Renal replacement anticoagulant management: Protocol and analysis plan for an observational comparative effectiveness study of linked data sources

Abstract

Acute kidney injury is common in critical illness. In patients with severe acute kidney injury, renal replacement therapy is needed to prevent harm from metabolic and electrolyte disturbances and fluid overload. In the UK, continuous renal replacement therapy (CRRT) is the preferred modality, which requires anticoagulation. Over the last decade, conventional systemic heparin anticoagulation has started being replaced by regional citrate anticoagulation for CRRT, which is now used in approximately 50% of ICUs. This shift towards regional citrate anticoagulation for CRRT is occurring with little evidence of safety or longer term effectiveness. Renal replacement anticoagulant management (RRAM) is an observational comparative effectiveness study, utilising existing data sources to address the clinical and cost-effectiveness of the change to regional citrate anticoagulation for CRRT in UK ICUs. The study will use data from approximately 85,000 patients who were treated in adult, general ICUs participating in the case mix programme national clinical audit between 1 April 2009 and 31 March 2017. A survey of health service providers’ anticoagulation practices will be combined with treatment and hospital outcome data from the case mix programme and linked with long-term outcomes from the Civil Registrations (deaths), Hospital Episodes Statistics for England, Patient Episodes Data for Wales, and the UK Renal Registry datasets. The primary clinical effectiveness outcome is all-cause mortality at 90-days. The study will incorporate an economic evaluation with micro-costing of both regional citrate anticoagulation and systemic heparin anticoagulation.

Study registration: NCT03545750

Keywords

Acute kidney injury anticoagulant citrate heparin renal replacement

Introduction

Acute kidney injury (AKI) is common in critically ill patients. In the UK, data from the Case Mix Programme (CMP – the national clinical audit of adult critical care) indicate that over 50% of all admissions in 2017–2018 were complicated by AKI during the 24 h following admission to an intensive care unit (ICU).¹ Of these admissions, approximately 10% require renal replacement therapy (RRT) to prevent harm from electrolyte and metabolic disturbance as well as fluid overload. Although some countries still use intermittent haemodialysis, or variations thereof, for acute RRT, continuous renal replacement therapy (CRRT) is preferred because of perceived improvements in haemodynamic stability² and is currently used for RRT in 95% of UK ICUs.³

A major limitation of any system that passes blood around an extracorporeal circuit is the need for anticoagulation to prevent the blood in the circuit from clotting.⁴ Traditionally, systemic heparin anticoagulation (SHA) was the most commonly prescribed anticoagulant for CRRT.⁵ During CRRT using SHA, heparin is added to blood to prevent within-circuit clotting. As heparin is only partially removed during the renal replacement process, patients may be at risk of bleeding and considerable resources are required to monitor clotting status.⁶

Regional citrate anticoagulation (RCA) is an alternative approach. RCA prevents blood clotting by chelating ionised calcium, necessary for the clotting process, using a citrate solution added to the patient’s blood as it enters the circuit. The citrate effect is reversed by infusing a calcium salt as it is returned to the circulation, restoring the blood's ability to clot and reducing the risk of bleeding.⁷ However, other risks associated with citrate anticoagulation, such as hypercalcemia and alterations in the blood’s acid–base balance, may lead to problems with muscle weakness, heart function, bone health, and breathing.^6,7

There is currently a rapid shift in practice towards citrate-based anticoagulation during CRRT in ICUs. However, this rapid adoption of RCA is occurring without robust evidence from large studies assessing both the clinical outcomes and cost-effectiveness. This paper reports the protocol (version 2.2 18 November 2019) for the Renal Replacement Anticoagulant Management (RRAM) observational comparative effectiveness study. The RRAM study utilises existing national audit and routine healthcare datasets to evaluate the clinical and cost-effectiveness of changing from SHA to RCA for CRRT in UK adult ICUs.

Aim and objectives

Our aim is to evaluate the clinical and health economic impacts of moving from SHA to RCA for CRRT in adult patients treated by non-specialist ICUs in England and Wales.

The study will address the following objectives:

Investigate the short-term benefits, risks, and costs of RCA compared to SHA.

Compare the long-term development of end-stage renal disease (ESRD) treated with RRT between RCA and SHA.

Establish efficient research techniques that, if successful, could be used to track the effects of any change in critical care practice occurring in ICUs in England and Wales over a reasonably short time scale.

Methods and analysis

Study design

This is an observational comparative effectiveness study using interrupted time series analysis techniques of individual patient data.^8,9 A cohort of patients who received CRRT for AKI in an ICU in England or Wales between 1 April 2009 and 31 March 2017 will be identified from the Case Mix Programme (CMP) national clinical audit dataset. Once the cohort has been identified, their individual patient data will be linked with, and data extracted from, other routinely collected datasets, to provide information on ICU and hospital resource usage and outcome, longer term mortality, subsequent hospitalisations, and longer term renal outcomes. These data will be combined with a survey of CRRT anticoagulation practices of ICUs in England and Wales participating in the CMP to separate the cohort according to which method of CRRT was in use by the relevant ICU at the time of treatment.

Study population

The study cohort will include critically ill adults who received at least one calendar day of CRRT while treated in an adult general ICU participating in the CMP national clinical audit between 1 April 2009 and 31 March 2017. Patients will be identified using the following criteria:

Inclusion criteria.

Age ≥16 years

Admitted to an adult, general ICU in England or Wales participating in the CMP between 1 April 2009 and 31 March 2017

Receipt of CRRT in ICU, identified by the recording of renal support, as defined by the Critical Care Minimum Dataset (CCMDS),¹⁰ on at least one calendar day during the ICU stay.

Exclusion criteria.

Patients with pre-existing ESRD, identified by the recording of a requirement for chronic renal replacement therapy in the CMP dataset (includes, but is not limited to, chronic haemodialysis, chronic haemofiltration and chronic peritoneal dialysis for irreversible renal disease, within six months prior to admission to ICU).

Patients admitted to an ICU after kidney or multi-organ transplantation including kidney.

Patients admitted with acute hepatic failure.

Interventions

Patients who received RCA for CRRT (exposed)

Patients receiving CRRT in an ICU after the date on which the ICU indicates that they completed the transition from SHA to RCA.

Patients who received SHA for CRRT (comparator)

Patients receiving CRRT in an ICU before the date on which the ICU indicates that they started to transition from SHA to RCA; or admission to an ICU that has not transitioned to RCA

Outcomes

Primary effectiveness outcome

The primary effectiveness outcome is all-cause mortality 90 days after first ICU admission in which CRRT was received.

Primary economic outcome

The primary economic outcome is incremental net monetary benefit gained at one year at a willingness-to-pay of £20,000 per quality-adjusted life year (QALY) associated with a change from SHA to RCA during CRRT.

Secondary outcomes

Secondary outcomes include:

all-cause mortality at hospital discharge, at 30 days after ICU admission, and at one year after ICU admission;

number of days of renal, cardiovascular, and advanced respiratory support while in ICU;

bleeding and thromboembolic episodes;

length of stay in ICU and in hospital;

development of ESRD treated by RRT at one year after ICU admission (among survivors);

estimated lifetime incremental net benefit associated with a change from heparin to citrate anticoagulation during CRRT.

Subgroup analyses

The clinical- and cost-effectiveness outcomes described above will be analysed in a pre-specified subgroup of patients with sepsis (defined according to the Sepsis-3 criteria).¹¹

Datasets and linkage

A brief description of the datasets and the key variables for can be found in Table 1.

Table 1.

Description of datasets for linkage and key variables.

Dataset	Description	Key variables
Case Mix Programme (CMP)	National clinical audit of patient outcomes from adult, general critical care units (intensive care and combined intensive care/high dependency units) in England, Wales and Northern Ireland which contains validated pooled case mix and outcome data from over 2 million consecutive critical care admissions. Since 2015, the CMP has had 100% coverage of adult, general ICUs.	Patient level information including patient identifiers required only for linkage; dates and methods of admission to hospital and ICU; patient case mix (including illness severity scores and mortality probability, derived using standard models for clinical audit); patient outcome at discharge from ICU and hospital; and ICU activity including the number, duration, and level (basic/advanced) of organs supported and length of stay in the ICU and hospital.
Civil Registrations (Mortality) dataset [previously Office for National Statistics]	Contains mortality information, for all deaths registered in England and Wales.	Data including date of death; cause of death; and place of death.
Hospital Episodes Statistics for England (HES)	Contains details of all admissions, accident and emergency attendances and outpatient appointments at NHS hospitals in England.	Patient level information including patient information (age, gender and ethnicity); diagnoses and operations; dates and methods of hospital admission and discharge; and geographical information (where patients were admitted/treated and live).¹²
Patient Episodes Data for Wales (PEDW)	Contains all inpatient and day case activity data for patients treated in NHS Wales and data on Welsh residents treated in English Trusts.	Patient level information including patient information (age, gender and ethnicity); diagnoses and operations; dates and methods of hospital admission and discharge; and geographical information (where patients were admitted/treated and live).
UK Renal Registry (UKRR)	Contains data on all patients on RRT (including patients receiving haemodialysis, peritoneal dialysis, kidney transplant for end stage renal disease (ESRD) and dialysis received on a renal unit for AKI)	Data include treatment data including date first seen by a renal physician; date of first ever RRT and modality (for ESRD patients); data on treatment timelines and haemodialysis sessions; and transplant information.

AKI: acute kidney injury; ICU: intensive care unit; RRT: renal replacement therapy; ESRD: end stage renal disease.

Linkage of individual patient data from the CMP with data held by the UK Renal Registry (UKRR), Hospital Episodes Statistics for England (HES) and Civil Registrations (deaths) datasets will be undertaken by NHS Digital acting as a ‘trusted third party’. Patient identifiers (with no associated clinical data) will be uploaded from the CMP and UKRR clinical audits to secure servers at NHS Digital, who will perform the data linkage and return to each data provider their local identifier (a field that is unique to the records within that dataset) together with a common key that will be used to link all records of the same patient across datasets. The UKRR will then transfer to ICNARC the agreed pseudonymised dataset (including the common key) for successfully linked patients. Similarly, NHS Digital will create a pseudonymised data extract of agreed fields from HES and Civil Registrations datasets and pass these to ICNARC (again including the common key for data linkage). Data linkage between CMP and Patient Episodes Data for Wales (PEDW) will be conducted separately by NHS Wales Informatics Service following a similar methodology. Researchers at ICNARC will then combine the pseudonymised linked data for analysis.

Health economics and quality of life

Micro-costing task analysis

Micro-costing of the set-up and running of CRRT using SHA and RCA will be conducted at a representative sample of sites identified from the survey of citrate uptake. Micro-costing will involve conducting a cognitive walk through (including hierarchical task analysis) with representative clinicians, where users mentally “walk through” the set-up and running of a CRRT device, allowing staff time and consumables for each task element to be estimated.¹³ The costing will be based on experience of delivering CRRT in a typical ICU. The cost of staff time will be obtained from the Unit Costs of Health and Social Care. Unit costs of anticoagulation drugs will be based on the NHS Business Services Authority Drug Tariff.¹⁴ CRRT fluid costs will be obtained from the manufacturers’ quoted prices. Consumable costs will be obtained from the NHS Supply Chain.¹⁵

CRRT set-up time and frequency

The system set-up time and the frequency with which the system (SHA or RCA) fails will be obtained via the Post Intensive Care Risk-Adjusted Alerting and Monitoring (PICRAM) study database which contains anonymised electronically held records of all patients treated on both Oxford general ICUs and the Royal Berkshire Hospital ICU in Reading from 2009 to 2015 and from electronically held data on the clinical information system for patients treated in Oxford following completion of PICRAM. From these data, the number and distribution of intervals between one CRRT system failing and the next being in place will be determined in addition to the running and recommissioning of CRRT for the periods in which citrate and heparin are in use. We will also determine the frequency of related events, such as transfusion episodes and quantities, when both citrate and heparin are in use.

Longer-term costs, health-related quality of life and quality-adjusted life years

The cost analysis will take a health services perspective and will measure the costs of delivering the CRRT (see Micro-costing task analysis and CRRT set-up time and frequency), ICU and hospital length of stay and RRT for ESRD (see The UK Renal Registry). For the cost of RRT for ESRD, patients identified from UKRR as receiving dialysis will have their costs estimated dependent on their mode of renal replacement therapy and time to transplant where applicable. Unit costs of ICU/hospital length of stay and dialysis will be obtained from the NHS Reference Costs 2015–16.¹⁶ The costs analysis will calculate total costs per patient up to one year since ICU admission.

EuroQuol EQ-5D-3L health-related quality of life (HRQoL) data for patients at three months and one year after ICU discharge will be obtained from the 8000 patient Intensive Care Outcome Network Study (ICON) study.¹⁷ Eligible patients meeting the inclusion criteria will be identified and divided into quartiles of age. Averaged EQ-5D-based utility weights by quartile at three months and one year will be calculated. These weights will be used as the measure of HRQoL. All patients developing ESRD and requiring dialysis will be assigned an appropriate utility weight based on European norms¹⁸ from the date of first RRT for ESRD forward. HRQoL at three months and one year will be combined with the survival data to calculate QALYs at one year.

Statistics and data analysis

Statistical analysis has been pre-specified in a statistical analysis plan which is publicly available on the ICNARC website.

Power calculation

Based on preliminary analysis of CMP data, we anticipate a total available sample size of approximately 85,000 patients from 184 ICUs. Prior to study commencement, UK suppliers indicated that 90 ICUs were currently using RCA. To assess the likely power of the available data to address the research question of interest, we simulated 1000 replications of the study using available CMP data under the following assumptions:

Thirty-five changes from SHA to RCA will be observed within the available data. This is a conservative assumption from the 90 ICUs across the UK reported to be using RCA, to allow for use in ICUs outside England, specialist ICUs and changes that occurred when ICUs were not participating in the CMP. In each simulation, 35 ICUs were selected at random to represent the observed changes.

Changes from SHA to RCA will be evenly distributed over the time period of the study. In the simulations, the changeover quarter for each of the 35 randomly selected ICUs was sampled from a uniform distribution from between their second and penultimate quarters.

Fifteen ICUs will have changed from SHA to RCA prior to the start of the study. In each simulation, 15 ICUs were selected at random to contribute data to the RCA group throughout. In the simulations, the indicator t_ij is used to indicate ICU i was using RCA in quarter j.

The distribution of risk of 90-day mortality for patients receiving RRT in UK ICUs will follow that of the ICNARC _H _-2015 model for acute hospital mortality in critical care (with mean 50%). This model has excellent discrimination (area under the receiver operating characteristic curve ∼0.9) and calibration in this population. In the simulation, the patient-level risk of death for patient k admitted in quarter j to ICU i, p_ijk, was calculated using this model.

The between ICU standard deviation for 90-day mortality will be 0.22. This value was estimated as the observed value for risk-adjusted acute hospital mortality in the CMP among patients receiving RRT and corresponds to an intraclass correlation coefficient (ICC) of 0.015. In each simulation, an ICU-level effect for ICU i, u_i, was sampled from a normal distribution with mean 0 and standard deviation 0.22. For the purpose of the simulations, no clustering of observations for patients within quarters in the same ICU was assumed.

Changing from SHA to RCA will be associated with an odds ratio for 90-day mortality of 0.9. For the purpose of simulation, only a change in level was considered with no change in slope.

In each simulation, the ‘observed’ outcome for each patient, y_ijk, was sampled from a Bernoulli distribution based on the following model:

logit (y_{ijk}) \sim logit (p_{ijk}) + ln (0.9) \times t_{ij} + u_{i}

The estimated treatment effect within each simulation was then estimated using a multilevel logistic regression with robust standard errors. Simulations were undertaken using Stata/SE version 14.2 (StataCorp LP, College Station, TX). The random number seed was set prior to analysis to ensure reproducibility of results.

The results of the simulations show this sample will have approximately 81% power (p<0.05) to detect a step change in 90-day mortality corresponding to an odds ratio of 0.9.

Analysis of clinical effectiveness

The analysis will use interrupted time series (ITS) analysis, where the interruption corresponds to the change from SHA to RCA for CRRT and will follow the eight quality criteria for ITS design and analysis described by Ramsay et al.¹⁹

Random effects multilevel generalised linear models (patients nested within time periods (quarters) nested within ICUs) will be used to estimate the ICU-level effect of transitioning to RCA on trends in patient-level outcomes. Logistic models will be used for binary outcomes and linear models will be used for continuous outcomes. The study will include periods both before and after the switch from SHA to RCA in individual units and a comparator group of ICUs that did not change treatment. The effect estimate will be the within-ICU change in trends with the control ICUs primarily improving estimates of patient-level confounders and underlying secular trend. Models will be fitted with robust standard errors to allow for model misspecification, including autocorrelation and heteroscedasticity. Doubly robust approaches will be considered should concerns about model misclassification arise.

The primary impact model for the effect of the change from SHA to RCA will allow for both a change in level and in slope. Linear trends will be assumed in both the pre-intervention and post-intervention periods. The quarter of data in which the change from SHA to RCA took place will be omitted from the model to allow for potential imprecision in the reporting of the time of change and time to transition from one modality to the other. Transition times will be collected in the initial survey. Where longer transition times occurred, these will be accounted for by excluding the corresponding window. If transition is reported to have taken more than a quarter in over 20% of participating units, we will amend the length of omitted time for all units accordingly. The potential for lagged and temporary effects will be explored in sensitivity analyses. The results of the regression models will be reported as the odds ratio (or for continuous outcomes, difference in means) with 95% confidence interval for the change in level and the odds ratio per year (difference in means per year) with 95% confidence interval for the change in slope associated with the change from SHA to RCA. The overall significance of the change from SHA to RCA will be assessed by the joint test of the two parameters for the change in level and change in slope.

Handling of missing data

Any ICUs for which it is not possible to establish whether/when a change from SHA to RCA for CRRT occurred will be excluded from the analysis. Missing values in individual patient covariates will be imputed using fully conditional specification implemented using the multivariate imputation by chained equations (MICE) algorithm.^20,21 The multiple imputation model will include all covariates planned to be included in the substantive model, plus the intervention and outcome measures.²² To ensure reproducibility of results, the random number seed will be set prior to producing the imputed datasets.

Cost effectiveness

The cost-effectiveness analysis (CEA) will report mean (95% confidence interval) incremental costs, and QALYs at one year associated with a change from SHA RCA for CRRT, overall and for pre-specified subgroups. The CEA will use multilevel generalised linear models that allow for clustering of patients in sites including random effects for both level and slope. Incremental net monetary benefits (INB) at one year associated with a change from SHA to RCA will be estimated valuing incremental QALYs according to a NICE recommended threshold willingness-to-pay for a QALY gain (£20,000) and subtract from this the incremental costs. Missing data will be addressed using the same approach followed for the primary clinical endpoints, assuming data are missing at random conditional on baseline covariates, resource use and observed endpoints.

The economic analysis will also project lifetime cost-effectiveness by encapsulating the relative effects of the alternative strategies on long-term survival and HRQoL, combining extrapolations from the patient survival data, with external evidence on long-term survival and HRQoL. We will consider alternative parametric extrapolation and choose the model based on model fit and plausibility when compared with age-gender-matched general population survival. Survival will then be extrapolated according to chosen parametric function for the duration of years that parametric curves predict excess mortality compared to age-gender-matched general population, after which we will assume that all cause death rates were those of the age-gender-matched general population. We will project lifetime costs by applying morbidity costs estimated at one year over the period of excess mortality. Sensitivity analyses will test whether the results are robust to methodological assumptions (e.g. lagged and temporary effects in the regression model (see clinical effectiveness analysis), extrapolation approach, and alternative HRQoL assumptions from published literature).

Study oversight and ethics

This study has received a favourable opinion in ethical review (REC reference: 18/SC/0204) and Section 251 approval (CAG reference: 18/CAG/0070) for the data collection and linkage as described.

Discussion

If RCA is found to be both clinically superior to SHA and less costly, it is likely the change from SHA to RCA in ICUs will continue to occur and may even accelerate. If RCA is clinically superior to SHA but more costly, the effect on uptake will depend on the cost of the clinical benefit gained, but it is likely the change to RCA will continue unless the cost to obtain the benefit is high. Should SHA be more clinically and cost-effective, the adoption of RCA should slow or stop. However, individual ICUs will have invested funds in the change from SHA to RCA, and individual clinicians will have championed the change, so reversal of the process would be likely to take time.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study is funded by the NIHR Health Technology Assessment (NIHR-HTA) Programme (Project No. 16/111/136). This publication presents independent research funded by the National Institute for Health Research (NIHR). The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care.

ORCID iDs

Doug W Gould

Mark Borthwick

David A Harrison

References

ICNARC. Case Mix Programme Summary Statistics 2017-2018. 2018, p, www.icnarc.org/Our-Audit/Audits/Cmp/Reports/Summary-Statistics (accessed 23 July 2019).

Rabindranath

Adams

Macleod

, et al. Intermittent versus continuous renal replacement therapy for acute renal failure in adults. Cochrane Database Syst Rev 2007; 18: Cd003773.

Jones

Devonald

. How acute kidney injury is investigated and managed in UK intensive care units – a survey of current practice. Nephrol Dial Transplant 2013; 28: 1186–1190.

Brandenburger

Dimski

Slowinski

, et al. Renal replacement therapy and anticoagulation. Best Pract Res Clin Anaesth 2017; 31: 387–401.

Uchino

Bellomo

Morimatsu

, et al. Continuous renal replacement therapy: a worldwide practice survey. The beginning and ending supportive therapy for the kidney (B.E.S.T. kidney) investigators. Intensive Care Med 2007; 33: 1563–1570.

Oudemans-van Straaten

Kellum

Bellomo

. Clinical review: anticoagulation for continuous renal replacement therapy – heparin or citrate? Crit Care 2011; 15: 202.

Morabito

Pistolesi

Tritapepe

, et al. Regional citrate anticoagulation for RRTs in critically ill patients with AKI. Clin J Am Soc Nephrol 2014; 9: 2173–2188.

Kontopantelis

Doran

Springate

, et al. Regression based quasi-experimental approach when randomisation is not an option: interrupted time series analysis. BMJ 2015; 350: h2750.

Wagner

Soumerai

Zhang

, et al. Segmented regression analysis of interrupted time series studies in medication use research. J Clin Pharm Ther 2002; 27: 299–309.

10.

Information Standards Board for Health and Social Care. Critical Care Minimum Data Set Overview, www.datadictionary.nhs.uk/data_dictionary/messages/supporting_data_sets/data_sets/critical_care_minimum_data_set_fr.asp?shownav=1 (accessed 23 July 2019).

11.

Shankar-Hari

Phillips

Levy

, et al. Developing a new definition and assessing new clinical criteria for septic shock: for the third international consensus definitions for sepsis and septic shock (sepsis-3). JAMA 2016; 315: 775–787.

12.

Herbert

Wijlaars

Zylbersztejn

, et al. Data resource profile: hospital episode statistics admitted patient care (HES APC). Int J Epidemiol 2017; 46: 1093–1093i.

13.

Lefrant

Garrigues

Pribil

, et al. The daily cost of ICU patients: a micro-costing study in 23 French Intensive Care Units. Anaesth Crit Care Pain Med 2015; 34: 151–157.

14.

NHS Business Services Authority Drug Tariff 2019, www.nhsbsa.nhs.uk/pharmacies-gp-practices-and-appliance-contractors/drug-tariff (accessed 19 September 2019).

15.

NHS Supply Chain 2019, www.supplychain.nhs.uk/ (accessed 19 September 2019).

16.

Department of Health. Department of Health. NHS reference costs. 2016.

17.

Griffiths

Morgan

Barber

, et al. Study protocol: the Intensive Care Outcome Network (‘ICON') study. BMC Health Serv Res 2008; 8: 132.

18.

Eriksson

Goldsmith

Teitsson

, et al. Cross-sectional survey in CKD patients across Europe describing the association between quality of life and anaemia. BMC Nephrol 2016; 17: 97.

19.

Ramsay

Matowe

Grilli

, et al. Interrupted time series designs in health technology assessment: lessons from two systematic reviews of behavior change strategies. Int J Technol Assess Health Care 2003; 19: 613–623.

20.

Bartlett

Seaman

White

, et al. Multiple imputation of covariates by fully conditional specification: accommodating the substantive model. Stat Methods Med Res 2014; 24: 462–487.

21.

White

Royston

Wood

. Multiple imputation using chained equations: issues and guidance for practice. Stat Med 2011; 30: 377–399.

22.

Moons

Donders

Stijnen

, et al. Using the outcome for imputation of missing predictor values was preferred. J Clin Epidemiol 2006; 59: 1092–1101.