Peripherally Inserted Central Catheters and Hemodialysis Outcomes

Abstract

Background and objectives

Use of peripherally inserted central catheters has expanded rapidly, but the consequences for patients who eventually require hemodialysis are undefined.

Design, setting, participants, & measurements

Our national, population–based analysis included 33,918 adult Medicare beneficiaries from the US Renal Data System who initiated hemodialysis with central venous catheters as their sole vascular access in 2010 and 2011. We used linked Medicare claims to identify peripherally inserted central catheter exposures and evaluate the associations of peripherally inserted central catheter placement with transition to working arteriovenous fistulas or grafts and patient survival using a Cox model with time-dependent variables.

Results

Among 33,918 individuals initiating hemodialysis with a catheter as sole access, 12.6% had received at least one peripherally inserted central catheter. Median follow-up was 404 days (interquartile range, 103–680 days). Among 6487 peripherally inserted central catheters placed, 3435 (53%) were placed within the 2 years before hemodialysis initiation, and 3052 (47%) were placed afterward. Multiple peripherally inserted central catheters were placed in 30% of patients exposed to peripherally inserted central catheters. Recipients of peripherally inserted central catheters were more likely to be women and have comorbid diagnoses and less likely to have received predialysis nephrology care. After adjustment for clinical and demographic factors, peripherally inserted central catheters placed before or after hemodialysis initiation were independently associated with lower likelihoods of transition to any working fistula or graft (hazard ratio for prehemodialysis peripherally inserted central catheter, 0.85; 95% confidence interval, 0.79 to 0.91; hazard ratio for posthemodialysis peripherally inserted central catheter, 0.81; 95% confidence interval, 0.73 to 0.89).

Conclusions

Peripherally inserted central catheter placement was common and associated with adverse vascular access outcomes. Recognition of potential long–term adverse consequences of peripherally inserted central catheters is essential for clinicians caring for patients with CKD.

Introduction

For patients who depend on hemodialysis (HD) to treat their kidney failure, a reliable arteriovenous fistula (AVF) is a lifeline. Using an AVF or even an arteriovenous graft (AVG) rather than a central venous catheter (CVC) reduces the risks of infection, hospitalizations, and mortality. However, >80% of Americans initiate HD with CVCs (1–3), despite evidence showing that failure to transition to a fistula strongly predicts poor outcomes and increased costs of care (4–8).

Peripherally inserted central catheters (PICCs) were introduced in 1975 to facilitate parenteral nutrition (9). PICCs are easily placed using portable ultrasound guidance and allow for home parenteral therapy. PICCs are convenient for nurses, physicians, and patients. Their use shortens the length of stay in hospitals and nursing facilities, and utilization has increased rapidly over the past 15 years (10,11). However, PICC use also is associated with stenosis, thrombosis, and obliteration of the central and peripheral veins in which they dwell (12,13), and these complications may have important long–term consequences for patients who subsequently require HD vascular access. Patients with CKD often have complex preexisting comorbid conditions that may result in PICC exposure, potentially increasing their risk for loss of veins (14). Practice guidelines that discourage PICC use in patients with CKD have received little attention outside the nephrology community (15–17).

Few studies examine exposures to PICC placement before and after HD initiation or the consequences of PICC exposure in patients receiving HD. We used US Renal Data System (USRDS) files to link Medicare claims from before and after enrollment in the ESRD Program. We report the PICC exposure of Medicare beneficiaries who began HD with CVC as their sole vascular access, a group of patients for whom the preservation of veins for AVF creation is crucial. We determined the associations of PICC exposure with vascular access outcomes and patient survival.

Materials and Methods

Data Sources and Study Population

The USRDS Standard Analytical Files were linked with Medicare claims for the years 2008–2012. This study was approved by the Institutional Review Board of Tufts Medical Center and deemed exempt from requirement for informed consent for the use of deidentified data.

The study population consisted of patients who initiated HD for the first time between April 1, 2010 and December 31, 2011 in whom CVC was the sole vascular access present at the first outpatient HD treatment. We restricted our analysis to Medicare beneficiaries ≥20 years of age having Medicare coverage 730 days before the day of first HD according to the USRDS Payer History File who had at least two predialysis Medicare claims (one or more of which occurred >365 days before the HD initiation date) and continuous coverage until the end of 2012 or a censoring event. Claims codes for HD catheters are identical to those used for other CVCs, and therefore, these could not be used to identify patients dialyzing with catheter access. Therefore, we instead determined the vascular access at HD initiation from the USRDS Medical Evidence Form Centers for Medicare and Medicaid Services (CMS) Form 2728 and then, verified catheter access by examination of HD treatment claims for the first 6 weeks. Patients were excluded for inconsistent vascular access data if HD claims showed AVF use <6 weeks or AVG use <3 weeks after HD initiation to eliminate patients whose AVF or AVG was already placed and maturing at the time of HD initiation.

Study Variables

PICC placements were ascertained by examining the institutional details, physician/supplier, and revenue center details files for the appropriate Common Procedural Terminology (CPT) codes (CPT codes 36569 and 36571), which are specific for peripherally inserted catheters. PICCs placed on or after the date of HD initiation were considered to be post-HD; PICCs placed within the 730 days before HD initiation were considered to be pre-HD. Although a claim for a CVC within 30 days of HD initiation was not recovered in 13% of patients, <4% of such patients had any claims for PICC, suggesting that these codes were not being used interchangeably with those for the placement of dialysis catheters. Vascular imaging was determined from codes specific for upper extremity, central venography, and vein mapping for dialysis access planning (CPT codes G0365, 36005, 75820, 75822, and 75827). Age at incident HD, sex, race, body mass index, primary causes of kidney failure, pre–HD nephrology care, pre–HD erythropoietin use, inability to ambulate, and pre–enrollment laboratory values (hemoglobin, albumin, and creatinine) were obtained from Medical Evidence Enrollment Forms (CMS Form 2728). Comorbid conditions were determined directly from predialysis Medicare claims over the 2 years before HD initiation according to the method by Liu et al. (18) with the exception of diabetes, because diabetes was represented as a primary cause of kidney failure as well as a comorbid condition. To prevent overlap between the two variables, the comorbid diagnosis variable for diabetes was scored as absent in patients for whom the primary cause of kidney failure was diabetic nephropathy. Claims for diabetes care were used to evaluate the presence or absence of diabetes in the subset of patients for whom the primary cause of kidney failure was not diabetic nephropathy.

Primary Outcomes

The USRDS Treatment History Files were used to ascertain dates of death, transplant, and changes in treatment modality. Patients were followed until death, transplant, transfer to another dialysis modality, or December 31, 2012, whichever came first. Vascular access was determined by the modifier codes on individual HD treatment claims. Conversion to fistula or graft access was defined as the first day of the first 30-day period in which all available vascular access modifiers on all HD treatment claims were for AVF only (modifier V7) or AVG only (modifier V6) with no intervening codes for catheter use (modifier V5) on the basis of established working definitions for a working AVF (19,20).

Statistical Analyses

Means and medians were used to summarize distributions of normally and non–normally distributed continuous variables, respectively, with t tests, Wilcoxon rank sum tests, and chi-squared tests used as appropriate to compare baseline values. Cox proportional hazard analyses were used to evaluate the associations of PICC placement with time to first AVF, first AVF or graft, or death.

PICC insertion events were pre-HD or post-HD depending on whether the date of insertion preceded or followed the date on which Form 2728 reported that regular chronic dialysis began. Cox models using time-dependent covariates evaluated the associations of PICC insertions after HD initiation with the three specified outcomes. To reduce the influence of a few patients having many PICC events, we dichotomized both pre- and postdialysis PICC recipients as having received no PICC or at least one PICC. For postdialysis PICC recipients, the earliest postdialysis PICC insertion defined the time of exposure for the time–dependent post–HD PICC variable. Models were adjusted for age at incident HD, sex, race, body mass index, primary causes of kidney failure, pre–HD nephrology care, pre–HD erythropoietin use, inability to ambulate, comorbid conditions, and pre–enrollment laboratory values (hemoglobin, albumin, and creatinine).

All analyses were conducted using SAS, version 9.4 (SAS Institute Inc., Cary, NC).

Results

Participant Characteristics

Over the study period, 178,787 patients new to HD enrolled in the USRDS. Of these potential participants, 144,869 were excluded. The major reason for exclusion was lack of predialysis Medicare coverage of sufficient duration in 69% of incident patients followed by the presence of a working or maturing AVF or AVG at the first HD treatment in 12%. Our study population consisted of 33,918 patients who satisfied inclusion criteria (Figure 1). The average age of the total study population was 72.6 years old, with 47.1% women, 73.2% white, and 22.5% black. Median follow-up after HD initiation was 404 days (interquartile range [IQR], 103–680 days).

Figure 1.

Flowchart of patients included in the study. AVF, arteriovenous fistula; AVG, arteriovenous graft; CVC, central venous catheter; HD, hemodialysis; PICC, peripherally inserted central catheter.

There was at least one exposure to PICC before or after HD initiation in 4257 (12.6%) patients. In total, 6487 PICCs were placed, 53% in the 2 years preceding dialysis initiation. Among patients receiving PICCs, 46% had predialysis PICC only, 42% had postdialysis PICC only, and 11% had both. More than one PICC was placed in 30% of these patients, and more than five PICCs were placed in 2.5% of patients (maximum of 14 PICCs). Table 1 compares the baseline characteristics of patients who received PICCs before or after starting HD with those who had none. PICC recipients were more likely to have each comorbid condition and be women, and they were less likely to have had predialysis nephrology care. Among 18,871 patients who had one or more vascular access surgeries, 49.8% had vascular imaging; 85.7% of the vascular imaging studies were performed after HD initiation.

Table 1.

Baseline patient characteristics at hemodialysis initiation by peripherally inserted central catheter status

Characteristic	All	Any PICCa	No PICC	P Value
N	33,918	4257	29,661
Age, yr, mean (SD)	72.6 (11.3)	71.2 (11.9)	72.8 (11.2)	<0.001
Women, %	47.1	51.8	46.5	<0.001
Race				<0.001
White	73.2	71.5	73.5
Black	22.5	25.4	22.1
Other	4.3	3.1	4.4
Body mass index, kg/m2, mean (SD)	28.8 (7.8)	29.8 (8.3)	28.7 (7.7)	<0.001
Pre–ESRD vascular imaging, %	7.4	8.2	7.3	0.04
Pre–ESRD nephrology care, %	47.3	41.2	48.2	<0.001
Pre–ESRD erythropoietin use, %	15.2	14.3	15.3	0.002
Primary ESRD diagnosis				<0.001
Diabetes	41.7	42.7	41.6
Hypertension	32.5	29.6	32.9
Primary GN	3.2	2.3	3.3
Other	22.6	25.4	22.2
Comorbid conditions, %
Atherosclerotic heart disease	49.6	59.4	48.2	<0.001
Congestive heart failure	45.7	55.7	44.3	<0.001
Other cardiac conditions	31.1	40.7	29.7	<0.001
Arrhythmia	29.8	37.8	28.6	<0.001
Peripheral vascular disease	28.6	38.5	27.2	<0.001
Pulmonary disease	26.9	35.3	25.7	<0.001
Diabetes not cause of ESRD	23.8	27.8	23.2	<0.001
Inability to ambulate	12.5	19.1	11.6	<0.001
Stroke	14.4	17.5	14.0	<0.001
Cancer	11.7	12.9	11.5	<0.01
Gastrointestinal disease	5.3	8.9	4.7	<0.001
Liver disease	4.2	6.3	3.9	<0.001
Pre–ESRD laboratory work, mean (SD)
Serum albumin, mg/dl	3.0 (0.7)	2.9 (0.7)	3.0 (0.7)	<0.001
Serum creatinine, mg/dl	5.3 (2.7)	4.8 (2.5)	5.4 (2.7)	<0.001
Hemoglobin, g/dl	9.8 (1.5)	9.7 (1.5)	9.8 (1.5)	0.004

PICC, peripherally inserted central catheter.

^aAny PICC includes all patients receiving any PICC either before or after hemodialysis initiation or both.

Vascular Access and Mortality Outcomes

Transition to a working AVF occurred in 1052 (24.7%) patients who received PICCs versus 9088 (30.6%) patients who did not (Figure 2). Transition to a working AVG occurred in 489 (11.5%) patients who received PICCs versus 3130 (10.6%) patients who did not. The overall probability of achieving any permanent vascular access was lower for patients exposed to PICCs (36.2% versus 41.2%; P<0.001). Overall, 20,159 patients never transitioned from CVC to any other working access. Among 13,759 patients who achieved a working AVF or AVG, PICC was associated with longer catheter exposure before the first working vascular access: 218 (IQR, 148–315) days versus 197 (IQR, 134–284) days (P<0.001).

Figure 2.

Cumulative incidence plots of death and vascular access as competing events. Delayed and lower achievement of arteriovenous fistula (AVF) or arteriovenous graft (AVG; P<0.001) and earlier and higher death (P<0.001) with peripherally inserted central catheter (PICC) exposure before hemodialysis initiation is shown.

Among 33,918 patients, 15,686 (46.2%) died; death was more frequent in those who had exposure to PICCs (56.4% versus 44.8%; P<0.001). Among those achieving AVF or AVG, 30.4% with PICC exposure versus 22.5% without PICC exposure died (P<0.001), whereas among those who did not transition from CVC, 71.2% with PICC exposure versus 60.4% without PICC exposure died (P<0.001).

In adjusted Cox models (Table 2) comparing patients with and without predialysis PICC exposure, patients who received PICCs had a 23% lower likelihood of achieving an AVF, a 15% lower likelihood of any AVF or AVG, and a 15% higher likelihood of death. Patients who received post-HD PICC had 19% lower likelihood of achieving an AVF and 19% lower likelihood of any AVF or AVG independent of the associations with pre–HD PICC placement. The hazard ratio (HR) for transition to any working AVF or AVG was 0.85 for pre-HD PICC (95% confidence interval [95% CI], 0.79 to 0.91) and 0.81 (95% CI, 0.73 to 0.89) for post-HD PICC. The adjusted HR for death was 1.05 (95% CI, 0.99 to 1.11) for PICC placed before HD initiation and 2.26 (95% CI, 2.13 to 2.39) for PICC placed thereafter. The association between post-HD PICC and death remained significant in an exploratory post hoc analysis, in which achievement of vascular access was included as a covariate in the time–dependent model for death (HR, 2.19; 95% CI, 2.07 to 2.32).

Table 2.

Relationships of peripherally inserted central catheter with first working arteriovenous fistula, first working arteriovenous fistula or graft, and death

Exposure Variables	First AV Fistula	First AV Fistula or Graft	Death
Standard Cox models
Pre-HD PICC (95% CI)	0.77 (0.71 to 0.85)	0.85 (0.79 to 0.92)	1.15 (1.09 to 1.21)
Time–dependent Cox models
Pre-HD PICC (95% CI)	0.78 (0.71 to 0.85)	0.85 (0.79 to 0.91)	1.05 (0.99 to 1.11)
Post-HD PICC (95% CI)	0.81 (0.72 to 0.90)	0.81 (0.73 to 0.89)	2.26 (2.13 to 2.39)

Adjusted for age, sex, race, predialysis nephrology care, predialysis erythropoietin, primary diagnosis for kidney failure, body mass index, hemoglobin, albumin, creatinine, and all comorbid conditions. AV, arteriovenous; HD, hemodialysis; PICC, peripherally inserted central catheter; 95% CI, 95% confidence interval.

Discussion

One in every eight patients who initiated HD with a CVC in the United States was exposed to PICCs while receiving HD or within the 2 years before HD initiation. PICC placement before or after HD initiation was strongly associated with failure to transition to a working fistula, and transition to any permanent access was lower, despite a small increase in the percentage of AVGs. PICC placed after HD initiation was associated with a higher risk of death, which remained significant, despite adjustment for baseline comorbid conditions, case mix factors, and achievement of vascular access.

Our findings confirm the very limited previously published observations about PICC use in patients on dialysis. In a case-control study of 282 patients receiving outpatient HD, 44% of patients dialyzing via HD catheters had a history of PICC use compared with <20% of those with working AVFs (21). The predialysis PICC exposure in that study, which assessed events occurring up to 14 years before HD, was higher than we observed; we examined claims for only the 2 years preceding dialysis. A recent cross–sectional study reported that hospitalized patients with GFR<45 ml/min per 1.73 m² were 30% more likely to receive PICCs than the overall hospital population (14). In both of these prior studies, 30%–50% of PICCs were placed for nonspecific indications, such as difficult vascular access (14,21). PICC placement enables blood sampling and continuous vascular access without frequent venipunctures or direct cannulation of central veins, and these conveniences contribute to their expanding popularity (22,23). Because PICC use has increased, potential concerns about bloodstream infections, thromboses, and stenoses of central and peripheral veins have emerged (11–13,24,25).

Our study expands the available findings to a large national sample of patients on dialysis with CVCs and provides additional evidence of the potential adverse consequences of PICC. We found that PICC exposure was common and associated with poorer vascular access outcomes. The association of PICC placement after HD initiation with death was not explained by baseline comorbid conditions or adjustment for achievement of vascular access, suggesting that this association may represent reverse causation, in which PICC placement is a marker for severity of illness and subsequent higher likelihood of death. Therefore, PICC placement in a catheter-dependent patient on HD may indicate clinical situations in which the short-term risk of death is particularly high.

Patients on HD are a vulnerable population. The central importance of durable dialysis vascular access alters the usual balance between the risks and benefits of PICCs. Premature exhaustion of veins needed for fistula construction may lead to prolonged or permanent use of HD catheters, which increases risks for bacteremia, endocarditis, metastatic infection, hospitalization, and death. In an observational study of 79,545 patients, conversion from CVC access to a working fistula or graft was associated with 30% decreases in both hospitalization and mortality (6,8). In a meta-analysis of 545,441 patients, CVC use was associated with 53% higher risk of death and more than twice as many fatal infections compared with working AVFs (26). Our findings also show a very high proportion of death among patients who fail to transition from CVC. PICC placement may precede the diagnosis of CKD, but our data suggest that frequent placement does not abate after recognition of either CKD or even ESRD.

We found extended predialysis nephrology care to be one of the few modifiable variables associated with decreased PICC exposure. Nephrologists are familiar with the consequences that occur when patients begin HD with depleted vascular access and need to accept responsibility for protecting veins for AVFs and AVGs before and after HD initiation. However, vein protection strategies can and should be practiced by all clinicians who provide care to patients with CKD. Individual programs have created successful vein protection protocols that use alternatives to PICCs. In our center, nurses responsible for PICC placement prompt clinicians to consult interventional radiologists to place small bore cuffed tunneled central catheters in the internal or external jugular veins. These small catheters are unlikely to injure large central veins, and the potential loss of an external or internal jugular is less problematic than the loss of a vein needed for an AVF. These alternatives require additional resources and expertise, making it important to provide convincing nonanecdotal evidence that the downstream risks of failed HD vascular access justify the additional efforts required to protect veins.

One strength of this study is the large population in which pre- and postdialysis PICC placement could be ascertained, allowing us to uncover robust associations that are potentially generalizable to large numbers of patients. Nonetheless, certain limitations warrant discussion. Our study was limited to individuals with Medicare coverage 2 years before the first HD, selecting an older group than the overall HD population. We minimized this potential bias by including all individuals with coverage, regardless of age. Our findings may not be generalizable to patients not on Medicare, patients who do not initiate HD with catheters, or children.

PICC placement was detected from claims, and we cannot rule out systematic underestimation of the PICC exposure of our population because of missing claims. Moreover, because ascertainment of claims was limited to the 20-year period before HD initiation, any patients who received PICC earlier than this may have been misclassified as unexposed, leading us to underestimate the association of PICC placement with lower transition from catheter access. The level of PICC exposure was nonetheless high, and the associations with vascular access outcomes were very strong, despite limitations that could potentially bias our findings to the null. Potentially important variables were unavailable in this dataset, including which arm was used for PICC placement and vascular access creation, indication for PICC, and results of preoperative vessel imaging, and the duration of follow-up was not sufficient to make meaningful conclusions about AVF/AVG longevity and failure rates. Finally, residual confounding by unmeasured covariates may be present, particularly for the association between post–HD initiation PICC placement and mortality.

In conclusion, patients undergoing HD have extensive exposure to PICC both before and after HD initiation. PICC exposure was associated with longer time to conversion from CVC to a working fistula or graft and shorter survival of patients on dialysis. Placement of PICC after HD initiation may indicate a poor short–term prognosis. Vein protection must begin long before dialysis treatment is needed, and therefore, clinicians caring for patients with CKD should consider when the potential long–term consequences of PICC outweigh their short-term convenience.

Disclosures

K.B.M., D.C.M., and D.E.W. all receive salary support from Dialysis Clinic, Inc., paid to their institution.