Contrast Medium Exposure During Computed Tomography and Risk of Development of End-Stage Renal Disease in Patients With Chronic Kidney Disease: A Nationwide Population-Based, Propensity Score-Matched, Longitudinal Follow-Up Study.

The aim of the study was to investigate the long-term association between contrast medium exposure during computed tomography (CT) and the subsequent development of end-stage renal disease (ESRD) in patients with chronic kidney disease (CKD). We conducted a population-based cohort study using Taiwan's National Health Insurance Research Database. A total of 7100 patients with nonadvanced CKD who underwent contrast medium-enhanced CT were identified and served as the study cohort. To avoid selection bias, we used the propensity score to match 7100 nonadvanced CKD patients, who underwent noncontrast medium-enhanced CT to serve as the comparison cohort. The age, sex, index year, and frequency of undergoing CTs were also matched between the study and comparison cohorts. Participants were followed until a new diagnosis of ESRD or December 31, 2011. Hazard ratios (HRs) with 95% confidence interval (95% CI) were calculated using the Cox proportional hazards regression. Contrast medium exposure was not identified as a risk factor for developing ESRD in nonadvanced CKD patients after confounders adjustment (adjusted HR = 0.91; 95% CI, 0.66-1.26; P = 0.580). We further divided the patients who underwent CTs with contrast medium use into ≤1 exposure per year on average, >1 and <2 exposure per year on average, and ≥2 exposure per year on average. After adjusting for confounders, we identified a much higher risk for developing ESRD in the 2 groups of >1 and <2 exposure per year on average and ≥2 exposure per year on average (adjusted HR = 8.13; 95% CI, 5.57-11.87 and adjusted HR = 12.08; 95% CI, 7.39-19.75, respectively) compared with the patients who underwent CTs without contrast medium use. This long-term follow-up study demonstrated that contrast medium exposure was not associated with an increased risk of ESRD development in nonadvanced CKD patients.

INTRODUCTION

Contrast-induced nephropathy (CIN) was a common cause of acute kidney injury (AKI).[1] The prevalence ranged from 2% to 30% because of different studied cohorts (underwent diagnostic or therapeutic procedures) and CIN definitions.[2,3] Patients who developed AKI after contrast medium exposure had markedly increased morbidity and mortality even after 1-year follow-up.[1,4,5]

With the increasing utilization of contrast medium in the intervention procedures and imaging modalities, CIN had become an important issue, particularly in chronic kidney disease (CKD) patients, who were more susceptible to CIN.

Taiwan had the highest prevalence of end-stage renal disease (ESRD) worldwide for >10 years before 2009 and remains high currently. A large Taiwanese cohort study showed that the prevalence of CKD was 11.9% in adults and was as high as 37.2% in the elderly.[6,7] With the gradual increase in Taiwan's elderly population, there has been a correspondingly steady rise in the prevalence of CKD. Furthermore, these CKD patients were prone to contrast medium exposure as they were more often required to undergo evaluation by computed tomography (CT).

CIN was generally thought to be a reversible form of AKI that happened soon after the administration of contrast medium.[8-16] However, it was increasingly being recognized that the impaired renal function might persist even following the return of serum creatinine to the baseline level.[5,17] This effect was particularly important in patients with CKD, among whom an occurrence of AKI might increase the risk of CKD progression, including to ESRD.

The long-term impact of contrast medium exposure for CT in CKD patients remains unknown. This study aimed to investigate the association between contrast medium exposure for CT in nonadvanced CKD patients and the development of ESRD using retrospective data of Taiwan's National Health Insurance Research Database (NHIRD). Because patients with advanced CKD were prone to developing ESRD even after a minor disease event, this study focused on patients with nonadvanced CKD.

METHODS

Data Sources and Study Participants

Taiwan's National Health Insurance program was promulgated on March 1, 1995, by the National Health Insurance Administration (NHIA) and covers >23.03 million residents in Taiwan (∼99.2% of the population). The NHIA releases deidentified data to the National Health Research Institute (NHRI), which maintains the NHIRD. The Longitudinal Health Insurance Database 2000 (LHID2000) used in this study contains medical information of 1 million National Health Insurance beneficiaries randomly sampled from the registry of all beneficiaries for the year 2000. Claims data in the LHID2000 were retrospectively collected for the period of January 1, 1996, to December 31, 2011. The distributions of sex and age in the original claims data and the sampled data did not differ significantly. The diagnosis codes of the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) are used in the NHIRD. The NHRI scrambles patient identification data and replaces them with surrogate numbers to ensure privacy. The data were also collected in accordance with the data regulations of the NHIA and the NHRI in order to maintain strict confidentiality. Because the NHIRD contains deidentified secondary data for research, the present study was waived from inform consent. This study was approved by the Institutional Review Board of China Medical University (CMUH104-REC2-115).

The patients were defined as having CKD if they had at least 3 outpatient service claims with a diagnosis of CKD or if they had a single hospitalization in which CKD was found in 1 of the 5 spaces used to report their diagnosis when hospitalized using the ICD-9-CM (581–584, 586–588, 403, 404, 285.21). The National Health Insurance (NHI) reimbursement regulations in Taiwan allowed patients with a serum creatinine level of >6 mg/dL (approximately equivalent to eGFR <15 mL/min/1.73 m2) and a hematocrit level of <28% to receive erythropoiesis-stimulating agent (ESA) treatment for anemia. Therefore, we defined patients with advanced CKD using the above criteria: that is, the patients being treated with ESA were considered to have advanced CKD. Further, we also excluded patients who received dialysis or kidney transplantation during the study period.

Propensity score matching could reduce selection bias because it allowed the bundling of many confounding covariates that might be present in an observation study.[18-20] In our study, multiple risk factors had might affect the clinical decision as to whether a patient would receive contrast medium for advanced image-enhancement or not while undergoing a CT. The propensity score indicated the possibility that contrast medium might be administered if the covariates were present. For each patient, we calculated the propensity score using the multivariate logistic regression by entering the baseline covariates, which also included important risk factors for CIN. We matched 1 comparison cohort patient (underwent noncontrast medium-enhanced CT) with each study cohort patient (underwent contrast medium-enhanced CT) according to the propensity score and obtained a dataset composed of matched patients who had a statistically identical likelihood of contrast medium exposure for CT. Also, as the propensity scores were composed of important underlying comorbidities, the propensity score matching made the study and comparison cohort patients stand on a comparable baseline condition and have equal opportunities to be affected by accidental disease events except the difference of contrast medium exposure or not.

Figure 1 illustrated the participant selection process of the study and comparison cohorts. Both the study and comparison cohort patients had undergone CTs with the only difference of contrast medium use or not. The frequencies of undergoing CTs between the study and comparison cohort patients were also further matched, that is, both the 2 cohort patients had undergone equal times of CTs. Patients with ESRD (ICD-9-CM 585) before the first CT (with or without contrast medium-enhanced) were excluded. The ICD-9 codes for ESRD used in this study, which were from Taiwan's NHIRD, were considered to be highly reliable, and many related studies that used these codes had been published.[21-23] Patients aged <20 or >100 years were excluded in this study. In addition, patients who underwent coronary angiography or angiographic embolization during the study period (1996–2012) were excluded. All the enrollees accepted follow-up of at least 1 year. A total of 7100 patients were identified in the NHIRD and served as the study cohort (underwent contrast medium-enhanced CT). The first CKD diagnosis was defined as the index time, and the index year was defined as the calendar year of the index time. An equal number of patients undergoing noncontrast medium-enhanced CT (n = 7100) were selected from the database after matching with the study cohort patient for age, sex, propensity score, frequencies of undergoing CTs, and the index year, and served as the comparison cohort. During the study period, >90% of the contrast medium was ionic contrast medium which was paid by the national health insurance program. The incidence of ESRD was evaluated until December 31, 2011.

FIGURE 1

Participants selection process for the study and comparison cohorts.

Statistical Analyses

Statistical analyses were performed using SAS 9.4 statistical package (SAS Institute Inc., Cary, NC), and the significance level was set at 0.05. Differences in demographic characteristics and comorbidities between the study and comparison cohorts were examined using chi-square and 2-sample t-tests. Hazard ratio (HR) with 95% confidence interval (95% CI) was calculated for each variable by Cox proportional hazards regression. The difference in the development of ESRD between the 2 cohorts was estimated using the Kaplan–Meier method and the log-rank test. Adjusted HRs for ESRD were obtained by Cox proportional hazards regression after adjustment for possible confounders, including age, sex, and underlying comorbidities. The adjusted underlying comorbidities were hypertension (HTN) (ICD-9-CM 401–405), diabetes mellitus (DM) (ICD-9-CM 250, 357.2, 362.01, 362.02, 366.41), ischemic heart disease (IHD) (ICD-9-CM 411–414), peripheral arterial occlusive disease (PAOD) (ICD-9-CM 440–444), congestive heart failure (CHF) (ICD-9-CM 428), and anemia (ICD-9-CM 280–285).[2] Diagnoses given ahead of or in concurrence with the diagnosis of CKD were considered to be underlying comorbidities.

RESULTS

A total of 7100 nonadvanced CKD patients and an equal number of matched patients were included in the study (underwent contrast medium-enhanced CT) and comparison cohort (underwent noncontrast medium-enhanced CT), respectively. The mean ages of the contrast medium exposure and noncontrast medium exposure cohorts were 65.31 ± 15.45 and 65.99 ± 15.80 years, respectively. The male to female ratio in the contrast medium exposure and noncontrast medium exposure cohorts were 1.34 and 1.36, respectively. Table 1 summarized the demographic characteristics and comorbidities of the study and comparison cohorts.

TABLE 1

Demographic Characteristics and Comorbidities of Nonadvanced CKD Patients who Underwent Contrast Medium-Enhanced or Nonenhanced CTs

The mean follow-up duration of patients in the contrast medium exposure cohort was 4.53 (±4.12) years and was 4.46 (±3.99) years for patients in the noncontrast medium exposure cohort. The mean duration between the first diagnosis of CKD and ESRD for patients in the contrast medium exposure cohort was 1.57 (±0.34) years and was 1.64 (±0.78) years for patients in the noncontrast medium exposure cohort. During the follow-up period, the incidence rates (IRs) of ESRD in the contrast medium exposure and noncontrast medium exposure cohorts were 3.77 and 3.67 per 1000 person-years, respectively. Kaplan–Meier analysis with log-rank test did not show an increased IR of ESRD development in the study cohort compared with the comparison cohort (Figure 2).

FIGURE 2

Kaplan–Meier analysis of the cumulative incidence of ESRD for the study and comparison cohorts. (X-axis: follow-up time in years; Y-axis: cumulative incidence per 1000 person-years). ESRD = end-stage renal disease.

In the univariate analysis, HTN, DM, IHD, and CHF increased the HR of ESRD development in nonadvanced CKD patients. In further multivariate analysis, DM and CHF significantly increased the adjusted HR of ESRD development (adjusted HR = 3.6; 95% CI, 2.51–5.17 and adjusted HR = 2.25; 95% CI, 1.48–3.40, respectively). However, the contrast medium exposure for CT did not show a significant correlation with ESRD development (adjusted HR = 0.91; 95% CI, 0.66–1.26; P = 0.580) (Table 2). In the subgroup analysis, stratified by sex and age (aged 20–39, 40–59, ≥60), we did not observe any correlation between contrast medium exposure and ESRD development in all subgroups (Table 3).

TABLE 2

HR of ESRD Associated With Contrast Medium Exposure in Nonadvanced CKD Patients

TABLE 3

Subgroup Analysis of IR and HR of ESRD After Contrast Medium Exposure

Because of varied follow-up time, the patients in the study cohort (underwent contrast medium-enhanced CT) were further divided into 3 groups: (1) underwent contrast medium-enhanced CT ≤1 time per year on average, (2) >1 and < 2 time per year on average, (3) ≥2 time per year on average. There were 5547, 971, and 582 patients in the 3 groups, respectively. The IRs of ESRD development in the ≤1, >1 and <2, and ≥2 exposures per year on average groups were 1.50, 30.94, and 45.99 per 1000 person-years, respectively. The adjusted HR of groups of ≤1, >1 and <2, and ≥2 contrast medium exposures per year on average was 0.51 (95% CI, 0.37–0.71), 8.13 (95% CI, 5.57–11.87), and 12.08 (95% CI, 7.39–19.75) compared with the patients who underwent noncontrast medium-enhanced CT (Table 4). The Kaplan–Meier analysis revealed a higher IR for developing ESRD in both the >1 and <2 exposure per year on average and ≥2 exposures per year on average groups compared with the noncontrast medium exposure group (log-rank test, P < 0.001) (Figure 3).

TABLE 4

Incidence for ESRD Among Patients who Underwent CTs With Different Frequency of Contrast Medium Exposure per Year on Average

FIGURE 3

Kaplan–Meier analysis of the cumulative incidence of ESRD for the 4 groups stratified by different frequencies of contrast medium exposures per year on average. (X-axis: follow-up time in years; X-axis: cumulative incidence per 1000 person-years). ESRD = end-stage renal disease.

DISCUSSION

To the best of our knowledge, this is the first nationwide cohort study to investigate the association between contrast medium exposure for CT and subsequent development of ESRD in nonadvanced CKD patients. Our study demonstrated that contrast medium exposure for CT was not associated with an increased risk of ESRD development in nonadvanced CKD patients. However, in the patients with a greater frequency of intensive contrast medium exposure (>1 contrast medium exposure per year on average), there was an increased risk of ESRD development in nonadvanced CKD patients compared with those underwent noncontrast medium-enhanced CT. In the present study, the follow-up time was prolonged for >1 year to determine the effect of contrast medium exposure on the risk of ESRD development in nonadvanced CKD patients. In previous studies, however, the analyses were primarily limited to intra-hospital or 1-year morbidity and mortality.[24]

CIN is defined as the impairment of renal function after contrast medium exposure with either a 25% increase in the serum creatinine level from baseline or a 0.5 mg/dL (44 μmol/L) increase in absolute value within 48 to 72 hours of intravenous contrast medium administration. In a typical course, the serum creatinine level begins to increase at 48 to 72 hours postcontrast medium exposure, peaks at 3 to 5 days, and returns to baseline within 3 to 5 days thereafter. Compared with patients with normal kidney function who are generally thought to be not at risk for CIN, patients with pre-existing CKD are much likely to develop this complication. Moreover, the risk increases with greater severity of underlying CKD.[9,14,25] Other risk factors include DM, CHF, anemia, administration of high-dose contrast medium, use of first-generation hyperosmolality contrast medium, and application of percutaneous coronary intervention.[2,8-13,15,26-28] In fact, hemodialysis is rarely required in patients who develop CIN.[29,30] However, a growing number of studies have indicated that CIN could be a harbinger of ESRD development.[5,17] In a study by Maioli et al, it was shown that persistent kidney damage could occur after CIN which highlights the potential of CIN to accelerate CKD.[5] In the present study, we observed that a greater frequency of intensive contrast medium exposure (>1 exposure per year on average) could be an important risk factor for ESRD development in nonadvanced CKD patients compared with noncontrast medium exposure.

In animal models, it is well accepted that CIN might be caused by acute tubular necrosis (ATN); however, the mechanism of ATN remains incompletely understood.[31-33] Two major theories have been proposed to explain the cause of ATN: (1) renal vasoconstriction mediated by nitric oxide, endothelin, and adenosine; and (2) the direct cytotoxic effect of contrast medium.[31-34] A variety of preventive measures have been suggested to reduce the risk of CIN according to its possible pathogenesis, including intravenous volume administration with isotonic saline, isotonic sodium bicarbonate, acetylcysteine administration, use of iodixanol or nonionic low-osmolal agents, such as iopamidol or ioversol, and reduced dose of contrast medium.[29,35-55] Other preventive measures include avoiding volume depletion, stopping nonsteroid anti-inflammatory drugs use, and withdrawal of angiotensin-converting enzyme inhibitors or angiotensin receptor blockers before contrast medium administration.[56] However, the best preventive measure is to avoid the use of contrast medium. In our study, although we could not examine the effects of preventive measures, we set ESRD as the primary endpoint of most concern to CKD patients and observed that if contrast medium was not administered in an intensive frequency, there was no increased risk of ESRD development.

According to previous studies, in the majority of CIN cases, AKI was transient, and a complete or near-complete recovery to baseline creatinine occurred within 3 months.[5,29] In a study by James et al, it was found that patients who developed AKI following contrast medium exposure had an increased risk of progressive long-term kidney function loss.[57] A number of mechanisms have been proposed as possible explanations of the association between CIN and progressive kidney function loss. The most widely accepted mechanism involves the impact of underlying comorbidities, such as HTN, DM, CHF, and pre-existing CKD. Patients who developed AKI following contrast medium exposure usually had multiple risk factors and these underlying comorbidities may predispose them to progression of CKD. In the present study, we used the propensity scores matching accompanied with the matching of CT frequencies to make the patients in the study and comparison cohorts comparable by standing on the same baseline of underlying conditions and they had equal opportunities to be affected by accidental disease events except the difference of contrast medium exposure or not. We observed a greater progression of CKD to ESRD after intensive contrast medium exposures for CTs. We inferred that the single random contrast medium exposure and underlying comorbidities may be not critically important risk factors for ESRD development in CKD patients.

McDonald et al proposed that contrast medium exposure for CT did not actually increase the risk of AKI in both groups of patients with normal and impaired renal function after single contrast medium exposure. We found a similar result to that reported by McDonald et al that a single random exposure to contrast medium for CT may not increase the risk of CKD progression to ESRD. However, there was an increased risk of CKD progression if the CKD patients were intensively exposed to contrast medium. It is reasonable to postulate that the patients who had intensive contrast medium exposure might have had multiple acute or severe disease events that required the use of contrast-enhanced imaging studies. Furthermore, these disease events might themselves predispose patients to progression of CKD. However, in contrast to previous studies, in the present study, the selection bias was particularly eliminated by using the propensity score matching between the study and comparison cohorts in order to minimize the impact of potential confounders from underlying comorbidities and accidental disease events, and thereby obtained a clearer understanding of the effects from contrast medium exposure for CT. We also excluded the patients with advanced CKD who were much more susceptible to developing ESRD, even after a minor disease event, compared with the general CKD patients. Our findings demonstrated that intensive contrast medium exposure may be associated with an increased risk of CKD progression to ESRD rather than single random exposure or nonexposure and should therefore be avoided.

The major strength of this study was the use of a large sample size from a nationwide database with a prolonged follow-up duration of >1 year. The most challenging aspect of this study was the elimination of selection bias for contrast medium use for CT in patients with different clinical conditions and severity of impaired kidney function. To overcome this selection bias, propensity score matching was used so that the 2 cohorts could be meaningfully compared. Also, the matched patients in the study and comparison cohorts had undergone equal times of CTs during the study period with the only difference of contrast medium use or not. Third, we excluded patients who underwent coronary angiography, which would have resulted in a greater risk of CIN compared with CT. Fourth, we excluded patients with advanced CKD as such patients are prone to developing ESRD and need dialysis even after a minor disease event.

The study had several limitations. First, the major limitation of our used database, NHIRD, was the lacking of clinical laboratory data, that is, the creatinine level of each patient, pre- and postcontrast medium exposure for CT. Also, some detailed information, such as the volume of used contrast medium, was not available. Although short-time kidney function change could not be evaluated, our study still provided a valuable long-term observation result. Nearly all the published articles focused on short-term kidney function change, that is, CIN, and long-term observation study was extremely rare because of complexity. Second, the enrolled patients in the study and comparison cohorts might have been in different stages of CKD. However, because of the large enrollee population, the patients with different levels of CKD severity would likely have been distributed equally. Third, preventive procedures before contrast medium exposure could not be evaluated in the present study.

CONCLUSION

Contrast medium exposure for CT was not associated with an increased risk of ESRD development in nonadvanced CKD patients. But when intensive contrast medium exposure for CT was mandatory, alternative imaging modality without contrast medium use should be considered.