Summary
Background and objectives
Initiation of dialysis on Sunday is limited by constraints that do not exist on other days of the week, which may lead to triaging of dialysis therapy. The study hypothesis was that patients with AKI on a Sunday would sustain higher mortality rates.
Design, setting, participants, & measurements
Study participants (n=4970) were part of the retrospective University of Pennsylvania Health AKI cohort, which is composed of patients with severe inpatient AKI (characterized by a doubling of admission creatinine) who were hospitalized from January 1, 2004, to August 31, 2010. Patient-days (n=15,995) were included if the patient had AKI severity of Acute Kidney Injury Network (AKIN) stage 2 or greater and had not yet begun receiving dialysis. The association of day of the week and inpatient mortality was assessed with logistic regression of data updated daily, using robust variance estimators.
Results
The rate (95% confidence interval [CI]) of initiation of dialysis on Sunday was 2.5 (1.8 to 3.1) per 100 patient-days, compared with 3.8 (3.5 to 4.1) per 100 patient-days on other days of the week (P=0.001). Inpatient mortality (95% CI) among patients with severe AKI present on a Sunday was 30% (28% to 32%), compared with 31% (31% to 32%) on other days of the week (P=0.08). Inpatient mortality among patients who initiated dialysis on Sunday was 65% (52% to 79%), compared with 65% (61% to 70%) among those who initiated dialysis from Tuesday through Saturday (P=0.79). Patients who initiated dialysis on Monday had a lower mortality than those who initiated it on another day of the week (52% [40% to 64%] versus 65% [61% to 70%]; P=0.03).
Conclusions
Despite a lower frequency of dialysis, patients with severe AKI on Sunday have mortality similar to that of patients with severe AKI on other days of the week.
Introduction
AKI is an increasingly common and highly morbid complication of hospitalization (1,2). The recent publication of the Kidney Disease: Improving Global Outcomes practice guidelines may help to further standardize the diagnosis and treatment of patients with AKI, but the decision of when to initiate dialysis remains highly subjective, save for a discrete number of life-threatening circumstances (3,4). Hospital practice patterns may differ during the weekend because of a variety of constraints, ranging from staffing differences to drug availability to patient load.
Patients with an admitting diagnosis of AKI who are admitted on the weekend carry a higher mortality risk than those admitted on other days of the week (5). One explanation for this observation is that a potentially useful therapy (dialysis) may not be initiated on a Sunday as a result of limitations in physician or nurse staffing or device availability (3,6). It remains unclear whether the lower rates of dialysis on the weekend (particularly Sundays) mediate the relationship between weekend admission and higher mortality in AKI. Further, AKI may develop after admission or evolve during the hospital stay. We set out to examine whether the characteristics of patients who develop AKI vary by days of the week and to characterize differences between patients initiated on dialysis on Sunday versus other days of the week. Finally, we sought to evaluate whether the day of the week when AKI is present modifies the relationship between AKI and mortality.
Methods
Patients
Patients in this study were part of the University of Pennsylvania Acute Kidney Injury (UPHS-AKI) cohort (7). Briefly, the UPHS-AKI cohort is a retrospective cohort study composed of patients with severe inpatient AKI during admissions to one of three urban hospitals from January 1, 2004, to August 31, 2010. All hospitals have a nephrology consult service that is capable of seeing consults and initiating dialysis 7 days a week, although weekend consults are typically covered by a single attending physician. Inpatient dialysis units are fully staffed Monday through Saturday during working hours, but emergency staff can be called in Sunday and off-hours when necessary. Adults aged ≥18 years were eligible for inclusion if the baseline creatinine (determined by the nadir creatinine within 48 hours of admission) was ≤1.4 mg/dl for men and ≤1.2 mg/dl for women and the creatinine subsequently doubled from that baseline during the hospital stay. Patients were excluded if they received a transplant during admission or received dialysis within 24 hours of admission.
In the current study, we included patients from the UPHS-AKI cohort who had AKI according to the Acute Kidney Injury Network (AKIN) stage 2 creatinine criteria: a doubling of creatinine that occurs within a 7-day period after an acute increase of 0.3 mg/dl or 50% within 48 hours (8). The unit of observation was the patient-day. All patient-days in which the AKIN stage was 2 or greater during the admission were included in analysis, but days after the initiation of dialysis were excluded. The study was approved by the Institutional Review Board at the University of Pennsylvania.
Covariate Ascertainment
Laboratory variables were retrieved from an electronic database and obtained in the course of usual clinical care. Comorbidity data were derived from admission International Classification of Diseases, Ninth Revision, codes as described by Quan et al. (9). Demographic data (including self-reported race) were obtained from the electronic medical record. Procedures and orders (including hemodialysis and continuous renal replacement therapy, intubation/extubation, ventilator settings, do-not-resuscitate orders) were obtained from an electronic order entry system. Death dates were determined by hospital administrative data. The Sequential Organ Failure Assessment (SOFA) score was calculated on a daily basis but did not include the renal or neurologic components (10). Estimated GFR was calculated using the CKD-Epidemiology Collaboration equation (11).
Propensity to be dialyzed was based on a logistic regression model using a stepwise approach, which retained the following covariates at P<0.20: serum concentration of creatinine, potassium, bicarbonate, BUN, and sodium; platelet count; intensive care unit (ICU) location; use of potassium-sparing diuretics, thiazide diuretics, paralytic agents, sedatives, antibiotics, aminoglycosides, and vasopressors; patient race (black versus nonblack) and sex; orders for blood cultures and transfusion of blood or fresh frozen plasma; history of diabetes mellitus, metastatic malignancy, and liver disease; do-not-resuscitate status; mechanical ventilation; fraction of inspired oxygen; and baseline creatinine concentration. Covariates dropped from the propensity model during stepwise regression included age; history of congestive heart failure, peptic ulcer disease, rheumatic disease, myocardial infarction, cerebrovascular disease, pulmonary disease, HIV infection, peripheral vascular disease, paraplegia, and malignancy; cryoglobulin or platelet transfusion; loop diuretic use; year of admission; use of angiotensin-converting enzyme inhibitors or angiotensin receptor blockers; surgical admitting service; serum hemoglobin concentration; and white blood cell count. Day of the week was deliberately excluded from the propensity model.
Day Categories
Patient-days were categorized as a Sunday, Monday, or Tuesday through Saturday, for a total of three categories. This categorization was based on the hypothesis that Sunday and Monday would differ from other days because patients triaged away from dialysis on Sunday would be present on Monday, potentially skewing that population in comparison with other weekdays.
Missing Data and Imputation
Missing laboratory data were allowed to be carried forward a maximum of 2 days. Variables with missing values after this procedure were not included in any analyses.
Statistical Analyses
Continuous covariates are described as means (±SD) or medians (interquartile range [IQR]) depending on normality. Proportions are expressed as percentages. Unadjusted and adjusted associations between categorical and continuous variables were performed using linear regression. Robust variance estimators were used to account for the fact that each patient could contribute multiple patient-days of data. Timing variables (including time in hospital, time from onset of AKI to doubling of creatinine, and time from doubling of creatinine to dialysis initiation) were log-transformed before analysis. Unadjusted and adjusted associations between categorical variables were performed using logistic regression with robust variance estimation. The primary analysis examined the association between exposure to Sundays and inpatient mortality using logistic regression. All statistical tests were performed in Stata software, version 12.1 (Stata Corp., College Station, TX).
Results
Of the 6119 patients in the UPHS-AKI cohort, 4970 met AKIN stage 2 criteria and were included in this analysis. Of those, 53% were male and 47%, female; 27% were identified as black. The mean age was 61±16 years. A majority (53%) of patients was admitted to a medical service, and the remainder were admitted to a surgical service. The most common comorbidity on admission was congestive heart failure (36% of patients); 26% of patients had diabetes. The mean baseline estimated GFR was 93±27 ml/min per 1.73 m2.
The rate (95% confidence interval [CI]) of initiation of dialysis for AKI was much lower on Sunday than other days of the week, with 2.5 (1.8 to 3.1) initiations per 100 patient-days on Sunday versus 3.3 (2.6 to 4.1) on Monday and 3.9 (3.5 to 4.2) on Tuesday through Saturday (P=0.06 for Sunday versus Monday and 0.001 for Sunday versus Tuesday through Saturday). Figure 1 depicts the rate of dialysis initiation by day of the week.
Figure 1.
Rate of dialysis initiation by day of the week. Bars represent 95% confidence intervals. P<0.001 for Sunday versus other days. No significant differences between rates of initiation on other days of the week were detected.
Population averages of patient characteristics, stratified by day of the week, are presented in Table 1. A total of 15,995 patient-days were available for analysis. The median number of days spent in AKIN stage 2 or greater was 2 (IQR, 1–4 days). The population of patients with AKIN stage 2 or greater on a Sunday was largely similar to the population present on other days of the week. However, the BUN concentration and the rate of change in creatinine were slightly higher among patients with AKI on Sunday versus those on Monday (although not compared with other days of the week). The serum potassium concentration was lower among patients with AKI present on a Sunday versus those present Tuesday through Saturday but did not differ from Monday patients.
Table 1.
Patient characteristics of patients with AKI stage 2 or greater, stratified by day of the week.
| Variable | Sunday (n=2320) | Monday (n=2263) | Tuesday–Saturday (n=11,412) | P Value (Sunday versus Monday) | P Value (Sunday versus Tuesday–Saturday) |
|---|---|---|---|---|---|
| Demographics | |||||
| Age (yr) | 60±16 | 60±16 | 60±16 | 0.61 | 0.19 |
| Men | 1258 (54) | 1210 (53) | 6117 (54) | 0.34 | 0.43 |
| Black | 633 (27) | 605 (27) | 3118 (27) | 0.44 | 0.96 |
| Hospital location | |||||
| Surgical service | 1078 (46) | 1062 (47) | 5326 (47) | 0.55 | 0.80 |
| Intensive care unit | 1141 (49) | 1095 (48) | 5597 (49) | 0.38 | 0.88 |
| Comorbid conditions (present at admission) | |||||
| HIV | 76 (3) | 66 (3) | 298 (3) | 0.14 | 0.002 |
| Congestive heart failure | 831 (36) | 833 (37) | 4175 (37) | 0.20 | 0.31 |
| Cardiovascular disease | 243 (10) | 223 (10) | 1115 (10) | 0.19 | 0.13 |
| Diabetes | 584 (25) | 575 (25) | 2885 (25) | 0.73 | 0.88 |
| Liver disease | 494 (21) | 491 (22) | 2382 (21) | 0.52 | 0.51 |
| Malignancy | 294 (13) | 266 (12) | 1409 (12) | 0.07 | 0.52 |
| Metastatic solid tumor | 228 (10) | 222 (10) | 1189 (10) | 0.97 | 0.20 |
| Myocardial infarction | 310 (13) | 318 (14) | 1503 (13) | 0.20 | 0.71 |
| Peptic ulcer disease | 81 (3) | 71 (3) | 364 (3) | 0.21 | 0.25 |
| Peripheral vascular disease | 336 (14) | 329 (15) | 1590 (14) | 0.92 | 0.31 |
| Pulmonary disease | 351 (15) | 325 (14) | 1606 (14) | 0.18 | 0.06 |
| Hemiplegia | 62 (3) | 64 (3) | 320 (3) | 0.51 | 0.60 |
| Rheumatologic disease | 93 (4) | 85 (4) | 368 (3) | 0.42 | 0.01 |
| Laboratory values | |||||
| Baseline creatinine (mg/dl) | 0.83±0.26 | 0.82±0.26 | 0.82±0.26 | 0.01 | 0.02 |
| Bicarbonate (mEq/L) | 22±6 | 22±6 | 22±6 | 0.77 | 0.10 |
| BUN (IQR) (mg/dl) | 43 (27–65) | 45 (28–67) | 42 (27–65) | 0.05 | 0.60 |
| Change in creatinine from day prior (IQR) (mg/dl) | 0.27 (0.00–0.7) | 0.20 (0.00–0.60) | 0.30 (0.00–0.78) | 0.01 | 0.04 |
| Creatinine (mg/dl) | 2.59±1.33 | 2.59±1.36 | 2.55±1.31 | 0.92 | 0.07 |
| Potassium (mEq/L) | 4.4±0.8 | 4.4±0.7 | 4.5±0.8 | 0.67 | <0.001 |
| AKI data | |||||
| Hospital stay prior to today (IQR) (d)a | 8 (4–15) | 8 (5–15) | 8 (5–15) | <0.001 | 0.33 |
| Onset of AKI to doubling of creatinine (IQR) (d)a | 2 (1–2) | 2 (1–2) | 2 (1–2) | <0.001 | 0.56 |
| Acuity of illness | |||||
| SOFA score | 5.9±3.6 | 6.0±3.6 | 5.9±3.7 | 0.15 | 0.89 |
| Ventilated | 742 (32) | 749 (33) | 3676 (32) | 0.18 | 0.78 |
| Pressor requirement | |||||
| None | 1715 (74) | 1663 (73) | 8417 (74) | 0.60 | 0.84 |
| 1 | 303 (13) | 295 (13) | 1429 (13) | 0.97 | 0.42 |
| ≥2 | 302 (13) | 305 (13) | 1566 (14) | 0.51 | 0.30 |
| Propensity to be dialyzed | 0.03±0.09 | 0.03±0.09 | 0.03±0.09 | 0.87 | 0.34 |
Data are mean ± SD or n (%) unless otherwise indicated (n=patient-days). P values are adjusted for intrapatient correlation because a patient may contribute data to more than one day category. IQR, interquartile range; SOFA, Sequential Organ Failure Assessment.
Patients with AKI stage 2 or greater present on a Sunday have slightly shorter hospital stays and more rapid doubling of creatinine from AKI onset than patients with AKI stage 2 or greater present on Monday.
The characteristics of patients initiated on dialysis on a Sunday differed slightly from those of patients initiating dialysis on other days of the week (Table 2). Those dialyzed on Sunday had significantly higher dialysis propensity scores, a proxy for severity of renal injury (mean of 0.33±0.27) than those dialyzed on Monday (0.24±0.18) or on Tuesday through Saturday (0.26±0.23) (P=0.02 for Sunday versus Monday; P=0.03 for Sunday versus Tuesday through Saturday). The BUN concentration was lower in those dialyzed Sunday (median, 54.5 [IQR, 40–90] mg/dl) versus those dialyzed Monday (69 [IQR, 52–103]) mg/dl; P=0.02), but not lower than those dialyzed during the remainder of the week (63 [IQR, 40–94] mg/dl; P=0.34). Only 15% of patients dialyzed on Sunday had no pressor requirement, compared with 28% of those dialyzed Monday (P=0.10) and 34% of those dialyzed during the rest of the week (P=0.01). Patients dialyzed Sunday had a median of 2 days from doubling of creatinine to initiation of dialysis compared with 3 days among those dialyzed Monday (P=0.05).
Table 2.
Patient characteristics, stratified by day of initiation of dialysis
| Variable | Sunday Initiation (n=52) | Monday Initiation (n=71) | Tuesday–Saturday Initiation (n=411) | P Value (Sunday versus Monday) | P Value (Sunday versus Tuesday–Saturday) |
|---|---|---|---|---|---|
| Demographics | |||||
| Age (yr) | 59±16 | 56±17 | 59±16 | 0.30 | 0.96 |
| Men | 31 (60) | 50 (70) | 259 (63) | 0.21 | 0.63 |
| Black | 10 (19) | 13 (18) | 70 (17) | 0.90 | 0.69 |
| Hospital location | |||||
| Surgical service | 28 (54) | 46 (65) | 234 (57) | 0.22 | 0.67 |
| Intensive care unit | 50 (96) | 65 (92) | 363 (88) | 0.32 | 0.10 |
| Dialytic modality | |||||
| CRRTa | 40 (77) | 56 (79) | 274 (67) | 0.80 | 0.14 |
| IHD | 12 (23) | 15 (21) | 137 (33) | ||
| Comorbid conditions (present at admission) | |||||
| HIV | 1 (2) | 0 (0) | 7 (2) | n/a | 0.91 |
| Congestive heart failure | 21 (40) | 30 (42) | 189 (46) | 0.84 | 0.45 |
| Cardiovascular disease | 10 (19) | 10 (14) | 39 (9) | 0.45 | 0.04 |
| Diabetes | 15 (29) | 20 (28) | 102 (25) | 0.93 | 0.53 |
| Liver disease | 19 (37) | 26 (37) | 176 (43) | 0.99 | 0.39 |
| Malignancy | 7 (13) | 6 (8) | 37 (9) | 0.38 | 0.31 |
| Metastatic solid tumor | 1 (2) | 2 (3) | 17 (4) | 0.75 | 0.45 |
| Myocardial infarction | 8 (15) | 16 (23) | 71 (17) | 0.33 | 0.73 |
| Peptic ulcer disease | 2 (4) | 4 (6) | 19 (5) | 0.65 | 0.80 |
| Peripheral vascular disease | 7 (13) | 18 (25) | 77 (19) | 0.11 | 0.36 |
| Pulmonary disease | 6 (12) | 6 (8) | 51 (12) | 0.57 | 0.86 |
| Hemiplegia | 1 (2) | 3 (4) | 10 (2) | 0.49 | 0.82 |
| Rheumatologic disease | 1 (2) | 4 (6) | 8 (2) | 0.33 | 0.99 |
| Laboratory values | |||||
| Baseline creatinine (mg/dl) | 0.96±0.27 | 0.94±0.27 | 0.94±0.25 | 0.75 | 0.68 |
| Bicarbonate (mEq/L) | 17±5 | 17±5 | 17±5 | 0.51 | 0.89 |
| BUN (IQR) (mg/dl) | 56 (44–96) | 74 (53–108) | 66 (41–96) | 0.02 | 0.34 |
| Change in creatinine from day prior (IQR) (mg/dl) | 0.80 (0.50–1.22) | 0.80 (0.44–1.14) | 0.83 (0.50–1.40) | 0.41 | 0.91 |
| Creatinine (mg/dl) | 4.01±2.03 | 4.26±1.61 | 4.18±1.81 | 0.45 | 0.54 |
| Potassium (mEq/L) | 5.2±0.9 | 5.1±1.0 | 5.3±1.1 | 0.59 | 0.75 |
| AKI data | |||||
| Hospital stay prior to today (IQR) (d) | 7.5 (4–12.5) | 8 (4–17) | 8 (4–14) | 0.30 | 0.63 |
| Onset of AKI to doubling of creatinine (IQR) (d) | 2 (1–2) | 2 (1–2) | 2 (1–2) | 0.69 | 0.65 |
| Time from AKI stage 2 to RRT (IQR) (d) | 2 (1–3) | 3 (2–5) | 2 (1–5) | 0.05 | 0.17 |
| Acuity of illness | |||||
| SOFA score | 11.8±2.9 | 10.8±2.8 | 10.8±3.4 | 0.08 | 0.06 |
| Ventilated | 41 (79) | 52 (73) | 269 (65) | 0.48 | 0.06 |
| Pressor requirement | |||||
| None | 8 (15) | 20 (28) | 141 (34) | 0.10 | 0.01 |
| 1 | 12 (23) | 12 (17) | 62 (15) | 0.39 | 0.14 |
| ≥2 | 32 (62) | 39 (55) | 208 (51) | 0.46 | 0.14 |
| Propensity to be dialyzed | 0.33±0.27 | 0.24±0.18 | 0.26±0.23 | 0.02 | 0.03 |
Data are mean ± SD or n (%) unless otherwise indicated (n=patients). CRRT, continuous renal replacement therapy; IHD, intermittent hemodialysis; IQR, interquartile range; SOFA, Sequential Organ Failure Assessment.
At the study institutions during the time course of this study, all CRRT was performed as continuous venovenous hemodialysis.
Inpatient Mortality
Inpatient mortality (95% CI) was 30% (28% to 32%) among patients with severe AKI present on a Sunday, compared with 31% (29% to 33%) for those with severe AKI on a Monday and 31% (31% to 32%) for those with severe AKI occurring on Tuesday through Saturday (odds ratio [OR] for Sunday versus other days, 0.94; 95% CI, 0.88 to 1.01; P=0.08). After adjustment for age, sex, race, BUN, creatinine, change in creatinine, potassium, ICU status, and modified SOFA score, this result was relatively unchanged (OR, 0.96; 95% CI, 0.89 to 1.03; P=0.29). We detected no interaction between day of the week and propensity to receive dialysis on mortality (P=0.76).
Inpatient mortality (95% CI) among those who initiated dialysis on a Sunday was 65% (52% to 79%) versus 52% (40% to 64%) for those initiated on a Monday and 65% (61% to 70%) for those initiated Tuesday through Saturday (P=0.79 for Sunday versus other days; P=0.03 for Monday versus other days).
The finding of a lower mortality among patients who initiated dialysis on Monday prompted an exploration of characteristics of these patients compared with those initiated on other days of the week (Supplemental Table 1). Interestingly, there were no statistically significant differences between patients initiated on Monday versus other days of the week across a range of laboratory, demographic, and clinical variables. After multivariable adjustment for dialysis propensity score, BUN, number of vasopressors, ICU status, surgical admission, SOFA score, and time between doubling of creatinine and initiation of renal replacement therapy the OR for inpatient mortality among patients initiated on dialysis Monday was 0.49 (95% CI, 0.27 to 0.87; P=0.02) versus other days of the week. Among the 71 patients in our study who initiated dialysis on Monday, 60 had been present in the hospital with AKI stage 2 or greater the preceding Sunday. Of this group, 31 (52%) died during the hospitalization. Among the 11 patients who initiated dialysis on Monday who did not have AKIN stage 2 or greater AKI on the antecedent Sunday, 6 (54%) died in the hospital. These proportions were not demonstrably different (P=0.14 for interaction).
To assess whether dialysis was initiated prophylactically on Friday or Saturday (in order to avoid a Sunday initiation), we compared BUN and creatinine levels among patients initiated Friday (n=88), Saturday (n=71), and Sunday (n=52). Mean BUN and creatinine were 65±41 mg/dl and 4.0±1.8 mg/dl on Friday, 72±39 mg/dl and 4.3±1.7 mg/dl on Saturday, and 64±36 mg/dl and 4.0±2.0 mg/dl on Sunday. None of these differences were statistically significant.
Sensitivity Analyses
Data analysis at the level of the patient-day allows patients to contribute information to multiple day-of-week categories using time-updated covariates. Although we account for within-patient correlation in our statistical modeling, we performed an additional analysis examining inpatient mortality based on the first day the patient met AKIN 2 criteria. Characteristics of patients stratified by the day they first met AKIN 2 appear in Supplemental Table 2. Patients first meeting AKIN 2 criteria on a Sunday had inpatient mortality (95% CI) of 31% (27% to 34%; P=0.36 versus other days); on a Monday, 30% (26% to 34%; P=0.73 versus other days); and on Tuesday through Saturday, 29% (27% to 30%; P=0.32 versus other days). Similar results were obtained when we examined the day patients (n=1873) first achieved AKIN stage 3 with inpatient mortality 37% (32% to 43%), 40% (34% to 46%), and 41% (39% to 44%) for Sunday, Monday, and Tuesday through Saturday, respectively (all P>0.05 versus other days).
Weekend admission was previously demonstrated to be associated with higher mortality in AKI (5). Of the 4970 patients in our sample, 792 were admitted on a Saturday or Sunday (16%). Inpatient mortality (95% CI) was 33% (30% to 36%) for those admitted on the weekend versus 28% (27% to 30%) for those admitted on a weekday (P=0.01). After adjustment for weekend admission, the OR for death among those with AKI stage 2 or greater present on a Sunday was essentially unchanged: 0.95 (0.89 to 1.01; P=0.10).
Discussion
Prior studies from multiple fields have demonstrated that hospital admission on the weekend is associated with adverse outcomes (12–18). Weekend admission with AKI was associated with higher inpatient mortality in a large national sample (5), but to our knowledge no studies have examined the effect of day of the week during the hospitalization on outcomes. Because AKI frequently arises during hospitalization (as opposed to on admission) and dialysis practices vary by day of the week, we performed this analysis of the effect of day of the week on the AKI-mortality relationship.
The rate of initiation of dialysis for AKI is much lower on Sunday than other days of the week. We suspect this is due to limited nursing and physician resources on Sundays, which leads to dialysis of only the most ill patients on that day of the week. Surprisingly, the clinical measures for those dialyzed on Sunday did not differ greatly from those of patients dialyzed the rest of the week, particularly compared with those dialyzed Tuesday through Saturday. We set out to determine whether the lower rate of dialysis on Sunday would lead to adverse outcomes.
If dialysis is a uniformly beneficial treatment for AKI, we would expect to see greater adverse outcomes among those with severe AKI on a Sunday versus those with severe AKI present other days of the week, assuming that the populations of patients are broadly similar. Finding no such result may indicate that policies and practices in effect on Sunday do not affect patients, on the whole, in an adverse fashion. It is possible, of course, that an unidentified population of patients is harmed by the deferral of dialysis, while a similarly sized group benefits from such a deferral—perhaps because they have an opportunity to recover renal function without dialysis.
It is likely that the relative benefit of dialysis for AKI changes as severity of AKI increases. Given that the dialysis propensity, our index of severity of renal injury, among dialyzed patients was higher on Sunday than other days of the week, it may be that clinicians are delaying or not pursuing dialysis in a population of patients who would be expected to derive less benefit from the therapy. If this is the case, the “Sunday model” can be entertained as one that could be realistically transferred to other days of the week, reducing resource consumption without compromising patient care. More broadly, these data may suggest that delay of dialysis in general is not harmful, and may even be beneficial in selected populations of patients. Although we cannot be sure that patients dialyzed Monday were systematically deferred on Sunday, we note that they had on average 1 more day of severe AKI and a higher BUN than those dialyzed Sunday, and their overall mortality was lower. This finding may add to our increasing understanding of the importance of timing of dialysis initiation in patients with AKI.
The finding of lower inpatient mortality among patients initiated on dialysis on a Monday versus other days of the week may reflect subtle differences in the Monday population. Although measured characteristics are very similar between patients who initiate dialysis on a Monday and those initiated other days of the week, it is conceivable that this is a healthier cohort overall. In particular, the willingness of a clinician to triage a patient with AKI on Sunday to possible dialysis initiation on Monday or later is likely to be a marker of less severe disease compared with patients dialyzed on Sunday. However, the lower mortality rate was also seen in patients dialyzed on Monday who had not been present with severe AKI the preceding Sunday. In addition, inpatient mortality among patients dialyzed Sunday was the same as those dialyzed Tuesday through Saturday. If only the sickest patients were initiated on Sunday, we’d expect to see a higher mortality on this day unless dialysis increases mortality among some of those with less severe AKI. Still, we cannot be certain whether the “Monday effect” is due to triaging from Sunday or whether dialysis as a procedure performs better on Monday than other days of the week. The distinction will need to be evaluated in larger cohorts.
We did not detect a significant interaction between propensity to receive dialysis and day of the week on inpatient mortality, suggesting that management of patients with severe kidney injury was appropriate regardless of day of the week. It is no worse to have AKI on Sunday than any other day; moreover, it is no worse to have severe AKI on Sunday versus another day. This again suggests that the patients with the most severe AKI are receiving the appropriate intervention regardless of day of the week. The patients in whom dialysis is deferred because the day is a Sunday appear to be able to be deferred safely.
The results of this study should be interpreted in the light of several limitations. First, the study was performed within a single health system; practices regarding the initiation of dialysis are likely to differ from those in other centers. It should be noted, however, that the study took place at three hospitals with three independent nephrology services. Second, the lack of reliable urine output data limits our ability to fully capture the severity of renal insult in these patients. It is unlikely, however, that urine output would differ systematically by day of the week, and, thus, this should not affect our primary analysis. Third, we are unable to determine which specific patients may have been deferred on any given day of the week. The interpretation of these findings should reassure physicians that practices occurring on a Sunday do not harm patients with AKI, but deferral itself may still be harmful (or helpful) in some situations. Finally, the study included only patients with a relatively normal baseline creatinine—the findings may not extend to patients with underlying CKD.
In conclusion, although the rate of dialysis initiation is lower on Sundays than on other days of the week, patients with severe AKI present on a Sunday fare no worse than those with severe AKI present on other days of the week. This strongly suggests that practices occurring on Sundays are not associated with deleterious outcomes among patients with severe AKI in the UPHS.
Disclosures
None.
Supplementary Material
Acknowledgments
We wish to thank Yulia Borovskiy and the Penn Data Store for their help in assembling this data set.
This study was funded in part by National Institute of Diabetes and Digestive and Kidney Diseases grant F32DK093223 awarded to F.P.W.
Footnotes
Published online ahead of print. Publication date available at www.cjasn.org.
This article contains supplemental material online at http://cjasn.asnjournals.org/lookup/suppl/doi:10.2215/CJN.03540413/-/DCSupplemental.
References
- 1.Chertow GM, Burdick E, Honour M, Bonventre JV, Bates DW: Acute kidney injury, mortality, length of stay, and costs in hospitalized patients. J Am Soc Nephrol 16: 3365–3370, 2005 [DOI] [PubMed] [Google Scholar]
- 2.Hsu RK, McCulloch CE, Dudley RA, Lo LJ, Hsu CY: Temporal changes in incidence of dialysis-requiring AKI. J Am Soc Nephrol 24:37–42, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Overberger P, Pesacreta M, Palevsky PM, VA/NIH Acute Renal Failure Trial Network : Management of renal replacement therapy in acute kidney injury: A survey of practitioner prescribing practices. Clin J Am Soc Nephrol 2: 623–630, 2007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Kidney Disease. Improving Global Outcomes (KDIGO) Acute Kidney Injury Work Group : KDIGO clinical practice guideline for acute kidney injury. Kidney Int Suppl 2: 1–138, 2012 [Google Scholar]
- 5.James MT, Wald R, Bell CM, Tonelli M, Hemmelgarn BR, Waikar SS, Chertow GM: Weekend hospital admission, acute kidney injury, and mortality. J Am Soc Nephrol 21: 845–851, 2010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Palevsky PM, Zhang JH, O’Connor TZ, Chertow GM, Crowley ST, Choudhury D, Finkel K, Kellum JA, Paganini E, Schein RM, Smith MW, Swanson KM, Thompson BT, Vijayan A, Watnick S, Star RA, Peduzzi P, VA/NIH Acute Renal Failure Trial Network : Intensity of renal support in critically ill patients with acute kidney injury. N Engl J Med 359: 7–20, 2008 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Wilson FP, Yang W, Feldman HI: Predictors of death and dialysis in severe AKI: The UPHS-AKI cohort. Clin J Am Soc Nephrol 8: 527–537, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Mehta RL, Kellum JA, Shah SV, Molitoris BA, Ronco C, Warnock DG, Levin A, Acute Kidney Injury Network : Acute Kidney Injury Network: Report of an initiative to improve outcomes in acute kidney injury. Crit Care 11: R31, 2007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Quan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi JC, Saunders LD, Beck CA, Feasby TE, Ghali WA: Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care 43: 1130–1139, 2005 [DOI] [PubMed] [Google Scholar]
- 10.Vincent JL, Moreno R, Takala J, Willatts S, De Mendonça A, Bruining H, Reinhart CK, Suter PM, Thijs LG: The SOFA (Sepsis-related Organ Failure Assessment) score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis-Related Problems of the European Society of Intensive Care Medicine. Intensive Care Med 22: 707–710, 1996 [DOI] [PubMed] [Google Scholar]
- 11.Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF, 3rd, Feldman HI, Kusek JW, Eggers P, Van Lente F, Greene T, Coresh J, CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) : A new equation to estimate glomerular filtration rate. Ann Intern Med 150: 604–612, 2009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Bell CM, Redelmeier DA: Mortality among patients admitted to hospitals on weekends as compared with weekdays. N Engl J Med 345: 663–668, 2001 [DOI] [PubMed] [Google Scholar]
- 13.Thompson RT, Bennett WE, Jr, Finnell SM, Downs SM, Carroll AE: Increased length of stay and costs associated with weekend admissions for failure to thrive. Pediatrics 131: e805–e810, 2013 [DOI] [PubMed] [Google Scholar]
- 14.Kostis WJ, Demissie K, Marcella SW, Shao Y-H, Wilson AC, Moreyra AE, Myocardial Infarction Data Acquisition System (MIDAS 10) Study Group : Weekend versus weekday admission and mortality from myocardial infarction. N Engl J Med 356: 1099–1109, 2007 [DOI] [PubMed] [Google Scholar]
- 15.Magid DJ, Wang Y, Herrin J, McNamara RL, Bradley EH, Curtis JP, Pollack CV, Jr, French WJ, Blaney ME, Krumholz HM: Relationship between time of day, day of week, timeliness of reperfusion, and in-hospital mortality for patients with acute ST-segment elevation myocardial infarction. JAMA 294: 803–812, 2005 [DOI] [PubMed] [Google Scholar]
- 16.Gallerani M, Imberti D, Bossone E, Eagle KA, Manfredini R: Higher mortality in patients hospitalized for acute aortic rupture or dissection during weekends. J Vasc Surg 55: 1247–1254, 2012 [DOI] [PubMed] [Google Scholar]
- 17.Deshmukh A, Pant S, Kumar G, Bursac Z, Paydak H, Mehta JL: Comparison of outcomes of weekend versus weekday admissions for atrial fibrillation. Am J Cardiol 110: 208–211, 2012 [DOI] [PubMed] [Google Scholar]
- 18.Ananthakrishnan AN, McGinley EL: Weekend hospitalisations and post-operative complications following urgent surgery for ulcerative colitis and Crohn’s disease. Aliment Pharmacol Ther 37: 895–904, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.