Visual Abstract
Keywords: epidemiology and outcomes, hospitalization, thrombosis
Abstract
Background
CKD has been implicated as a risk factor of venous thromboembolism, but the evidence is limited to relatively healthy populations. The objective of this study was to discern whether parameters of kidney function and damage are associated with the occurrence of venous thromboembolism after hospitalization.
Methods
We conducted a retrospective study including 23,899 and 11,552 adult individuals hospitalized within Geisinger Health System and New York University (NYU) Langone Health from 2004 to 2019 and 2012 to 2022, respectively. A Poisson model was used to evaluate adjusted incidence rates of venous thromboembolism according to eGFR and albuminuria categories in each cohort. Cox proportional hazards models were used to analyze associations of eGFR and urinary albumin-to-creatinine ratio (UACR) with venous thromboembolism, and hazard ratios (HRs) were meta-analyzed across cohorts.
Results
Both lower eGFR and higher UACR were associated with higher risks of venous thromboembolism. In the Geisinger cohort, the incidence of venous thromboembolism after hospital discharge ranged from 10.7 (95% confidence interval [CI], 9.2 to 12.6) events per 1000 person-years in individuals in G1A1 (eGFR >90 ml/min per 1.73 m2 and UACR <30 mg/g) to 27.7 (95% CI, 20.6 to 37.2) events per 1000 person-years in individuals with G4-5A3 (eGFR <30 ml/min per 1.73 m2 and UACR >300 mg/g). A similar pattern was observed in the NYU cohort. Meta-analyses of the two cohorts showed that every 10 ml/min per 1.73 m2 reduction in eGFR below 60 ml/min per 1.73 m2 was associated with a 6% higher risk of venous thromboembolism (HR 1.06 [95% CI, 1.02 to 1.11], P = 0.01), and each two-fold higher UACR was associated with a 5% higher risk of venous thromboembolism (HR 1.05 [95% CI, 1.03 to 1.07], P < 0.001).
Conclusions
Both eGFR and UACR were independently associated with higher risk of venous thromboembolism after hospitalization. The incidence rate was higher with greater severity of CKD.
Podcast
This article contains a podcast at https://dts.podtrac.com/redirect.mp3/www.asn-online.org/media/podcast/CJASN/2023_12_14_CJN0000000000000352.mp3
Introduction
The estimated incidence of venous thromboembolism in adults ranges from 104 to 183 per 100,000 person-years.1,2 Venous thromboembolism, which includes both deep vein thrombosis and pulmonary embolism, is associated with substantial morbidity and mortality. The 1-year case fatality rate has been estimated to be as high as 21.6% after a first venous thromboembolic event.3 Other serious complications include risk of recurrent venous thromboembolism and chronic morbidities, such as post-thrombotic syndrome and chronic thromboembolic pulmonary hypertension.4–10
Traditional risk factors of venous thromboembolism, including older age, obesity, history of venous thromboembolism, cancer, prolonged bed rest, major surgery, fracture (hip or leg), multiple traumas, and hereditary predisposition (e.g., factor V Leiden mutation, prothrombin gene mutation, protein S deficiency, protein C deficiency, and antithrombin deficiency), are well established.11 Among kidney diseases, nephrotic syndrome and nephrotic-range proteinuria may lead to a hypercoagulable state and increase the risk of venous thromboembolism.12–16 Although CKD in general has been implicated as a risk factor of venous thromboembolism, the evidence is limited to relatively healthy populations in prospective cohorts.16–19 A study of 19,073 middle-aged and older adults showed that individuals with eGFR between 30 and 60 ml/min per 1.73 m2 were more than twice as likely to develop venous thromboembolism as those with preserved kidney function.17 However, no information on albuminuria was available, and CKD was defined solely on the basis of eGFR. Conversely, in the Prevention of Renal and Vascular End-stage Disease (PREVEND) study, an association between microalbuminuria and venous thromboembolism was observed.16–18 However, the number of individuals with advanced CKD (eGFR <30 ml/min per 1.73 m2) was limited. Similarly, individuals with advanced CKD (eGFR <30 ml/min per 1.73 m2) were excluded from the more recent meta-analysis, which assessed the joint associations of eGFR and albuminuria with venous thromboembolism.19
Hospitalized patients are at higher risk of developing venous thromboembolism, and CKD is associated with a broad range of diseases requiring hospitalization. However, the independent contributions of reduced eGFR and albuminuria on venous thromboembolism risk remain uncertain. We conducted an analysis of records from two large health systems to better understand the association between eGFR and albuminuria and venous thromboembolism.
Methods
Study Population and Design
We conducted a retrospective study using electronic health record (EHR) data from two large health care systems: Geisinger Health System and New York University (NYU) Langone Health. The Geisinger Health System is a large, regional health care system in central and northeastern Pennsylvania. NYU Langone Health is a large health care system serving the New York (NY) Metropolitan area. The EHR provides information on patient demographics, inpatient and outpatient encounters, inpatient and outpatient prescriptions, vitals, laboratory measurements, and diagnostic/billing codes. We included adult individuals (aged 18 years or older) hospitalized for over 1 day and with at least 6 months of prior engagement with the health system. Hospitalizations occurred between January 1, 2004, and December 31, 2019, within the Geisinger cohort and between November 1, 2010, and November 26, 2022, at NYU. Individuals were followed from hospitalization discharge (index date) until incident venous thromboembolism. Those who did not experience an outcome were followed to their last encounter with the health care system and censored at that date. For patients who were hospitalized more than once, we selected a random hospitalization per person as baseline. Individuals with a history of venous thromboembolism before or during index hospitalization; those pregnant at the time of index hospitalization; or those without available information on serum creatinine, urinary albumin–creatinine ratio (UACR), or urinary protein–creatinine ratio were excluded from the study.
Exposures of Interest
Primary exposures of interest were eGFR and UACR. eGFR was calculated from serum creatinine measures using the 2021 Chronic Kidney Disease Epidemiology Collaboration equation.20 The latest inpatient serum creatinine measure before discharge was used to estimate GFR. The most recent outpatient creatinine measure within 2 years before hospitalization was used when inpatient measures were not available. The latest outpatient measures before index hospitalization were used for UACR because inpatient measures were rare. For individuals without available information on UACR, we converted their urinary protein–creatinine ratio to UACR using a validated formula.21 We defined CKD per Kidney Disease Improving Global Outcomes 2012 guidelines as eGFR <60 ml/min per 1.73 m2 or UACR ≥30 mg/g. In sensitivity analysis, we used only outpatient measures of serum creatinine obtained before the index hospitalization to estimate GFR.
Outcomes of Interest
The primary outcome was incident venous thromboembolism after the index hospitalization. Incident venous thromboembolism was defined by diagnostic/billing codes that indicated pulmonary embolism and other venous thromboembolism using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) or ICD-10-CM (Supplemental Table 1). As secondary outcomes, we investigated venous thromboembolism that occurred within 1 and 3 years of hospitalization.
Other Covariates of Interest
Baseline characteristics such as age, sex, race, body mass index (BMI), BP, and cigarette smoking status (current, former, and never) were collected for analysis. Sex and race were self-reported by the individuals. BMI was calculated using the last outpatient measure of weight within 2 years before hospitalization. Baseline comorbidities, such as history of cancer, prior fracture, stroke, heart failure, coronary heart disease, and diabetes mellitus (types 1 and 2), were defined by the presence of ICD codes (ICD-9 and ICD-10) before index hospitalization, as listed in Supplemental Table 1. Medication use was defined as having an active prescription on the day of discharge. Anticoagulants included warfarin or direct oral anticoagulant (DOAC). Antiplatelet agents consisted of aspirin or P2Y12 receptor blockers.
Statistical Analyses
We described the baseline characteristics of the overall study cohorts (Geisinger Health System and NYU Langone Health) and by CKD status (CKD versus no CKD).
We evaluated the number and percentage of events and raw incidence rate per 1000 person-years by eGFR (G) and UACR (A) categories and compared incidence rates in each combined G and A categories with the G1A1 (eGFR ≥90 ml/min per 1.73 m2 and UACR <30 ml/min per 1.73 m2) category using an unadjusted Poisson model and by calculating the differences in the rates.
We also evaluated the adjusted incidence rates of incident venous thromboembolism by eGFR and UACR categories in both cohorts using a Poisson model. We reported adjusted incidence rates for the average level of covariates in the Geisinger cohort. Cox proportional hazards models were used to determine the independent associations of eGFR and UACR with venous thromboembolism. The results were expressed as hazard ratios (HRs) with 95% confidence intervals (CIs) and P values. We first modeled eGFR and log-transformed UACR as cubic splines with three knots to evaluate for nonlinear associations. We then modeled eGFR as a linear spline with a knot at 60 ml/min per 1.73 m2 and log-transformed UACR as a continuous variable. Analyses were performed separately in each cohort and then meta-analyzed using random-effects models. All models were adjusted for age, sex, eGFR, log-transformed UACR, BMI, smoking status, systolic BP, comorbidities, medication use, and length of hospitalization, as listed in Table 1.
Table 1.
Baseline characteristics of the study populations
| Variable | Geisinger Health System | NYU Langone Health | ||||||
|---|---|---|---|---|---|---|---|---|
| Overall (n=23,889) | No CKD (n=10,762) | CKDa (n=13,127) | Missing n (%) |
Overall (n=11,552) | No CKD (n=4765) | CKDa (n=6787) | Missing n (%) |
|
| Age, mean (SD), yr | 67 (15) | 63 (14) | 71 (14) | 66 (15) | 62 (15) | 69 (15) | ||
| Female sex, n (%) | 12,190 (51) | 5682 (53) | 6508 (50) | 5695 (49) | 2555 (54) | 3140 (46) | ||
| Black race, n (%) | 483 (2) | 244 (2) | 239 (2) | 1915 (17) | 775 (16) | 1140 (17) | ||
| BMI, mean (SD) | 33 (8) | 34 (9) | 32 (8) | 1830 (8) | 30 (8) | 31 (8) | 30 (7) | 205 (2) |
| Systolic BP, mean (SD), mm Hg | 130 (20) | 129 (18) | 130 (21) | 487 (2) | 129 (19) | 126 (17) | 131 (20) | 192 (2) |
| Diastolic BP, mean (SD), mm Hg | 71 (11) | 72 (11) | 70 (12) | 487 (2) | 73 (11) | 75 (10) | 73 (12) | 192 (2) |
| Cigarette smoking, n (%) | ||||||||
| Current | 3287 (14) | 1683 (16) | 1604 (12) | 802 (7) | 361 (8) | 441 (7) | ||
| Former | 10,844 (46) | 4607 (43) | 6237 (48) | 4564 (40) | 1733 (36) | 2831 (42) | ||
| Never | 9758 (41) | 4472 (41) | 5286 (40) | 6186 (54) | 2671 (56) | 3515 (52) | ||
| eGFR (inpatient or outpatient), mean (SD), ml/min per 1.73 m2 b | 71 (29) | 90 (16) | 56 (27) | 69 (31) | 91 (17) | 54 (30) | ||
| UACR, median (IQR), mg/gc | 17 (7–73) | 8 (4–14) | 60 (22–222) | 26 (9–131) | 9 (5–16) | 91 (35–401) | ||
| Comorbidity, n (%) | ||||||||
| Cancer | 6252 (26) | 2418 (23) | 3834 (29) | 3090 (27) | 1145 (24) | 1945 (29) | ||
| Fracture | 5451 (23) | 2221 (21) | 3230 (25) | 2357 (20) | 917 (19) | 1440 (21) | ||
| Stroke | 5167 (22) | 1793 (17) | 3374 (26) | 2337 (20) | 713 (15) | 1624 (24) | ||
| Heart failure | 6161 (26) | 1377 (13) | 4784 (36) | 2966 (26) | 582 (12) | 2384 (35) | ||
| Coronary heart disease | 10,771 (45) | 3741 (35) | 7030 (54) | 5362 (46) | 1594 (34) | 3768 (56) | ||
| Diabetesd | 18,448 (77) | 8407 (78) | 10,041 (77) | 7884 (68) | 3002 (63) | 4882 (72) | ||
| Medication use, n (%) e | ||||||||
| Warfarin | 2318 (10) | 783 (7) | 1535 (12) | 332 (3) | 76 (2) | 256 (4) | ||
| DOAC | 354 (2) | 125 (1) | 229 (2) | 1411 (12) | 401 (8) | 1010 (15) | ||
| Antiplatelet | 9742 (41) | 3608 (34) | 6134 (47) | 4539 (39) | 1873 (39) | 2666 (39) | ||
| Hospitalization duration, median (IQR), d | 4 (2–6) | 3 (2–5) | 4 (3–7) | 4 (3–7) | 3 (2–6) | 4 (3–8) | ||
BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); DOAC, direct oral anticoagulant; IQR, interquartile range; KDIGO, Kidney Disease Improving Global Outcomes; NYU, New York University; UACR, urinary albumin-to-creatinine ratio.
CKD is defined as eGFR <60 ml/min per 1.73 m2 or UACR ≥30 mg/g per KDIGO 2012 guidelines.
The latest inpatient serum creatinine before discharge was used to estimate GFR. The most recent outpatient creatinine within 2 years before hospitalization was used when inpatient measures were not available.
Outpatient UACR was used since inpatient measures were very rare.
Includes both type 1 and type 2 diabetes mellitus.
Medication use was defined as having an active prescription on the day of discharge.
The extent of missing data for each covariate is also presented in Table 1. Missing data were imputed using the multiple imputation by chained equations method with a linear model using age, sex, eGFR, log-transformed UACR, BMI, smoking status, systolic BP, comorbidities, medication use, and the venous thromboembolism outcome as predictors. Five imputed datasets were created, and the estimates from these data for each analysis were pooled within each cohort.
Because of the large number of urine dipstick measures performed in Geisinger, in a sensitivity analysis, we included individuals with outpatient dipstick measures, converting these measures to UACR.21 Analyses were conducted using Stata/SE version 17 (College Station, TX). P < 0.05 was considered significant.
Results
Study Population and Baseline Characteristics
Overall, there were 23,899 individuals admitted to one of the Geisinger hospitals and 11,552 individuals admitted to one of the NYU hospitals who met inclusion criteria (Figure 1).
Figure 1.
Patient selection. (A) Geisinger Health System and (B) New York University (NYU) Langone Health. UACR, urinary albumin-to-creatinine ratio; UPCR, urinary protein–creatinine ratio; VTE, venous thromboembolism.
The mean (SD) ages of the study cohorts (Geisinger and NYU) were 67 (15) and 66 (15) years, respectively (Table 1). Women accounted for 51% and 49% of participants in Geisinger and NYU, respectively. The NYU cohort included a higher proportion of Black individuals compared with the Geisinger cohort (17% versus 2%). Obesity was prevalent in both study cohorts, with an average BMI of 33 (8) at Geisinger and 30 (8) at NYU. Warfarin was more commonly used than DOAC in the Geisinger cohort, whereas the opposite was true for NYU.
Incident Venous Thromboembolism
In the Geisinger cohort, 5% of individuals developed incident venous thromboembolism after index hospitalization over a mean follow-up of 3.4 years, with 3% developing venous thromboembolism within 3 years of index hospitalization. In the NYU cohort, 4% of the individuals developed incident venous thromboembolism after index hospitalization over a mean of 2.4 years, with 3% developing venous thromboembolism within 3 years of index hospitalization.
The rate of incident venous thromboembolism was higher with greater severity of CKD (both G and A categories) in both cohorts. In the Geisinger cohort, individuals in the G1A1 (eGFR >90 ml/min per 1.73 m2 and UACR <30 mg/g) category had an adjusted incidence rate of 10.7 (95% CI, 9.2 to 12.6) events per 1000 person-years and 27.7 (95% CI, 20.6 to 37.2) events per 1000 person-years in G4-5A3 (eGFR <30 ml/min per 1.73 m2 and UACR >300 mg/g) (Table 2). In NYU, the adjusted incidence rate ranged from 11.7 (95% CI, 9.1 to 14.9) events per 1000 person-years to 27.1 (95% CI, 20.2 to 36.5) events per person-years (Table 2). The absolute number of incident venous thromboembolic events and the unadjusted incidence rates and rate differences are provided in Supplemental Table 2.
Table 2.
Adjusted incidence rates (1000 person-years) of venous thromboembolism by eGFR (G) and UACR (A) categories
| G/A Category | A1 (0–30 mg/g) | A2 (30–300 mg/g) | A3 (>300 mg/g) | |||
|---|---|---|---|---|---|---|
| eGFR (G)a | IR | P Value | IR | P Value | IR | P Value |
| Geisinger Health System | ||||||
| ≥90 (G1) | 10.7 (9.2–12.6) | REF | 13.9 (10.7–18.1) | 0.08 | 24.3 (15.9–37.3) | <0.001 |
| 60–89 (G2) | 13.4 (11.8–15.2) | 0.04 | 13.6 (11.0–16.8) | 0.09 | 21.5 (15.2–30.4) | <0.001 |
| 30–59 (G3a and G3b) | 14.2 (11.9–16.9) | 0.03 | 13.4 (10.6–16.9) | 0.15 | 22.9 (17.0–30.9) | <0.001 |
| <30 (G4 and G5) | 14.0 (8.5–22.9) | 0.33 | 28.7 (20.8–39.8) | <0.001 | 27.7 (20.6–37.2) | <0.001 |
| NYU Langone Health | ||||||
| ≥90 (G1) | 11.7 (9.1–14.9) | REF | 14.5 (10.0–20.9) | 0.33 | 6.2 (2.0–19.3) | 0.28 |
| 60–89 (G2) | 11.5 (8.9–14.9) | 0.93 | 15.7 (11.6–21.4) | 0.14 | 17.7 (10.2–30.8) | 0.17 |
| 30–59 (G3a and G3b) | 16.2 (12.0–21.8) | 0.10 | 13.8 (9.9–19.3) | 0.42 | 18.3 (12.1–27.9) | 0.07 |
| <30 (G4 and G5) | 21.7 (11.9–39.6) | 0.06 | 17.0 (10.5–27.7) | 0.18 | 27.1 (20.2–36.5) | <0.001 |
Models were adjusted for age, sex, smoking, BMI, systolic BP, history of coronary heart disease, diabetes, hypertension, heart failure, cancer, fracture, anticoagulant use, antiplatelet use, and duration of hospitalization. Incidence rates are reported with covariates set at the average of the Geisinger population as reported in Table 1. BMI, body mass index; IR, incidence rate; NYU, New York University; UACR, urinary albumin-to-creatinine ratio.
Measures in ml/min per 1.73 m2.
Association between CKD and Venous Thromboembolism
When analyzed as cubic splines, lower eGFR and higher UACR were associated with higher risks of venous thromboembolism in both the Geisinger and NYU cohorts (Figure 2). In meta-analyses, every 10 ml/min per 1.73 m2 reduction in eGFR below 60 ml/min per 1.73 m2 was associated with a 6% higher risk of venous thromboembolism (HR 1.06 [95% CI, 1.02 to 1.11], P = 0.01). However, above eGFR 60 ml/min per 1.73 m2, the association was not statistically significant (HR 1.01 [95% CI, 0.97 to 1.05], P = 0.57). Similarly, in meta-analyses, each two-fold higher UACR was independently associated with a higher risk of venous thromboembolism (HR 1.05 [95% CI, 1.03 to 1.07], P < 0.001). The results were qualitatively similar between the Geisinger and NYU cohorts but not statistically significant in the latter (Table 3).
Figure 2.
Adjusted hazard ratios of venous thromboembolism with eGFR and (log) UACR modeled as cubic splines. Models were adjusted for age, sex, smoking, body mass index, systolic BP, history of coronary heart disease, diabetes, hypertension, heart failure, cancer, fracture, anticoagulant use, antiplatelet use, and duration of hospitalization. (A) and (B) Associations with eGFR and (C) and (D) associations with UACR in Geisinger and NYU, respectively. CI, confidence interval.
Table 3.
Hazard ratios (95% confidence intervals) for the risk of venous thromboembolism on the basis of eGFR and urinary albumin-to-creatinine ratio
| Risk Factor | Geisinger Health System | NYU Langone Health | Pooled | |||
|---|---|---|---|---|---|---|
| HR (95% CI) | P Value | HR (95% CI) | P Value | HR (95% CI) | P Value | |
| eGFR <60 ml/min per 1.73 m2 (every 10 ml/min per 1.73 m2 lower) | 1.07 (1.01 to 1.14) | 0.02 | 1.05 (0.97 to 1.13) | 0.22 | 1.06 (1.02 to 1.11) | 0.01 |
| eGFR >60 ml/min per 1.73 m2 (every 10 ml/min per 1.73 m2 lower) | 1.00 (0.95 to 1.05) | 0.94 | 1.04 (0.97 to 1.12) | 0.25 | 1.01 (0.97 to 1.05) | 0.57 |
| UACR (per two-fold higher) | 1.05 (1.03 to 1.08) | <0.001 | 1.04 (1.00 to 1.08) | 0.05 | 1.05 (1.03 to 1.07) | <0.001 |
Models were adjusted for age, sex, smoking, BMI, systolic BP, history of coronary heart disease, diabetes, hypertension, heart failure, cancer, fracture, anticoagulant use, antiplatelet use, and duration of hospitalization. The results were pooled by meta-analysis using random effects. BMI, body mass index; CIs, confidence intervals; HRs, hazard ratios; NYU, New York University; UACR, urinary albumin-to-creatinine ratio.
Associations of other risk factors with venous thromboembolism are presented in Supplemental Table 3.
Sensitivity Analyses
The results were similar to the primary findings in analyses restricted to venous thromboembolism outcomes that occurred within 3 years and within 1 year of the index hospitalization (Supplemental Table 4), although no longer statistically significant at the 1-year time horizon, likely because of the relatively low number of events. In the Geisinger cohort, including patients with dipstick measures of urine protein increased the population size to 58,152. Associations with eGFR and UACR with venous thromboembolism were consistent in this population (Supplemental Table 5). Finally, estimating GFR solely on the basis of outpatient measures obtained before hospitalization resulted in similar findings (Supplemental Table 6).
Discussion
CKD has been increasingly recognized as a risk factor of venous thromboembolism in community-dwelling populations.16–19 This study of more than 35,000 hospitalized adults in two large health care systems extends these findings to evaluate the independent and combined contribution of eGFR and albuminuria to the development of venous thromboembolism. We found that lower eGFR below 60 ml/min per 1.73 m2 and higher UACR were independently associated with a higher risk of incident venous thromboembolism. The observed association was robust with qualitatively similar findings in two different cohorts and across multiple sensitivity analyses.
Previous studies16–19 have identified an association between CKD and incident venous thromboembolism, but with limitations. For example, Watanakit et al. identified CKD as a potential risk factor of venous thromboembolism in their prospective cohort study using the data from the Longitudinal Investigation of Thromboembolism Etiology study.17 However, because no data on albuminuria were available in this cohort, the authors were unable to analyze associations of albuminuria with venous thromboembolism. Conversely, analyses of the PREVEND study demonstrated associations of albuminuria but not eGFR with venous thromboembolism, although there were relatively few individuals with reduced eGFR.16,18 For example, there were only eight individuals with CKD category G4 and three with CKD category G5 in the baseline PREVEND cohort. The analysis of the PREVEND study by Ocak et al. excluded all of these 11 individuals. Similarly, a meta-analysis19 summarizing studies of the association of kidney function and venous thromboembolism identified a total of only 181 individuals with eGFR <30 ml/min per 1.73 m2 combined among five studies, which were excluded from final analyses.
Our study was consistent with prior observations of CKD and venous thromboembolism, but extended upon them in several ways. In particular, we evaluated the associations of both eGFR and UACR with the risk of venous thromboembolism, finding that lower eGFR below 60 ml/min per 1.73 m2 and higher UACR were independently associated with risk of venous thromboembolism, with higher risk with both worsening G and A CKD categories. Our results extend evidence of the association to individuals with advanced CKD (eGFR <30 ml/min per 1.73 m2) for which prior evidence is limited.18,19
Similar to the proposed mechanism in individuals with nephrotic-range proteinuria, the higher risk of venous thromboembolism with CKD, particularly in individuals with albuminuria, may be secondary to loss of anticoagulant proteins.15 Endothelial dysfunction22 or related changes in procoagulant proteins and inflammatory markers, such as increased levels of D-dimer, von Willebrand factor, fibrinogen, factor VII, factor VIII, plasminogen activator inhibitor-1, IL-6, TNF-α, and C-reactive protein, in the setting of CKD have also been observed in individuals with kidney dysfunction and might also contribute to thrombosis risk.23,24 Further research with direct assessment of the levels of coagulation factors and inflammatory biomarkers in individuals with albuminuria and the association with development of venous thromboembolism could provide important insights into the underlying pathophysiology of venous thromboembolism in CKD. Platelet dysfunction might underlie the higher risk of venous thromboembolism in those with low eGFR but without albuminuria because some studies suggest that gut-derived uremic toxins may increase platelet activation and promote thrombosis.25–29
Notable strengths of our study include use of a large representative community-based sample of hospitalized patients with consistent results in a second independent cohort and comprehensive data pertaining to potential confounders. There were also several limitations to our study. First, we used administrative data to identify venous thromboembolism and baseline comorbidities. Although there is some risk of misclassification of comorbidities, the algorithms we used to identify venous thromboembolism have been previously validated.30,31 Phlebitis and superficial vein thrombosis, conditions that do not typically warrant anticoagulation, were excluded from the outcome.
Because clinical data were used, laboratory measurements were at the discretion of health care providers rather than according to uniform protocols. Measurements of urine creatinine, urine protein, and urine albumin were likely not standardized and could be obtained at random times of the day rather than during the recommended first- or second-morning urine sample. Immobility and hereditary predisposition, such as factor V Leiden mutation, prothrombin gene mutation, protein S deficiency, protein C deficiency, and antithrombin deficiency, are well established to be risk factors of venous thromboembolism. Ideally, these covariates should be included in the analyses. However, mobility status is not reliably assessed using EHR data. Given the low incidence of hereditary thrombophilia, it is unlikely that failure to adjust for these conditions qualitatively affected our findings. Finally, we classified eGFR on the basis of the value at hospital discharge, in an effort to model what the discharging provider would base decisions on. That said, this value may not reflect the true level of a patient's kidney function. Reassuringly, the results were similar when only outpatient measures were used.
Warfarin was more commonly used than DOACs in the Geisinger cohort, while the opposite was true for NYU. This likely reflects differences in the eras covered in each of the data sources. The Geisinger cohort contains earlier data than the NYU cohort, when DOACs were not frequently used. Despite the difference, the associations between eGFR and venous thromboembolism and UACR and venous thromboembolism were similar across two distinct health systems.
In conclusion, both lower eGFR and higher UACR were independently associated with higher risk of venous thromboembolism after hospitalization, with higher incidence in the most severe categories of CKD. These results should motivate enhanced attention to venous thromboembolism prophylaxis in the CKD population, particularly among those with more advanced categories of CKD.
Supplementary Material
Acknowledgments
Because Dr. David M. Charytan is an Associate Editor of CJASN, he was not involved in the peer-review process for this manuscript. Another editor oversaw the peer-review and decision-making process for this manuscript.
Disclosures
A.R. Chang reports employment with Geisinger Health System; consultancy for Amgen, Medscape, Novartis, and Reata; research funding from Bayer, Novartis, and Novo Nordisk; advisory or leadership roles for Amgen and Reata; and grant support from National Kidney Foundation. D.M. Charytan reports consultancy for Allena Pharmaceuticals (DSMB), Amgen, AstraZeneca, CSL Behring, Eli Lilly/Boehringer Ingelheim, Fresenius, Gilead, GSK, Medtronic, Merck, Novo Nordisk, Renalytix, and Zogenix; research funding from Medtronic-clinical trial support, Amgen, Gilead, and Novo Nordisk; royalties from UpToDate for authorship/edits on reviews; advisory or leadership role as an Associate Editor of CJASN; and expert witness fees related to proton pump inhibitors. M.E. Grams reports advisory or leadership roles for American Journal of Kidney Diseases, CJASN, JASN Editorial Fellowship Committee, NKF Scientific Advisory Board, KDIGO Executive Committee (co-chair elect), USRDS Scientific Advisory Board, and ASN Publication Committee; grant funding from NKF, which receives funding from multiple pharmaceutical companies; grant funding from NIH; payment from academic institutions for grand rounds; payment from NephSAP; and travel reimbursement from KDIGO and the Korean Society of Nephrology. J.-I. Shin reports research funding from NIH and Merck. All remaining authors have nothing to disclose.
Funding
This work is supported by NIH from K24HL155861 and NIDDK from R01DK115534 (M.E. Grams).
Author Contributions
Conceptualization: David M. Charytan, Morgan E. Grams, Zhong Zheng.
Data curation: Alex R. Chang, Morgan E. Grams, Krutika Pandit, Aditya Surapaneni.
Formal analysis: Aditya Surapaneni.
Investigation: Aditya Surapaneni.
Methodology: Morgan E. Grams, Aditya Surapaneni.
Supervision: David M. Charytan, Morgan E. Grams.
Writing – original draft: Zhong Zheng.
Writing – review & editing: Alex R. Chang, David M. Charytan, Morgan E. Grams, Krutika Pandit, Jung-Im Shin, Aditya Surapaneni.
Data Sharing Statement
Data cannot be shared. The investigation is based on patient data from health care provided at two health systems in the United States. So we are unable to deposit the data in a public database. But we will be happy to respond to any reasonable inquiry with summary statistics, etc.
Supplemental Material
This article contains the following supplemental material online at http://links.lww.com/CJN/B836.
Supplemental Table 1. Diagnostic/billing codes.
Supplemental Table 2A and 2B. Unadjusted incidence of venous thromboembolism by eGFR and UACR categories and incidence rate differences by eGFR and UACR categories.
Supplemental Table 3. Other factors associated with venous thromboembolism.
Supplemental Table 4A and 4B. Adjusted hazard ratios for eGFR and UACR associated with incident venous thromboembolism within 3 years and 1 year of hospitalization.
Supplemental Table 5. Adjusted hazard ratios for eGFR and UACR associated with venous thromboembolism when additionally using dipstick measures converted to UACR.
Supplemental Table 6. HRs (95% CIs) for eGFR and UACR associated with venous thromboembolism using only outpatient measures of eGFR.
References
- 1.Heit JA. The epidemiology of venous thromboembolism in the community: implications for prevention and management. J Thromb Thrombolysis. 2006;21(1):23–29. doi: 10.1007/s11239-006-5572-y [DOI] [PubMed] [Google Scholar]
- 2.Heit JA, Spencer FA, White RH. The epidemiology of venous thromboembolism. J Thromb Thrombolysis. 2016;41(1):3–14. doi: 10.1007/s11239-015-1311-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Naess IA, Christiansen SC, Romundstad P, Cannegieter SC, Rosendaal FR, Hammerstrøm J. Incidence and mortality of venous thrombosis: a population-based study. J Thromb Haemost. 2007;5(4):692–699. doi: 10.1111/j.1538-7836.2007.02450.x [DOI] [PubMed] [Google Scholar]
- 4.Dartevelle P Fadel E Mussot S, et al.. Chronic thromboembolic pulmonary hypertension. Eur Respir J. 2004;23(4):637–648. doi: 10.1183/09031936.04.00079704 [DOI] [PubMed] [Google Scholar]
- 5.Pengo V Lensing AW Prins MH, et al.. Incidence of chronic thromboembolic pulmonary hypertension after pulmonary embolism. N Engl J Med. 2004;350(22):2257–2264. doi: 10.1056/NEJMoa032274 [DOI] [PubMed] [Google Scholar]
- 6.Kahn SR Kearon C Julian JA, et al.. Predictors of the post-thrombotic syndrome during long-term treatment of proximal deep vein thrombosis. J Thromb Haemost. 2005;3(4):718–723. doi: 10.1111/j.1538-7836.2005.01216.x [DOI] [PubMed] [Google Scholar]
- 7.Roumen-Klappe EM, den Heijer M, Janssen MCH, van der Vleuten C, Thien T, Wollersheim H. The post-thrombotic syndrome: incidence and prognostic value of non-invasive venous examinations in a six-year follow-up study. Thromb Haemost. 2005;94(4):825–830. doi: 10.1160/TH05-03-0146 [DOI] [PubMed] [Google Scholar]
- 8.Baldwin MJ, Moore HM, Rudarakanchana N, Gohel M, Davies AH. Post-thrombotic syndrome: a clinical review. J Thromb Haemost. 2013;11(5):795–805. doi: 10.1111/jth.12180 [DOI] [PubMed] [Google Scholar]
- 9.Guérin L Couturaud F Parent F, et al.. Prevalence of chronic thromboembolic pulmonary hypertension after acute pulmonary embolism. Prevalence of CTEPH after pulmonary embolism. Thromb Haemost. 2014;112(3):598–605. doi: 10.1160/TH13-07-0538 [DOI] [PubMed] [Google Scholar]
- 10.Ende-Verhaar YM Cannegieter SC Vonk Noordegraaf A, et al.. Incidence of chronic thromboembolic pulmonary hypertension after acute pulmonary embolism: a contemporary view of the published literature. Eur Respir J. 2017;49(2):1601792. doi: 10.1183/13993003.01792-2016 [DOI] [PubMed] [Google Scholar]
- 11.Anderson FA, Jr, Spencer FA. Risk factors for venous thromboembolism. Circulation. 2003;107(23 suppl 1):I9–I16. doi: 10.1161/01.CIR.0000078469.07362.E6 [DOI] [PubMed] [Google Scholar]
- 12.Llach F. Hypercoagulability, renal vein thrombosis, and other thrombotic complications of nephrotic syndrome. Kidney Int. 1985;28(3):429–439. doi: 10.1038/ki.1985.149 [DOI] [PubMed] [Google Scholar]
- 13.Rabelink TJ, Zwaginga JJ, Koomans HA, Sixma JJ. Thrombosis and hemostasis in renal disease. Kidney Int. 1994;46(2):287–296. doi: 10.1038/ki.1994.274 [DOI] [PubMed] [Google Scholar]
- 14.Crew RJ, Radhakrishnan J, Appel G. Complications of the nephrotic syndrome and their treatment. Clin Nephrol. 2004;62(4):245–259. doi: 10.5414/cnp62245 [DOI] [PubMed] [Google Scholar]
- 15.Singhal R, Brimble KS. Thromboembolic complications in the nephrotic syndrome: pathophysiology and clinical management. Thromb Res. 2006;118(3):397–407. doi: 10.1016/j.thromres.2005.03.030 [DOI] [PubMed] [Google Scholar]
- 16.Mahmoodi BK Gansevoort RT Veeger NJ, et al.. Microalbuminuria and risk of venous thromboembolism. JAMA. 2009;301(17):1790–1797. doi: 10.1001/jama.2009.565 [DOI] [PubMed] [Google Scholar]
- 17.Wattanakit K, Cushman M, Stehman-Breen C, Heckbert SR, Folsom AR. Chronic kidney disease increases risk for venous thromboembolism. J Am Soc Nephrol. 2008;19(1):135–140. doi: 10.1681/ASN.2007030308 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Ocak G Verduijn M Vossen CY, et al.. Chronic kidney disease stages 1-3 increase the risk of venous thrombosis. J Thromb Haemost. 2010;8(11):2428–2435. doi: 10.1111/j.1538-7836.2010.04048.x [DOI] [PubMed] [Google Scholar]
- 19.Mahmoodi BK Gansevoort RT Næss IA, et al.. Association of mild to moderate chronic kidney disease with venous thromboembolism: pooled analysis of five prospective general population cohorts. Circulation. 2012;126(16):1964–1971. doi: 10.1161/CIRCULATIONAHA.112.113944 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Inker LA Eneanya ND Coresh J, et al.. New creatinine- and cystatin C-based equations to estimate GFR without race. N Engl J Med. 2021;385(19):1737–1749. doi: 10.1056/NEJMoa2102953 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Sumida K Nadkarni GN Grams ME, et al.. Conversion of urine protein-creatinine ratio or urine dipstick protein to urine albumin-creatinine ratio for use in chronic kidney disease screening and prognosis: an individual participant-based meta-analysis. Ann Intern Med. 2020;173(6):426–435. doi: 10.7326/M20-0529 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Clausen P, Feldt-Rasmussen B, Jensen G, Jensen JS. Endothelial haemostatic factors are associated with progression of urinary albumin excretion in clinically healthy subjects: a 4-year prospective study. Clin Sci. 1999;97(1):37–43. doi: 10.1042/cs0970037 [DOI] [PubMed] [Google Scholar]
- 23.Dubin R Cushman M Folsom AR, et al.. Kidney function and multiple hemostatic markers: cross sectional associations in the multi-ethnic study of atherosclerosis. BMC Nephrol. 2011;12:3. doi: 10.1186/1471-2369-12-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Keller C, Katz R, Cushman M, Fried LF, Shlipak M. Association of kidney function with inflammatory and procoagulant markers in a diverse cohort: a cross-sectional analysis from the Multi-Ethnic Study of Atherosclerosis (MESA). BMC Nephrol. 2008;9:9. doi: 10.1186/1471-2369-9-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Yang K Du C Wang X, et al.. Indoxyl sulfate induces platelet hyperactivity and contributes to chronic kidney disease-associated thrombosis in mice. Blood. 2017;129(19):2667–2679. doi: 10.1182/blood-2016-10-744060 [DOI] [PubMed] [Google Scholar]
- 26.Chang MC Wang TM Yeung SY, et al.. Antiplatelet effect by p-cresol, a uremic and environmental toxicant, is related to inhibition of reactive oxygen species, ERK/p38 signaling and thromboxane A2 production. Atherosclerosis. 2011;219(2):559–565. doi: 10.1016/j.atherosclerosis.2011.09.031 [DOI] [PubMed] [Google Scholar]
- 27.Karbowska M Kaminski TW Marcinczyk N, et al.. The uremic toxin indoxyl sulfate accelerates thrombotic response after vascular injury in animal models. Toxins (Basel). 2017;9(7):229. doi: 10.3390/toxins9070229 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Bazilinski N Shaykh M Dunea G, et al.. Inhibition of platelet function by uremic middle molecules. Nephron. 1985;40(4):423–428. doi: 10.1159/000183512 [DOI] [PubMed] [Google Scholar]
- 29.Davis JW, McField JR, Phillips PE, Graham BA. Guanidinosuccinic acid on human platelet. Effects of exogenous urea, creatinine, and aggregation in vitro. Blood. 1972;39(3):388–397. doi: 10.1182/blood.v39.3.388.388 [DOI] [PubMed] [Google Scholar]
- 30.Kim S Martin C White J, et al.. Validation of an algorithm to identify venous thromboembolism in health insurance claims data among patients with rheumatoid arthritis. Clin Epidemiol. 2023;15:671–682. doi: 10.2147/CLEP.S402360 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Fang MC Fan D Sung SH, et al.. Validity of using inpatient and outpatient administrative codes to identify acute venous thromboembolism: the CVRN VTE study. Med Care. 2017;55(12):e137–e143. doi: 10.1097/MLR.0000000000000524 [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
Data Availability Statement
Data cannot be shared. The investigation is based on patient data from health care provided at two health systems in the United States. So we are unable to deposit the data in a public database. But we will be happy to respond to any reasonable inquiry with summary statistics, etc.