Introduction
In the United States, >110,000 individuals with ESRD initiate maintenance dialysis each year (1 2 , 3 4 , 5
One of the persistent barriers to greater adoption of home HD is the need for patients or caregivers to perform frequent cannulation of arteriovenous (AV) access (fistula or graft) (6 – 8 9 central venous catheters (CVCs) as long–term vascular access for patients treated with home HD. In patients undergoing in-center HD, use of CVCs is associated with greater risk of adverse clinical outcomes, including higher rates of mortality and hospitalization , at least in part because of greater incidence of infection-related complications (10 – 14 vascular access type with clinical outcomes for patients undergoing home HD, individuals in whom incidence of nosocomial infection may be lower (15
Using nationally representative data from a large dialysis provider in the United States, we undertook this study to examine the null hypothesis that, in patients undergoing home HD, use of a CVC compared with use of an AV access is not associated with higher risk of mortality, time to first hospitalization , or transfer to in-center HD. In addition, we sought to juxtapose our findings for patients undergoing home HD with those for patients undergoing conventional three times per week HD.
Materials and Methods
Data Source
The study population comprised patients ≥18 years of age who started maintenance dialysis from 2007 to 2011 and received care at one of the facilities operated by DaVita, Inc. for at least 60 days (16 n =323) in the primary cohort, the data from patients with an AV graft or fistula were pooled into a single group for the primary analysis. Characteristics of patients included in the cohorts compared with those excluded because of missing vascular access type are reported in Supplemental Tables 1 and 2.
Data from dialysis facility electronic medical records were used to determine demographics and comorbidities. Dialysis facility experience with home HD was defined as the number of 91-day patient periods in which patients on home HD were treated at that facility. Similarly, facility home HD CVC experience was defined as the number of 91-day periods in which patients on home HD were treated with a CVC at that facility. Laboratory measurements were averaged for each patient for each 91-day period. All laboratory values were measured in the central DaVita, Inc. laboratory (Deland, FL). The Institutional Review Boards at the Los Angeles Biomedical Research Institute and the University of Washington approved the study as exempt from informed consent.
Statistical Analyses
Data were complete for age, sex, diabetes, cause of ESRD, primary health insurance, and cardiovascular comorbidities. Data for race; dialysis facility region; serum calcium, potassium, and bicarbonate; and erythropoietin dose were missing for <1% of the cohort. Data for serum albumin, parathyroid hormone, phosphorus, alkaline phosphatase and creatinine, blood hemoglobin, total iron–binding capacity, ferritin, and white blood cell count were missing for 1%–7% of the cohort. Missing covariate data were imputed using multiple imputation with five repetitions (17 Vascular access type was complete for 95% of patients. Missing data for vascular access type were not imputed.
For the primary analysis, a propensity score–matched cohort was constructed to minimize the influence of bias caused by confounding by indication. A logistic regression model was built with CVC as initial access type as the outcome and the following variables as predictors: age, sex, race, diabetes status, primary health insurance, cause of ESRD, cardiovascular comorbidity, year of start of maintenance dialysis, time from start of dialysis to start of home HD, facility geographic region, body mass index, serum albumin, calcium, potassium, creatinine, bicarbonate, alkaline phosphatase, parathyroid hormone, blood hemoglobin, ferritin, transferrin saturation, white blood cell count, percentage of lymphocytes, cumulative iron dose, and weekly dose of erythropoietin. This model was used to calculate the probability of each patient being treated with a CVC at the time of start of home HD (propensity scores). Propensity scores were used to identify one patient initiating home HD with an AV access for each patient with a CVC using a greedy matching algorithm with a caliper width of 0.1 SDs (18
Unadjusted time to event competing risks survival analyses (19 hospitalization , and transfer to in-center HD. For each analysis, the referent group comprised patients treated with an AV access. Censoring reasons included transplant, discharge to a facility operated by another dialysis provider, discontinuation of dialysis or recovery of renal function, and end of follow-up (Supplemental Table 4). For each outcome, the presence of effect modification by facility experience with incident home HD overall and use of CVC for home HD was tested through assessment of the significance of the first–order interaction term.
To further examine the robustness of our findings and incorporate information on changes in vascular access and time-varying confounders during follow-up, we performed competing risk regression analyses within the entire unmatched cohort (n =2481), modeling vascular access as a time-varying exposure that was updated at the start of each 91-day period of follow-up. Three nested hierarchical models were examined: (1 ) unadjusted; (2 ) adjusted for age, sex, race/ethnicity, and diabetes status; and (3 ) additionally adjusted for duration of dialysis treatment before start of home HD, dialysis facility home HD experience, body mass index, time–varying serum albumin and creatinine, and time–varying blood hemoglobin.
Two sets of sensitivity analyses were performed. In the first set of analyses, vascular access type at the start of home HD was modeled as a static baseline exposure in the survival analyses. In the second set, AV fistula and AV graft were assessed as distinct exposures rather than pooled together into a single AV access group. As an exploratory analysis, we performed competing risk regression to examine the association of vascular access type with first bacteremia, defined as first incidence of a positive outpatient blood culture. Additionally, we assessed for evidence of effect modification by duration of CVC use during home HD on the association between initial vascular access type and outcomes through assessment of significance of the first–order interaction term.
Finally, for comparison and to enhance the external validity of our findings, we performed a propensity score–matched analysis of patients only treated with conventional three times per week in–center HD (22,431 pairs of patients). Time to event competing risk survival analyses were performed to determine the associations of CVC use at the start of in-center HD with all-cause mortality and first hospitalization .
Statistical analyses were performed using SAS, version 9.3 (SAS Institute Inc., Cary, NC).
Results
Home HD Study Population
Between January 1, 2007 and December 31, 2011, 2543 patients started home HD in 464 facilities in 43 states. Of these, 2481 patients had at least one 91-day period in which there was available information on vascular access type. For the unmatched cohort (2481 patients), compared with patients initiating home HD with an AV access, patients with a CVC were older, had a shorter interval from date of first-ever dialysis to home HD, were more likely to be black, were more likely have diabetes or cardiovascular comorbidity, and had lower baseline serum albumin and creatinine (Table 1 ). The propensity score–matched cohort comprised 526 pairs of individuals with CVC or AV access at time of start of home HD. In contrast to the unmatched cohort, there were no meaningful differences between the groups with CVC or AV access in any of the measured baseline demographic characteristics or laboratory variables.
Table 1: Characteristics of unmatched (n =2481) and propensity score–matched (n =1052) study cohorts stratified by initial vascular access type at the time of initiation of home hemodialysis
The median duration of maintenance dialysis treatment before start of home HD was 157 days (interquartile range [IQR] =45–369) for patients with a CVC and 308 days (IQR=124–583) for patients with AV access (Table 1 ). Median treatment time per session and mean number of treatments per week were 168 (IQR=150–191) minutes and 4.1±1.4 treatments for patients with a CVC and 161 (IQR=146–180) minutes and 4.2±1.3 treatments for patients with AV access, respectively. Among all study patients, 72% started home HD with an AV access; the prevalence of AV access increased to 86% by end of year 1 and remained constant over the remainder of the follow-up period (Figure 1 ); <10% of patients who initiated home HD with an AV access switched to use of a CVC (Figure 2 ). In contrast, for patients initiating home HD with CVC, >50% after 1 year had switched to use of an AV access (Figure 2 ).
Figure 1: Vascular access over study follow-up amongpatients undergoing home hemodialysis (HD; n=2481).
Figure 2: Vascular access over study follow-up among patients undergoing home hemodialysis (HD) stratified by initial vascular access type.vascular access as central venous catheter (CVC; n =579). (B) Patients with initial vascular access as arteriovenous (AV) access (n =1794).
Overall, 174 patients with known initial vascular access type died, 1199 were hospitalized at least one time, and 386 patients transferred to in-center HD over 2549 person-years of follow-up. Within the propensity score–matched cohort, the crude mortality rate for patients with CVC was 12.1 (95% confidence interval [95% CI], 9.5 to 62.9) per 100 person-years compared with 6.9 (95% CI, 5.1 to 9.2) per 100 person-years for patients with AV access (Table 2 ). The hospitalization and transfer to in–center HD rates were 56.3 (95% CI, 50.4 to 62.9) and 18.1 (95% CI, 14.9 to 22.0) per 100 person-years for patients using a CVC compared with 46.8 (95% CI, 41.7 to 52.6) and 16.3 (95% CI, 13.2 to 19.9) for patients using AV access. The proportions of patients who underwent kidney transplantation, regained kidney function, or transferred to a nonaffiliated dialysis facility were similar, irrespective of initial vascular access type (Supplemental Table 4).
Table 2: Follow-up, number of events, and event rate by initial vascular access type in propensity score–matched (n =1052) and unmatched (n =2481) cohorts
Association of Vascular Access Type with All-Cause Mortality in Home HD
Within the propensity score–matched cohort, patients undergoing home HD with CVCs had greater risk for all-cause mortality compared with patients using an AV access (hazard ratio [HR], 1.73; 95% CI, 1.18 to 2.54) (Figure 3 ). There was no effect modification by dialysis facility home HD experience (P value for interaction =0.81) or facility home HD CVC experience (P value for interaction =0.28).
Figure 3: Association of central venous catheter (CVC) use with risk for all-cause mortality, hospitalization , and transfer to in-center hemodialysis in patients undergoing home hemodialysis . The reference group was arteriovenous (AV) access. The propensity score–matched cohort included 1052 patients (CVC, n =526; AV access, n =526); the unmatched cohort included 2481 patients. Hazard ratios are subdistribution hazard ratios from competing risks regression. The minimally adjusted model was adjusted for age, sex, diabetes, and race and/or ethnicity. The fully adjusted model was further adjusted for duration of dialysis treatment before the start of home hemodialysis , dialysis facility home hemodialysis experience, body mass index, serum albumin, serum creatinine, and blood hemoglobin.
The results of secondary analyses within the entire unmatched cohort were similar to analyses within the propensity score–matched cohort (Figure 3 ). Using an unadjusted model, time–varying CVC use was associated with higher risk for all-cause mortality (HR, 2.36; 95% CI, 1.73 to 3.22). After adjustment for potential confounders, the association with all-cause mortality was substantially attenuated (HR, 1.39; 95% CI, 0.99 to 1.95) (Figure 3 ).
Association of Vascular Access Type with Hospitalization in Home HD
Compared with patients undergoing home HD with an AV access, patients using a CVC had a higher risk for hospitalization (HR, 1.19; 95% CI, 1.02 to 1.39). Analyses in the unmatched cohort were similar to those in the propensity score–matched cohort, regardless of level of adjustment. As for mortality, there was no evidence for effect modification by either dialysis facility home HD experience (P value for interaction =0.61) or facility home HD CVC experience (P value for interaction =0.10).
Association of Vascular Access Type with Transfer to In-Center HD in Home HD
There was no statistically significant association between vascular access type and risk for transfer to in-center HD in the propensity score–matched cohort or the unmatched cohort after full adjustment for potential confounders (Figure 3 ). However, the relationship between CVC use and risk for transfer to in-center HD was significantly modified by the cumulative dialysis facility CVC experience (P value for interaction <0.01), with a trend toward stronger risk for patients undergoing dialysis in facilities with lower home HD CVC experience (Supplemental Figure 3, Supplemental Table 5). There was no evidence of effect modification by facility home HD experience (P value for interaction =0.28).
Association of Vascular Access Type with All-Cause Mortality and Hospitalization in Conventional In–Center HD
The propensity score–matched cohort of patients undergoing in-center HD comprised 22,431 pairs of individuals with CVC or AV access at time of start of dialysis. There were no meaningful differences between the groups with CVC or AV access in any of the measured baseline demographic characteristics or laboratory variables (Supplemental Tables 6 and 7). Individuals with a CVC had a significantly higher risk for death (HR, 1.30; 95% CI, 1.25 to 1.35) and hospitalization (HR, 1.37; 95% CI, 1.34 to 1.40) compared with individuals with an AV access (Figure 4 ).
Figure 4: Association of central venous catheter (CVC) use with risk for all-cause mortality and hospitalization among a propensity score–matched cohort of patients undergoing conventional three times per week in–center hemodialysis. The reference group was arteriovenous (AV) access. Results are presented for the propensity score–matched cohort of 44,862 patients (CVC, n =22,431; AV access, n =22,431). Hazard ratios are subdistribution hazard ratios from competing risks regression.
Sensitivity and Exploratory Analyses
In sensitivity analyses modeling vascular access type at the start of home HD as the primary exposure within the entire unmatched cohort, results were similar to those from the primary analyses (Supplemental Table 8). In a second set of sensitivity analyses assessing AV fistula or AV graft as distinct exposures, patients dialyzing with a graft did not have a higher risk for all-cause mortality, hospitalization , or transfer to in-center HD compared with those using a fistula, regardless of the level of covariate adjustment. Results of an exploratory analysis examining the association of initial vascular access type with first bacteremia events showed higher risk for bacteremia with use of CVC within the propensity score–matched cohort (HR, 2.59; 95% CI, 1.60 to 4.20) as well as within the unmatched cohort, irrespective of the level of statistical adjustment (HR, 3.22; 95% CI, 2.21 to 4.68 for the fully adjusted model). Finally, there was no evidence of effect modification by duration of CVC use on the association of initial vascular access type with outcomes.
Discussion
In this large study of patients on incident home HD, we found that treatment with a CVC was associated with a higher risk for death and hospitalization but not transfer to in-center HD compared with treatment with an AV access. These findings were observed in an analysis of a propensity score–matched cohort as well as after adjustment for potential confounders in a larger unmatched cohort using baseline or time–varying vascular access . Additionally, the magnitude of higher risk for death and hospitalization observed with use of a CVC in home HD was similar to that observed in a parallel analysis of patients undergoing conventional in–center HD.
Numerous studies over the past two decades have shown associations between CVC use and higher risk for adverse clinical outcomes in patients undergoing in-center HD (10 – 14 , 20 – 22 23 – 25 26 , 27
Although these differences provide a rationale for why CVC risk may be different for in-center HD and home HD, the results of our study show that higher risks for mortality and hospitalization associated with CVC use also exist for patients undergoing home HD and are similar in magnitude (22 hospitalization , associated with CVCs among patients undergoing daily or nocturnal HD compared with those undergoing in-center HD (28 , 29 13 , 30 31
The results of our study support current guidelines that encourage patients and their providers to work diligently to establish and maintain AV access, including when dialyzing at home. We found that that over one half of patients who remained on home HD for at least 1 year and initially started home HD with a CVC switch to use of AV access. This finding emphasizes the importance of continued engagement with patients regarding optimal vascular access , even after the initial transition to maintenance dialysis.
Our study has multiple strengths. First, we used data from patients in >400 facilities across >40 states. This approach enhances external validity of our findings and also, provides substantial power to detect differences in outcomes among patients with differing vascular access . Second, we performed the primary analysis in a propensity score–matched cohort, a statistical technique shown to reduce bias from nonrandom exposure assignment (32 – 34 35 vascular access type as a time-varying exposure or a baseline exposure at the time of start of home HD, an important consideration to ensure accuracy in attribution of risk to vascular access type.
Despite its strengths, our study has several limitations. Although we identified a number of important associations, given the observational nature of our study, there remains the possibility of residual confounding. Additionally, data were not available on cause-specific mortality or hospitalizations, limiting our ability to determine associations between vascular access type and outcomes, such as access-related hospitalizations. We were also not able to account for whether patients used high–frequency, low–dialysate volume equipment or conventional HD machines or whether treatments were administered during the day or at night. Furthermore, the overall number of events limited power to examine whether the associations that we observed vary in patient subgroups.
In conclusion, in a large, nationally representative cohort of patients on incident home HD, use of a CVC compared with an AV access was associated with higher risk for death and hospitalization . Future studies should investigate whether interventions exist that may reduce the risk associated with use of CVCs among patients undergoing home HD or whether there are key subgroups among whom this higher risk does not exist.
Disclosures
A.R.N. is an employee of DaVita, Inc., El Segundo, CA.
Acknowledgments
This work was supported by National Institutes of Health/National Institute of Diabetes and Digestive and Kidney Diseases Grants T32DK007467 (supporting M.B.R.), R21AG047306 (to M.Z.M., K.K.-Z., and R.M.), and R01DK095668 (to K.K.-Z. and R.M.).
Published online ahead of print. Publication date available at www.cjasn.org
See related editorial, “The Burden of Harm—What Is the Ideal Vascular Access for Home Hemodialysis ?,” on pages 205–206.
This article contains supplemental material online at http://cjasn.asnjournals.org/lookup/suppl/doi:10.2215/CJN.06570615/-/DCSupplemental
References
1. United States Renal Data System: USRDS Annual Data Report. An Overview of the Epidemiology of Kidney Disease in the United States, Bethesda, MD, National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases, 2014
2. Grassmann A, Gioberge S, Moeller S, Brown G: ESRD patients in 2004: Global overview of patient numbers, treatment modalities and associated trends. Nephrol Dial Transplant 20: 2587–2593, 2005
3. Rivara MB, Mehrotra R: The changing landscape of home dialysis in the United States. Curr Opin Nephrol Hypertens 23: 586–591, 2014
4. Centers for Medicare & Medicaid Services (CMS), HHS: Medicare program;
end-stage renal disease prospective payment system. Final rule. Fed Regist 75: 49029–49214, 2010
5. Kohn OF, Coe FL, Ing TS: Solute kinetics with short-daily
home hemodialysis using slow dialysate flow rate. Hemodial Int 14: 39–46, 2010
6. Tennankore KK, Chan CT, Curran SP: Intensive home haemodialysis: Benefits and barriers. Nat Rev Nephrol 8: 515–522, 2012
7. Young BA, Chan C, Blagg C, Lockridge R, Golper T, Finkelstein F, Shaffer R, Mehrotra RASN Dialysis Advisory Group: How to overcome barriers and establish a successful home HD program. Clin J Am Soc Nephrol 7: 2023–2032, 2012
8. McLaughlin K, Manns B, Mortis G, Hons R, Taub K: Why patients with ESRD do not select self-care dialysis as a treatment option. Am J Kidney Dis 41: 380–385, 2003
9. Suri RS, Larive B, Sherer S, Eggers P, Gassman J, James SH, Lindsay RM, Lockridge RS, Ornt DB, Rocco MV, Ting GO, Kliger ASFrequent Hemodialysis Network Trial Group: Risk of
vascular access complications with frequent hemodialysis. J Am Soc Nephrol 24: 498–505, 2013
10. Astor BC, Eustace JA, Powe NR, Klag MJ, Fink NE, Coresh JCHOICE Study: Type of
vascular access and survival among incident hemodialysis patients: The Choices for Healthy Outcomes in Caring for ESRD (CHOICE) Study. J Am Soc Nephrol 16: 1449–1455, 2005
11. Pisoni RL, Arrington CJ, Albert JM, Ethier J, Kimata N, Krishnan M, Rayner HC, Saito A, Sands JJ, Saran R, Gillespie B, Wolfe RA, Port FK: Facility hemodialysis
vascular access use and mortality in countries participating in DOPPS: An instrumental variable analysis. Am J Kidney Dis 53: 475–491, 2009
12. Lacson E Jr., Wang W, Lazarus JM, Hakim RM: Change in
vascular access and
hospitalization risk in long-term hemodialysis patients. Clin J Am Soc Nephrol 5: 1996–2003, 2010
13. Banerjee T, Kim SJ, Astor B, Shafi T, Coresh J, Powe NR:
Vascular access type, inflammatory markers, and mortality in incident hemodialysis patients: The Choices for Healthy Outcomes in Caring for
End-Stage Renal Disease (CHOICE) Study. Am J Kidney Dis 64: 954–961, 2014
14. Polkinghorne KR, McDonald SP, Atkins RC, Kerr PG:
Vascular access and all-cause mortality: A propensity score analysis. J Am Soc Nephrol 15: 477–486, 2004
15. Hayes WN, Tennankore K, Battistella M, Chan CT:
Vascular access -related infection in nocturnal
home hemodialysis . Hemodial Int 18: 481–487, 2014
16. Kuttykrishnan S, Kalantar-Zadeh K, Arah OA, Cheung AK, Brunelli S, Heagerty PJ, Katz R, Molnar MZ, Nissenson A, Ravel V, Streja E, Himmelfarb J, Mehrotra R: Predictors of treatment with dialysis modalities in observational studies for comparative effectiveness research. Nephrol Dial Transplant 30: 1208–1217, 2015
17. Cummings P: Missing data and multiple imputation. JAMA Pediatr 167: 656–661, 2013
18. Austin PC: An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 46: 399–424, 2011
19. Fine JP, Gray RJ: A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 94: 496–509, 1999
20. Zhang JC, Al-Jaishi AA, Na Y, de Sa E, Moist LM: Association between
vascular access type and patient mortality among elderly patients on hemodialysis in Canada. Hemodial Int 18: 616–624, 2014
21. Pastan S, Soucie JM, McClellan WM:
Vascular access and increased risk of death among hemodialysis patients. Kidney Int 62: 620–626, 2002
22. Ravani P, Palmer SC, Oliver MJ, Quinn RR, MacRae JM, Tai DJ, Pannu NI, Thomas C, Hemmelgarn BR, Craig JC, Manns B, Tonelli M, Strippoli GFM, James MT: Associations between hemodialysis access type and clinical outcomes: A systematic review. J Am Soc Nephrol 24: 465–473, 2013
23. Zacharioudakis IM, Zervou FN, Ziakas PD, Rice LB, Mylonakis E: Vancomycin-resistant enterococci colonization among dialysis patients: A meta-analysis of prevalence, risk factors, and significance. Am J Kidney Dis 65: 88–97, 2015
24. Zacharioudakis IM, Zervou FN, Ziakas PD, Mylonakis E: Meta-analysis of methicillin-resistant Staphylococcus aureus colonization and risk of infection in dialysis patients. J Am Soc Nephrol 25: 2131–2141, 2014
25. Pop-Vicas A, Strom J, Stanley K, D’Agata EMC: Multidrug-resistant gram-negative bacteria among patients who require chronic hemodialysis. Clin J Am Soc Nephrol 3: 752–758, 2008
26. Ellingson KD, Palekar RS, Lucero CA, Kurkjian KM, Chai SJ, Schlossberg DS, Vincenti DM, Fink JC, Davies-Cole JO, Magri JM, Arduino MJ, Patel PR:
Vascular access hemorrhages contribute to deaths among hemodialysis patients. Kidney Int 82: 686–692, 2012
27. Tennankore KK, d’Gama C, Faratro R, Fung S, Wong E, Chan CT: Adverse technical events in
home hemodialysis . Am J Kidney Dis 65: 116–121, 2015
28. Perl J, Lok CE, Chan CT: Central venous catheter outcomes in nocturnal hemodialysis. Kidney Int 70: 1348–1354, 2006
29. Lindsay RM, Leitch R, Heidenheim AP, Kortas CLondon Daily/Nocturnal Hemodialysis Study: The London Daily/Nocturnal Hemodialysis Study-- design, morbidity, and mortality results. Am J Kidney Dis 42[1 Suppl]: 5–12, 2003
30. Hung AM, Ikizler TA: Hemodialysis
central venous catheters as a source of inflammation and its implications. Semin Dial 21: 401–404, 2008
31. Vesely TM: Central venous catheter tip position: A continuing controversy. J Vasc Interv Radiol 14: 527–534, 2003
32. Rubin DB: Estimating causal effects from large data sets using propensity scores. Ann Intern Med 127: 757–763, 1997
33. Winkelmayer WC, Owen WF Jr., Levin R, Avorn J: A propensity analysis of late versus early nephrologist referral and mortality on dialysis. J Am Soc Nephrol 14: 486–492, 2003
34. Winkelmayer WC, Glynn RJ, Mittleman MA, Levin R, Pliskin JS, Avorn J: Comparing mortality of elderly patients on hemodialysis versus peritoneal dialysis: A propensity score approach. J Am Soc Nephrol 13: 2353–2362, 2002
35. Noordzij M, Leffondré K, van Stralen KJ, Zoccali C, Dekker FW, Jager KJ: When do we need competing risks methods for survival analysis in nephrology? Nephrol Dial Transplant 28: 2670–2677, 2013
