A Retrospective Cohort Study of the Effect of Gout on Mortality Among Patients with a History of Kidney Transplantation

Background

Gout is a common form of inflammatory arthritis, with an estimated lifetime prevalence of 4% in the USA [1].Gout is caused by elevated serum uric acid (sUA) levels that promote deposition of urate crystals in joints and other tissues [2]. Gout has a range of clinical manifestations that can include painful acute joint inflammation (or flares) and visible tophaceous deposits [3]. Left uncontrolled, gout can lead to bone erosions and joint destruction and has been linked to several comorbid conditions, including cardiovascular disease and chronic kidney disease (CKD) [2,4,5].

High rates of gout have been reported among kidney transplantation patients [6]. In a study using data from 2016, the prevalence of gout was 13.1% among kidney transplantation patients [7]. Emerging evidence also suggests that, among nephrology patients with gout, those with a history of kidney transplantation are more likely to experience symptoms of severe, uncontrolled gout [8].

The treatment of gout in the context of kidney transplantation aftercare poses several challenges. Hyperuricemia, as well as new-onset and recurrent gout, can be induced by use of the calcineurin inhibitors, which form the basis of most prophylactic immunosuppressive regimens prescribed to this population [9]. Additionally, xanthine oxidase-inhibiting urate-lowering therapies are contraindicated in patients receiving azathioprine [1], a second-line antiproliferative agent in certain immunosuppressive regimens used to protect against graft loss.

Several recent studies investigated the relationship between hyperuricemia and graft outcomes [11,12]. However, despite evidence of gout contributing an independent risk associated with CKD and end-stage renal disease (ESRD) [13,14], cardiovascular morbidity and mortality [15], and all-cause mortality in other adult populations [16], to the best of our knowledge only 1 study has investigated the risk of gout in kidney transplant recipient populations [6]. Abbott et al. found that development of new-onset gout following kidney transplantations performed in the 1990s was an independent predictor of graft loss and decreased survival (HR 1.26, 95% CI 1.08–1.47) [6].

This study was conducted to examine mortality risk among patients with a history of kidney transplantation with vs. without gout. A preliminary abstract of this study was previously presented at the Annual European Congress of Rheumatology, EULAR 2019, Madrid, 12–15 June 2019 [17].

Material and Methods

STUDY DESIGN, DATA, AND POPULATION:

A retrospective cohort study was conducted using Medicare Fee-for-Service Limited Data Set administrative claims obtained from the Centers for Medicare and Medicaid Services (CMS). This publicly available dataset contains a representative, de-identified 5% sample of beneficiaries enrolled in Medicare Parts A (inpatient care) and B (physician services and outpatient care). Data on patient demographics and enrollment information, as well as on services and procedures provided to the beneficiaries, are available at the claim level. While this dataset captures specific diagnostic and laboratory tests performed, the results of the tests are not available.

To be included in the study, patients must have: 1) been continuously enrolled in Medicare Parts A and B in calendar year 2012; 2) be alive as of January 1, 2013; 3) had a kidney transplant status claim (International Classification of Diseases, Ninth Revision, Clinical Modification code V42.0) in 2012; and 4) not received a solid-organ transplant procedure in 2012. Patients with incomplete information for US Census Bureau region, sex, or race were excluded from the analysis.

All patients were indexed on January 1, 2013 (index date) and tracked forward until end of their Medicare A and B enrollment, death, or end of the study period (December 31, 2016). A look-back period from 2007 to 2011 was used to identify the kidney transplant procedure date, if available. Baseline demographic and clinical characteristics of patients were based on 2012 claims.

OUTCOME AND PREDICTORS OF INTEREST:

The main outcome of interest was time to all-cause mortality. Patients whose death was not observed within the study period were censored on the date that their enrollment ended or the end of the study period, whichever occurred first. The main predictor of interest was gout diagnosis, assessed both in the 2012 look-back year (“baseline gout”) and during the follow-up period (“incident gout”). Gout status was defined by the presence of at least 1 gout-related claim identified using International Classification of Diseases, Ninth and Tenth Revision (ICD-9 and ICD-10) code prefixes (274 in ICD-9, M10 and M1A in ICD-10). The high concordance between physician documentation of gout in electronic health records and claims with gout ICD diagnosis codes was previously demonstrated [18]. To adjust for incident gout, gout diagnosis was treated as time-varying for all regression analyses, in which a patient was considered a gout patient starting from the date that the first gout-related service claim was observed. Those who did not have a gout-related claim in 2012 or during follow-up were considered non-gout patients. As serum urate levels are not available, patients in the gout and non-gout groups may have had elevated uric acid levels.

COVARIATES:

Baseline demographic covariates included age, sex, race, and US Census Bureau region as of January 1, 2012.

Kidney transplant year was treated as a categorical variable and was determined using 2007–2011 claims. For patients who were continuously enrolled but did not have a kidney transplant reported during the period, transplant year was classified as “Prior to 2007.” Patients who were not continuously enrolled and did not report a kidney transplant procedure during the period were coded as having an “Unknown” transplant year.

Baseline (2012) health status of patients was assessed via the Charlson Comorbidity Index (CCI), using claims-coding algorithms as previously described [19]. Although all patients had a history of kidney transplantation, patients were not assigned CCI points for ‘moderate to severe renal disease’ unless they met corresponding criteria from the claims-coding algorithm published by Deyo et al. [19] Importantly, this algorithm does not consider history of kidney transplant alone to qualify a patient as currently having ‘moderate to severe renal disease’. Among patients not meeting the CCI claims-coding algorithm criteria for ‘moderate to severe renal disease’, it was assumed that successful renal transplantation had mitigated the pre-transplantation renal disease diagnosis as of the study baseline timepoint (i.e., calendar year 2012).

CONFOUNDING AND EFFECT MODIFICATION:

The presence of confounding distorts the estimated measure of association between gout and time to death. To eliminate any confounding by baseline demographics, analyses were controlled for age, sex, race, and Census Bureau region in all multivariable regression models.

In contrast to patient demographics, the causal relationship between gout and specific comorbidities is less well-established; therefore, 3 analyses were conducted to explore this relationship. Under the assumption that gout and several of the conditions that make up the CCI were associated with one another by way of a common pathogenic root and patients’ baseline health status therefore confounded the association between gout and death, regression models were constructed to control for gout, demographics, and CCI score. Alternatively, under the assumption that gout was upstream of these comorbidities and CCI score was therefore an intermediate (mediating) step in the causal pathway between gout and death, regression models were fit without the CCI score. Lastly, to understand how patients’ baseline health modified the effect of gout on death, stratified analyses were performed at various levels of CCI score.

STATISTICAL ANALYSES:

All statistical analyses were performed using SAS version 9.4 (SAS Institute, Inc., Cary NC, USA).

The effect of gout on time to death was first examined in a Cox proportional hazards regression model without adjustment for potential confounders (unadjusted). Multivariable models were then constructed, adjusting for age, sex, race, region, and year of kidney transplant procedure, with and without controlling for baseline CCI score.

To explore the possibility that CCI modifies the effect of gout on death, the relationship between gout and other comorbid conditions at baseline was assessed. Mean CCI scores by gout status were compared using the t test and the rates of comorbidities by gout status were compared using the chi-square test. The results suggested the presence of effect modification by baseline CCI score, so stratified multivariable Cox proportional hazards models were constructed based on CCI scores of 0 vs. 1–4 vs. ≥5.

Results

Of 2 220 575 Medicare beneficiaries contained in the 5% Limited Data Set from January 2012 to January 2013, there were 5046 patients with a history of kidney transplantation who met the eligibility criteria for this study (Figure 1). Of these patients, 1405 (27.8%) either had gout at baseline (13.4%) or developed gout during follow-up (14.4%). Gout patients were older, more likely to be male, and more likely to have received their kidney transplant in earlier years compared to non-gout patients (Table 1). Both among gout and non-gout cohorts, the transplant year was unknown for 12.7% of patients. Furthermore, gout patients’ mean baseline CCI score was higher compared to that of non-gout patients. The median follow-up time for the overall cohort was 1461 days, or 4 years.

The unadjusted model revealed that kidney transplant recipients with gout had a 44% greater risk of death from all causes compared to patients without gout (hazard ratio: 1.44, 95% CI: 1.27–1.63) (Table 2). Controlling for age, sex, race, region, and year of kidney transplant procedure, the risk of death was attenuated (hazard ratio: 1.16, 95% CI: 1.02–1.32). When CCI score at baseline was added to the model, gout was no longer a significant predictor of death (hazard ratio: 1.03, 95% CI: 0.90–1.17).

There was a strong association between gout status and CCI score at baseline. Kidney transplant patients with gout at baseline had a mean CCI score of 4.9, while those without gout at baseline had a mean score of 4.0 (p<0.001) (data not shown). Furthermore, gout patients were characterized by significantly higher rates of 9 out of the 17 comorbidity categories that comprise the CCI score (Table 3). The rate of myocardial infarction was 2.1 times higher among those with gout at baseline compared to those without gout at baseline (p<0.001). The rate of congestive heart failure, connective tissue disease (e.g., rheumatoid arthritis and osteoarthrosis), and any malignancy were 62% (p<0.001), 56% (p=0.003), and 54% (p≤0.001) greater, respectively, among gout patients compared to non-gout patients. Additionally, patients with gout were 40% more likely to have peripheral vascular disease, chronic obstructive pulmonary disease, and cerebrovascular disease (all p<0.001) compared to those without gout. Finally, gout patients were also 14% more likely to have diabetes without chronic complication and 12% more likely to have moderate to severe renal disease, despite history of kidney transplantation.

Stratified survival analysis revealed that the CCI score modifies the relationship between gout and risk of death (Table 4). Among 248 healthy patients (CCI=0), the presence of gout increased the risk of death by 3.5-fold (HR: 3.48, 95% CI: 1.27–9.57). However, among patients with CCI=1–4 and CCI ≥5, gout did not appear to increase the risk of death (HR: 1.08, 95% CI: 0.86–1.36; and HR: 0.99, 95% CI: 0.848–1.16, respectively).

Discussion

LIMITATIONS:

This study has several limitations worth noting. As with any study that leverages administrative claims data, medical claims lack detailed clinical information such as sUA levels, a definitive diagnosis for hyperuricemia, and living vs. deceased donor status. It is possible that some patients were misdiagnosed with gout. Malik et al. found that, in practice, 33% of patients meeting American College of Rheumatology (ACR) gout clinical diagnostic criteria did not meet the criterion standard for gout diagnosis (i.e., presence of urate crystals in synovial fluid) [34]. Further, data from a medical records study suggests that, considering the criteria of the present study and the distribution of patients with 1 vs. 2 or more gout claims from a prior study in the kidney transplantation population, ~43% of patients identified with gout ICD codes would not have evidence meeting ACR criteria in their records [7,35]. However, a similar study, which followed up with additional patient interviews, found that the majority of such errors were due to a lack of documentation in medical records rather than due to an absence of qualifying signs and symptoms [36]. Nonetheless, considering the possibility for misdiagnosis of gout, and given the overlap between gout and hyperuricemia, one might ask whether it would be appropriate to refer to the predictor of interest in this study as “hyperuricemia” rather than “gout”. However, the US National Health and Nutrition Examination Survey has consistently found that the proportion of the general population with any history of a gout diagnosis is many times smaller than the population with hyperuricemia – 3.9% vs. 20.1% in 2015–2016 [37]. The proportion of patients with hyperuricemia who have a current gout diagnosis (i.e., within a calendar year rather than at any point in the patient’s medical history) could be even lower considering a recent study that estimated the prevalence of a current gout diagnosis at 1.1% in the USA [7]. Thus, referring to the predictor of interest in the present study as “hyperuricemia” rather than “gout” would result in the misclassification of a far greater proportion of patients, even in light of the limitations of gout coding and clinical diagnostic criteria.

eGFR values were not available in the dataset. Variability in eGFR is possible even among patients with CCI=0, either because of the wide range of possible values above the threshold for a CKD diagnosis (≥90 mL/min/1.73 m²) or if the diagnosis coding capture is less consistent for patients with lower severity CKD (e.g., stage 1–2) even after transplantation. If so, it is also possible that lower eGFR values are responsible for the higher rates of gout and the increased mortality among a subset of patients with CCI=0. Data on prescription drugs were not available in this dataset, which meant that the use of urate-lowering therapy to treat gout could not be controlled for in the models. Similarly, because the immunosuppressant regimen used was not available in the dataset, one cannot rule out the possibility that the minority of patients treated with cyclosporine had high rates of gout as well as poorer outcomes compared to the rest of the population, most of whom would have received tacrolimus. For patients undergoing transplantation prior to 2007, lack of a specific transplant date limits the model’s ability to control for time since transplantation.

Conclusions

In our unadjusted model of gout and mortality among patients with a history of kidney transplantation from the Medicare Fee-for-Service database, greater mortality risk was observed in the cohort of patients with gout. However, gout was not shown to be an independent predictor of mortality after baseline risk-adjustments that included CCI. In stratifying baseline CCI, gout was observed to be an independent predictor among a subset of patients without comorbidities (CCI=0). Given the nuanced pathogenesis and varied clinical manifestations of gout, an independent association of gout in kidney transplant recipients with comorbid disease cannot be ruled out. The possibility that gout behaves similarly in the larger prevalent kidney transplant population with comorbidities as it does in a healthy subset should concern physicians, as the presence of gout can serve as an early indicator of subsequent declining renal function and overall health.