a Japanese health insurance claims database. The inclusion criteria were patients who (1) intravenously received second- or third-generation cephalosporins between April 2010 and March 2017 and (2) received warfarin during cephalosporin therapy. Patients were administered either cephalosporins with the NMTT side chain (NMTT group) or those without NMTT (non-NMTT group). After matching patient data by propensity score, the following outcomes were compared between the two groups: (1) proportion of patients administered vitamin K, (2) proportion of bleeding events, and (3) changes in the daily dose of warfarin. Among 203 patients, 100 patients (50 per group) were matched by the propensity score. The proportion of patients administered vitamin K was 6.0% in both groups. These patients intravenously received a single dose of menatetrenone; no bleeding was observed. The proportion of patients subjected to a reduction in the daily dose of warfarin was 6.5% and 4.3% in the NMTT and non-NMTT groups, respectively. As our study had a small sample size, we could not determine whether the risk of over anticoagulation of warfarin is affected by cephalosporins with or without NMTT side chain. However, we showed the bleeding risk was sufficiently low regardless of the presence/absence of the NMTT side chain.

New-onset diabetes after renal transplantation in a patient with a novel HNF1B mutation.

CAKUT are the most frequent causes of ESRD in children. Mutations in the gene encoding HNF1B, a transcription factor involved in organ development and maintenance, cause a multisystem disorder that includes CAKUT, diabetes, and liver dysfunction. Here, we describe the case of a patient with renal hypodysplasia who developed NODAT presenting with liver dysfunction. The NODAT was initially thought to be steroid and FK related. However, based on the patient's clinical features, including renal hypodysplasia and recurrent elevations of transaminase, screening for an HNF1B mutation was performed. Direct sequencing identified a novel splicing mutation of HNF1B, designated c.344 + 2T>C. Because CAKUT is the leading cause of ESRD in children and HNF1B mutations can cause both renal hypodysplasia and diabetes, HNF1B mutations may account for a portion of the cases of NODAT in pediatric patients who have undergone kidney transplantation. NODAT is a serious and major complication of solid organ transplantation and is associated with reduced graft survival. Therefore, for the appropriate management of kidney transplantation, screening for HNF1B mutations should be considered in pediatric patients with transplants caused by CAKUT who develop NODAT and show extra-renal symptoms.

Urinary Cross-linked N-terminal Telopeptide of Type I Collagen Levels of Infants with Osteogenesis Imperfecta and Healthy Infants.

The urinary cross-linked N-terminal telopeptide of type I collagen (uNTx) levels in infantile osteogenesis imperfecta (OI) have not been well studied. Here we investigated the levels of uNTx in infants with OI and healthy infants. We collected spot urine samples from 30 infants with OI (male/female, 14/16; Sillence classification, I/II/III/IV: 15/3/6/6; age, 5.2±4.4 months) and 120 healthy infants (male/female, 75/45; age, 5.1±4.1 months) for the measurement of uNTx levels. The uNTx levels of the OI infants were significantly lower than those of the healthy infants (mean±SD, 1,363.7±530.1 vs. 2,622.2±1,202.6 nmol BCE/mmol Cr; p＜0.001). The uNTx levels of the infants with type I OI were significantly lower than those of the age-matched healthy infants, although an overlap was observed between the 2 groups. Among the 1-month-old infants, the uNTx levels of the infants with types I, III or IV OI were significantly lower than those of the healthy infants, without overlap (1,622.5±235.8 vs. 3,781.0±1,027.1 nmol BCE/mmol Cr; p＜0.001). These results indicate that uNTx levels are significantly lower in infants with OI than in healthy infants, and they suggest that uNTx might be useful as a reference for diagnosing OI.

A cross-sectional survey of hospitalization and blood tests implementation status in patients who received tolvaptan under 75 years of age using a Japanese claims database.

BACKGROUND: Hypernatremia and liver injury are typical adverse effects of tolvaptan. Therefore, hospitalization and frequent monitoring of serum sodium concentration and liver function are necessary for tolvaptan initiation. We performed a cross-sectional survey to evaluate these situations. RESEARCH DESIGN AND METHODS: We employed the Japanese claims database, which contains data of patients aged < 75 years. Patients who were newly prescribed tolvaptan for fluid accumulation induced by chronic heart failure (FA-CHF) or liver cirrhosis (FA-LC) from January 2011 to June 2017 were included. We evaluated the hospitalization status and implementation of serum sodium and liver function tests in the evaluation period, based on the Japanese package insert. RESULTS: Of 1,173 patients, 347 and 117 were enrolled in FA-CHF and FA-LC groups, respectively. Among them, 10.7% (FA-CHF group) and 5.13% (FA-LC group) were prescribed tolvaptan without hospitalization. In the FA-CHF group, 11.0% and 17.6% did not undergo serum sodium and liver function tests even once in the evaluation period, respectively, compared with 12.0% and 12.8% in the FA-LC group. CONCLUSIONS: Our results highlight the deviation from Japanese package insert recommendations. This approach can be applied to other drugs and provides important perspectives on pharmacovigilance research.

A cross-sectional exploratory survey on occurrence of triple-whammy prescription pattern in Japan.

Background The concurrent use of nonsteroidal anti-inflammatory drugs, renin-angiotensin-aldosterone system blockers, and diuretics, known as a "triple-whammy," is related to the occurrence of acute kidney injury. However, there are few reports regarding the prescription pattern of the triple-whammy. Objective To elucidate the patterns of the triple-whammy prescription in Japan. Methods A cross-sectional study was performed using a health-insurance-claims database that included Japanese people under 75 years of age, and enrolled outpatients that were prescribed any nonsteroidal anti-inflammatory drugs, renin-angiotensin-aldosterone system blockers, and diuretics between April 2017 and June 2017. As an outcome, the proportion of triple-whammy prescriptions was evaluated. Among the patients who received triple-whammy prescriptions, we evaluated the prevalence of chronic kidney disease and the proportion of prescriptions provided for these three drugs from different clinical departments and institutions. Results Overall, 730 of 246,721 (0.3%) patients received triple-whammy prescriptions. Among these patients, 13.3% had underlying chronic kidney disease. The proportions of any of the three drug types prescribed by different clinical departments and institutions was 48.2% and 61.8%, respectively. Conclusions We examined the patterns of triple-whammy prescriptions and concluded that pharmacists need to pay attention to triple-whammy prescriptions if the prescriptions are provided by multiple clinical departments or institutions.

Validation of the usefulness of artificial neural networks for risk prediction of adverse drug reactions used for individual patients in clinical practice.

Artificial neural networks are the main tools for data mining and were inspired by the human brain and nervous system. Studies have demonstrated their usefulness in medicine. However, no studies have used artificial neural networks for the prediction of adverse drug reactions. We aimed to validate the usefulness of artificial neural networks for the prediction of adverse drug reactions and focused on vancomycin -induced nephrotoxicity. For constructing an artificial neural network, a multilayer perceptron algorithm was employed. A 10-fold cross validation method was adopted for evaluating the resultant artificial neural network. In total, 1141 patients who received vancomycin at Hokkaido University Hospital from November 2011 to February 2019 were enrolled. Among these patients, 179 (15.7%) developed vancomycin -induced nephrotoxicity. The top three risk factors of vancomycin -induced nephrotoxicity which are relatively important in the artificial neural networks were average vancomycin trough concentration ≥ 13.0 mg/L and concomitant use of piperacillin-tazobactam and vasopressor drugs. The predictive accuracy of the artificial neural network was 86.3% and that of the multiple logistic regression model (conventional statistical method) was 85.1%. Moreover, area under the receiver operating characteristic curve (AUROC) of the artificial neural network was 0.83. In the 10-fold cross-validation, the accuracy obtained was 86.0% and AUROC was 0.82. The artificial neural network model predicting the vancomycin -induced nephrotoxicity showed good predictive performance. This appears to be the first report of the usefulness of artificial neural networks for an adverse drug reactions risk prediction model.

Association of the ward pharmacy service with active implementation of therapeutic drug monitoring for vancomycin and teicoplanin-an epidemiological surveillance study using Japanese large health insurance claims database.

BACKGROUND: Ward pharmacists are required for the active implementation of therapeutic drug monitoring (TDM). This epidemiological study verified whether Japanese ward pharmacists contribute to improving the TDM implementation proportions of anti-methicillin-resistant Staphylococcus aureus (MRSA) agents using the large health insurance claims database. METHODS: The patients who received intravenous anti-MRSA agents from April 2012 to March 2017 were enrolled. We defined ward pharmacy service as the "drug management and guidance fee" and/or "inpatient pharmaceutical services premium". In addition, implementation of TDM was identified by "the specific drug treatment management fee". We compared the proportions of TDM implementation for vancomycin (VCM), teicoplanin (TEIC), and arbekacin (ABK) in the ward and non-ward pharmacy service groups. To avoid confounding, the propensity score method was employed. Moreover, the clinical variables affecting TDM implementation in each anti-MRSA agent were analyzed by using a multiple logistic regression model. RESULTS: The following number of patients were included in the study: VCM (n = 2138), TEIC (n = 596), and ABK (n = 142). After propensity score matching, the proportions of TDM implementation for VCM and TEIC were higher in the ward pharmacy service group than in the non-ward pharmacy service group (VCM: 69.2% vs 60.3%, TEIC: 51.4% vs 34.7%), while no significant difference was observed for ABK (21.2% vs 23.1%). As independent clinical variables affecting TDM implementation for VCM and TEIC, several clinical variables, including ward pharmacy services, were extracted. In contrast, no clinical variables were extracted for ABK. CONCLUSIONS: We found that the ward pharmacy service is associated with the active implementation of TDM for anti-MRSA agents, such as VCM and TEIC.