Community prescribing of potentially nephrotoxic drugs and risk of acute kidney injury requiring renal replacement therapy in critically ill adults: A national cohort study

Abstract

Background

Acute kidney injury demonstrates a high incidence in critically ill populations, with many requiring renal replacement therapy. Patients may be at increased risk of acute kidney injury if prescribed certain potentially nephrotoxic medications. We aimed to evaluate this association in ICU survivors.

Methods

Study design – secondary analysis of national cohort of ICU survivors to hospital discharge linked to Scottish healthcare datasets. Outcomes: primary – renal replacement therapy in ICU; secondary – early acute kidney injury (calculated using urine output and relative change from estimated baseline serum creatinine within first 24 h of ICU admission using modified-RIFLE criteria). Primary exposure: pre-admission community prescribing of at least one potential nephrotoxin: angiotensin-converting-enzyme inhibitors/angiotensin-receptor blockers, diuretics or nonsteroidal anti-inflammatory drugs. Statistical analyses: unadjusted associations – univariable logistic regression; confounder adjusted: multivariable logistic regression.

Results

During 2011–2013, 12,838 of 23,116 patients (55.5%) were prescribed at least one community prescription of at least one nephrotoxin; 1330 (5.8%) patients received renal replacement therapy; 3061 (15.7%) had acute kidney injury. Patients exposed to at least one examined nephrotoxin experienced higher incidence of renal replacement therapy (6.8% vs 4.5%; adjOR 1.46, 95%CI 1.24, 1.72, p < 0.001) and acute kidney injury (19.8% vs 10.9%; adjOR 1.61, 1.44, 1.80, p < 0.001). Increased risk of RRT was also found for angiotensin-converting-enzyme inhibitors/angiotensin-receptor blockers (adjOR 1.65, 1.40, 1.94), non-steroidal anti-inflammatory drugs (adjOR 1.12, 1.02, 1.44) and diuretics (adjOR 1.35, 1.14, 1.59).

Conclusions

Community prescribing of potential nephrotoxins increases the risk of renal replacement therapy/early acute kidney injury in ICU populations. Analyses were limited by the survivor dataset and potential residual confounding. Findings add consistency to previous research improving understanding of the harmful potential of these important medications and their timely cessation in acute illness.

Keywords

Renal replacement therapy critical care nephrotoxin community prescribing acute kidney injury

Introduction

Acute kidney Injury (AKI) is a condition typified by a sudden loss of kidney function demonstrating high incidence during hospitalisation in critically ill populations (67%) and continues to increase nationally.¹ It is associated with high mortality of 20–40% in ICU populations^2–5 and may necessitate renal replacement therapy (RRT).⁶ The National Institute for Health and Care Excellence (NICE) estimates that AKI costs the NHS between £434 and 620 million annually, higher than breast, skin and lung cancer combined.⁷

Nephrotoxic medications are the suspected aetiology in approximately 20% of cases, rising to 66% in elderly populations.^8–10 This is of interest, as through anticipation and early intervention, drug-induced AKI should be largely preventable.

Due to extensive use,¹¹ three classes of potentially nephrotoxic medications are of particular interest: angiotensin-converting-enzyme inhibitors (ACE-i) and angiotensin-II receptor blockers (ARBs), diuretics, and non-steroidal anti-inflammatory drugs (NSAIDs). The Medicine Sick Day Rules Card scheme by the Scottish Patient Safety Programme (SPSP) aimed to raise patient and clinician awareness of the risks of these medicines within primary care.¹² The mechanisms of nephrotoxicity of these medicines have been examined in the literature. Diuretics increase susceptibility to insults through hypovolaemia; ACE-i/ARBs cause efferent arteriolar vasodilation causing intraglomerular haemodynamic reduction; whereas, NSAIDs cause vasoconstriction of the afferent renal arterioles.^13–16

Strong evidence exists for an increased association of AKI with both NSAIDs and “triple-therapy” (the concurrent use of ACE-i/ARBs, diuretics and NSAIDs).^15,17–19 However, conclusive evidence does not exist for ACE-I/ARBs,^20–22 diuretics²³ or “dual-therapy.”^18,19,24,25 Furthermore, prior populations were poorly defined or from primary care databases without clear temporality for reported AKI. Particularly noticeable is the lack of ICU-specific research, the setting for management of the sickest patients and costly interventions such as RRT.

We aimed to assess the association of AKI requiring RRT with the community prescribing of ACE-i/ARBs, diuretics and NSAIDs in critically ill patients across Scotland. These drugs were examined both alone and in combination.

Methods

Study design, setting and databases

We undertook a retrospective population-based cohort study using linked Scottish healthcare datasets. This comprised a complete, national three-year cohort of patients admitted to Scottish adult general ICUs (01 Jan 2011 to 31 Dec 2013) from the Scottish Intensive Care Society Audit Group (SICSAG) database, linked to national hospital records (SMR01), death records and the Scottish community prescribing database. The SICSAG database routinely and prospectively collects detailed information for audit and research. Data undergo checks for completeness and validity including at data entry, periodic case-note validations and central validation.²⁶ The study was a secondary-analysis of a subset of data used in the PROFILE study.²⁷ The PROFILE study aimed to identify drivers for unplanned readmissions to hospital in a survivor cohort of ICU patients. The PROFILE cohort comprised patients admitted during 2005–2013. Prescribing data were only complete from 01 Jan 2011 onwards. Analyses were restricted to these patients.

The Scottish community prescribing database contains all items dispensed in Scotland by community pharmacy or dispensing doctors. It does not include medicines prescribed and dispensed within hospital.

As the prescribing database includes only community-dispensed medication, patients who were prescribed medicines but did not have them dispensed were classified as unexposed, reducing exposure misclassification.

Appropriate permissions for data were achieved including SICSAG Steering Committee and Public Benefit and Privacy Panel for Health and Social Care (PBPP²⁸). Electronic data storage and analysis took place within the NHS National Services Scotland safe haven.

Participants

Participants included all patients admitted to Scottish adult general ICUs during the study period who survived to hospital discharge. Patients were excluded if they were as follows:

receiving chronic RRT pre-admission, defined as a patient undergoing haemodialysis, haemofiltration or peritoneal dialysis for end-stage renal disease (ESRD) on admission to ICU.²⁹

missing data for primary outcome (receipt of RRT).

missing a unique patient identifier for linkage to corresponding Scottish prescribing information records.

Attempt was made to obtain data for the complete cohort including non-survivors, but financial restrictions made this unfeasible.

Variables

Outcomes

The primary outcome was receipt of RRT at any point during ICU stay identified by SICSAG database coding. The secondary outcome was early AKI derived from urine output and change in serum creatinine relative to an estimated baseline during the first 24 h of ICU admission, using modified-RIFLE criteria defined by the Australian and New Zealand Intensive Care Society (ANZICS).³⁰ This definition has been previously used in studies using UK ICU data.³⁰ This outcome captures patients who did not receive RRT. Measurements used for serum creatinine and urine output are recorded from the initial 24 h of ICU admission. The ANZICS definition was modified to include patients receiving RRT within the initial 24 h of ICU admission in the failure category. All outcome variables were from the SICSAG database. All derivations are detailed (eFigures 1 to 3).

Exposures

Exposures were community prescription of ACE-I/ARBs, diuretics or NSAIDs on admission from the Scottish community prescribing database. The primary exposure was community-prescribing of at least one of any examined nephrotoxin (i.e. the primary exposed drug class listed may be in combination with exposure to an additional class – for example, NSAIDs = Participants exposed to at least an NSAID +/− additional drug class). In additional analyses, drugs were assessed alone as monotherapy, and in combination. Identification was by British National Formulary (BNF) chapter/subchapter (ACE-i and ARBs: 2.5.5; Diuretics: 2.2; NSAIDs: 10.1.1).³¹

Potential confounders

Potential confounders were identified from the SICSAG and SMR01 databases consisting of age, sex, socioeconomic status (Scottish Index of Multiple Deprivation³²), previous number of hospital admissions, severity-of-disease (Acute Physiology and Chronic Health Evaluation II [APACHE-II] score), ICU admission type (elective surgery, emergency surgery, non-operative), number of comorbidities and receipt of cardiopulmonary resuscitation prior to admission. These were adjusted for during analysis. Potential unmeasured confounders include iodinated contrast agents, dehydration status, urological obstructive disorders and/or medicines prescribed during the hospital admission.

Statistical methods

Analyses were undertaken using Stata MP 14.1 (StataCorp LP, College Station, TX). Baseline characteristics for both exposed and non-exposed groups were described using number and percentage, or median and interquartile range (IQR). Univariable associations involving exposures, outcomes and baseline variables of interest were assessed using Chi-square or Wilcoxon rank-sum tests. Unadjusted results were expressed as odds ratio derived from univariable logistic regression. Adjustment for potential confounders was by multivariable logistic regression. Appropriate model diagnostics demonstrated no significant deviations.³³ Continuous variables (age and APACHE II score) were evaluated for linearity assumptions. APACHE II was included in models as a continuous term. Age violated linearity assumptions and was entered as a categorical term (10 categories). We evaluated model diagnostics for the logistic regression models (values for primary analysis in brackets) reporting c-index (0.907), Brier’s Score (0.042) and inspecting calibration plots of observed vs predicted values in deciles, which demonstrated good calibration across the range of predicted risk.

Sensitivity analyses

As the dataset was restricted to a cohort of ICU patients who survived to hospital discharge, we undertook a series of sensitivity analyses to assess the robustness of findings to the inclusion of a hypothetical non-survivor cohort. We reported the unadjusted odds ratios for this combined (survivor + non-survivor) cohort for exposure-outcome (RRT) associations. Four parameters were required to conduct these analyses: (1) the size of the non-survivor cohort (derived from the hospital mortality rate of all patients admitted to ICU); (2) the prevalence of RRT in the non-survivor cohort; (3) the prevalence of community prescriptions of the examined drugs and (4) the association between community prescriptions of the examined drugs and RRT. We obtained an estimate of hospital mortality rate of 20%, and RRT prevalence in this non-survivor cohort (25%) from SICSAG for 2011–2013 (SICSAG, personal communication). As the prevalence of community prescriptions of ACE-I/ARBs, diuretics and NSAIDS was unknown in non-survivors, we modelled a range of prevalances 25–50% greater than and less than the prevalence in the survivor cohort. In addition, the association between community prescription of the examined drugs and RRT was unknown. We therefore modelled a range of risk ratios, varying from 0.5 (i.e. a 50% reduction in relative risk of RRT) and 3.0 (a 3-fold increase in relative risk of RRT).

Results

During the three-year study period, 24,504 patients were admitted to ICU and survived to hospital discharge. After exclusion criteria, 23,116 (94.3%) remained (Figure 1). Secondary outcome analysis of early AKI defined by RIFLE classification was completed for patients with complete biochemistry and length of stay data (n = 19,509; 79.6%).

Figure 1.

Patient flow diagram. Flow diagram describing reasons for exclusion from primary and secondary analyses. RRT: Renal replacement therapy.

Baseline characteristics

Baselines characteristics for the cohort are shown in Table 1 (extended version: eTable 1). Patients were stratified by those exposed to any of the three examined drugs, and those not exposed. The exposed group was significantly older, had a higher proportion of females, had greater previous admissions, were more likely to have been admitted after elective surgery and had higher illness severity scores. Additionally, they had increased polypharmacy and greater overall comorbidity – notably cardiovascular disease, peripheral vascular disease, diabetes mellitus and renal disease.

Table 1.

Baseline characteristics summary. A summary of key demographic and clinical baseline characteristics including total numbers examined stratified by prescription of at least any one of an ACE-i/ARB, diuretic, or NSAID. Extended descriptive characteristics are presented in supplementary Appendix (eTable 1).

Baseline characteristics	Total	Any community nephrotoxin prescription			p Value
Baseline characteristics	Total	No	Yes	Total	p Value
Total		10,278 (44.5%)	12,838 (55.5%)	23,116
Demographics
Sex (Male)	23,116	6075 (59.1%)	7,169 (55.8%)	13,289 (57.3%)	<0.001
Age (Median [IQR])	23,116	53 (38, 66)	66 (55, 74)	62 (47, 72)	<0.001
SIMD Quintile	23,094				0.002
1		2747 (26.8%)	3,239 (25.3%)	5999 (25.9%)
2		2256 (22.0%)	2,989 (23.3%)	5258 (22.7%)
3		1982 (19.3%)	2,601 (20.3%)	4597 (19.9%)
4		1869 (18.2%)	2,195 (17.1%)	4082 (17.6%)
5		1410 (13.7%)	1,806 (14.1%)	3226 (13.9%)
Comorbidity
Number of prescribed drugs [Median (IQR)]	23,116	5 (2, 10)	12 (8, 17)	9 (5, 15)	<0.001
Number of comorbidities	23,116				<0.001
0		5185 (50.5%)	4,578 (35.66%)	9763 (42.2%)
1		3264 (31.8%)	4,008 (31.22%)	7272 (31.5%)
2		1229 (12.0%)	2,291 (17.9%)	3520 (15.2%)
3+		600 (5.8%)	1,961 (15.3%)	2561 (11.1%)
Admissions
Previous hospital admissions [Median (IQR)]	23,116	0 (0, 1)	1 (0, 2)	1 (0, 2)	<0.001
ICU
Admission type	23,184
Elective surgery		3051 (29.9%)	5,433 (42.7%)	8484 (37.0%)	<0.001
Emergency surgery		2148 (21.0%)	2,479 (19.5%)	4627 (20.1%)	<0.001
Non-operative		5018 (49.1%)	4,808 (37.8%)	9826 (42.8%)	<0.001
Severity of illness scores [Median (IQR)]
APACHE II	20,088	13 (9, 18)	15 (11, 20)	14 (10, 19)	<0.001
APS	20,077	10 (7, 15)	11 (7, 15)	10 (7, 15)	<0.001
Length of stay [Median (IQR)]	23,116
ICU stay		2 (1, 4)	2 (1, 4)	2 (1, 4)	<0.001
Pre-ICU stay		1 (0, 1)	1 (0, 2)	1 (0, 2)	<0.001
Post-ICU stay		7 (3, 16)	8 (4, 18)	8 (4, 17)	<0.001
Total hospital stay		12 (6, 24)	14 (8, 26)	13 (7, 26)	<0.001

ACE-i: Angiotensin-converting-enzyme inhibitor; APACHE II: Acute Physiologic Assessment and Chronic Health Evaluation II Scoring System; APS: Acute Physiological Score; ARB: Angiotensin-II receptor blocker; NSAID: Non-steroidal anti-inflammatory drug; SIMD: Scottish Index of Multiple Deprivation. Data presented as n (%) unless otherwise stated.

Exposures and outcomes

Over the three-year study period, 12,838 of 23,116 patients (55.5%) examined were prescribed at least one examined nephrotoxin (Table 2). Within this population, 4733 (20.5%), 5961 (25.8%) and 7637 (33.0%) were prescribed NSAIDs, diuretics and ACE-i/ARBs, respectively. Monotherapy and combinations are detailed in eTable 2. RRT was received by 5.8% of patients (1330/23,116). Biochemically-defined AKI was evident in 15.7% patients (3061/19,509) (Table 2).

Table 2.

Association between primary exposures (at least one potential nephrotoxin) and outcomes. Exposed drug class listed may be in combination with exposure to an additional class – for example, NSAIDs = participants exposed to at least an NSAID +/- additional drug class.

		Outcome
		RRT			Early AKI
Exposure	Number	Exposed	Not exposed	p Value^a	Exposed	Not exposed	p Value^a
Any	12,838 (55.5%)	872 (6.8%)	458 (4.5%)	<0.001	2094 (19.8%)	967 (10.9%)	<0.001
NSAID	4733 (20.5%)	300 (6.3%)	1030 (5.6%)	0.053	610 (15.0%)	2451 (15.9%)	0.168
Diuretic	5961 (25.8%)	453 (7.6%)	877 (5.1%)	<0.001	1171 (24.0%)	1890 (12.9%)	<0.001
ACE-i/ARB	7637 (33.0%)	605 (7.9%)	725 (4.7%)	<0.001	1433 (23.2%)	1628 (12.2%)	<0.001

ACE-i: Angiotensin-converting-enzyme inhibitor; early AKI: acute kidney injury occurring within 24 h of ICU admission; ARB: Angiotensin-II receptor blocker; NSAID: Non-steroidal Anti-inflammatory drug; RRT: renal replacement therapy.

p Value represents unadjusted analyses.

Main results

A community prescription for any of the three examined drugs was associated with increased RRT (6.8% vs 4.5%, p < 0.001), persisting after adjustment (adjOR 1.46, 95%CI 1.24, 1.72, p < 0.001) (Figure 2).

Figure 2.

Forest plot of association between primary exposures (at least one potential nephrotoxin) and outcomes: unadjusted and adjusted results. Exposed drug class listed may be in combination with exposure to an additional class – for example, NSAIDs = Participants exposed to at least an NSAID +/− additional drug class. ACE-i: angiotensin-converting-enzyme inhibitor; AKI: early acute kidney injury occurring within 24 hours of ICU admission; ARB: angiotensin-II receptor blockers; NSAID: non-steroidal anti-inflammatory drug; OR: odds ratio; RRT: renal replacement therapy; 95% CI: 95% confidence interval.

Significant univariable associations were found between RRT and those prescribed a diuretic (7.6% vs 5.1%, p < 0.001) or ACE-i/ARB (7.9% vs 4.7%, p < 0.001), but not for NSAIDs (6.34% vs 5.60%, p = 0.053). After adjustment, prescription of each of the three examined drugs was significantly associated with increased RRT (Figure 2).

A community prescription for any of the three examined drugs was associated with increased risk of early AKI (19.8% vs 10.9%, p < 0.001), which persisted after adjustment (adjOR 1.61, 95%CI 1.44,1.80, p < 0.001) (Figure 2).

After adjustment for confounders, increased risk of early AKI was associated with patients prescribed a diuretic (adjOR 1.52, 95%CI 1.35,1.70, p < 0.001) or ACE-i/ARB (adjOR 1.75, 95%CI 1.56, 1.95, p < 0.001) but not NSAIDs (adjOR 0.96, 95%CI 0.85, 1.08, p = 0.467) (Figure 2).

Additional analyses

Monotherapy

Monotherapy with each of the three examined drugs was not associated with RRT. Monotherapy with diuretics was associated with reduced RRT. Similar associations were found after adjustment (eFigure 4).

Early AKI was significantly associated with monotherapy for each of the examined drugs, showing an increased risk of AKI for diuretics and ACE-i/ARBs but a reduced risk of AKI with NSAIDs. After adjustment, associations for NSAIDs and ACE-i/ARBs remained significant (eFigure 4).

Dual-therapy

The combinations of an ACE-i/ARB plus an NSAID and an ACE-i/ARB plus a diuretic were associated with increased RRT. After adjustment, dual-therapy with an ACE-i/ARB plus diuretic remained significant (eFigure 4).

Increased risk of early AKI was found for dual-therapy with ACE-i/ARBs plus an NSAID and ACE-i/ARBs plus a diuretic. After adjustment, dual-therapy with an ACE-i/ARB plus diuretic remained significant (eFigure 4).

Triple-therapy

Of the 655 patients prescribed triple-therapy, almost 1 in 10 required RRT (n = 63, 9.6%). Triple-therapy was associated with increased RRT, persisting after adjustment (eFigure 4).

Additionally, almost 1 in 4 of patients prescribed triple-therapy developed AKI (n = 136, 24.2%). Triple-therapy was associated with an increased risk of early AKI, remaining after adjustment (eFigure 4).

Sensitivity analyses

Under a series of scenarios for a hypothetical non-surviving ICU cohort, the sensitivity analyses demonstrated that the unadjusted association between any of the three examined drugs and RRT was likely to be underestimated, ranging from 1.61 to 2.78 (eFigure 5). This assumed that the prevalence of potential nephrotoxic drug prescription was at least as common in the non-survivor group as the survivor group (i.e. at least 55.5%), and the association between exposure and outcome was at least as large (OR at least 1.56, equivalent to RR 1.52). We demonstrated similar findings for associations for each of the three drug classes individually and RRT (eFigure 5).

Discussion

ICU survivors receiving a community prescription pre-admission for any one of ACE-i/ARBs, NSAIDs or diuretics experienced a higher incidence of both RRT and early AKI during their ICU stay. After adjustment, this equated to a 46% and 61% increase in odds of RRT and AKI, respectively. Increased risk of RRT and early AKI was associated with community prescription of at least one of ACE-i/ARBs or diuretics. Prescription of at least an NSAID was associated with increased risk of RRT but not AKI. These associations need to be interpreted in the context of a survivor cohort and the likelihood of residual confounding; however, they add to increasing evidence of potential harm associated with these drugs in acute illness.

Strengths and limitations

We used a complete, national cohort comprising all ICU survivors during a three-year period linked to community prescribing data with population coverage. Furthermore, we comprehensively evaluated all three chosen nephrotoxic drugs, both alone and in combination. However, the dataset did not describe compliance, length of exposure, and over-the-counter self-medication. Additionally, prescribing patterns may have changed in the last five years from when data was collected. However, reassuringly, prescribing rates of the drug classes has remained consistent from community pharmacy remuneration data.³⁴

As outcomes are defined objectively with strict definitions, receipt of RRT and biochemistry, the risk of detection or misclassification bias was minimised. The secondary outcome was chosen as, despite RRT being a useful surrogate for AKI and a costly intervention, it has been estimated as not received by 48% of AKI patients.³⁰ Prior research has also suggested use of biochemistry in identifying disease omitted by hospital coding.^19,35 Whilst useful, this biochemically-defined outcome assumes normal renal function pre-admission and likely overestimates AKI. In addition, creatinine and urine output measurements were only available for the first 24 h of ICU stay, potentially underestimating AKI if this developed after 24 h.

This study had a number of important limitations. The association between the examined drugs and RRT/early AKI is likely to be confounded. However, we were only able to adjust these associations using the limited range of potential confounders recorded in the database. This means that the likelihood of residual confounding remains, and the findings should be viewed as hypothesis generating rather than causal. Whilst randomisation would not be feasible to evaluate a causal relationship between the exposures and outcomes in this study, access to databases with more detailed recording of potential confounders and prospective study designs may reduce the impact of residual confounding on these associations.

Data relating to CKD or pre-admission creatinine were unavailable for the cohort. However, patients with end-stage renal disease requiring RRT were coded in the dataset and could therefore be excluded from the cohort. ACE-i/ARBs and diuretics are frequently prescribed to patients with CKD and stratifying analyses by CKD-status may have provided valuable additional insights.

Notably, the overall provision of RRT across Scottish ICUs decreased during the study period from 12% to 9%, from 2011 to 2013, respectively.³⁶ This may suggest changes in culture regarding the decision to administer RRT and was consistent with our data.

It is important to note that the cohort excluded patients that died during ICU admission. It would be reasonable to assume that those receiving RRT or with AKI would be at a greater risk of mortality.⁶ Patients with an increased risk of mortality would likely have greater comorbidity, and consequently be prescribed more medications. Therefore, it is likely the associations demonstrated in this survivor cohort would underestimate the true association in a cohort comprising survivors and non-survivors, which may explain the lack of association or negative association seen for NSAIDs and diuretics. Crude hospital mortality rates across Scottish hospitals for ICU patients during the study period were 21.8%, 19.6% and 19.0% for 2011, 2012 and 2013, respectively.^29,36,37

This is a major limitation of the study. We conducted sensitivity analyses to provide a plausible range of values for the “true” association between exposure and outcome in a cohort of ICU survivors and non-survivors.

We were unable to report if medications had been discontinued during hospital admission. This would have reduced the potential nephrotoxic effects in the exposed cohort, biasing associations towards the null. However, the short pre-ICU hospital length of stay (Table 1) reduces the likelihood that drugs would have been stopped, or if stopped, would be unlikely to have reached sufficiently low concentrations to discount influence.

Interpretation

This research contributes to evidence that patients exposed to community prescribing of ACE-i/ARBs, NSAIDs and/or diuretics are at increased risk of requiring RRT and/or developing an early AKI in ICU. An increased risk persisted when ACE-i/ARBs were prescribed in combination with either diuretics or NSAIDs, and when all three drugs were prescribed concomitantly.

We found a previously undescribed association between RRT and early AKI with community prescribing of any prescribed number of the examined nephrotoxins – justifying the drugs selected in the Scottish Patient Safety Programme “Sick Day Rules” campaign.¹² Furthermore, we found that each of the drug classes examined were individually associated with increased risk of RRT – and all, except NSAIDs, were significantly associated with increased biochemical AKI.

Lafrance and Miller²⁴ and Huerta et al.²⁵ assessed NSAIDs individually and in combination – finding increased risk of AKI with concomitant diuretic use. Dreischulte et al.¹⁹ found increased risk of AKI for NSAIDs plus diuretics or ACE-i/ARBs. However, Lapi¹⁸ found no association with AKI when examining NSAIDs in both combinations. Interestingly, the described associations involving NSAIDs and ACE-i/ARBs or diuretics^19,24,25 were not found in this study. In fact, unexpectedly, NSAIDs were associated with reduced risk of early AKI. NSAIDs are not essential medications for survival and may have been prescribed more cautiously in higher risk patients. NSAIDs were also prescribed in a younger and less comorbid population – however, these factors were adjusted for in the analyses. An alternate hypothesis is that these patients had higher mortality and were therefore not present in the survivor cohort.

We elicited a significant association between both RRT and early AKI with concomitant community prescribing of ACE-i/ARBs and diuretics. This association appears to have been unexamined in prior studies, highlighting a potentially understated risk in community prescribing.

A decreased association with risk of RRT and AKI was found for diuretics when prescribed alone (eFigure4), compared to when evaluated with potential for co-exposure to other examined drugs (Figure 2). This may be due to an additive risk in the presence of other examined drugs – particularly, ACE-i/ARBs, where a relatively large association was found for RRT/AKI and the co-prescribing of both drugs.

This study further demonstrated the “triple whammy” effect coined by Thomas,¹⁷ which proposed a negative synergistic effect with concurrent ACE-i/ARBs, diuretics and NSAIDs. This association between triple-therapy and increased risk of AKI has been elicited using various study designs and AKI definitions.^18,19,38

Generalisability

The inclusive eligibility criteria and the population-based nature of the study minimise selection bias, improving external validity. Additionally, unlike some prior studies,^18,19 patients with pre-existing renal disease (not requiring RRT) or history of AKI were included. These are an important patient group with increased risk of RRT. However, generalisability beyond the ICU-survivor population is difficult. Additionally, the national population has risk of potential residual confounding due to regional and inter-site variability in practice and population.

Implications

An ideal future dataset would include baseline creatinine data to improve the accuracy of classifying AKI, and contain additional confounder information such as contrast agent exposure, dehydration and hospital-prescribing. Additionally, more detailed prescribing information would allow evaluation of treatment dose and duration effects, and differences within drug classes. The study demonstrates novel linkage of national prescribing and ICU datasets, serving as proof of concept for further work. This type of linkage has significant potential, with increasing availability of data suitable for this approach, particularly with respect to prescribing and drug-related outcomes.

Conclusions

This population-based study quantifies the relationship between community prescribing and risk of early AKI requiring RRT in an ICU survivor population. Despite limitations of the survivor dataset, this study highlights the harmful potential of these commonly prescribed medications in a critically ill population including the importance of timely cessation during acute illness. Further studies in a complete cohort, including non-survivors, would allow risk estimates to be generalisable to a wider population.

Supplemental Material

INC900099 Supplemental Material - Supplemental material for Community prescribing of potentially nephrotoxic drugs and risk of acute kidney injury requiring renal replacement therapy in critically ill adults: A national cohort study

Supplemental material, INC900099 Supplemental Material for Community prescribing of potentially nephrotoxic drugs and risk of acute kidney injury requiring renal replacement therapy in critically ill adults: A national cohort study by Steven Tominey, Alan Timmins, Robert Lee, Timothy S Walsh and Nazir I Lone in Journal of the Intensive Care Society

Footnotes

Acknowledgements

The authors thank the Scottish Intensive Care Society Audit Group for providing data and the staff at participating hospitals. The authors acknowledge the support of the eDRIS Team (National Services Scotland) for their involvement in obtaining approvals, provisioning and linking data and the use of the secure analytical platform within the National Safe Haven. ST thanks Kidney Research UK for funding in the form of an Intercalated Degree Award. NL is a Deputy Director for Research for the Intensive Care Foundation.

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: ST was awarded a Kidney Research UK Intercalated Degree Award to fund himself during this project.

ORCID iDs

Steven Tominey https://orcid.org/0000-0002-3925-3801 Robert Lee https://orcid.org/0000-0001-9379-3748 Nazir I Lone

References

Levi

De Souza

De Magalhães

, et al. Comparison of the RIFLE, AKIN and KDIGO criteria to predict mortality in critically ill patients. Rev Bras Ter Intensiva 2013; 25: 290–296.

Czempik

Cieśla

Knapik

, et al. Mortality of patients with acute kidney injury requiring renal replacement therapy. Adv Clin Exp Med 2018; 27: 327–333.

Hwang

Park

Kim

, et al. Changes in acute kidney injury epidemiology in critically ill patients: a population-based cohort study in Korea. Ann Intensive Care 2019; 9: 65.

Gammelager

Christiansen

Johansen

, et al. One-year mortality among Danish intensive care patients with acute kidney injury: a cohort study. Crit Care 2012; 16: R124.

Truche

Ragey

Souweine

, et al. ICU survival and need of renal replacement therapy with respect to AKI duration in critically ill patients. Ann Intensive Care 2018; 8: 127.

Uchino

. Acute renal failure in critically ill patients. A multinational, multicenter study. JAMA 2005; 294: 813.

National Institute for Health and Care Excellence. CG 169: Acute kidney injury. 2013.

Kaufman

Dhakal

Patel

, et al. Community-acquired acute renal failure. Am J Kidney Dis 1991; 17: 191–198.

Abdel-Kader

Palevsky

. Acute Kidney Injury in the Elderly. Clin Geriatr Med 2009; 25: 331–358.

10.

Bellomo

. The epidemiology of acute renal failure: 1975 versus 2005. Curr Opin Crit Care 2006; 12: 557–560.

11.

NHS Scotland ISD. Dispensing reimbursement, remuneration and volume. 2016.

12.

Scottish Patient Safety Programme. Medicine Sick Day Rules Card.

13.

Schetz

Dasta

Goldstein

, et al. Drug-induced acute kidney injury. Curr Opin Crit Care 2005; 11: 555–565.

14.

Ishiguro

Fujita

Omori

, et al. Assessing the effects of non-steroidal anti-inflammatory drugs on antihypertensive drug therapy using post-marketing surveillance database. J Epidemiol 2008; 18: 119–124.

15.

Loboz

Shenfield

. Drug combinations and impaired renal function – The ‘triple whammy’. Br J Clin Pharmacol 2005; 59: 239–243.

16.

Greger

. Physiology of renal sodium transport. Am J Med Sci 2000; 319: 51–62.

17.

Thomas

. Diuretics, ACE inhibitors and NSAIDs – the triple whammy. Med J Aust 2000; 172: 184–185.

18.

Lapi

Azoulay

Yin

, et al. Risk of acute renal failure likely due to concurrent use of ace-inhibitors, angiotensin receptors blockers, diuretics and anti-inflammatory drugs. Pharmacoepidemiol Drug Saf 2012; 21: 163.

19.

Dreischulte

Morales

Bell

, et al. Combined use of nonsteroidal anti-inflammatory drugs with diuretics and/or renin-angiotensin system inhibitors in the community increases the risk of acute kidney injury. Kidney Int 2015; 88: 396–403.

20.

Chaumont

Pourcelet

van Nuffelen

, et al.

Acute kidney injury in elderly patients with chronic kidney disease: do angiotensin-converting enzyme inhibitors carry a risk?

J Clin Hypertens (Greenwich) 2016; 18: 514–521.

21.

Stirling

Houston

Robertson

, et al. Diarrhoea, vomiting and ACE inhibitors: – an important cause of acute renal failure. J Hum Hypertens 2003; 17: 419–423.

22.

Umruddin

Moe

Superdock

. ACE inhibitor or angiotensin II receptor blocker use is a risk factor for contrast-induced nephropathy. J Nephrol 2012; 25: 776–781.

23.

Zhang

Ren

, et al. Diuretics associated acute kidney injury: clinical and pathological analysis. Ren Fail 2014; 36: 1051–1055.

24.

Lafrance

J-P

Miller

. Selective and non-selective non-steroidal anti-inflammatory drugs and the risk of acute kidney injury. Pharmacoepidemiol Drug Saf 2009; 18: 923–931.

25.

Huerta

Castellsague

Varas-Lorenzo

, et al. Nonsteroidal anti-inflammatory drugs and risk of ARF in the general population. Am J Kidney Dis 2005; 45: 531–539.

26.

Scottish Intensive Care Society Audit Group. Data Collection, Methodology, http://www.sicsag.scot.nhs.uk/Data/methodology.html (2018, accessed 17 November 2018).

27.

Walsh

Salisbury

Donaghy

, et al. PReventing early unplanned hOspital readmission aFter critical ILlnEss (PROFILE): protocol and analysis framework for a mixed methods study. BMJ Open 2016; 6: e012590.

28.

NHS Scotland. Public Benefit and Privacy Panel for Health and Social Care, http://www.informationgovernance.scot.nhs.uk/pbpphsc/ (2018, accessed 22 November 2018).

29.

Scottish Intensive Care Society Audit Group. Audit of Critical Care in Scotland 2013 Reporting on 2012. 2013.

30.

Harris

Lewington

AJP

Harrison

, et al. Relationship between patients’ outcomes and the changes in serum creatinine and urine output and RIFLE classification in a large critical care cohort database. Kidney Int 2015; 88: 369–377.

31.

Joint Formulary Committee. British national formulary, 74th ed. London: BMJ Group and Pharmaceutical Press, 2017.

32.

Government S. SIMD Scottish Index of Multiple Deprivation 2009 General Report, http://www.gov.scot/resource/doc/289599/0088642.pdf (2009, accessed 29 November 2018).

33.

Betty

Jonathan

. Essential medical statistics, 2nd ed. Oxford: Blackwell Publishing Ltd, 2003.

34.

NHS Scotland ISD. Community Dispensing, https://www.isdscotland.org/Health-topics/Prescribing-and-medicines/Community-Dispensing/ (accessed 9 April 2019).

35.

Waikar

. Validity of International Classification of Diseases, Ninth Revision, Clinical Modification Codes for Acute Renal Failure. J Am Soc Nephrol 2006; 17: 1688–1694.

36.

Scottish Intensive Care Society Audit Group. Audit of Critical Care in Scotland 2014 Reporting on 2013. 2014.

37.

Scottish Intensive Care Society Audit Group. Audit of Critical Care in Scotland 2012: Reporting on 2011. 2012.

38.

Loboz

Shenfield

. Drug combinations and impaired renal function – the ‘triple whammy’. Br J Clin Pharmacol 2005; 59: 239–243.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.57 MB