Int J Curr Pharm Res, Vol 15, Issue 1, 63-67Original Article
PRESCRIPTION PATTERN OF ANTIBIOTICS AND THEIR APPROPRIATENESS IN PATIENTS WITH CHRONIC KIDNEY DISEASE-AN OBSERVATIONAL STUDY IN A TERTIARY CARE TEACHING HOSPITAL IN SOUTH INDIA
PALLAVI CHALIVENDRA1, UMAMAHESWARA RAO K.2, J SUCHARITHA3, VURIMI BHOPAL CHANDRA4*, ANJANI DEVI NELAVALA5
1,2,3SVIMS, Sri Padmavathi Medical College for Women, Tirupathi, 4Rama Medial College–Hospital and Research Center, Mandhana, Kanpur, UP, 5Faculty of Nursing, Rama University, Mandhana, Kanpur, UP
Email: bhopalchandra@gmail.com
Received: 15 Oct 2022, Revised and Accepted: 22 Dec 2022
ABSTRACT
Objective: The primary objective of the study was to assess the prescribing pattern of antimicrobial agents in patients with chronic kidney disease. The secondary objectives of this study are to assess antibiotic appropriateness and dose optimization in patients with chronic kidney disease in relation to their comorbidities.
Methods: A retrospective study was conducted, and medical records of all patients with CKD who were admitted in the nephrology department of Sri Venkateswara Institute of Medical Sciences, Tirupati, during Jan 2018-Dec 2018 were reviewed for antibiotic prescriptions. A total of 200 medical records were selected and assessed for antimicrobial prescriptions. A p-value <0.05 was considered significant throughout the statistical analysis.
Results: Analysis showed that overall 163 drugs were prescribed to CKD patients, of which nearly 96 (58.9%) required dosage adjustment. Of those 163 drugs, the majority N= 25 (26%), were unadjusted, and the remaining N = 71 (74%) were properly adjusted. The length of hospitalization of CKD patients was below 7 was 13.5%, above 7 was 86.5%. Mean and SD was 10.27±7.18 d, (Range: 1–35 d). The Chi-square analysis confirmed that out of the seven studied variables, two i.e. Length of stay days; p<0.001.
Conclusion: It is concluded that the occurrence of medication dosing errors was moderate in hospitalized chronic kidney disease patients in our study. Nearly 20% of patients who had prolonged stays were prescribed antibiotics for a prolonged period. The predictors of medication dosing errors in CKD patients were the severe-to-end stages of chronic kidney disease, the number of prescribed antibiotics, and the length of hospitalization.
Keywords: Chronic kidney disease, GFR, Antibiotics, Prescription
© 2023 The Authors. Published by Innovare Academic Sciences Pvt Ltd. This is an open access article under the CC BY license (https://creativecommons.org/licenses/by/4.0/)
DOI: https://dx.doi.org/10.22159/ijcpr.2023v15i1.2077 Journal homepage: https://innovareacademics.in/journals/index.php/ijcpr
INTRODUCTION
Chronic Kidney Disease (CKD) is a serious public health concern that affects nearly 10 to 15%of the adult population worldwide. CKD is defined as a reduced GFR (glomerular filtration rate) <60 ml/minute for>3 mo that may or may not be coupled with kidney damage and classified into several stages based on GFR [1]. When GFR falls below 15 ml/min, it is Kidney failure and the patient requires dialysis. More than 800 million people worldwide, more than 10% of the overall population, have chronic kidney disease [2], between 1990 and 2017; in India, there was a 38% increase in deaths attributable to kidney failure between 2001–03 and 2010–13 [3]. The global all-age prevalence and mortality from CKD increased by 29.3 and 41.5%, respectively [4]. CKD alters the pharmacokinetic and pharmacodynamics response of various drugs mainly excreted from the body through the renal route [5]. Reduction in GFR causes accumulation of drugs that are excreted unchanged via the kidney, accumulation of active metabolites of drugs causing prolonged action, change in drug distribution-protein binding and a decrease in renal drug metabolism. Kidney dysfunction influences the pharmacokinetic parameters of at least 50% of all essential drugs. The response to a drug can be altered by pathologic conditions (acute and chronic illness and kidney disease) by Decreased receptor number (Emax) and sensitivity (EC50), Decreased receptor binding and altered signal transduction. It was found that vancomycin with a time-dependent antibacterial effect showed increased efficacy and decreased toxicity when administered by prolonged infusion rather than short-term infusion [6]. Similar effect is noted with Meropenem and Piperacillin, where the dosage was based on pharmacodynamics regimens. The complications of CKD include increased susceptibility to Adverse drug reactions, metabolic and endocrine complications, including anemia and acidosis, increased risk for cardiovascular diseases, and a variety of other complications, including infections, frailty, and cognitive impairment. Complications may occur at any stage. Complications may also arise from adverse effects of interventions to prevent or treat the disease and associated comorbidities.
Hence normal kidney function is of utmost importance for several drugs and their metabolites to be eliminated from the body. Doses have to be reduced or adjusted in patients with renal impairment. This is often overlooked by the healthcare professions and could lead to adverse drug reactions (ADRs), drug toxicities and therapeutic failure [7].
Several studies conducted worldwide have shown that the doses of drugs that are excreted unchanged in Urine are not being adjusted appropriately among CKD patients. The most commonly prescribed medications among CKD patients are Antibiotics which account for nearly 97% of the prescriptions in patients undergoing Chronic Renal Replacement Therapy in the hospital setting. Thus the present study was conducted to assess the prescribing pattern of antimicrobial agents in patients with chronic kidney disease. The secondary objectives of this study are to assess antibiotic appropriateness and dose optimization in patients with chronic kidney disease in relation to their comorbidities.
MATERIALS AND METHODS
Study setting and study design
A retrospective study was conducted, and medical records of all patients with chronic kidney disease admitted to the nephrology department of Sri Venkateswara Institute of Medical Sciences (SVIMS), Tirupati, during Jan 2018-Dec 2018 were reviewed for antibiotic prescriptions after obtaining Ethics Committee Approval for the study (Roc. No. AS/11/IEC/SVIMS/2017/849). The study data was anonymized and deidentified to ensure the confidentiality of the patient’s details. Chronic kidney disease patients were selected based on estimating the Glomerular filtration rate (eGFR) using the Modification of Diet in Renal Disease (MDRD) equation. Patients with eGFR<60 ml/min/1.73m2 will be included in the study. The formula for eGFR using the MDRD equation:
eGFR =175 x (SCr)-1.[15]4x (age)- °.[20]3 x 0.742 [if female] x 1.212 [if African American]
Medical records of all patients with chronic kidney disease who were admitted to the nephrology unit for a minimum period of 24 h, having eGFR<60 ml/min/1.73m2, Which means patients included in the study had stages of chronic kidney disease as follows: stage 3 (CrCl 30–59 ml/min), stage 4 (CrCl 15–29 ml/min), and stage 5 (CrCl<15 ml/min) and those prescribed Antibiotics during the hospital stay were studied. All those medical charts not meeting the criteria mentioned above were excluded. A total of 200 medical records were selected and assessed for antimicrobial prescriptions. A special data collection form was designed and used to collect the required data.
MATERIALS AND METHODS
Data on the Patient’s age, gender, length of Hospital stay, associated comorbidities, serum creatinine, estimated GFR, antibiotics prescribed, type and dosing pattern were recorded from the medical records. The antimicrobials were classified based on chemistry and mode of action.
The Antimicrobial prescriptions were assessed for the appropriateness and rationality in chronic kidney disease patients for their class, dose and duration of therapy according to the standard guidelines of the National Formulary of India 2021. The doses of drugs were assessed for appropriateness individually for every patient using the dose adjustment guidelines (table 1) [8].
Table 1: Dose adjustment guidelines for antimicrobials prescribed in the study
| S. No. | Antimicrobials | Usual dose (Normal renal function) | CrCl (ml/min) | Dosage adjustment (In renal insufficiency) |
| 1. | Ceftriaxone sodium | 1-2g q24h in equally two divided doses as 30 min infusion Max dose: 4g | - | No dose adjustment is necessary, but serum monitoring is required in severe renal impairment and in patients with both renal and hepatic dysfunction. |
| 2. | Piperacillin/tazobactam | Nosocomial pneumonia: 4.5g q6h plus an aminoglycoside Other Infections: 3.375g q6h as 30 min infusion | >40 | 3.375g q6h and for nosocomial pneumonia 4.5g q6h |
| 20-40 | 2.25g q6h and for nosocomial pneumonia 3.375g q6h | |||
| <20 | 2.25g q8h and for nosocomial pneumonia 2.25g q6h | |||
| HD | 2.25g q12h and for nosocomial pneumonia 2.25g q8h | |||
| CAPD | 2.25g q12h and for nosocomial pneumonia 2.25g q8h | |||
| 3. | Doxycycline | 100 mg/day | - | No dose change is needed with kidney/liver damage |
| 4. | Cefixime | 200 mg q12h | >50 | 200 mg q12h |
| 10-50 | Decrease dose by 75% | |||
| <10 | Decrease dose by 50% | |||
| 5. | Metronidazole | Prophylaxis: 500 mg before surgery; repeated 8 hourly Treatment: 500 mg q8-12h | Mild, | No dose |
| moderate | adjustment required | |||
| 6. | CEF+Salbactum | 3.0-4.5g q12h as 15-60 min infusion (Maximum recommended dose of cefoperazone is 8g and sulbactam is 4g) | 15-30 | 3g q12h |
| <15 | 1.5g q12h | |||
| HD | Dose to be given after dialysis | |||
| 7. | Vancomycin | 500 mg q6h or 1g q12h as 1h infusion |
>50-90 | 1g q12h |
| 10-50 | 1g q24-96h | |||
| <10/HD/CAPD | 1g q4-7days | |||
| 8. | Amoxicillin/clavulanic acid | 1.2g q6-8h | >30 No dose adjustment necessary |
>30 No dose adjustment necessary |
| 10 | 30 1.2g IV stat., followed by 600 mg q12h | |||
| <10 | 1.2g IV stat., followed by 600 mg q24h | |||
| 9. | Ciprofloxacin | 400 mg q8-12h depending on the severity of infection as 1h infusion | >30 | No dose adjustment required |
| 5-29 | 200-400 mg q18-24h |
Data analysis
Statistical analysis was carried out using IBM SPSS version 20. Descriptive statistics were applied on independent variables such as age, gender, stages of chronic kidney disease, the number of prescribed antibiotics, the length of hospitalization, comorbidities of patients, and the pattern of medication dosing errors. A p-value<0.05 was considered significant throughout the statistical analysis.
RESULTS
In this study, nearly 192 medical charts of chronic kidney disease (CKD) patients were assessed. Of which, nearly 71.4% of patients were males and 28.6% were females. Nearly 34 (17.7%) patients were aged between 21–40 y, followed by 98 (51%) who were aged between 41–60 y, and only 26% who had an age above 60 y. The length of hospitalization of CKD patients was below 7 was 13.5%, above 7 was 86.5%. Mean and SD was 10.27±7.18 d, (Range: 1–35 d). Furthermore, it was observed that the majority N = 160 (83.3%) of patients had CKD stage 5, followed by nearly 9.9% who had CKD stage 4, and the remaining 5.2% of patients had CKD stage 3. Hypertension 72 (37.5%), Diabetes with Hypertension 37 (19.3%) and Respiratory Illness (3.1%) were observed as the three topmost comorbidities in CKD patients (table 2). Further analysis showed that overall, 163 drugs were prescribed to CKD patients, and nearly 96 (58.9%) required dosage adjustment, of those 163 drugs, the majority N= 25 (26%) were unadjusted, and the remaining N = 71 (74%) were properly adjusted (table 3). Moreover, the descriptive statistics revealed that the most common unadjusted drugs were Inj. Amoxicillin+Clavulanic acid, Inj. Metronidazole (0.6%), Inj. Cefoperazone+Sulbactam, Tab. Amoxicillin+Clavulanic acid (0.6%), Inj. Amoxicillin+Clavulanic acid (0.6%), Tab. Amoxicillin+Clavulanic acid (4.9%), Tab. Cefixime (3.7%), Tab. Metronidazole (0.6%), Tab. Norfloxacin (0.6%). For further details, please see table 4. The Chi-square analysis was carried out to assess the predictors of medication dosing errors in CKD patients. The analysis confirmed that out of the seven studied variables, two i.e. Length of stay days; p<0.001, Antibiotics prescribed; p<0.05 was associated with medication dosing errors. reaming Age (p-0.9706), Gender (p-0.3633), CKD Stages (p-0.6275), Comorbidity present (p-0.6159), Comorbidities (p-0.2307) were not associated with the medication dosing errors at p<0.05 level (table 5).
Table 2: Patient characteristics (n = 192)
| Variables | Frequency | Percentage | |
| Age | |||
| <20 | 10 | 5.2% | |
| 21-40 | 34 | 17.7% | |
| 41-60 | 98 | 51.0% | |
| >61 | 50 | 26.0% | |
| Gender | |||
| Female | 55 | 28.6% | |
| Male | 137 | 71.4% | |
| Length of stay days | <7 | 26 | 13.5% |
| >8 | 166 | 86.5% | |
| CKD Stage | |||
| Stage 1 | 1 | 0.5% | |
| Stage 2 | 2 | 1.0% | |
| Stage 3 | 10 | 5.2% | |
| Stage 4 | 19 | 9.9% | |
| Stage 5 | 160 | 83.3% | |
| Comorbidity present | |||
| Yes | 125 | 65.1% | |
| No | 67 | 34.9% | |
| Antibiotics prescribed | |||
| Yes | 163 | 84.9% | |
| No | 29 | 15.1% | |
| Comorbidities | |||
| HT | 72 | 37.5% | |
| T2DM | 3 | 1.6% | |
| HTN, T2DM | 37 | 19.3% | |
| HTN, Respiratory Illness | 3 | 1.6% | |
| HTN, Tropical Fever | 1 | 0.5% | |
| Respiratory Illness | 6 | 3.1% | |
| Tropical Fever | 1 | 0.5% | |
| Av Fistula Tract Infection | 1 | 0.5% | |
| GE | 1 | 0.5% |
Table 3: Frequency of adjusted and unadjusted prescribed drugs
| Variable | Frequency | Percentage |
| Total drugs prescribed | 163 | 100 % |
| Number of drugs requiring dose adjustment | 96/163 | 58.9% |
| Number of drugs Not requiring dose adjustment | 67/163 | 41.1% |
| Number of drugs properly adjusted | 71/96 | 74.0% |
| Number of drugs unadjusted | 25/96 | 26.0% |
Table 4: Pattern of medication dosing errors in chronic kidney disease patients
| Generic name | Frequency | Percentage | Dose adjusted for eGFR |
| Inj. Amoxicillin+Clavulanic acid, Inj. Metronidazole | 1 | 0.6% | Un Adjusted |
| Inj. Cefoperazone+Sulbactam, Inj. Piperacillin | 1 | 0.6% | Adjusted |
| Inj. Cefoperazone+Sulbactam, Tab. Amoxicillin+Clavulanic acid | 1 | 0.6% | Un adjusted |
| Inj. Piperacillin+Tazobactam | 8 | 4.9% | Adjusted |
| Inj. Amoxicillin+Clavulanic acid | 1 | 0.6% | Un adjusted |
| Inj. Cefoperazone+Sulbactam | 54 | 33.1% | Adjusted |
| Inj. Cefoperazone+Sulbactam, Inj. Doxycycline | 1 | 0.6% | Adjusted |
| Inj. Cefoperazone+Sulbactam, Inj. Metronidazole | 1 | 0.6% | Adjusted |
| Inj. Ceftriaxone | 12 | 7.4% | Not required |
| Inj. Ceftriaxone+Sulbactam | 48 | 29.4% | Not required |
| Inj. Ceftriaxone+Sulbactam, Inj. Ceftriaxone | 1 | 0.6% | Not required |
| Inj. Ceftriaxone, Inj. Doxycycline | 1 | 0.6% | Not required |
| Inj. Doxycycline | 4 | 2.5% | Not required |
| Inj. Imipenem+Cilastatin | 1 | 0.6% | Not required |
| Inj. Piperacillin+Tazobactam | 5 | 3.1% | Adjusted |
| Inj. Vancomycin | 1 | 0.6% | Adjusted |
| Inj. Vancomycin, Inj. Ceftriaxone | 1 | 0.6% | Adjusted |
| Tab. Amoxicillin+Clavulanic acid | 10 | 6.1% | Un adjusted |
| Tab. Cefixime | 9 | 5.5% | Un adjusted |
| Tab. Doxycycline | 1 | 0.6% | Un adjusted |
| Tab. Metronidazole | 1 | 0.6% | Un adjusted |
| Tab. Norfloxacin | 1 | 0.6% | Un adjusted |
Table 5: Predictors of medication dosing errors in chronic kidney disease patients
| Variables | Frequency | Patients with unadjusted drugs | p-value | |
| Age | <20 | 10 | 1 | 0.9706 |
| 21-40 | 34 | 4 | ||
| 41-60 | 98 | 14 | ||
| >61 | 50 | 6 | ||
| Gender | Female | 55 | 5 | 0.3633 |
| Male | 137 | 20 | ||
| Length of stay days | <7 | 26 | 14 | 0.0001 |
| >8 | 166 | 11 | ||
| CKD Stage | Stage 3 | 13 | 3 | 0.6275 |
| Stage 4 | 19 | 2 | ||
| Stage 5 | 160 | 20 | ||
| Comorbidity present | Yes | 125 | 15 | 0.6159 |
| No | 67 | 10 | ||
| Antibiotics prescribed | Yes | 163 | 25 | 0.0368 |
| No | 29 | - | ||
| Comorbidities | HT | 72 | 10 | 0.2307 |
| HTN,T2DM | 37 | 4 | ||
| GE | 1 | 1 |
Chi-square; Significance: p<0.05.
DISCUSSION
The aim of the present study is to assess the type of antibiotics prescribed to CKD patients with moderate to severe renal dysfunction. Additionally, the appropriateness and the dose adjustment for these antibiotics was also studied in relation to the patient’s condition and comorbidities. To date, very few studies have explored antibiotic use and appropriateness in patients having severe renal dysfunction. One hospital-based study was conducted by Hui et al. in Australia [9] to assess the patterns of use and appropriateness of antibiotics. The results of our study were similar to their study in terms of prescribing antibiotics like Vancomycin and Amoxicillin and their dose adjustments. It was noted that 99% (n=189) of these patients had undergone hemodialysis and antibiotics were given prophylactically as these patients are at high risk of developing infections. The antibiotics were chosen according to the ICMR treatment guidelines for antimicrobial Use in common syndromes and local antimicrobial resistance pattern. The most common antibiotics prescribed were found to be injectable Cephalosporins i.e., Ceftriaxone+Sulbactam and Inj. Cefoperazone+Sulbactam. The appropriateness included choosing the right dose and frequency for the given indication and adjusting according to the patients' estimated Glomerular Filtration Rate. The number of prescriptions having appropriate dose adjustments was 74%, and those having inappropriateness was 26%.
However, the length of stay in the hospital has matched with the duration of administration of antibiotics in our study. The average length of hospitalization was 10.27 d. Out of 192 patients, 40 (20%) had prolonged hospital stay of more than 14 d and received antibiotics throughout the stay, which is of great concern. Prolonged administration of antibiotics contributes to the selection of resistant mutants and antibiotic resistance. Out of seven variables tested, two i.e. Length of stay in days; p<0.001, Antibiotics prescribed; p<0.05 was associated with the medication dosing errors in the multivariate analysis of linear regression. It was found that patients having a comorbid illness such as respiratory illness, AV fistula tract infections were prescribed with multiple antibiotics. They were prescribed initially with unadjusted doses and dose decrement done after 3-4 d. However, the information on the reason for the change of antibiotic from one class to another could not be retrieved from the medical records. The dosing interval of orally administered antibiotics is inadequate for nearly 11 % of patients, which is similar to that of Hui et al. and needs further research in this aspect. The percentage of medication dosing errors in antimicrobials is found to be less in our study i.e., 26% compared to the results of a similar study done by Saleemand Masood in 2016 [10]. Establishing antimicrobial stewardship in dialysis centers has been discussed by D’Agata [11]. It may be worthwhile to provide further continuing education to doctors, community pharmacists or nurses on antibiotic prescribing and use in this particular population.
LIMITATIONS
The present study has several strengths. Firstly, the pattern and predictors of medication dosing errors of antimicrobials were studied and identified in moderate to severe renal disease patients. Despite these strengths, the study has a few limitations. First, a retrospective study design was employed, which restricted us from suggesting interventions and observing actual adverse drug reactions. Second, the MDRD equation was used due to a lack of data regarding patients’ weight, which is less suitable for patients with higher muscle mass and those suffering from serious disorders such as cancer. Third, due to the lack of data, the most responsible diagnosis for hospital admission could not be identified.
CONCLUSION
It is concluded that the occurrence of medication dosing errors was moderate in hospitalized chronic kidney disease patients in our study. Nearly 20% of patients who had prolonged stay were prescribed antibiotics. The predictors of medication dosing errors in CKD patients were the severe-to-end stages of chronic kidney disease, the number of prescribed antibiotics, and the length of hospitalization. Therefore, physicians are advised to take care of these predictors of medication dosing errors while prescribing drugs to chronic kidney disease patients to minimize the risk of drug-related toxicities and limit the spread of antimicrobial resistance.
ACKNOWLEDGEMENT
Dr. V. Siva Kumar, Professor in Nephrology and Dr. R. Ram, Professor and HOD in Nephrology for their support and encouragement.
FUNDING
Nil
AUTHORS CONTRIBUTIONS
All the authors have contributed equally.
CONFLICT OF INTERESTS
Declared none
REFERENCES
Bukar AA, Sulaiman MM, Ladu AI, Abba AM, Ahmed MK, Marama GT. Chronic kidney disease amongst sickle cell anaemia patients at the University of Maiduguri Teaching Hospital, Northeastern Nigeria: a study of prevalence and risk factors. Mediterr J Hematol Infect Dis. 2019;11(1):e2019010. doi: 10.4084/MJHID.2019.010, PMID 30671216.
Kovesdy CP. Epidemiology of chronic kidney disease: an update 2022. Kidney Int Suppl (2011). 2022 Apr 1;12(1):7-11. doi: 10.1016/j.kisu.2021.11.003, PMID 35529086.
GBD Chronic Kidney Disease Collaboration. Global, regional, and national burden of chronic kidney disease, 1990-2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet. 2020;395(10225):709-33. doi: 10.1016/S0140-6736(20)30045-3, PMID 32061315.
Kumar V, Yadav AK, Sethi J, Ghosh A, Sahay M, Prasad N. The Indian chronic kidney disease (ICKD) study: baseline characteristics. Clin Kidney J. 2022 Jan;15(1):60-9. doi: 10.1093/ckj/sfab149, PMID 35035937.
Scheen AJ. Pharmacokinetics, pharmacodynamics and clinical use of SGLT2 inhibitors in patients with type 2 diabetes mellitus and chronic kidney disease. Clin Pharmacokinet. 2015 Jul;54(7):691-708. doi: 10.1007/s40262-015-0264-4, PMID 25805666.
Ozoux ML, Briand V, Pelat M, Barbe F, Schaeffer P, Beauverger P. Potential therapeutic value of urotensin II receptor antagonist in chronic kidney disease and associated comorbidities. J Pharmacol Exp Ther. 2020 Jul 1;374(1):24-37. doi: 10.1124/jpet.120.265496, PMID 32332113.
Waikar SS, Srivastava A, Palsson R, Shafi T, Hsu CY, Sharma K. Association of urinary oxalate excretion with the risk of chronic kidney disease progression. JAMA Intern Med. 2019 Apr 1;179(4):542-51. doi: 10.1001/jamainternmed.2018.7980, PMID 30830167.
Saleem A, Masood I. Pattern and predictors of medication dosing errors in chronic kidney disease patients in Pakistan: a single center retrospective analysis. Plos One. 2016 Jul 1;11(7):e0158677. doi: 10.1371/journal.pone.0158677, PMID 27367594.
Hui K, Nalder M, Buising K, Pefanis A, Ooi KY, Pedagogos E. Patterns of use and appropriateness of antibiotics prescribed to patients receiving haemodialysis: an observational study. BMC Nephrol. 2017 Dec;18(1):156. doi: 10.1186/s12882-017-0575-9, PMID 28499421.
Saleem A, Masood I, Khan TM, Nawaz M. Potential drug-drug interactions in renal impairment patients in Pakistan. Value Health. 2016 May 1;19(3):A12. doi: 10.1016/j.jval.2016.03.262.
D’Agata EMC, Lindberg CC, Lindberg CM, Downham G, Esposito B, Shemin D. The positive effects of an antimicrobial stewardship program targeting outpatient hemodialysis facilities. Infect Control Hosp Epidemiol. 2018 Dec;39(12):1400-5. doi: 10.1017/ice.2018.237, PMID 30253815.