Abstract
Objectives:
To evaluate whether the incidence of infectious diseases increases the long-term risk of incident end-stage renal disease (ESRD) in the general population.
Patients and Methods:
In 10,290 participants of the Atherosclerosis Risk in Communities Study who attended visit 4 (1996-1998), we evaluated the association of incident hospitalization with major infections (pneumonia, urinary tract infection, bloodstream infection, and cellulitis/osteomyelitis) with subsequent risk of ESRD through September, 2015. Hospitalization with major infection was entered into multivariable Cox models as a time-varying exposure to estimate the hazard ratios (HRs).
Results:
The mean age was 63 years, 56% were female, 22% were black, and 7% had an estimated glomerular filtration rate <60 ml/min/1.73m2. During a median follow-up of 17.4 years, there were 2,642 incident hospitalizations with major infection and 281 cases of ESRD (132 cases after hospitalizations with major infection). The risk of ESRD was higher following major infection compared to while free of major infection (crude incidence rate, 10.9 vs. 1.0 per 1,000 person-years). In multivariable time-varying Cox analysis, hospitalization with major infection was associated with a 3.3-fold increased risk of ESRD (HR, 3.34 [95%CI, 2.56-4.37]). The association was similar across pneumonia, urinary tract infection, bloodstream infection, and cellulitis/osteomyelitis and remained significant across subgroups of age, sex, race, diabetes, history of cardiovascular disease, and chronic kidney disease.
Conclusions:
Hospitalization with major infection was independently and robustly associated with subsequent risk of ESRD. Whether preventive approaches against infection have beneficial impacts on kidney outcomes may deserve future investigations.
Keywords: Chronic kidney disease, Chronic renal Insufficiency, Chronic kidney failure, Glomerular filtration rate, Albuminuria, Proteinuria, End-stage renal disease, Hospitalization, Infection, Infectious disease
Introduction
The incidence of end-stage renal disease (ESRD) has been more than doubled from 51,000 cases in 1990 to 124,000 cases in 2015.1 Patients with ESRD are at a greater risk of cardiovascular disease and mortality, posing a significant social and economic burden.1 Currently, there are few interventions to reverse reduced kidney function. Epidemiological studies to identify modifiable risk factors of ESRD have important public health implications, since interventions to those modifiable risk factors potentially reduce the burden of ESRD.2
Infectious disease may play an important role in the development of ESRD in this context since animal studies demonstrated that infection could cause fibrosis of the kidney,3 ischemia,4,5 and unstable atherosclerosis6 leading to kidney injury. Infection can also cause hemodynamic changes leading to the development of tissue hypo-perfusion.7,8 Medical interventions for treating infection (e.g., antibiotics) may also contribute to worsening kidney function through renal tubule and interstitium injury.9–11 A limited number of previous studies have investigated the association between infection and risk of ESRD.12–14 However, previous investigations have studied selected populations (i.e., Asian population, military conscripted men, advanced CKD [chronic kidney disease] patients), analyzed limited infection types (e.g., pneumonia), relied on ESRD ascertainment using hospital discharge records alone, and/or did not account for some major confounders (e.g., kidney function, albuminuria).
We hypothesized that major acute infections such as pneumonia, urinary tract infection, bloodstream infection, and cellulitis/osteomyelitis that were severe enough to warrant hospitalization (hereafter, major infection) would increase subsequent risk of ESRD, regardless of infection type and across demographic and clinical subgroups, after rigorously accounting for risk factors for developing ESRD. To test our hypothesis, we analyzed data from the Atherosclerosis Risk in Communities (ARIC) Study, a community-based cohort of mostly black and white men and women.
Methods
Study population
The ARIC Study is a prospective cohort of 15,792 individuals who were aged 45-64 years at the time of enrollment in 1987-1989 (visit 1) from four US communities (Forsyth County, North Carolina; Jackson, Mississippi; suburbs of Minneapolis, Minnesota; and Washington County, Maryland).15 Three short-term follow-up visits were conducted approximately every 3 years in 1990-1992 (visit 2), 1993-1995 (visit 3), and 1996-1998 (visit 4). For the present study, we used data at visit 4 as the baseline since estimated glomerular filtration rate (eGFR) and urinary albumin-to-creatinine ratio (ACR), two major risk factors for ESRD,16–19 were simultaneously assessed for the first time in ARIC at this visit. Of 11,656 individuals who attended visit 4, we excluded those who reported race other than white or black (n=31), had prevalent ESRD or eGFR <30 ml/min/1.73m2 (n=69), had a history of hospitalization with major infection (n=719), or had missing covariates (n=547), leaving 10,290 individuals included in the present analysis. Written informed consent was obtained from all participants, and the institutional review board at each study site approved the study.
Exposure
The primary exposure of interest was incident hospitalization with major infection. We focused on acute infections, since chronic infections such as human immunodeficiency virus and hepatitis C infection may interact with kidney through different mechanisms20,21 and are quite rare. According to the previous literature,22 we included the four most frequent types of infection: pneumonia, urinary tract infection, bloodstream infection, and cellulitis/osteomyelitis. In ARIC, investigators identify all hospitalizations through annual telephone calls to participants and their proxies, and active surveillance collects hospital discharge information including the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes. ICD-9CM codes 480-486 (pneumonia), 590.1, 599.0, and 601.0 (urinary tract infection), 038, 054.5, 785.52, 790.7, 995.91, and 995.92 (bloodstream infection), and 681, 682, 730.0-2, and 040.0 (cellulitis/osteomyelitis) were used to identify infection events. All major infection events were ascertained regardless of their diagnostic position.
Outcome
The primary outcome of interest was incident ESRD. Cases of ESRD were identified through the linkage to the United States Renal Data System. Participants who did not develop incident ESRD were censored when they were lost to follow-up or died, whichever came first. The follow-up was continued until September 30, 2015. This date was chosen because of the change in the coding system from ICD-9CM to ICD-10 occurred on October 1, 2015.
Covariates
All covariates except for years of education were assessed at visit 4 (1996-1998). Years of education were based on visit 1 (1987-1989) data. Smoking (ever vs. never) and alcohol history (ever vs. never drinker) were based on self-report questionnaires. Diabetes was defined as a self-reported physician diagnosis, a fasting glucose ≥126 mg/dL, a random glucose ≥200 mg/dL, or taking antidiabetic drugs. GFR was estimated using the Chronic Kidney Disease Epidemiology Collaboration equation.23 ACR was calculated by dividing urinary albumin by urinary creatinine. History of cardiovascular disease (coronary heart disease, heart failure, or stroke) was defined as prevalent cases at visit 1 or incident cases between visit 1 and visit 4.24 History of chronic obstructive pulmonary disease (COPD) and cancer were defined as hospitalization with ICD-9CM 490-492, 494, and 496 (COPD) and 140-165, 170-176, 179-209, and 235-239 (cancer) between visit 1 and visit 4, respectively. We also used annually updated covariates on diabetes and hypertension (semi-annually after 2012) through telephone calls to participants; as well as body mass index, ever smoking, ever drinking, systolic blood pressure, anti-hypertensive drugs, eGFR, ACR, and history of COPD and cancer at visit 5 (2011-2013).
Statistical analysis
Baseline characteristics were compared between those who did and did not have hospitalization with major infection using chi-square tests for categorical variables and Student’s t-tests for continuous variables. Crude incidence rates and their 95% confidence intervals (95%CIs) were estimated using Poisson regression models.
To evaluate the association of hospitalization with infection with subsequent risk of ESRD, we used multivariable Cox proportional hazard models, where we entered the incident hospitalization with infection into the model as a binary time-varying exposure (yes=1 vs. no=0). In other words, participants contributed person-time to the “no infection (=0)” category until they had the incident hospitalization with infection, when they began contributing person-time to the “infection (=1)” category for the remaining period of follow-up. For those who had multiple events of different infection subtypes (e.g., pneumonia as the first event and urinary tract infection as the second event), the first event (pneumonia in this example) was used for the analysis of all major infection and multiple events contributed to the analysis for each relevant infection subtype (pneumonia and urinary tract infection in this example as appropriate). In multivariable Cox analyses, Model 1 adjusted for age, sex, and race. Model 2 additionally adjusted for body mass index, ever smoking, ever drinking, years of education, systolic blood pressure, antihypertensive medication use, eGFR (continuous), ACR (continuous), history of cardiovascular disease, diabetes, COPD and cancer, with relevant covariates updated as described above. Model 3 additionally adjusted for incident cardiovascular disease as a time-varying covariate, since previous studies suggested that infection may increase the risk of cardiovascular disease,25 which may contribute to subsequent CKD progression.26,27 The proportional hazards assumption was confirmed by plotting the log of the negative log of the estimated survival probability vs. the log of follow-up time. We performed the analysis for all four major infection subtypes combined (i.e., first incident hospitalization with any infection subtype) and each subtype (i.e., pneumonia, urinary tract infection, bloodstream infection, and cellulitis and osteomyelitis) separately.
We also explored the association of hospitalization with osteoarthritis (ICD-9CM, 715) as a non-infection related hospitalization comparator, since hospitalized patients may generally be sick and have a higher risk of ESRD regardless of infection. We selected osteoarthritis, since it is the most common cause of hospitalization among middle to older age adults in the US,28 but presumably not as closely linked to the ESRD risk as infectious disease.
For subgroup analyses, we analyzed the association separately in a priori determined categories of age (<65 vs. ≥65 years), sex (male vs. female), race (white vs. black), diabetes (yes vs. no), history of cardiovascular disease (yes vs. no), and CKD (yes vs. no). CKD was defined as eGFR <60 ml/min/1.73m2 or ACR ≥30 mg/g.
For sensitivity analysis, we repeated the analysis restricting the ascertainment of major infections to those occurred within the first five years of follow-up, because infections that occurred years after the baseline may be less related to the baseline characteristics. In addition, infection hospitalizations can be complicated by acute kidney injury, which may affect subsequent ESRD risk. Thus, we performed an exploratory analysis of whether the association differed by a concurrent diagnosis of acute kidney injury during hospitalization with infection. ICD-9CM 584 (acute kidney injury) was used to identify cases of acute kidney injury.
Finally, we quantified the absolute risk of ESRD following hospitalization with infection. For this analysis, we used the incidence density sampling methods to identify two participants without infection as a comparator at every occurrence of hospitalization with infection, matched on age, sex, and race and sampled from those who remained in the risk set.29 We used Kaplan Meier curves to depict the cumulative incidence of ESRD (i.e., 1 - S(t)) between participants with and without infection. We took this approach because Kaplan Meier curves cannot be drawn in time-varying exposure analysis due to different timings at which individuals are entered into the risk set. A 2-sided p <0.05 was considered statistically significant. All statistical analyses were performed using Stata version 15.
Results
Baseline characteristics
The mean age at baseline was 63 years, 56% were female, 22% were black, and 7% had an eGFR 30-59 ml/min/1.73m2. During a median follow-up of 17.4 years, 2,642 (24%) had incident hospitalization with major infection. Table 1 shows baseline characteristics for those who developed major infection during follow-up and those who did not. Individuals with major infections were more likely to be older and smokers, less likely to have received 12 years or more education, and more likely to have hypertension, diabetes, lower eGFR, higher ACR, cardiovascular disease, COPD, and cancer.
Table1:
Participant characteristics overall, and according to whether the participant was hospitalized with major infection during follow-up: ARIC 1996-2015
|
|
||||
|---|---|---|---|---|
| Overall | Hospitalization with major infection | |||
| Characteristics | (n=10,290) | No (n=7,648) | Yes (n=2,642) |
P-value |
| Age, years, mean (SD) | 63 (5.7) | 62 (5.5) | 64 (5.7) | <0.001 |
| Female, no (%) | 5781 (56) | 4275 (56) | 1506 (57) | 0.32 |
| Black race, no (%) | 2252 (22) | 1655 (22) | 597 (23) | 0.31 |
| Body mass index, kg/m2, mean (SD) | 29 (5.5) | 28 (5.4) | 30 (6.0) | <0.001 |
| Ever smoke, no (%) | 5999 (58) | 4360 (57) | 1639 (62) | <0.001 |
| Ever drink, no (%) | 8183 (80) | 6138 (80) | 2045 (77) | 0.002 |
| Received education 12 years or more, no (%) | 8367 (81) | 6354 (83) | 2013 (76) | <0.001 |
| Systolic blood pressure, mmHg, mean (SD) | 128 (19) | 127 (19) | 130 (19) | <0.001 |
| Diastolic blood pressure, mmHg, mean (SD) | 71 (10) | 71 (10) | 71 (11) | 0.002 |
| Antihypertensive drugs, no (%) | 4365 (42) | 3012 (39) | 1353 (51) | <0.001 |
| Diabetes, no (%) | 1654 (16) | 1075 (14) | 579 (22) | <0.001 |
| Laboratory tests | ||||
| eGFR, ml/min/1.73m2, no (%) | ||||
| ≥90 | 4398 (43) | 3453 (45) | 945 (36) | <0.001 |
| 60-89 | 5226 (51) | 3777 (49) | 1449 (55) | |
| 45-59 | 494 (5) | 311 (4) | 183 (7) | |
| 30-44 | 172 (2) | 107 (1) | 65 (2) | |
| ACR, mg/g, no (%) | ||||
| <10 | 8267 (80) | 6258 (82) | 2009 (76) | <0.001 |
| 10-29 | 1222 (12) | 883 (12) | 339 (13) | |
| 30-299 | 655 (6) | 419 (5) | 236 (9) | |
| ≥300 | 146 (1) | 88 (1) | 58 (2) | |
| Past medical history, no (%) | ||||
| Cardiovascular disease | 1297 (13) | 829 (11) | 468 (18) | <0.001 |
| COPD | 304 (3.0) | 167 (2.2) | 137 (5.2) | <0.001 |
| Cancer | 570 (5.5) | 397 (5.2) | 173 (6.5) | 0.009 |
Abbreviation: ESRD, end-stage renal disease; eGFR, estimated glomerular filtration rate; ACR, urinary albumin-to-creatinine ratio; COPD, chronic obstructive pulmonary disease
Major infection and subsequent risk of ESRD
During follow-up, there were 281 cases of incident ESRD (crude incidence rate, 1.76 per 1,000 person-years [95%CI, 1.56-1.98]). Of those, 149 cases occurred while free of major infection, and 132 cases occurred following major infection (Table 2). The incidence rate of ESRD was nearly 11-fold higher following major infection, compared to the incidence rate while free of major infection (1.0 vs. 10.9 per 1,000 person-years). In the age-, sex-, and race-adjusted Cox model, hospitalization with major infection was associated with an 8.3-fold higher risk of ESRD (HR, 8.26 [95%CI, 6.45-10.58]). After adjustment for other confounders including eGFR and ACR, the association remained significant, with the HR greater than 5-fold (HR, 5.22 [95%CI, 4.03-6.75]). The association remained strong after additionally accounting for incident cardiovascular disease as a time-varying covariate (HR, 3.34 [95%CI, 2.56-4.37]).
Table 2:
The hazard ratios for incident ESRD comparing the risk during person-years following hospitalization with major infection, type-specific infections, and non-infection related hospitalization to the risk during person-years while free of hospitalization: ARIC 1996-2015
|
|
||||||
|---|---|---|---|---|---|---|
| Outcomes | Infection-related hospitalization | Non-infection-related Hospitalization | ||||
|
|
||||||
| All major infection (2,642 events) |
Pneumonia (1,278 events) |
Urinary tract infection (1,341 events) |
Bloodstream infection (685 events) |
Cellulitis and osteomyelitis (559 events) |
Osteoarthritis (2,484 events) |
|
| Person-years while free of hospitalization (×103) | 147.7 | 154.6 | 154.5 | 157.7 | 157.2 | 141.4 |
| ESRD cases | 149 | 216 | 214 | 245 | 244 | 217 |
| IR per 1,000 person-years | 1.0 (0.9-1.2) | 1.4 (1.2-1.6) | 1.4 (1.2-1.6) | 1.6 (1.4-1.8) | 1.6 (1.4-1.8) | 1.6 (1.4-1.8) |
| Person-years following hospitalization (×103) | 12.2 | 5.3 | 5.3 | 2.1 | 2.6 | 18.4 |
| ESRD cases | 132 | 65 | 67 | 36 | 37 | 64 |
| IR per 1,000 person-years | 10.9 (9.2-12.9) | 12.5 (9.8-15.9) | 12.6 (9.9-16.0) | 16.9 (12.2-23.5) | 14.1 (10.2-19.5) | 3.5 (2.8-4.5) |
| Hazard ratio (95%CI)* | ||||||
| Model 1 | 8.26 (6.45-10.58) | 6.33 (4.76-8.42) | 6.59 (4.95-8.79) | 7.46 (5.20-10.70) | 6.18 (4.35-8.79) | 1.70 (1.27-2.26) |
| Model 2 | 5.22 (4.03-6.75) | 3.92 (2.92-5.28) | 3.90 (2.90-5.24) | 4.18 (2.89-6.05) | 3.72 (2.60-5.32) | 1.24 (0.93-1.66) |
| Model 3 | 3.34 (2.56-4.37) | 2.57 (1.91-3.47) | 2.55 (1.89-3.44) | 2.63 (1.81-3.80) | 2.47 (1.72-3.55) | 0.96 (0.71-1.28) |
Model 1 adjusted for age, sex, and race. Model 2 additionally adjusted for body mass index, ever smoking, ever drink, years of education, systolic blood pressure, antihypertensive drugs, diabetes, eGFR (continuous), ACR (continuous), and history of cardiovascular disease, COPD and cancer. Covariates were updated when available. Model 3 additionally adjusted for incident cardiovascular disease. ESRD indicates end-stage renal disease. IR indicates incidence rate. HR indicates hazard ratio. CVD indicates cardiovascular disease.
Reference, person-years in which there was no hospitalization with major infection.
When assessing separately for the risk of ESRD associated with pneumonia (1,278 events), urinary tract infection (1,341 events), bloodstream infection (685 events), and cellulitis and osteomyelitis (559 events), the results were consistent and significant with the similar HRs across types of infection (HRs in Model 3, 2.57 [95%CI, 1.91-3.47] for pneumonia; 2.55 [1.89-3.44] for urinary tract infection; 2.63 [1.81-3.80] for bloodstream infection, and 2.47 [1.72-3.55] for cellulitis/osteomyelitis) (Table 2).
When assessing non-infection related hospitalization (i.e., hospitalization with osteoarthritis), the HR was not significant in either Model 2 (HR, 1.24 [95%CI, 0.93-1.66]) or Model 3 (0.96 [0.71-1.28]).
When splitting the follow-up period to <5, 5 to <10, and ≥10 years following hospitalization with major infection, the association was strongest in the first 5 years, but remained significant thereafter with similar HRs between 5 to <10 years and ≥10 years (e.g., HRs in Model 3, 5.73 [2.22-14.80] for <5 years, 2.96 [1.74-5.04] for 5 to <10 years, and 2.80 [2.01-3.89] for >10 years) (Table 3). When analyzing the association of major infections that occurred in the first five years of follow-up (547 events) with the risk of ESRD, the association was attenuated but remained strong and significant (HR, 4.19 [95%CI, 3.00-5.86] for Model 1; 2.66 [1.89-3.75] for Model 2; and 1.76 [1.24-2.49] for Model 3). When multiple events of hospitalization with infection per person were taken into account, the HRs were higher following 3 or more hospitalizations with major infection compared to the person-time following 1 or 2 hospitalizations with major infection (e.g., HRs in Model 3, 3.19 [2.42-4.21] vs. 4.54 [2.87-7.19]) (Table 4).
Table 3:
The hazard ratios for incident ESRD comparing the risk during person-years following hospitalization with major infection to the risk during person-years while free of hospitalization by years following hospitalization with major infection: ARIC 1996-2015
|
|
||||
|---|---|---|---|---|
| Hazard ratio (95%CI)* |
Full cohort | Years following hospitalization with major infection | ||
| <5 years | 5 to <10 years | ≥10 years | ||
| Model 1 | 8.26 (6.45-10.58) | 10.35 (4.53-23.63) | 6.54 (4.12-10.37) | 6.49 (4.81-8.76) |
| Model 2 | 5.22 (4.03-6.75) | 6.25 (2.50-15.64) | 4.69 (2.82-7.80) | 4.13 (3.02-5.65) |
| Model 3 | 3.34 (2.56-4.37) | 5.73 (2.22-14.80) | 2.96 (1.74-5.04) | 2.80 (2.01-3.89) |
Model 1 adjusted for age, sex, and race. Model 2 additionally adjusted for body mass index, ever smoking, ever drink, years of education, systolic blood pressure, antihypertensive drugs, diabetes, eGFR (continuous), ACR (continuous), and history of cardiovascular disease, COPD and cancer. Covariates were updated when available. Model 3 additionally adjusted for incident cardiovascular disease.
Reference, person-years in which there was no hospitalization with major infection.
Table 4:
The hazard ratios for incident ESRD comparing the risk during person-years following hospitalization with major infection to the risk during person-years while free of hospitalization by number of hospitalization with major infection: ARIC 1996-2015
|
|
|||
|---|---|---|---|
| Hazard ratio (95%CI) |
Number of hospitalization with major infection | ||
| 0 | 1 or 2 | 3 or more | |
| Model 1 | 1 [Reference] | 7.22 (5.56-9.38) | 21.38 (13.98-32.70) |
| Model 2 | 1 [Reference] | 4.88 (3.73-6.38) | 8.49 (5.39-13.36) |
| Model 3 | 1 [Reference] | 3.19 (2.42-4.21) | 4.54 (2.87-7.19) |
Model 1 adjusted for age, sex, and race. Model 2 additionally adjusted for body mass index, ever smoking, ever drink, years of education, systolic blood pressure, antihypertensive drugs, diabetes, eGFR (continuous), ACR (continuous), and history of cardiovascular disease, COPD and cancer. Covariates were updated when available. Model 3 additionally adjusted for incident cardiovascular disease.
In subgroup analyses, the HRs were significant across subgroups of age, sex, race, diabetes, history of cardiovascular disease, and CKD (Figure 1). When interaction was assessed, the association was stronger in white than black participants (HR, 4.76 [95%CI, 3.37-6.71] vs. 2.22 [1.43-3.46]; p-for-interaction, 0.003), and slightly stronger in participants without CKD than those with CKD (HR, 4.23 [95%CI, 2.86-6.26] vs. 2.45 [1.69-3.55]; p-for-interaction, 0.04).
Figure 1: Subgroup analysis for the association of major infection with incident ESRD by age, sex, race, diabetes, history of cardiovascular disease, and CKD: ARIC 1996-2015.
Model was adjusted for age, sex, race, body mass index, ever smoking, ever drink, years of education, systolic blood pressure, antihypertensive drugs, diabetes, eGFR (continuous), ACR (continuous), history of cardiovascular disease, COPD and cancer, and incidence of cardiovascular disease as a time-varying covariate. Covariates were updated when available. ESRD indicates end-stage renal disease. HR indicates hazard ratio. CVD indicates cardiovascular disease. CKD indicates chronic kidney disease.
Among hospitalizations with major infection, a concurrent diagnosis of acute kidney injury was observed in 503 events (17%) (Table 5). When assessing by a concurrent diagnosis of acute kidney injury, the risk of ESRD was significant in both those with and without acute kidney injury, although the HR was higher among those with acute kidney injury than in those without (HR, 4.39 [95%CI, 3.06-6.30] vs. 2.03 [1.54-2.50] in Model 3). When assessing separately for type-specific infections, the HRs were significant regardless of acute kidney injury across types of infection (e.g., HR, 5.00 [3.14-7.95] in pneumonia with acute kidney injury vs. 1.77 [1.26-2.50] in pneumonia without acute kidney injury).
Table 5:
The hazard ratios of end-stage renal disease for hospitalization with infection with and without a concurrent diagnosis of acute kidney injury: ARIC 1996-2015
|
|
|||||
|---|---|---|---|---|---|
| Outcomes | Infection-related hospitalization |
||||
| All major infection (2,642 events) |
Pneumonia (1,278 events) |
Urinary tract infection (1,341 events) |
Bloodstream infection (685 events) |
Cellulitis and osteomyelitis (559 events) |
|
| Infection with acute kidney injury | 503 events | 279 events | 289 events | 323 events | 94 events |
| Hazard ratio (95%CI)* | |||||
| Model 1 | 19.29 (13.59-27.37) | 20.96 (13.33-32.94) | 20.23 (13.33-30.72) | 12.15 (7.12-20.74) | 16.88 (7.91-36.04) |
| Model 2 | 6.81 (4.73-9.81) | 8.04 (5.03-12.86) | 6.34 (4.09-9.82) | 6.91 (4.04-11.84) | 7.01 (3.24-15.13) |
| Model 3 | 4.39 (3.06-6.30) | 5.00 (3.14-7.95) | 3.84 (2.48-5.95) | 3.86 (2.25-6.61) | 5.21 (2.41-11.27) |
| Infection without acute kidney injury | 2,139 events | 999 events | 1,052 events | 362 events | 465 events |
| Hazard ratio (95%CI)* | |||||
| Model 1 | 4.74 (3.65-6.15) | 4.06 (2.91-5.66) | 3.97 (2.82-5.59) | 5.35 (3.40-8.41) | 5.27 (3.59-7.74) |
| Model 2 | 3.17 (2.43-4.15) | 2.66 (1.89-3.74) | 2.66 (1.88-3.76) | 3.01 (1.90-4.78) | 3.25 (2.20-4.80) |
| Model 3 | 2.03 (1.54-2.66) | 1.77 (1.26-2.50) | 1.81 (1.28-2.56) | 2.01 (1.26-3.19) | 2.14 (1.44-3.17) |
Model 1 adjusted for age, sex, and race. Model 2 additionally adjusted for body mass index, ever smoking, ever drink, years of education, systolic blood pressure, antihypertensive drugs, diabetes, eGFR (continuous), ACR (continuous), and history of cardiovascular disease, COPD and cancer. Covariates were updated when available. Model 3 additionally adjusted for incident cardiovascular disease. ESRD indicates end-stage renal disease.
Reference, person-years in which there was no hospitalization with major infection.
Cumulative incidence of ESRD following major infection
Figure 2 shows Kaplan Meier curves depicting the cumulative incidence of ESRD among participants with and without hospitalization with major infection. Of note, because this was the matched cohort analysis using the incidence density sampling, the analytic sample was restricted to 7,901 for major infection, 3,820 for pneumonia, 4,014 for urinary tract infection, 2,052 for bloodstream infection, 1,671 for cellulitis/osteomyelitis, and 6,523 for osteoarthritis. In general, the risk separation was consistently observed throughout the time following hospitalization with infection across pneumonia, urinary tract infection, bloodstream infection, and cellulitis/osteomyelitis (Figure 2A–2E). When assessing non-infection related hospitalization, the cumulative incidence of ESRD was largely comparable between participants with and without hospitalization with osteoarthritis (Figure 2F).
Figure 2: Kaplan Meier curves depicting the cumulative incidence of ESRD in matched cohort analysis using the incidence density sampling: ARIC 1996-2015.
The analytic sample was restricted to (A) 7,901 for major infection, (B) 3,820 for pneumonia, (C) 4,014 for urinary tract infection, (D) 2,052 for bloodstream infection, (E) 1,671 for cellulitis/osteomyelitis, and (F) 6,523 for osteoarthritis. Numbers are not equal to three times the number of events due to the inability to find matched controls. ESRD indicates end-stage renal disease.
Discussion
In this cohort of middle-aged to older black and white adults residing in the community, hospitalization with major infection was independently and robustly associated with a higher risk of ESRD, with the similar HRs across pneumonia, urinary tract infection, bloodstream infection, and cellulitis/osteomyelitis. The association was consistent in demographic and clinical subgroups, robust to sensitivity analyses, and regardless of whether concurrent acute kidney injury during hospitalization with infection was documented.
Our findings are consistent with few previous reports investigating CKD outcomes following infection among an Asian population, military conscripted men, and advanced CKD patients.12–14 We extend the literature to a broad spectrum of major infection types in a racially diverse cohort of middle to older adults; the age range at greatest risk for ESRD.1 We were also able to adjust for major ESRD risk factors including eGFR and ACR,19 and account for incident cardiovascular disease during follow-up, a potential mediator for the risk of ESRD.26,27 Furthermore, we confirmed the consistent association across demographic and clinical subgroups. Finally, as a ‘negative control’ comparator, we showed a null result for the association for hospitalization with osteoarthritis, a major cause of non-infection related hospitalization.28
Although our data alone cannot establish causality, there are several plausible biological mechanisms through which infection may increase the risk of ESRD. In animal models, bacterial and viral infections triggered the accumulation of the extracellular matrix through the activation of fibroblasts and macrophages, leading to kidney fibrosis.30,31 Septic rats challenged by cecal ligation and perforation showed impaired microvascular perfusion,32 a major cause of ischemia-reperfusion injury.4,5 In addition, inflammation also activated the expression of monocyte-derived macrophages33 and endothelial adhesion molecules,34 which may contribute to unstable atherosclerosis.6 Importantly, these pathways are orchestrated by a series of inflammatory response irrespective of a primary etiology, and often persist for an extended period of time.35 Thus, our findings where the risk of ESRD was increased regardless of infection type may corroborate these mechanisms.
Increased risk of ESRD following infection may also be explained by various medical conditions or interventions associated with infection. For example, the use of antibiotics can lead to the injury of renal tubules and interstitium.9 In addition, radiocontrast agents are often used to diagnose the source of infection, which may cause radiocontrast-induced nephropathy.10 Furthermore, infectious diseases can induce myocardial dysfunction leading to tissue hypoxia.7 Finally, fluid resuscitation is often required to maintain tissue perfusion, but excess fluid resuscitation may result in renal congestion and decreased renal blood flow.11
In the present study, acute kidney injury was diagnosed in 17% of hospitalizations with major infection, and the risk of ESRD following major infection with acute kidney injury was 4.4-fold higher compared to the risk while free of infection. A previous meta-analysis reported that hospitalized patients with acute kidney injury had a greater risk of ESRD.36 Our findings are consistent, but uniquely extend the literature by analyzing across major types of infection, using time-varying exposure analysis, and having an extended period of follow-up. We also found that the association was significant even without a concurrent diagnosis of acute kidney injury, with the risk being 2-fold higher. Thus, hospitalization with major infection may increase the risk of ESRD regardless of a diagnosis of acute kidney injury, although some missed diagnoses should be taken into account due to the low sensitivity of ICD-9CM codes for acute kidney injury.37
The present study has several important clinical implications. Infection should be recognized as an important risk factor for ESRD. In primary care settings, it seems important to collect information on the history of major infection along with other risk factors of ESRD such as diabetes and hypertension. Although infection is recognized as an important cause of acute kidney injury, our study demonstrated the long-term adverse kidney outcomes following infection. Thus, close monitoring for kidney function following major infection is important in both those with and without a concurrent diagnosis of acute kidney injury. Finally, infection prevention may be a potential target for reducing the burden of ESRD. For example, influenza and pneumococcal vaccinations are representative prevention approaches, and already recommended in the guidelines for many adults for both kidney disease prevention and general health.38–40 However, we recently reported the suboptimal prevalence of vaccination among both those with and without CKD.41,42 Thus, increasing uptake of vaccinations is critical for preventing relevant infections but may also simultaneously help reduce the risk of ESRD. Nonetheless, future studies should explore whether infection prevention measures can contribute to better kidney outcomes.
The present study has several limitations. First, hospitalizations with major infections were ascertained through ICD-9CM codes. However, we focused on acute infections, for which diagnosis is often based on acute symptoms, and thus likely less susceptible to misclassification.43,44 Second, we did not include non-hospitalized infection. Thus, future studies are needed to explore whether infections in the outpatient setting are associated with the risk of ESRD. However, those studies need to be done carefully since the reliability of outpatient diagnosis of infection is uncertain. In this context, our focus on hospitalization with infection is still valuable since it is the most relevant consideration from a public health perspective given its impact on morbidity and mortality.45 Third, during a hospitalization, patients may experience various clinical conditions and interventions (e.g., hemodynamic instability, multiple nephrotoxic medication exposures, volume shifts) that can also contribute to the increased risk of ESRD. However, many of these factors can be consequences of infection and thus do not necessarily limit our study. Nonetheless, future investigations are required to explore mediators linking infection to ESRD since such an investigation will have implications on the prevention of ESRD. Fourth, the possibility of residual confounding may not be excluded. However, we extensively accounted for known major ESRD risk factors, and confirmed the consistent associations across key demographic and clinical subgroups and in several sensitivity analyses. Finally, our study population consisted of a cohort of middle to older age black and white adults, and therefore, the generalizability of our findings to other races and age groups should be interpreted with caution.
Conclusion
In conclusion, in this community-based cohort of middle to older age black and white adults, hospitalization with major infection was independently associated with the risk of ESRD. These findings suggest the adverse consequences of infectious disease on kidney function. Individuals following an episode of infection should be regarded as at high risk for ESRD. Future studies should investigate whether infection prevention measures help reduce the burden of ESRD by reducing the risk of preventable infections.
Acknowledgements
The Atherosclerosis Risk in Communities study has been funded in whole or in part with Federal funds from the National Heart, Lung, and Blood Institute, National Institutes of Health, Department of Health and Human Services, under Contract nos. (HHSN268201700001I, HHSN268201700002I, HHSN268201700003I, HHSN268201700005I, HHSN268201700004I).
The authors thank the staff and participants of the ARIC study for their important contributions.
Abbreviations:
- ACR
urinary albumin-to-creatinine ratio
- ARIC
Atherosclerosis Risk in Communities
- CKD
chronic kidney disease
- COPD
chronic obstructive pulmonary disease
- eGFR
estimated glomerular filtration rate
- ESRD
end-stage renal disease
References
- 1.Saran R, Robinson B, Abbott KC, et al. US Renal Data System 2016 Annual Data Report: Epidemiology of Kidney Disease in the United States. American journal of kidney diseases : the official journal of the National Kidney Foundation. 2017;69(3S1):A7–A8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Levin A. Identification of patients and risk factors in chronic kidney disease--evaluating risk factors and therapeutic strategies. Nephrology, dialysis, transplantation : official publication of the European Dialysis and Transplant Association - European Renal Association. 2001;16 Suppl 7:57–60. [DOI] [PubMed] [Google Scholar]
- 3.Lee SB, Kalluri R. Mechanistic connection between inflammation and fibrosis. Kidney Int Suppl. 2010(119):S22–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Kinsey GR, Li L, Okusa MD. Inflammation in acute kidney injury. Nephron Experimental nephrology. 2008;109(4):e102–107. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Akcay A, Nguyen Q, Edelstein CL. Mediators of inflammation in acute kidney injury. Mediators of inflammation. 2009;2009:137072. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Hansson GK, Robertson AK, Soderberg-Naucler C. Inflammation and atherosclerosis. Annual review of pathology. 2006;1:297–329. [DOI] [PubMed] [Google Scholar]
- 7.Kakihana Y, Ito T, Nakahara M, Yamaguchi K, Yasuda T. Sepsis-induced myocardial dysfunction: pathophysiology and management. J Intensive Care. 2016;4:22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Zanotti Cavazzoni SL, Dellinger RP. Hemodynamic optimization of sepsis-induced tissue hypoperfusion. Crit Care. 2006;10 Suppl 3:S2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Markowitz GS, Perazella MA. Drug-induced renal failure: a focus on tubulointerstitial disease. Clinica chimica acta; international journal of clinical chemistry. 2005;351(1-2):31–47. [DOI] [PubMed] [Google Scholar]
- 10.Kidney Disease: Improving Global Outcomes (KDIGO) AKI work Group. KDIGO Clinical Practice Guideline for Acute Kidney Injury. Kidney Int Suppl. 2012;2(1):1–138. [Google Scholar]
- 11.J FG, von Haehling S, Anker SD, Raj DS, Radhakrishnan J. The relevance of congestion in the cardio-renal syndrome. Kidney international. 2013;83(3):384–391. [DOI] [PubMed] [Google Scholar]
- 12.Sundin PO, Udumyan R, Fall K, Montgomery S. Hospital admission with pneumonia and subsequent persistent risk of chronic kidney disease: national cohort study. Clinical epidemiology. 2018;10:971–979. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Cheikh Hassan HI, Tang M, Djurdjev O, Langsford D, Sood MM, Levin A. Infection in advanced chronic kidney disease leads to increased risk of cardiovascular events, end-stage kidney disease and mortality. Kidney international. 2016;90(4):897–904. [DOI] [PubMed] [Google Scholar]
- 14.Huang ST, Lin CL, Chang YJ, et al. Pneumococcal pneumonia infection is associated with end-stage renal disease in adult hospitalized patients. Kidney international. 2014;86(5):1023–1030. [DOI] [PubMed] [Google Scholar]
- 15.The Atherosclerosis Risk in Communities (ARIC) Study: design and objectives. The ARIC investigators. American journal of epidemiology. 1989;129(4):687–702. [PubMed] [Google Scholar]
- 16.Ishigami J, Grams ME, Chang AR, Carrero JJ, Coresh J, Matsushita K. CKD and Risk for Hospitalization With Infection: The Atherosclerosis Risk in Communities (ARIC) Study. American journal of kidney diseases : the official journal of the National Kidney Foundation. 2017;69(6):752–761. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Xu H, Gasparini A, Ishigami J, et al. eGFR and the Risk of Community-Acquired Infections. Clinical journal of the American Society of Nephrology : CJASN. 2017;12(9):1399–1408. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Ishigami J, Matsushita K. Clinical epidemiology of infectious disease among patients with chronic kidney disease. Clinical and experimental nephrology. 2019;23(4):437–447. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Coresh J, Turin TC, Matsushita K, et al. Decline in estimated glomerular filtration rate and subsequent risk of end-stage renal disease and mortality. JAMA : the journal of the American Medical Association. 2014;311(24):2518–2531. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Liangpunsakul S, Chalasani N. Relationship between hepatitis C and microalbuminuria: results from the NHANES III. Kidney international. 2005;67(1):285–290. [DOI] [PubMed] [Google Scholar]
- 21.Atta MG, Estrella MM, Kuperman M, et al. HIV-associated nephropathy patients with and without apolipoprotein L1 gene variants have similar clinical and pathological characteristics. Kidney international. 2012;82(3):338–343. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Christensen KL, Holman RC, Steiner CA, Sejvar JJ, Stoll BJ, Schonberger LB. Infectious disease hospitalizations in the United States. Clinical infectious diseases : an official publication of the Infectious Diseases Society of America. 2009;49(7):1025–1035. [DOI] [PubMed] [Google Scholar]
- 23.Inker LA, Schmid CH, Tighiouart H, et al. Estimating glomerular filtration rate from serum creatinine and cystatin C. The New England journal of medicine. 2012;367(1):20–29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Cheng S, Claggett B, Correia AW, et al. Temporal trends in the population attributable risk for cardiovascular disease: the Atherosclerosis Risk in Communities Study. Circulation. 2014;130(10):820–828. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Cowan LT, Alonso A, Pankow JS, et al. Hospitalized Infection as a Trigger for Acute Ischemic Stroke: The Atherosclerosis Risk in Communities Study. Stroke; a journal of cerebral circulation. 2016;47(6):1612–1617. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Braam B, Joles JA, Danishwar AH, Gaillard CA. Cardiorenal syndrome--current understanding and future perspectives. Nature reviews Nephrology. 2014;10(1):48–55. [DOI] [PubMed] [Google Scholar]
- 27.Ishigami J, Cowan LT, Demmer RT, et al. Incident Hospitalization with Major Cardiovascular Diseases and Subsequent Risk of ESKD: Implications for Cardiorenal Syndrome. Journal of the American Society of Nephrology : JASN. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Healthcare Cost and Utilization Project. Trends in Hospital Inpatient Stays in the United States, 2005–2014. Available: https://www.hcup-us.ahrq.gov/reports/statbriefs/sb225-Inpatient-US-Stays-Trends.pdf. Accessed 21 Jan 2020.
- 29.Langholz B, Thomas DC. Nested case-control and case-cohort methods of sampling from a cohort: a critical comparison. American journal of epidemiology. 1990;131(1):169–176. [DOI] [PubMed] [Google Scholar]
- 30.Ebrahimi B, Dutia BM, Brownstein DG, Nash AA. Murine gammaherpesvirus-68 infection causes multi-organ fibrosis and alters leukocyte trafficking in interferon-gamma receptor knockout mice. The American journal of pathology. 2001;158(6):2117–2125. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Cosma CL, Sherman DR, Ramakrishnan L. The secret lives of the pathogenic mycobacteria. Annual review of microbiology. 2003;57:641–676. [DOI] [PubMed] [Google Scholar]
- 32.Lam C, Tyml K, Martin C, Sibbald W. Microvascular perfusion is impaired in a rat model of normotensive sepsis. The Journal of clinical investigation. 1994;94(5):2077–2083. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Smith JD, Trogan E, Ginsberg M, Grigaux C, Tian J, Miyata M. Decreased atherosclerosis in mice deficient in both macrophage colony-stimulating factor (op) and apolipoprotein E. Proceedings of the National Academy of Sciences of the United States of America. 1995;92(18):8264–8268. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Cybulsky MI, Iiyama K, Li H, et al. A major role for VCAM-1, but not ICAM-1, in early atherosclerosis. The Journal of clinical investigation. 2001;107(10):1255–1262. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Scrivo R, Vasile M, Bartosiewicz I, Valesini G. Inflammation as “common soil” of the multifactorial diseases. Autoimmunity reviews. 2011;10(7):369–374. [DOI] [PubMed] [Google Scholar]
- 36.Coca SG, Singanamala S, Parikh CR. Chronic kidney disease after acute kidney injury: a systematic review and meta-analysis. Kidney international. 2012;81(5):442–448. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Grams ME, Waikar SS, Macmahon B, Whelton S, Ballew SH, Coresh J. Performance and Limitations of Administrative Data in the Identification of AKI. Clinical journal of the American Society of Nephrology : CJASN. 2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Kidney Disease: Improving Global Outcomes (KDIGO) CKD Work Group. KDIGO 2012 Clinical Practice Guideline for the Evaluation and Management of Chronic Kidney Disease. Kidney inter, Suppl. 2013;3:1–150. [Google Scholar]
- 39.Kim DK, Hunter P. Advisory Committee on Immunization Practices Recommended Immunization Schedule for Adults Aged 19 Years or Older - United States, 2019. MMWR Morbidity and mortality weekly report. 2019;68(5):115–118. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Ishigami J, Sang Y, Grams ME, Coresh J, Chang A, Matsushita K. Effectiveness of Influenza Vaccination Among Older Adults Across Kidney Function: Pooled Analysis of 2005-2006 Through 2014-2015 Influenza Seasons. American journal of kidney diseases : the official journal of the National Kidney Foundation. 2019. [DOI] [PubMed] [Google Scholar]
- 41.Ishigami J, Padula WV, Grams ME, et al. Cost-effectiveness of Pneumococcal Vaccination Among Patients With CKD in the United States. American journal of kidney diseases : the official journal of the National Kidney Foundation. 2019;74(1):23–35. [DOI] [PubMed] [Google Scholar]
- 42.Ishigami J, Sang Y, Grams ME, Coresh J, Chang A, Matsushita K. Effectiveness of Influenza Vaccination among Older Adults across Kidney Function: Pooled Analysis of 2005-2006 through 2014-2015 Influenza Seasons (in press). American journal of kidney diseases : the official journal of the National Kidney Foundation. 2019. [DOI] [PubMed] [Google Scholar]
- 43.Barber C, Lacaille D, Fortin PR. Systematic review of validation studies of the use of administrative data to identify serious infections. Arthritis care & research. 2013;65(8):1343–1357. [DOI] [PubMed] [Google Scholar]
- 44.Drahos J, Vanwormer JJ, Greenlee RT, Landgren O, Koshiol J. Accuracy of ICD-9-CM codes in identifying infections of pneumonia and herpes simplex virus in administrative data. Annals of epidemiology. 2013;23(5):291–293. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Hall MJ, De Frances CJ, Williams SN, Golosinskiy A, Schwartzman A. National Hospital Discharge Survey: 2007 summary. National health statistics reports; no29. Hyattsville, MD: National Center for Health Statistics. Available: http://www.cdc.gov/nchs/data/nhsr/nhsr029.pdf. Accessed 21 Jan 2020. [Google Scholar]