Key Points
Applying four different commonly used baseline creatinine definitions revealed AKI cohorts of comparable size, AKI rate, and mortality.
Including patients with no baseline creatinine yields cohorts with considerably lower AKI rate and higher mortality.
Consistent use and reporting of methods for defining baseline creatinine is important, especially in studies of community-acquired AKI.
Keywords: AKI and ICU nephrology, AKI, creatinine, data management, laboratories
Visual Abstract
Abstract
Background
The baseline creatinine level is central in the Kidney Disease Improving Global Outcomes (KDIGO) criteria of AKI, but baseline creatinine is often inconsistently defined or unavailable in AKI research. We examined the rate, characteristics, and 30-day mortality of AKI in five AKI cohorts created using different definitions of baseline creatinine.
Methods
This nationwide cohort study included all individuals aged ≥18 years in Denmark with a creatinine measurement in 2017. Applying the KDIGO criteria, we created four AKI cohorts using four different baseline definitions (most recent, mean, or median value of outpatient creatinine 365–368 days before, or median value 90–98 days before, if available, otherwise median value 365–391 days before) and one AKI cohort not using a baseline value. AKI rate and the distribution of age, sex, baseline creatinine, and comorbidity were described for each AKI cohort, and the 30-day all-cause mortality was estimated using the Kaplan–Meier method.
Results
The study included 2,095,850 adults with at least one creatinine measurement in 2017. The four different baseline definitions identified between 61,189 and 62,597 AKI episodes. The AKI rate in these four cohorts was 13–14 per 1000 person-years, and 30-day all-cause mortality was 17%–18%. The cohort created without using a baseline creatinine included 37,659 AKI episodes, corresponding to an AKI rate of 8.2 per 1000 person-years and a 30-day mortality of 23%. All five cohorts were similar regarding age, sex, and comorbidity.
Conclusions
In a population-based setting with available outpatient baseline creatinine, different baseline creatinine definitions revealed comparable AKI cohorts, whereas the lack of a baseline creatinine when defining AKI led to a smaller AKI cohort with a higher mortality. These findings underscore the importance of availability and consistent use of an outpatient baseline creatinine, particulary in studies of community-acquired AKI.
Introduction
AKI is associated with increased morbidity, mortality, and subsequent development or progression of chronic kidney disease (1–3). According to the Kidney Disease Improving Global Outcomes (KDIGO) criteria (4), AKI is defined by one of the following: an increase in creatinine to at least 1.5 times baseline creatinine that is presumed to have occurred within 7 days, an increase in creatinine by 26.5 µmol/L within 48 hours, or urine volume <0.5 ml/kg per hour for 6 hours. Thus, baseline creatinine should reflect the expected kidney function of the last 7 days if creatinine is not measured in this period. Although baseline creatinine has an important effect on the subsequent detection of AKI, baseline creatinine remains inconsistently defined and is often unavailable in studies of AKI (5–7). Correct estimation of baseline creatinine might be important to identify community-acquired AKI because patients are admitted with elevated creatinine.
The presumption that a 1.5-fold increase happens within 7 days is often an issue because many people do not have a measured creatinine within the latest 7 days. Defining the baseline creatinine then becomes a matter of using the creatinine measurements that are available. Some studies use the first or the lowest creatinine measurement during the hospital admission as baseline or estimate baseline creatinine, assuming eGFR to be 75 ml/min per 1.73 m2 (6,8–10). Other studies use preadmission outpatient creatinine to estimate baseline creatinine, but disagreement among these studies also exists (7,10–14). This inconsistent use of baseline creatinine definitions hampers comparability in observational AKI studies (7,10,12–15). To improve comparability between studies of AKI, a common baseline definition of kidney function could be important. As long as there is no consensus on baseline creatinine, researchers have to consider how different baseline definitions will affect the detection of AKI in their study. We aimed to compare five AKI cohorts, of which four were sampled using different definitions of baseline creatinine and the fifth was sampled without a baseline creatinine. Across these five AKI cohorts, we compared the estimated incidence, characteristics, and 30-day all-cause mortality.
Materials and Methods
Study Design and Setting
We conducted this nationwide, population-based study using Danish medical data collected during routine clinical practice from all hospital contacts, laboratory tests, and deaths (16,17). All Danish inhabitants have a unique civil registration number, which facilitates individual-level linkage between databases (17–22). The Danish health care system is tax funded and provides equal and free access to medical care, including laboratory testing in primary, secondary, and tertiary care. Private hospitals account for <1% of all hospital admission, and all acute care is provided by public hospitals (17).
We retrieved data on plasma creatinine from the regional Clinical Laboratory Information System Research Database and the National Register of Laboratory Results for Research (21,22). Together, these databases contain both in- and outpatient creatinine measurements, i.e., measurements from general practitioners, outpatient clinics, emergency rooms, and admissions, covering all of Denmark. Creatinine measurements were identified through the International System of Nomenclature, Properties, and Units (see Supplemental Table 1) (23).
The study period was from January 1, 2017 to December 31, 2017, including all creatinine measurements from January 1, 2016 to December 31, 2017, classified as either in- or outpatient measurements to define baseline creatinine.
The study was reported to the Danish Data Protection Agency by Aarhus University (Aarhus University record no. 2016–051–000001/812). Ethical permission was not required for this noninterventional registry-based study.
AKI Study Populations Using Different Baseline Definitions
According to the guidelines from KDIGO, AKI is defined as an increase in creatinine of at least 26.5 µmol/L within 48 hours or a 1.5-fold increase in creatinine, which is known or presumed to have occurred within 7 days (4). We implemented this definition with the following three steps. Step 1 was an absolute increase of at least 26.5 µmol/L within the last 48 hours. If AKI was not detected in step 1, we continued with step 2, which was a relative increase in creatinine of at least 1.5 compared with lowest creatinine measurement within the last 7 days. If AKI still was not detected in step 2, we continued to step 3, which was a relative increase in creatinine of at least 1.5 times the baseline that was presumed to represent creatinine level 7 days before. In step 3, four different baseline definitions were used. We did not have information on urine output, but it is generally accepted to leave out this criterion in studies of AKI in a nonintensive-care setting (24).
We applied the following different baseline creatinine definitions to each creatinine measurement in 2017 from people aged ≥18 years. For the first (recent) baseline definition, we used the most recent outpatient creatinine measurement 365–368 days before each measurement (14,15). The second (mean) definition used the mean value of outpatient creatinine 365–368 days before each measurement (7). The third (median) definition used the median value of outpatient creatinine 365–368 days before each measurement (10). Finally, the fourth (twostep) definition used the median value of outpatient creatinine 90–98 days before each measurement if there were any outpatient measurements in this period, and otherwise used the median value of outpatient creatinine 365–391 days before each measurement (13). See Supplemental Figure 1 for a graphic description of the four baseline definitions. We also created a fifth (no baseline) cohort identified without applying step 3, i.e., this only included AKI episodes identified by an absolute increase of µmol/L within 48 hours (step 1) or by an observed 1.5-fold increase in creatinine within 7 days (step 2).
Because baseline creatinine preferably should present stable kidney function, only outpatient measurements were used in baseline estimation. When creating each cohort, we included only the first AKI episode per person in the study period and only AKI episodes not preceded by kidney failure with replacement therapy within 1 year.
Patient Characteristics
Comorbidity was identified from the Danish National Patient Registry, which has recorded diagnoses of patients in all hospital admissions since 1977, using the International Classification of Disease 10th version since 1994 (20). This registry has furthermore recorded all emergency room and outpatient clinic visits since 1995 and medical treatments, including dialysis, since 1999.
Information on diabetes was also obtained through any filled prescriptions of antidiabetic medication recorded in the Danish National Prescription Registry (25). We also used this database to obtain information on prescriptions of diuretics, statins, angiotensin-converting enzyme inhibitors, and angiotensin II receptor blockers (25). Information on kidney failure with replacement therapy was identified using procedure codes and diagnosis codes in the Danish National Patient Registry. We included comorbidities and comedication in the 10 years before inclusion. Information on sex, date of birth, and death was retrieved from the Danish Civil Registration System (18,19).
Statistical Analyses
We described the distribution of age, sex, baseline creatinine, number of creatinine measurements within 365–368 days before AKI, Charlson comorbidity index score (0, 1–2, and ≥3), and specific comorbidities for the five AKI cohorts (26) (Table 1). Median baseline creatinine was calculated in people with an available baseline creatinine.
Table 1.
Characteristics, comorbidity, medication, rate, and 30-day mortality of the five AKI cohorts with different baseline creatinine definitions
| AKI Cohorts Identified by Different Baseline Definitions | |||||
|---|---|---|---|---|---|
| Recent | Mean | Median | Two-Step | No Baseline | |
| AKI episodes, n | 61,759 | 61,189 | 62,597 | 61,399 | 37,659 |
| AKI cases identified in step 3,a n (%) | 30,496 (49) | 30,245 (49) | 31,914 (51) | 30,207 (49) | — |
| Age, yr, median (IQR) | 73 (63–82) | 73 (63–82) | 73 (63–82) | 73 (63–82) | 74 (65–83) |
| Men, n (%) | 31,803 (51) | 31,845 (52) | 32,469 (52) | 31,727 (52) | 20,824 (55) |
| AKI episodes with measured baseline creatinine, n (%) | 56,332 (91) | 55,762 (91) | 57,170 (91) | 55,972 (91) | — |
| Baseline creatinine, median (IQR)b | 82 (64–108) | 84 (66–110) | 83 (66–109) | 82 (65–109) | — |
| Baseline creatinine in step 3, median (IQR)c | 77 (60–99) | 80 (64–103) | 79 (63–102) | 78 (62–100) | — |
| Baseline eGFR,d median (IQR) | 72 (50–91) | 70 (49–89) | 71 (50–89) | 71 (49–90) | — |
| Number of creatinine measurements during the preceding 365–368 days, median (IQR) | |||||
| Identified in step 1 or 2 | 6 (2–15) | 6 (2–15) | 6 (2–15) | 6 (2–15) | 7 (3–16) |
| Identified in step 3 | 5 (2–11) | 5 (2–12) | 5 (2–12) | 5 (2–11) | — |
| Identified by an inpatient creatinine measurement, n (%) | 47,687 (77) | 47,032 (77) | 47,297 (76) | 47,449 (77) | 35,557 (94) |
| Charlson comorbidity index score, n (%) | |||||
| 0 | 13,543 (22) | 13,001 (21) | 13,302 (21) | 13,281 (22) | 6142 (16) |
| 1–2 | 31,771 (51) | 31,596 (52) | 32,367 (52) | 31,667 (52) | 19,655 (52) |
| 3+ | 16,445 (27) | 16,592 (27) | 16,928 (27) | 16,451 (27) | 11,862 (31) |
| Comorbidity, n (%) | |||||
| Myocardial infarction | 5245 (8) | 5287 (9) | 5364 (9) | 5239 (9) | 3837 (10) |
| Congestive heart failure | 10,466 (17) | 10,706 (17) | 10,903 (17) | 10,533 (17) | 7506 (20) |
| Diabetes | 16,590 (27) | 16,528 (27) | 16,955 (27) | 16,487 (27) | 10,052 (27) |
| Hypertension | 27,883 (45) | 27,847 (46) | 28,487 (46) | 27,765 (45) | 18,066 (48) |
| Cerebrovascular disease | 10,629 (17) | 10,519 (17) | 10,696 (17) | 10,572 (17) | 7050 (19) |
| Liver disease | 3196 (5) | 3205 (5) | 3267 (5) | 3184 (5) | 2144 (6) |
| Hospital-diagnosed obesity | 5877 (10) | 5765 (9) | 5936 (9) | 5814 (9) | 3459 (9) |
| Medications, n (%) | |||||
| Diuretics (thiazides and loop) | 38,031 (62) | 38,137 (62) | 39,028 (62) | 37,928 (62) | 23,997 (64) |
| Statins | 32,218 (52) | 32,107 (52) | 32,913 (53) | 32,100(52) | 19,927 (53) |
| ACEI/ARB | 38,495 (62) | 38,489 (63) | 39,454 (63) | 38,246 (62) | 23,599 (63) |
| Rate per 1000 person-years, n (95% CI) | 13.5 | 13.4 | 13.7 | 13.4 | 8.2 |
| (13.4 to 13.6) | (13.3 to 13.5) | (13.6 to 13.8) | (13.3 to 13.5) | (8.1 to 8.3) | |
| 30-day mortality, % (95% CI) | 18 | 18 | 17 | 18 | 23 |
| (17 to 18) | (17 to 18) | (17 to 18) | (18 to 18) | (23 to 24) | |
| Identified in step 1 or 2, % (95% CI) | 22 | 22 | 22 | 22 | — |
| (21 to 22) | (21 to 22) | (21 to 22) | (21 to 22) | ||
| Identified in step 3, % (95% CI) | 14 | 14 | 13 | 14 | — |
| (13 to 14) | (12 to 14) | (13 to 14) | (13 to 14) | ||
IQR, interquartile range; CI, confidence interval; ACE, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker.
AKI episodes identified by a 1.5-fold increase from baseline creatinine value and not caught in step 1 or 2 in detection of AKI.
Among AKI episodes with a defined baseline creatinine.
Among AKI episodes identified by a 1.5-fold increase from baseline creatinine.
Estimated from baseline creatinine.
We made an upset plot to provide an overview of the overlap of the five cohorts and calculated the rate of AKI per 1000 person-years in 2017 for each definition. The denominator was the number of Danish inhabitants aged ≥18 years in 2017 according to census numbers from Statistics Denmark (N=4,580,547) (16). Finally, we used the Kaplan–Meier method to estimate 30-day all-cause mortality (with 95% confidence intervals [CIs]) for all cohorts. To examine if 30-day all-cause mortality differed between those identified in step 1 or 2 and those identified in step 3, we also plotted 30-day all-cause mortality curves separately for those identified in step 1 or 2 and step 3 for each of the four cohorts with a baseline creatinine measurement.
To address the effect of unavailable baseline creatinine, we conducted an additional analysis replacing all missing baselines with a creatinine value estimated from the Chronic Kidney Disease Epidemiology Collaboration equation, assuming eGFR to be 75 ml/min per 1.73 m2 according to the recommendation from KDIGO (4,27). As a post hoc analysis, we also tabulated characteristics of the people who were identified with AKI when applying the recent definition but not by the other definitions.
Results
Characteristics of the Five AKI Cohorts
Among the 2,095,850 people with at least one creatinine measurement in 2017, the median definition found the highest number of AKI episodes (n=62,597) and the mean definition found the lowest number of AKI episodes (n=61,189) among the four cohorts using a baseline creatinine (Figure 1, Table 1). The no baseline cohort found a substantially lower number of AKI episodes (n=37,659; Table 1). The median age, baseline creatinine, and number of creatinine measurements 365–368 days before AKI were comparable between the five cohorts. Likewise, there were no considerable differences in the distribution of comorbidity and use of prescription medicine. All cohorts contained a higher proportion of men than women, and the no baseline cohort contained a slightly higher percentage of men compared with the other cohorts. The percentages of AKI episodes identified during hospitalization were similar in the four cohorts using a baseline definition, but in the no baseline cohort, the vast majority of AKI episodes were inpatients. In the recent, mean, median, and twostep cohorts, median baseline eGFR ranged from 70 to 72 ml/min per 1.73 m2.
Figure 1.
Flow chart of cohort inclusion and exclusion.
We estimated the incidence rates of AKI to be between 8.2 and 13.7 AKI cases per 1000 person-years in the five cohorts, with the lowest rate in the no baseline cohorts and the highest rate in the median cohort.
The 30-day mortality was similar for the recent, mean, median, and twostep cohorts, ranging from 17% to 18% (Table 1). People identified with step 1 or 2 of the AKI definition had considerably higher 30-day mortality (approximately 22%) than those identified in step 3 (approximately 13%–14%; Figure 2, Table 1). The 30-day all-cause mortality for the no baseline cohort was 23% (95% CI, 23 to 24; Table 1).
Figure 2.
Kaplan–Meier plot of 30-day all-cause mortality after AKI for the recent, mean, median, and two-step cohorts.
The upset plot illustrates that of 67,031 distinct individuals identified by any of the definitions, 37,659 people were identified in all five cohorts and additionally 19,358 people were identified by the four cohorts using a baseline definition (Figure 3). The recent definition identified 2240 people who were only identified in this cohort (Table 2). Compared with the cohorts in Table 1, those 2240 people had a lower median age, a lower percentage of men, and fewer comorbidities. Baseline creatinine and percentage of people identified during hospitalization were also substantially lower in this group. The 30-day all-cause mortality for this group was 7% (95% CI, 6 to 8; Table 2).
Figure 3.
Upset plot showing how many people were included in one or more of the AKI cohorts sampled using different definitions for baseline kidney function.
Table 2.
Characteristics, comorbidity, medication, rate, and 30-day mortality of people only identified with an AKI episode when applying the recent definition for estimating baseline creatinine
| Only Identified in Recent Definition | |
|---|---|
| AKI episodes, n | 2240 |
| AKI episodes identified in step 3,a n (%) | 2240 (100) |
| Age, yr, median (IQR) | 70 (57–79) |
| Men, n (%) | |
| Episodes with a defined baseline, n (%) | 2240 (100) |
| Baseline creatinine, median (IQR) | 64 (51–83) |
| Baseline eGFR,b median (IQR) | 89 (68–103) |
| Number of creatinine measurements during the preceding 365–368 days, median (IQR) | 6 (3–13) |
| Identified by an inpatient creatinine measurement, n (%) | 828 (37) |
| Charlson comorbidity index score, n (%) | |
| 0 | 643 (29) |
| 1–2 | 1132 (51) |
| 3+ | 465 (21) |
| Comorbidity | |
| Myocardial infarction | 111 (5) |
| Congestive heart failure | 251 (11) |
| Diabetes | 640 (29) |
| Hypertension | 904 (40) |
| Cerebrovascular disease | 332 (15) |
| Liver disease | 95 (4) |
| Hospital-diagnosed obesity | 265 (12) |
| Medications | |
| Diuretics (thiazides and loop) | 1276 (57) |
| Statins | 1123 (50) |
| ACEI/ARB | 1331 (60) |
| 30-day all-cause mortality, % (95% CI) | 7 (6 to 8) |
IQR, interquartile range; CI, confidence interval; ACE, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker.
AKI episodes identified by a 1.5-fold increase from baseline creatinine value and not caught in step 1 or 2 in detection of AKI.
Estimated from baseline creatinine.
Additional Analyses
Table 3 shows that many more AKI episodes were found by all four different baseline definitions when assuming eGFR to be 75 ml/min per 1.73 m2 in people without a baseline creatinine. All covariates were comparable between the four cohorts. The analysis assuming eGFR to be 75 ml/min per 1.73 m2 identified cohorts that were slightly older and had fewer comorbidities, had lower 30-day all-cause mortality, and a lower percentage was identified by an inpatient measurement (Table 3). A higher proportion of episodes were identified in step 3 from the baseline creatinine value, and they had a lower baseline creatinine.
Table 3.
Characteristics, comorbidity, medication, rate, and 30-day mortality of four AKI cohorts with different baseline creatinine definitions assuming eGFR to be 75 ml/min per 1.73 m2 when baseline was missing
| AKI Cohorts Identified by Different Baseline Definitions | ||||
|---|---|---|---|---|
| Recent | Mean | Median | Two-Step | |
| AKI episodes, n | 81,686 | 81,137 | 82,525 | 81,331 |
| AKI episodes identified in step 3,a n (%) | 52,582 (64) | 52,357 (65) | 54,007 (65) | 52,303 (64) |
| Age, yr, median (IQR) | 76 (66–85) | 76 (66–85) | 76 (66–85) | 76 (66–85) |
| Men, n (%) | 39,555 (48) | 39,613 (49) | 40,228 (49) | 39,488 (49) |
| Baseline creatinine, median (IQR)b | 78 (64–95) | 79 (65–96) | 79 (65–95) | 78 (64–95) |
| Baseline creatinine in step 3, median (IQR)c | 73 (63–87) | 75 (64–89) | 75 (64–89) | 73 (63–88) |
| Baseline eGFR,d median (IQR) | 75 (61–83) | 75 (60–81) | 75 (61–82) | 75 (61–82) |
| Number of creatinine measurements during the preceding 365–368 days, median (IQR) | ||||
| Identified in step 1 or 2 | 7 (3–16) | 7 (3–15) | 7 (3–15) | 7 (3–16) |
| Identified in step 3 | 5 (2–11) | 5 (2–11) | 5 (2–12) | 5 (2–11) |
| Identified by an inpatient creatinine measurement, n (%) | 52,733 (65) | 52,090 (64) | 52,345 (63) | 52,500 (65) |
| Charlson comorbidity index score, n (%) | ||||
| 0 | 22,168 (27) | 21,635 (27) | 21,932 (27) | 21,913 (27) |
| 1–2 | 40,900 (50) | 40,734 (50) | 41,491 (50) | 40,794 (50) |
| 3+ | 18,618 (23) | 18,768 (23) | 19,102 (23) | 18,624 (23) |
| Comorbidity, n (%) | ||||
| Myocardial infarction | 6435 (8) | 6479 (8) | 6553 (8) | 6429 (8) |
| Congestive heart failure | 12,397 (15) | 12,642 (16) | 12,836 (16) | 12,466 (15) |
| Diabetes | 19,460 (24) | 19,405 (24) | 19,828 (24) | 19,362 (24) |
| Hypertension | 35,746 (44) | 35,718 (44) | 36,348 (44) | 35,630 (44) |
| Cerebrovascular disease | 13,875 (17) | 13,767 (17) | 13,940 (17) | 13,820 (17) |
| Liver disease | 3344 (4) | 3353 (4) | 3414 (4) | 3331 (4) |
| Hospital-diagnosed obesity | 6714 (8) | 6602 (8) | 6771 (8) | 6650 (8) |
| Medications, n (%) | ||||
| Diuretics (thiazides and loop) | 50,820 (62) | 50,943 (63) | 51,821 (63) | 50,724 (62) |
| Statins | 41,923 (51) | 41,821 (52) | 42,621 (52) | 41,808 (51) |
| ACEI/ARB | 51,521 (63) | 51,532 (64) | 52,484 (64) | 51,379 (63) |
| Rate per 1000 person-years, n (95% CI) | 17.8 (17.7 to 18.0) | 17.7 (17.6 to 17.8) | 18.0 (17.9 to 18.1) | 17.8 (17.6 to 17.9) |
| 30-day mortality, % (95% CI) | 14 (14 to 14) | 14 (14 to 14) | 14 (14 to 14) | 14 (14 to 14) |
IQR, interquartile range; CI, confidence interval; ACE, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker.
AKI episodes identified by a 1.5-fold increase from baseline creatinine value and not caught in step 1 or 2 in detection of AKI.
Among AKI episodes with a defined baseline creatinine.
Among AKI episodes identified by a 1.5-fold increase from baseline creatinine.
Estimated from baseline creatinine.
Discussion
Key Findings
In this study, we described four AKI cohorts created using different baseline definitions and one cohort created without applying a baseline creatinine. Overall, the four AKI cohorts using preadmission baseline creatinine had similar distributions of baseline characteristics and yielded similar rates of AKI and similar 30-day mortality. Without preadmission baseline creatinine, fewer AKI cases were included, yielding a lower rate of AKI and higher 30-day mortality. Many people were identified with an AKI episode in all five cohorts, and a large majority was identified by all four cohorts using a baseline definition.
Limitations
Some limitations should be considered when interpreting the findings from this nationwide, population-based study. First, although we consider the databases to be complete, we may have underestimated the true rate of AKI in the Danish population because we could not detect AKI in people without creatinine measurements (e.g., subclinical cases of AKI). As the aim of this study was to compare the cohorts, we do not expect that this possible underestimation has biased our results because the same denominator was used in all cohorts. Second, we cannot be sure that measured baseline creatinine reflects the real stable baseline kidney function because some outpatient creatinine measurements may be taken during acute disease in people without hospital admissions. Thus, we might have underestimated the incidence of AKI because measurements during outpatient disease most likely would yield higher baseline creatinine estimates, although many outpatient creatinine measurements are routine measurements reflecting stable kidney function, and we assume that estimation of baseline kidney function from outpatient creatinine measurements makes a more precise baseline estimation compared with, for example, assuming a baseline eGFR of 75 ml/min per 1.73 m2 (6). Third, we estimated baseline creatinine from four different definitions using measurements during the preceding 365–368 days. Consequently, some people will not have a baseline value because of no outpatient measurements in this period. One could consider a solution of assuming these people’s eGFR to be 75 ml/min per 1.73 m2. However, that might in some cases misclassify people with a higher baseline eGFR as not having AKI and people with a lower baseline eGFR as having AKI (6,8–10). A longer or a shorter period would certainly influence the baseline value, but 365 days is a commonly used time frame to estimate baseline creatinine (7,12–14). Using a longer look-back period may lead to an underestimate of the baseline creatinine, especially in people with a decreasing kidney function. On the other hand, a shorter period would lead to fewer people having a baseline creatinine measurement, and it would increase the effect of outliers among the people with baseline creatinine measurements.
Interpretation
Although we only found small differences in characteristics between the four cohorts using a baseline definition, the continued use of different baseline definitions limits accuracy and comparability in studies of AKI. This is particularly an issue in a setting with less complete data. Despite similarities, all baseline definitions identified some people with AKI who were not identified by the other definitions or only identified by some of the other definitions. In particular, the recent definition identified a large group of people (n=2240) who were not identified by the other definitions. These seemed different from the other episodes identified with fewer comorbidities and lower 30-day mortality. This high number, only identified by the recent definition, could be due to regression to the mean because a single spurious low measurement may lead to detection of AKI at the next measurement. The most recent outpatient creatinine measurement may be the most precise up-to-date estimate for baseline kidney function, but only in people with stable kidney function where the creatinine value is not affected by any condition that influences creatinine. Using the median or the mean value up to 1 year before the measurement could underestimate baseline creatinine in people with a rapid decline in kidney function. We were also concerned that using the median or mean definitions could overestimate baseline creatinine if multiple repetitive outpatient measurements were taken in a period with high creatinine, but that does not appear to be highly problematic because the median definition found the highest number of AKI episodes. The no baseline cohort found far fewer AKI episodes, and these episodes were more frequently identified during hospitalization, which was expected because steps 1 and 2 in AKI identification required measurements within 48 hours or 7 days, respectively.
Much research has been done on AKI, but it remains unclear how to define baseline creatinine (5). Many AKI episodes are community acquired (i.e., developing before hospital admission), and detection of such AKI episodes may be hindered by a lack of pre-AKI creatinine measurements in the community setting within the last week (28). It may be problematic to use the first or the lowest creatinine measurement during the hospital admission as baseline because this baseline creatinine presumably is higher than the actual stable baseline kidney function (6,8,15,28). Another study examined misclassification of AKI when defining baseline creatinine solely by measurements during admission or imputed from an eGFR of 75 ml/min per 1.73 m2 (6). Compared with a known preadmission outpatient baseline value, they found that these frequently used methods for estimating baseline creatinine resulted in bidirectional misclassification of incidence and prognosis of AKI. We are aware that many researchers lack access to outpatient or preadmission measurements but examining the best solution for missing preadmission creatinine is outside the scope of this study. Therefore, no inferences can be made about estimating baseline creatinine if data on preadmission outpatient creatinine measurements are lacking. The optimal definition of baseline creatinine may depend on the setting and which creatinine measurements are available. Because preadmission creatinine is not always available, it is even more important to unify the definition of baseline creatinine and use the same definition, when preadmission creatinine is available, to promote comparability between these studies.
In conclusion, we suggest that the definition of baseline kidney function should be standardized in AKI research. We show that using a baseline creatinine value is important in the identification of biochemically defined AKI, especially to ensure identification of community-acquired AKI. Although the four definitions of baseline creatinine yielded comparable cohorts in our setting, it would be beneficial for AKI research to standardize the baseline definition because it would improve comparability between AKI studies. If no common baseline definition exists, it is important that researchers carefully evaluate which baseline definition to use for a specific research question and are aware of which people might be misclassified because of the chosen baseline definition.
Disclosures
C. Christiansen reports that various companies provided research grants to Aarhus University. U. Heide-Jørgensen reports that the Department of Clinical Epidemiology at Aarhus University Hospital is involved in studies with funding from various companies as research grants to (and administered by) Aarhus University but that none of these studies are related to the current study. S. Vestergaard reports that the Department of Clinical Epidemiology is involved in studies with funding from various companies as research grants to (and administered by) Aarhus University. All remaining authors have nothing to disclose.
Funding
The study was supported by the Independent Research Fund Denmark (grant number 0134-00407B).
Author Contributions
C. Christiansen conceptualized the study, curated the data, and was responsible for the formal analysis, investigation, methodology, supervision, and visualization, and for reviewing and editing the article. H. Graversen conceptualized the study, curated the data, and was for responsible for the formal analysis, investigation, methodology, project administration, and visualization, and for writing the original draft of the article. S. Jensen and U. Heide-Jorgensen conceptualized the study, curated the data, and were responsible for the formal analysis, investigation, methodology, supervision, and visualization, and for reviewing and editing the article. S. Vestergaard was responsible for data curation, investigation, supervision, and visualization, and for reviewing and editing the article.
Supplemental Material
This article contains the following supplemental material online at http://kidney360.asnjournals.org/lookup/suppl/doi:10.34067/KID.0006082021/-/DCSupplemental.
Download Supplemental Figure 1, PDF file, 677 KB (676.3KB, pdf)
Download Supplemental Table 1, PDF file, 677 KB (676.3KB, pdf)
References
- 1.Lameire NH, Bagga A, Cruz D, De Maeseneer J, Endre Z, Kellum JA, Liu KD, Mehta RL, Pannu N, Van Biesen W, Vanholder R: Acute kidney injury: An increasing global concern. Lancet 382: 170–179, 2013. 10.1016/S0140-6736(13)60647-9 [DOI] [PubMed] [Google Scholar]
- 2.Chawla LS, Eggers PW, Star RA, Kimmel PL: Acute kidney injury and chronic kidney disease as interconnected syndromes. N Engl J Med 371: 58–66, 2014. 10.1056/NEJMra1214243 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Thakar CV, Christianson A, Freyberg R, Almenoff P, Render ML: Incidence and outcomes of acute kidney injury in intensive care units: A Veterans Administration study. Crit Care Med 37: 2552–2558, 2009. 10.1097/CCM.0b013e3181a5906f [DOI] [PubMed] [Google Scholar]
- 4.Kidney Disease Improving Global Outcomes : KDIGO Clinical Practice Guideline for Acute Kidney Injury. Section 2: AKI Definition. Available at https://kdigo.org/wp-content/uploads/2016/10/KDIGO-2012-AKI-Guideline-English.pdf. Accessed December 7, 2021
- 5.Ostermann M, Bellomo R, Burdmann EA, Doi K, Endre ZH, Goldstein SL, Kane-Gill SL, Liu KD, Prowle JR, Shaw AD, Srisawat N, Cheung M, Jadoul M, Winkelmayer WC, Kellum JA; Conference Participants : Controversies in acute kidney injury: Conclusions from a Kidney Disease: Improving Global Outcomes (KDIGO) Conference. Kidney Int 98: 294–309, 2020. 10.1016/j.kint.2020.04.020 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Siew ED, Matheny ME, Ikizler TA, Lewis JB, Miller RA, Waitman LR, Go AS, Parikh CR, Peterson JF: Commonly used surrogates for baseline renal function affect the classification and prognosis of acute kidney injury. Kidney Int 77: 536–542, 2010. 10.1038/ki.2009.479 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Siew ED, Ikizler TA, Matheny ME, Shi Y, Schildcrout JS, Danciu I, Dwyer JP, Srichai M, Hung AM, Smith JP, Peterson JF: Estimating baseline kidney function in hospitalized patients with impaired kidney function. Clin J Am Soc Nephrol 7: 712–719, 2012. 10.2215/CJN.10821011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Hoste EA, Clermont G, Kersten A, Venkataraman R, Angus DC, De Bacquer D, Kellum JA: RIFLE criteria for acute kidney injury are associated with hospital mortality in critically ill patients: A cohort analysis. Crit Care 10: R73, 2006. 10.1186/cc4915 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Závada J, Hoste E, Cartin-Ceba R, Calzavacca P, Gajic O, Clermont G, Bellomo R, Kellum JA; AKI6 investigators : A comparison of three methods to estimate baseline creatinine for RIFLE classification. Nephrol Dial Transplant 25: 3911–3918, 2010. 10.1093/ndt/gfp766 [DOI] [PubMed] [Google Scholar]
- 10.Kashani K, Shao M, Li G, Williams AW, Rule AD, Kremers WK, Malinchoc M, Gajic O, Lieske JC: No increase in the incidence of acute kidney injury in a population-based annual temporal trends epidemiology study. Kidney Int 92: 721–728, 2017. 10.1016/j.kint.2017.03.020 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Crellin E, Mansfield KE, Leyrat C, Nitsch D, Douglas IJ, Root A, Williamson E, Smeeth L, Tomlinson LA: Trimethoprim use for urinary tract infection and risk of adverse outcomes in older patients: Cohort study. BMJ 360: k341, 2018. 10.1136/bmj.k341 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Liu KD, Yang J, Tan TC, Glidden DV, Zheng S, Pravoverov L, Hsu CY, Go AS: Risk factors for recurrent acute kidney injury in a large population-based cohort. Am J Kidney Dis 73: 163–173, 2019. 10.1053/j.ajkd.2018.08.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Sawhney S, Marks A, Fluck N, Levin A, Prescott G, Black C: Intermediate and long-term outcomes of survivors of acute kidney injury episodes: A large population-based cohort study. Am J Kidney Dis 69: 18–28, 2017. 10.1053/j.ajkd.2016.05.018 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Graversen HV, Nørgaard M, Nitsch D, Christiansen CF: Preadmission kidney function and risk of acute kidney injury in patients hospitalized with acute pyelonephritis: A Danish population-based cohort study. PLoS One 16: e0247687, 2021. 10.1371/journal.pone.0247687 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Liu KD, Hsu CY, Yang J, Tan TC, Zheng S, Ordonez JD, Go AS: Acute kidney injury ascertainment is affected by the use of first inpatient versus outpatient baseline serum creatinine. Kidney Int Rep 3: 211–215, 2017. 10.1016/j.ekir.2017.08.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Statistics Denmark : Population in Denmark. Available at: https://www.dst.dk/en/Statistik/emner/befolkning-og-valg/befolkning-og-befolkningsfremskrivning/folketal. Accessed June 26, 2021
- 17.Schmidt M, Schmidt SAJ, Adelborg K, Sundbøll J, Laugesen K, Ehrenstein V, Sørensen HT: The Danish health care system and epidemiological research: From health care contacts to database records. Clin Epidemiol 11: 563–591, 2019. 10.2147/CLEP.S179083 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Schmidt M, Pedersen L, Sørensen HT: The Danish Civil Registration System as a tool in epidemiology. Eur J Epidemiol 29: 541–549, 2014. 10.1007/s10654-014-9930-3 [DOI] [PubMed] [Google Scholar]
- 19.Pedersen CB, Gøtzsche H, Møller JO, Mortensen PB: The Danish Civil Registration System. A cohort of eight million persons. Dan Med Bull 53: 441–449, 2006 [PubMed] [Google Scholar]
- 20.Schmidt M, Schmidt SA, Sandegaard JL, Ehrenstein V, Pedersen L, Sørensen HT: The Danish National Patient Registry: A review of content, data quality, and research potential. Clin Epidemiol 7: 449–490, 2015. 10.2147/CLEP.S91125 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Grann AF, Erichsen R, Nielsen AG, Frøslev T, Thomsen RW: Existing data sources for clinical epidemiology: The clinical laboratory information system (LABKA) research database at Aarhus University, Denmark. Clin Epidemiol 3: 133–138, 2011. 10.2147/CLEP.S17901 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Arendt JFH, Hansen AT, Ladefoged SA, Sørensen HT, Pedersen L, Adelborg K: Existing data sources in clinical epidemiology: Laboratory information system databases in Denmark. Clin Epidemiol 12: 469–475, 2020. 10.2147/CLEP.S245060 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Pontet F, Magdal Petersen U, Fuentes-Arderiu X, Nordin G, Bruunshuus I, Ihalainen J, Karlsson D, Forsum U, Dybkaer R, Schadow G, Kuelpmann W, Férard G, Kang D, McDonald C, Hill G; Joint Committee on Nomenclature, Properties and Units (C-SC-NPU) of the IFCC and IUPAC : Clinical laboratory sciences data transmission: The NPU coding system. Stud Health Technol Inform 150: 265–269, 2009 [PMC free article] [PubMed] [Google Scholar]
- 24.Levey AS, Eckardt K-U, Dorman NM, Christiansen SL, Hoorn EJ, Ingelfinger JR, Inker LA, Levin A, Mehrotra R, Palevsky PM, Perazella MA, Tong A, Allison SJ, Bockenhauer D, Briggs JP, Bromberg JS, Davenport A, Feldman HI, Fouque D, Gansevoort RT, Gill JS, Greene EL, Hemmelgarn BR, Kretzler M, Lambie M, Lane PH, Laycock J, Leventhal SE, Mittelman M, Morrissey P, Ostermann M, Rees L, Ronco P, Schaefer F, St. Clair Russell J, Vinck C, Walsh SB, Weiner DE, Cheung M, Jadoul M, Winkelmayer WC: Nomenclature for kidney function and disease: Report of a Kidney Disease: Improving Global Outcomes (KDIGO) Consensus Conference. Kidney Int 97: 1117–1129, 2020. 10.1016/j.kint.2020.02.010 [DOI] [PubMed] [Google Scholar]
- 25.Pottegård A, Schmidt SAJ, Wallach-Kildemoes H, Sørensen HT, Hallas J, Schmidt M: Data resource profile: The Danish National Prescription Registry. Int J Epidemiol 46: 798–798f, 2017. 10.1093/ije/dyw213 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Brusselaers N, Lagergren J: The Charlson comorbidity index in registry-based research. Methods Inf Med 56: 401–406, 2017. 10.3414/ME17-01-0051 [DOI] [PubMed] [Google Scholar]
- 27.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. 10.7326/0003-4819-150-9-200905050-00006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Wonnacott A, Meran S, Amphlett B, Talabani B, Phillips A: Epidemiology and outcomes in community-acquired versus hospital-acquired AKI. Clin J Am Soc Nephrol 9: 1007–1014, 2014. 10.2215/CJN.07920713 [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.
Supplementary Materials
Download Supplemental Figure 1, PDF file, 677 KB (676.3KB, pdf)
Download Supplemental Table 1, PDF file, 677 KB (676.3KB, pdf)