Utility of Patient-Reported Risk Factors for Identifying Advanced Chronic Kidney Disease Before Outpatient CT: Comparison With Recent ACR/NKF Consensus Criteria
Abstract
Please see the Editorial Comment by Shao-Bin Duan discussing this article.
BACKGROUND. Practices vary for screening patients for risk of renal dysfunction before administration of iodinated contrast medium. A 2020 American College of Radiology/National Kidney Foundation (ACR/NKF) consensus statement provided streamlined screening criteria.
OBJECTIVE. The purpose of this study was to assess the yield of patient-reported risk factors for identifying estimated glomerular filtration rate (eGFR) less than 30 mL/min/1.73 m2 before outpatient CT.
METHODS. This retrospective study was performed at a health system that implemented an electronic screening form for patients to complete before outpatient CT encounters to report undergoing dialysis, taking cancer-treating medications, having kidney disease, undergoing prior kidney surgery, having diabetes mellitus treated with medication, having hypertension treated with medication, or having multiple myeloma. Patients with any risk factor were required to undergo eGFR testing before CT. Of 44,708 patients completing the form from June 1, 2020, through February 28, 2021, 10,256 patients (5315 men, 4941 women; mean age, 66.8 ± 11.9 [SD] years; range, 21–98 years) underwent eGFR testing on the day of CT. Multivariable regression analysis for predicting reduced eGFR was performed. Findings were compared with those from theoretic use of the ACR/NKF criteria.
RESULTS. Same-day testing yielded eGFR less than 30 mL/min/1.73 m2 in 1.4% (144/10,256) of patients. The only significant independent predictors of low eGFR were dialysis (odds ratio [OR], 203.30], kidney disease (OR, 12.55), and diabetes mellitus treated with medication (OR, 2.44). If the ACR/NKF criteria (only kidney disease, defined as dialysis, kidney disease, or prior kidney surgery) had been followed as a trigger for eGFR testing, the number of patients needing testing would have decreased 89.7%, from 10,256 to 1059; yield would have increased to 7.2% (76/1059); and 47.2% (68/144) of patients with low eGFR would have been missed. If the ACR/NKF criteria had been followed but diabetes mellitus been considered a required rather than an optional criterion, the number of patients needing testing would have decreased 77.1%, to 2353; yield would have increased to 4.0% (95/2353); and 34.0% (49/144) of patients with low eGFR would have been missed.
CONCLUSION. Using patient-reported risk factors resulted in frequent eGFR testing but low yield of low eGFR. Commonly applied risk factors were not independently associated with low eGFR.
CLINICAL IMPACT. Application of ACR/NKF criteria would substantially reduce eGFR testing, but patients with renal dysfunction would be missed. The statement should consider omitting kidney surgery as a trigger for eGFR testing and including diabetes mellitus as a required trigger.
HIGHLIGHTS
Key Finding
▪ Patient-reported risk factors had low yield (1.4%) of eGFR < 30 mL/min/1.73 m2 before outpatient CT encounters. Dialysis (OR, 203.30), kidney disease (OR, 12.55), and medication-treated diabetes (OR, 2.44) independently predicted low eGFR. Cancer medications, prior kidney surgery, medication-treated hypertension, and multiple myeloma were not independently associated with low eGFR.
Importance
▪ Applying the ACR/NKF criteria but treating diabetes mellitus as a required rather than optional trigger may decrease unnecessary eGFR testing before IV contrast medium administration.
Exposure to IV contrast medium was historically viewed as a common cause of acute kidney injury (AKI) after CT examinations, often leading to withholding of IV contrast media for CT examinations of patients with renal insufficiency [1–3]. Controversy has existed around optimal strategies for identifying patients at risk of contrast-induced AKI and determining which patients should receive contrast media and from which patients it should be withheld [4, 5]. Results of large retrospective studies performed with propensity score matching have suggested that if preexisting renal insufficiency is accepted as a risk factor for contrast-associated AKI, then the increased risk of contrast-associated AKI is most likely present for patients with an estimated glomerular filtration rate (eGFR) less than 30 mL/min/1.73 m2 [6, 7]. Therefore, according to recommendations from the American College of Radiology (ACR) contrast media manual [8], if an eGFR cutoff is used to determine whether to administer contrast media, a cutoff of 30 mL/min/1.73 m2 is supported by the strongest evidence.
To reduce unnecessary laboratory testing and increase the yield of eGFR assessment, studies have explored risk factors predictive of reduced eGFR [9–11], so that patients could be screened for these risk factors, and eGFR testing would be performed only in those at risk. Although it is now standard practice to screen for risk factors to select patients for eGFR testing before CT examinations, approaches for such screening vary widely in terms of the specific risk factors used for screening. In 2020, a joint consensus statement from the ACR and National Kidney Foundation (ACR/NKF) [12] provided a standardized approach to handling administration of iodinated contrast medium to patients with kidney disease. This consensus statement proposed screening patients for only a personal history of kidney disease (defined in the statement as known chronic kidney disease, remote history of AKI, dialysis, kidney surgery, kidney ablation, or albuminuria) and, optionally, history of diabetes mellitus. The ACR contrast media manual had also previously recommended using age 60 years or older or a history of hypertension as a trigger for obtaining an eGFR level [12, 13], but these were removed after release of the consensus statement. Various other risk factors have also been used in clinical practice.
Since the release of the ACR/NKF statement, a paucity of research has explored the impact of screening based on the statement recommendations versus other combinations of risk factors used in clinical practice. Moreover, the previously noted studies were performed before recognition of 30 mL/min/1.73 m2 as the threshold below which patients are at increased risk of contrast-associated AKI and instead assessed for risk factors predictive of other thresholds, such as eGFR less than 45 mL/min/1.73 m 2 or a creatinine level greater than values ranging from 1.7 to 2.0 mg/dL [9–11]. Greater insight into the specific risk factors most strongly predictive of eGFR below the currently accepted threshold would help standardize screening practices. The purpose of this study was to assess the yield of patient-reported risk factors for identifying eGFR less than 30 mL/min/1.73 m2 before administration of iodinated IV contrast medium for CT.
Methods
This HIPAA-compliant retrospective study was approved by the institutional review board of Mass General Brigham. The requirement for written informed consent was waived. The study was conducted at a large health system comprising 12 sites (two large academic institutions, nine community hospitals, and a large outpatient cancer center).
Screening for Risk of Contrast-Induced Acute Kidney Injury in Outpatient CT Encounters
An electronic screening form for assessment of risk of contrast-associated AKI was developed by consensus of an institutional (Mass General Brigham) multidisciplinary expert panel. The form was embedded in the electronic health record (EHR; Epic Hyperspace, November 2021, Epic Systems) on June 1, 2020, for implementation throughout the health system. Before the launch date, introduction of the form was communicated to leadership at each site.
Patients could optionally complete the form before arrival for any outpatient CT encounter using the patient-facing component of the EHR. Because the patient portal did not interface with the radiology protocoling module, patients were invited to complete the online screening form before all outpatient CT encounters whether or not the CT examination was protocoled with or without IV contrast medium administration. If the patient completed the electronic form before the encounter, the CT technologist reviewed the responses at the time of the encounter before potential IV contrast medium administration to determine whether eGFR measurement was required and, if so, whether an eGFR value was present in the EHR or had to be obtained. If the form was not completed before the encounter, the technologist conducted the screening by asking the patient the same questions that appeared on the form and recording the responses in the EHR or by completing a paper version of the form that was then scanned into the EHR. Thus, all patients ultimately underwent screening before the CT examination, whether or not the electronic form was completed in advance of the encounter.
The electronic screening form contained multiple questions intended to identify patients at risk of stage 4 or 5 chronic kidney disease (eGFR < 30 mL/min/1.73 m2), representing patients considered at increased risk of development of contrast-associated AKI. All questions were optional. Patients were asked to report whether they were currently undergoing dialysis (Yes or No), whether they were currently using any medications to treat cancer (Yes or No), and whether they had any of the following conditions (options to indicate presence of each condition or None of the above): kidney disease, history of kidney surgery, diabetes mellitus treated with medication, hypertension treated with medication, and multiple myeloma. Patients who responded affirmatively to any survey item (except for the question regarding cancer-treating medication) were required to undergo eGFR testing within 30 days before the date of the CT encounter, reflecting recommendations in the ACR contrast media manual [13].
At three of the sites in the health system (all with large populations of patients with cancer), patients were also required to undergo eGFR testing within 30 days before CT if they responded affirmatively to the question regarding cancer-treating medications. Five sites in the health system also required eGFR measurement within 6 months before the encounter if the patient was at least 60 years old (based on a manual check of the patient's age in the EHR by the technologist at the time of the encounter); the other seven sites did not require eGFR testing based on an age threshold.
If an eGFR measurement was required but no eGFR result was available in the EHR within the specified timeline (1 month in the presence of an affirmative response to any self-reported risk factor or 6 months in the presence of age 60 years or older), the technologist initiated the process of the patient's undergoing eGFR testing on the same day as the CT encounter. This assessment was performed by point-of-care testing at most sites, which returned results in approximately 15 minutes. A small number of sites without point-of-care testing available required the patient to visit a central laboratory, which returned results in approximately 1–2 hours. The same processes were used for same-day eGFR assessment of patients who underwent risk factor screening by the CT technologist on the day of the encounter.
Nine of the 12 sites in the health system adopted the electronic form on the implementation date. Three sites did not implement the form and continued using the paper version only. Sites adopting the form may have conducted local efforts to encourage form use by patients scheduled for CT encounters at the site.
Study Sample
Figure 1 shows the patient selection flow diagram. The institutional clinical data warehouse was searched for all CT encounters performed at the health system from June 1, 2020 (the date of implementation of the electronic screening form in the EHR), through February 28, 2021. The search identified 264,709 CT encounters in 166,038 patients. Individual encounters could potentially include CT examinations of multiple anatomic regions (e.g., chest CT and abdominopelvic CT). Of these, encounters were excluded for the following reasons: duplicate, patient already included in the sample (n = 42,056); inpatient or emergency department setting (n = 114,467). After these exclusions, 108,186 patients remained who underwent an outpatient CT encounter during the study period (50.3% [54,459] with IV contrast medium administration; 49.7% [53,727] performed without IV contrast media administration).
The electronic screening form was completed by 44,708 (41.3%) patients (including 35,805 [65.7%] who underwent contrast-enhanced examinations, maximum form use across sites of 7759 of 9511 [81.6%], and minimal form use across sites that adopted the form of 0 of 3607 [0%]). The 63,478 patients who did not complete the electronic screening form before the encounter were excluded. Of those who completed the electronic screening form, 10,256 (22.9%) underwent same-day eGFR testing; the other 34,452 patients did not undergo same-day eGFR testing and were excluded. The final study sample comprised the 10,256 patients (5315 men, 4941 women; mean age, 66.8 ± 11.9 [SD] years; age range, 21–98 years) who completed electronic screening forms in the EHR before outpatient CT encounters and who also underwent same-day eGFR testing. Of the patients in the final study sample, 94.2% (9662/10,256) underwent CT with IV contrast medium administration, and 5.8% (595/10,256) underwent CT without IV contrast medium administration.
To avoid bias introduced by a potential change in protocol on the day of the encounter from a contrast-enhanced to an unenhanced protocol, eligible encounters in which the patient completed the electronic form were not excluded on the basis of whether IV contrast medium was ultimately administered. For all included patients, responses to the electronic screening form questions and the results of same-day eGFR testing were extracted from the clinical data warehouse. Presence of age 60 years or older was also recorded for all patients, whether or not the patient underwent CT at a site that included age 60 years or older as a trigger for eGFR measurement.
Manual Review of Medical Records
A single investigator (N.A., a research fellow) reviewed the EHR for all patients with same-day eGFR less than 30 mL/min/1.73 m2 who received IV contrast medium for the CT examination (n = 37) to assess for the development of contrast-associated AKI. Contrast-associated AKI was defined as an absolute increase in serum creatinine level of 0.3 mg/dL or greater or a percentage increase in serum creatinine level of 50% or greater, either occurring within 48 hours of contrast material exposure [14]. The same investigator reviewed the EHRs of 40 randomly assigned patients to assess the accuracy of patient-reported risk factors for stages 4 and 5 chronic kidney disease with the electronic records as the reference standard.
Statistical Analysis
Data were summarized descriptively by use of counts and percentages. Results for the initial screening questions with separate Yes and No responses were calculated only among patients who selected either of these options. Results for the final question regarding various conditions were calculated only among patients who selected at least one condition or who selected None of the above. The yield of same-day eGFR testing was calculated as the percentage of patients undergoing same-day eGFR testing who had an eGFR less than 30 mL/min/1.73 m2. Univariable logistic regression analysis was performed to identify associations between patient-reported predictors and eGFR less than 30 mL/min/1.73 m2. The risk factors were used as the explanatory variable and the presence of eGFR less than 30 mL/min/1.73 m2 as the outcome measure. Multivariable logistic regression analysis was performed with significant factors identified in the univariable regression analyses. Regression coefficients and odds ratios (ORs) with 95% CIs were reported.
Potential differences were calculated in the number of same-day eGFR tests performed and in the number and percentage of same-day eGFR tests yielding eGFR less than 30 mL/min/1.73 m2 between the applied set of risk factors and a theoretic alternative approach whereby only a history of kidney disease would be used as a trigger for obtaining an eGFR, as recommended by the ACR/NKF consensus statement. For this assessment, kidney disease was defined as including patient reports of current dialysis treatment and patient reports of having kidney disease or undergoing kidney surgery, consistent with the definition in the consensus statement. Risk factors among patients with eGFR less than 30 mL/min/1.73 m2 who had contrast-associated AKI after contrast medium administration and results of the manual EHR review of the randomly selected patients were summarized. Values of p ≤ .05 were considered statistically significant. All analyses were performed with Matlab Statistics Toolbox (version 2021a, MathWorks).
Results
Risk Factors and Results of Same-Day eGFR Testing
With respect to patient-reported risk factors, 0.5% (45/9379) of patients were currently undergoing dialysis, 16.6% (1542/9302) were receiving cancer-treating medications, 7.2% (667/9317) had kidney disease, 6.0% (555/9317) had undergone kidney surgery, 19.3% (1796/9317) had diabetes mellitus treated with medication, 55.2% (5140/9317) had hypertension treated with medication, and 0.3% (25/9317) had multiple myeloma. A total of 76.7% (7867/10,256) of patients were 60 years old or older. A total of 90.1% (9237/10,256) of patients had at least one patient-reported risk factor as the reason for same-day eGFR testing; 29.3% (3005/10,256) had age 60 years or older as the sole reason for same-day eGFR testing; and 9.9% (1019/10,256) underwent same-day eGFR testing despite having no documented risk factors or being 60 years old or older. The same-day eGFR testing was performed as point-of-care testing for 92.2% (9455/10,256) and as central laboratory evaluation for 7.8% (801/10,256). Same-day eGFR testing revealed eGFR less than 30 mL/min/1.73 m2 in 1.4% (144/10,256) of patients.
Associations Between Risk Factors and eGFR Less Than 30 mL/min/1.73 m2
The frequency of eGFR less than 30 mL/min/1.73 m2 among patients 60 years old or older was 1.4% (109/7867), among patients currently undergoing dialysis was 84.4% (38/45), among patients currently using cancer-treating medication was 1.2% (18/1542), among patients with kidney disease was 10.3% (69/667), among patients who had undergone prior kidney surgery was 2.5% (14/555), among patients with diabetes mellitus treated with medications was 2.4% (43/1796), among patients with hypertension treated with medication was 1.7% (85/5140), and among patients with multiple myeloma was 4.0% (1/25).
In univariable logistic regression analysis (Table 1), eGFR less than 30 mL/min/1.73 m2 was significantly associated with the following in order of decreasing OR): currently undergoing dialysis (OR, 597.79; p < .001), kidney disease (OR, 18.36; p < .001), diabetes mellitus treated with medication (OR, 2.28; p < .001), prior kidney surgery (OR, 2.05; p = .01), and hypertension treated with medication (OR, 1.83; p = .02). Age 60 years or older, currently taking cancer-treating medications, and multiple myeloma were not significantly associated with eGFR less than 30 mL/min/1.73 m2 in univariable logistic regression analyses (all p > .05). The multivariable logistic regression analysis (Table 2) included 9302 patients who responded to all screening items. In multivariable logistic regression analysis of the risk factors that were significant univariable predictors, significant independent predictors were as follows in order of decreasing OR): dialysis (OR, 203.30; p < .001), kidney disease (OR, 12.55; p < .001), and diabetes mellitus treated with medication (OR, 2.44; p < .001). Prior kidney surgery and hypertension treated with medication were not significant independent predictors (both p > .05).
| Screening Questions and Responses | eGFR < 30 mL/min/1.73 m2a | Odds Ratiob | p |
|---|---|---|---|
| Currently undergoing dialysis | |||
| Yes | 84.4 (38/45) | 597.79 (259.55–1376.80) | < .001 |
| No | 0.9 (84/9334) | Reference | Reference |
| Currently using cancer-treating medications | |||
| Yes | 1.2 (18/1542) | 0.86 (0.52–1.42) | .56 |
| No | 1.4 (105/7760) | Reference | Reference |
| Any of the following medical conditions: | |||
| Kidney disease | 10.3 (69/667) | 18.36 (12.74–26.47) | < .001 |
| Kidney surgery | 2.5 (14/555) | 2.05 (1.17–3.60) | .01 |
| Diabetes mellitus treated with medication | 2.4 (43/1796) | 2.28 (1.56–3.31) | < .001 |
| Hypertension treated with medication | 1.7 (85/5140) | 1.83 (1.24–2.69) | .002 |
| Multiple myeloma | 4.0 (1/25) | 3.13 (0.42–23.33) | .26 |
| None of the above | 0.3 (10/3100) | Reference | Reference |
| Age ≥ 60 yc | |||
| Yes | 1.4 (109/7867) | 0.95 (0.64–1.39) | .77 |
| No | 1.5 (35/2389) | Reference | Reference |
Note—Boldface type indicates statistically significant finding.
a
Values are percentage of patients with numerator and denominator in parentheses.
b
Values in parentheses are 95% CIs.
c
Not included in electronic screening form but manually checked by CT technologists.
| Risk Factor | Odds Ratio (95% CI) | p |
|---|---|---|
| Currently on dialysis | 203.30 (81.89–504.68) | < .001 |
| Kidney disease | 12.55 (8.06–12.54) | < .001 |
| Diabetes mellitus treated with medication | 2.44 (1.55–3.83) | < .001 |
| Hypertension treated with medication | 1.48 (0.94–2.33) | .09 |
| Kidney surgery | 0.78 (0.37–1.65) | .52 |
Note—Model includes significant predictors in univariable regression analysis. Boldface type indicates statistically significant finding.
Comparison With ACR/NKF Consensus Statement
If the health system had followed the ACR/NKF consensus statement and used only a history of kidney disease (defined as patient reports of currently undergoing dialysis or of having kidney disease or undergoing prior kidney surgery) as a basis for requiring eGFR testing, then the number of patients needing same-day eGFR testing would have decreased 89.7%, from 10,256 to 1059, and the percentage yield of same-day eGFR testing (i.e., yield of eGFR < 30 mL/min/1.73 m2) would have increased from 1.4% (144/10,256) to 7.2%(76/1059). However, 47.2% (68/144) of patients with eGFR less than 30 mL/min/1.73 m2 would have been missed.
The risk factors from the electronic screening forms of patients in whom eGFR less than 30 mL/min/1.73 m2 would have been missed if the ACR/NKF consensus statement had been followed were as follows: hypertension treated with medication (n = 31), diabetes mellitus treated with medication (n = 22), multiple myeloma (n = 1), and age 60 years or older (n = 58). If the health system had followed the ACR/NKF consensus statement in using a history of kidney disease (defined as patient reports of currently undergoing dialysis or of having kidney disease or undergoing prior kidney surgery) as a basis for requiring eGFR testing but also considered diabetes mellitus treated with medication a required rather than an optional criterion for eGFR testing, then the number of patients needing same-day eGFR testing would have decreased 77.1%, from 10,256 to 2353, the percentage yield of same-day eGFR testing would have increased from 1.4% to 4.0% (95/2353), and 34.0% (49/144) of patients with eGFR less than 30 mL/min/1.73 m2 would have been missed.
Contrast-Associated Acute Kidney Injury in Patients With eGFR Less Than 30 mL/min/1.73 m2 Who Received Contrast Material
A total of 24.3% (35/144) of patients with same-day eGFR less than 30 mL/min/1.73 m2 received IV contrast media for the CT examination. Of these patients, 20.0% (7/35) reported undergoing dialysis on the electronic screening form. One of the 35 patients had a documented eGFR value in the EHR within 48 hours of the contrast material administration and had contrast-associated AKI. This patient reported the following risk factors on the electronic screening form: dialysis, kidney disease, kidney surgery, and hypertension treated with medication. The patient met the criteria for contrast-associated AKI on the basis of an increase in creati-nine level from 6.5 mg/dL before administration to 7.66 mg/dL observed 24 hours after administration.
Manual Electronic Hospital Record Review of Patient-Reported Risk Factors
The manual review of the records of 40 randomly assigned patients identified discordant information. In 27.5% (11/40) of cases, risk factors documented in the EHR were not reported by the patient in the electronic screening form. The additional risk factors identified by chart review were hypertension treated with medication (n = 10), kidney disease (n = 5), diabetes mellitus treated with medication (n = 2), and dialysis (n = 1). All patient-reported risk factors were also identified in manual EHR review.
Discussion
In this retrospective study based on an electronic screening form completed by patients before outpatient CT encounters at a large health system, 22.9% of patients who completed the screening form underwent same-day eGFR testing based on identified risk factors, and the yield of such testing (i.e., identifying an eGFR < 30 mL/min/1.73 m2) was very low at 1.4%. Only three patient-reported risk factors were independently associated with increased risk of an eGFR less than 30 mL/min/1.73 m2 on same-day testing: current dialysis, history of kidney disease, and diabetes mellitus treated with medication. Other patient-reported risk factors (hypertension, kidney surgery, multiple myeloma, use of cancer-treating medications) and patient age at least 60 years were not independently predictive of eGFR less than 30 mL/min/1.73 m2. Given the low yield of same-day eGFR testing before CT and the lack of association between some traditional risk factors and eGFR less than 30 mL/min/1.73 m2, a more streamlined approach to patient screening for renal dysfunction is likely warranted.
As stated in the ACR/NKF consensus statement, using multiple risk factors, such as age, hypertension, and others, as triggers for eGFR assessment is overly sensitive and leads to many false-positive tests. Our findings corroborate the recommendation in the ACR/NKF consensus statement to use a history of kidney disease (including dialysis) as a trigger for laboratory assessment of renal function. The ACR/NKF consensus statement, however, includes a history of kidney surgery in its definition of kidney disease; this risk factor was not independently associated with an eGFR less than 30 mL/min/1.73 m2 in our study. Diabetes mellitus treated with medication, however, was independently associated with an eGFR less than 30 mL/min/1.73 m2.
If the health system had followed the ACR/NKF consensus statement and used only a history of kidney disease as a trigger for eGFR measurement rather than the expanded criteria in use during the study period, then the number of same-day eGFR tests would have declined 89.7%, but 47.2% of patients with eGFR less than 30 mL/min/1.73 m2 would have been missed. If the health system had followed the ACR/NKF consensus statement but omitted kidney surgery and included diabetes mellitus treated with medication as a trigger for same-day eGFR measurement, then a potentially better balance would have been achieved, reducing the number of tests 77.1% while only missing 34.0% of patients with an eGFR less than 30 mL/min/1.73 m2. The observations support requiring eGFR testing based on the presence of diabetes mellitus treated with medication (rather than optional testing in accordance with the current consensus statement).
The association between iodinated IV contrast medium administration in patients with an eGFR less than 30 mL/min/1.73 m 2 and the development of contrast-associated AKI is controversial; results of several studies [15–18] have suggested the absence of a causal relationship. Despite the uncertain impact of administering iodinated IV contrast media to patients with an eGFR less than 30 mL/min/1.73 m2, screening of patients to identify those potentially at risk has been a standard radiology practice.
Historically, a spectrum of risk factors have been used for screening of patients for increased risk of contrast-associated AKI. Choyke et al. [9] conducted an early study of the predictive performance of a patient survey to identify patients with renal dysfunction. Their study had an endpoint of a serum creatinine level greater than 1.7 mg/dL, reflecting a departmental cutoff to withhold iodinated contrast media. A total of 92% of patients who indicated the absence of all risk factors had normal creatinine levels. However, only 28% of all patients indicated the absence of all risk factors. If the questionnaire had been limited to only six questions, the proportion of completely negative responses would have increased to 67%. A retrospective analysis [19] of the questionnaire proposed by Choyke et al. showed the survey to be 92.9% sensitive but only 65.3% specific in identifying an eGFR less than 45 mL/min/1.73 m2. In that study, only a history of “renal problems” and a history of diabetes mellitus were statistically significant predictors of eGFR less than 45 mL/min/1.73 m2. The authors of that study did not explore identifying patients with an eGFR less than 30 mL/min/1.73 m2.
Moos et al. [10] proposed several models for identifying patients with eGFRs less than 60 and 45 mL/min/1.73 m2. In their study, a model that incorporated diabetes mellitus, history of urologic or nephrologic disease, age older than 75 years, and congestive heart failure was most effective in lowering the percentage of normal eGFR measurements while maintaining detection of patients with a reduced eGFR. This approach was subsequently validated in a later study of the same risk factors [20]. The smaller number of questions supported by our results may reflect our endpoint of an eGFR less than 30 mL/min/1.73 m2 rather than less than 45 mL/min/1.73 m2.
The launch of the systemwide screening form followed best practices for health information technology and EHR implementation, including content approval by a multidisciplinary expert panel and upfront communication to local leadership at each site. Nonetheless, adoption of the desired workflow (consistent use of the electronic screening form) was heterogeneous; only 41.3% of patients completed the form before outpatient CT encounters, including only 65.7% before encounters with IV contrast medium administration. Not only was adoption heterogeneous among sites but also several sites did not use the electronic form at all. These findings suggest that enterprise efforts to use harmonized tools in the EHR may not yield the desired outcome despite ease of access to the tool for patients and staff.
Effective change management for this initiative may have benefited from the following additional considerations: clearer communication to all stakeholders about the reason for and urgency of the new workflow; addressing concerns regarding potential workflow issues involving interactions between technologists and patients as a result of form completion before scheduled CT examinations; establishing predefined metrics and targets for monitoring adoption of the expected work-flow (i.e., target percentage usage of the form for each site); and timely feedback to CT technologists and site managers in the event of deviations from the desired workflow (e.g., low local use of the form). In addition, simply providing easy patient access to an intervention through an electronic portal is not sufficient to attain high adoption, given barriers to portal use by patients [21]. For example, a national survey [22] revealed that 63% of patients had not used a portal in the prior year for a range of reasons, including a desire to speak directly with care practitioners and privacy concerns. Thus, additional efforts to engage patients may have helped increase usage.
The use of patient self-reported risk factors itself may not be the optimal strategy for identifying patients potentially at risk of contrast-induced AKI, considering the observed frequency of patients who did not report risk factors that were present in the EHR (27.5%). Although the current study assessed for such discrepancies on the basis of a manual chart review in a small subset of patients, prior work has likewise shown that patient-reported data may be inaccurate or viewed as unreliable [23, 24].
Our study had limitations. First, it was retrospective. Second, although the electronic form was implemented at a heterogeneous combination of sites, including academic medical centers and community hospitals, all sites were part of a single health system. Third, the analysis for associations with renal dysfunction was performed solely with patient-reported risk factors, despite the recognized potential inaccuracy of this approach. However, the approach mirrors present common clinical practice, and manual chart review of all 10,256 patients would not have been practical. Fourth, only same-day eGFR values were used in the analysis, introducing possible bias from eGFR tests ordered by referring providers in advance of the imaging encounters for patients with known or suspected renal dysfunction. Fifth, a cost-benefit analysis of the impact of reducing the number of screened risk factors and thus the number of eGFR assessments before outpatient CT encounters was not performed. Future work is needed to quantify the cost impact of adherence to the ACR/NKF consensus statement (considering diabetes mellitus treated with medication as both a required and an optional risk factor). Finally, a subset of patients who completed the electronic screening form underwent unenhanced CT and therefore did not receive IV contrast media. However, the focus of our study was on prediction of renal dysfunction rather than of contrast-associated AKI, and this approach avoided bias by exclusion of patients in whom the decision regarding contrast media administration was impacted by the results of the same-day eGFR testing.
In conclusion, use of a patient-driven questionnaire to identify those with any of a range of risk factors for contrast-induced AKI resulted in frequent performance of same-day eGFR testing, the results of which were rarely less than 30 mL/min/1.73 m2. An optimal strategy may be to limit triggers for eGFR testing to a history of kidney disease or diabetes mellitus treated with medication, because assessing renal function in patients with other risk factors, including hypertension, kidney surgery, age 60 years or older, multiple myeloma, or receiving cancer-treating medications, likely leads to substantially increased unnecessary eGFR testing with limited overall benefit. Our findings are concordant with the recent ACR/NKF consensus statement with the exceptions that our findings suggest that a patient-reported history of kidney surgery can be excluded as a risk factor and that diabetes mellitus treated by medication should be a required rather than an optional trigger of eGFR testing. Further investigation is needed to confirm our findings in other settings and to assess the impact of multifaceted information technology–enabled interventions to better identify patients at risk of contrast-induced AKI.
Acknowledgment
We thank Laura Peterson for manuscript editing assistance.
References
1.
Parfrey PS, Griffiths SM, Barrett BJ, et al. Contrast material-induced renal failure in patients with diabetes mellitus, renal insufficiency, or both: a prospective controlled study. N Engl J Med 1989; 320:143–149
2.
Nash K, Hafeez A, Hou S. Hospital-acquired renal insufficiency. Am J Kidney Dis 2002; 39:930–936
3.
Luk L, Steinman J, Newhouse JH. Intravenous contrast-induced nephropathy: the rise and fall of a threatening idea. Adv Chronic Kidney Dis 2017; 24:169–175
4.
Davenport MS, Cohan RH, Ellis JH. Contrast media controversies in 2015: imaging patients with renal impairment or risk of contrast reaction. AJR 2015; 204:1174–1181
5.
Elicker BM, Cypel YS, Weinreb JC. IV contrast administration for CT: a survey of practices for the screening and prevention of contrast nephropathy. AJR 2006; 186:1651–1658
6.
Davenport MS, Khalatbari S, Cohan RH, Dillman JR, Myles JD, Ellis JH. Contrast material–induced nephrotoxicity and intravenous low-osmolality iodinated contrast material: risk stratification by using estimated glomerular filtration rate. Radiology 2013; 268:719–728
7.
Ellis JH, Khalatbari S, Yosef M, Cohan RH, Davenport MS. Influence of clinical factors on risk of contrast-induced nephrotoxicity from IV iodinated low-osmolality contrast material in patients with a low estimated glomerular filtration rate. AJR 2019; 213:[web]W188–W193
8.
ACR Committee on Drugs and Contrast Media. ACR manual on contrast media. American College of Radiology website. www.acr.org/-/media/ACR/Files/Clinical-Resources/Contrast_Media.pdf. Published 2021. Accessed April 1, 2022
9.
Choyke PL, Cady J, DePollar SL, Austin H. Determination of serum creati-nine prior to iodinated contrast media: is it necessary in all patients? Tech Urol 1998; 4:65–69
10.
Moos SI, Stoker J, Nagan G, de Weijert RS, van Vemde DNH, Bipat S. Prediction of presence of kidney disease in a general patient population undergoing intravenous iodinated contrast enhanced computed tomography. Eur Radiol 2014; 24:1266–1275
11.
Tippins RB, Torres WE, Baumgartner BR, Baumgarten DA. Are screening serum creatinine levels necessary prior to outpatient CT examinations? Radiology 2000; 216:481–484
12.
Davenport MS, Perazella MA, Yee J, et al. Use of intravenous iodinated contrast media in patients with kidney disease: consensus statements from the American College of Radiology and the National Kidney Foundation. Radiology 2020; 294:660–668
13.
American College of Radiology Committee on Drugs and Contrast Media. ACR manual on contrast media. American College of Radiology, 2015
14.
Lameire NH, Levin A, Kellum JA, et al. Harmonizing acute and chronic kidney disease definition and classification: report of a Kidney Disease: Improving Global Outcomes (KDIGO) consensus conference. Kidney Int 2021; 100:516–526
15.
McDonald RJ, McDonald JS, Bida JP, et al. Intravenous contrast material–induced nephropathy: causal or coincident phenomenon? Radiology 2013; 267:106–118
16.
McDonald JS, McDonald RJ, Carter RE, Katzberg RW, Kallmes DF, Williamson EE. Risk of intravenous contrast material–mediated acute kidney injury: a propensity score–matched study stratified by baseline-estimated glomerular filtration rate. Radiology 2014; 271:65–73
17.
McDonald JS, McDonald RJ, Comin J, et al. Frequency of acute kidney injury following intravenous contrast medium administration: a systematic review and meta-analysis. Radiology 2013; 267:119–128
18.
Goulden R, Rowe BH, Abrahamowicz M, Strumpf E, Tamblyn R. Association of intravenous radiocontrast with kidney function: a regression discontinuity analysis. JAMA Intern Med 2021; 181:767–774
19.
Too CW, Ng WY, Tan CC, Mahmood MI, Tay KH. Screening for impaired renal function in outpatients before iodinated contrast injection: comparing the Choyke questionnaire with a rapid point-of-care-test. Eur J Radiol 2015; 84:1227–1231
20.
Schreuder SM, Stoker J, Bipat S. Prediction of presence of kidney disease in patients undergoing intravenous iodinated contrast enhanced computed tomography: a validation study. Eur Radiol 2017; 27:1613–1621
21.
Kooij L, Groen WG, van Harten WH. Barriers and facilitators affecting patient portal implementation from an organizational perspective: qualitative study. J Med Internet Res 2018; 20:e183
22.
Anthony DL, Campos-Castillo C, Lim PS. Who isn't using patient portals and why? Evidence and implications from a national sample of US adults. Health Aff (Millwood) 2018; 37:1948–1954
23.
Wuerdeman L, Volk L, Pizziferri L, et al. How accurate is information that patients contribute to their electronic health record? AMIA Annu Symp Proc 2005; 2002:834–838
24.
Jacobson RP, Kang D, Houck J. Can Patient-Reported Outcomes Measurement Information System® (PROMIS) measures accurately enhance understanding of acceptable symptoms and functioning in primary care? J Patient Rep Outcomes 2020; 4:39
Study Guide
Utility of Patient-Reported Risk Factors for Identifying Advanced Chronic Kidney Disease Before Outpatient CT: Comparison With Recent ACR/NKF Consensus Criteria
Alan Mautz, MD1, Joseph J. Budovec, MD2
1Northern Light AR Gould Hospital, Presque Isle, ME.
2Medical College of Wisconsin, Milwaukee, WI.
*Please note that the authors of the Study Guide are distinct from those of the companion article.
Introduction
1. According to the American College of Radiology (ACR) contrast media manual, what risk factor for developing contrast-induced acute kidney injury is supported by the strongest evidence?
2. What risk factors do the consensus statement from the ACR and National Kidney Foundation (NKF) propose to screen patients for reduced estimated glomerular filtration rate (eGFR) before CT examinations with iodinated contrast media?
3. What is the stated objective of this study?
Methods
4. What study design was used? What were the inclusion criteria? What were the exclusion criteria?
5. How were data gathered for this study? How uniform was eGFR testing across the sites included in the study?
6. How was the yield of same-day eGFR testing evaluated?
Results
7. How often did same-day eGFR testing result in an eGFR of less than 30 mL/min/1.73 m2?
8. What risk factors had the strongest association with eGFR less than 30 mL/min/1.73 m2?
9. How would use of ACR/NKF recommendations regarding screening for kidney disease have altered the frequency of eGFR testing?
Discussion
10. What are the limitations of this study? Are these adequately discussed?
11. What current screening criteria do you use for which patients undergo screening eGFR testing before CT with contrast medium? Do your clinical colleagues advise patients differently from the ACR/NKF guidelines or your local guidelines for eGFR screening?
12. How might you design a follow-up study?
Suggested Reading
1.
Davenport MS, Perazella MA, Yee J, et al. Use of intravenous iodinated contrast media in patients with kidney disease: consensus statements from the American College of Radiology and the National Kidney Foundation. Radiology 2020; 294:660–668
2.
Ellis JH, Khalatbari S, Yosef M, Cohan RH, Davenport MS. Influence of clinical factors on risk of contrast-induced nephrotoxicity from IV iodinated low-osmolality contrast material in patients with a low estimated glomerular filtration rate. AJR 2019; 213:[web]W188–W193
Information & Authors
Information
Published In
Copyright
© American Roentgen Ray Society.
History
Submitted: February 3, 2022
Revision requested: February 17, 2022
Revision received: March 3, 2022
Accepted: March 25, 2022
Version of record online: April 6, 2022
Keywords
Authors
Metrics & Citations
Metrics
Citations
Export Citations
To download the citation to this article, select your reference manager software.