Fistula Placement in Incident Patients, 2005-2006,
As Viewed from Medical Evidence Forms
Northwest Renal Network
7 February 2007
Since June 2005, Medical Evidence Form 2728, which the Centers for Medicare and Medicaid Services (CMS) uses to initiate Medicare payment for chronic dialysis, has included questions on the vascular access in use and in place by incident end stage renal disease (ESRD) patients, and on whether these patients were under the care of a nephrologist prior to their initiation of chronic dialysis. In this report, we examine the first eighteen months of experience with these questions, particularly as they apply to the placement of arteriovenous fistulae (AVF) in hemodialysis patients.
In all cases in the following we refer only to incident hemodialysis patients. At times in this report we will refer to the percentage of patients who had been under the care of a nephrologist prior to initiating dialysis as the “early-referral rate,” and to these patients as “early-referral patients.” We’ll refer to the percentage of patients who began chronic dialysis with an AVF in place as the “AVF-in-place rate.” Unless otherwise indicated, the “AVF-in-place rate” refers to all incident hemodialysis patients.
When overall AVF-in-place rates show statistical significance, we also examine the AVF-in-place rate for only those patients who had been under a nephrologist’s care prior to beginning chronic dialysis. As we will see, comparing the overall AVF-in-place rate to the early-referral AVF-in-place rate allows us to draw some tentative conclusions about how the effects of early referral and the effects of AVF placement differ.
We focus on AVF-in-place rather than AVF-in-use because AVF-in-place is a more meaningful measure for incident patients. Incident-patient AVF rates vary greatly depending on how incidence is defined. For instance, Fistula First (FF) data measures the access in place at the end of the calendar month in which an incident patient began chronic dialysis. FF data is collected at the facility level. Clinical Performance Measures (CPM) data consider the access in place at the end of the calendar year in which a patient began chronic dialysis. CPM data is collected at the patient level, but only on a 5% Network sample. The Medical Evidence data we consider here measures the access in place at the time when a patient began chronic dialysis. This Form 2728 data is collected on the patient level.
Early Referral
Surprisingly, many of these patients – 68% – had been under a nephrologist’s care prior to their initiation of chronic dialysis. Slightly more than a third of all new Network ESRD patients had been cared for by a nephrologist for more than two years before they began chronic dialysis. This table provides details.
Number
Percent of All Incident Hemodialysis Patients
Percent of Incident Patients Under Nephrologist Care Prior to Initiating Chronic Dialysis
Under Care of a Nephrologist for < 6 Months Prior to Initiating Chronic Dialysis
241
5.8%
8.6%
Under Care of a Nephrologist for 6-12 Months Prior to Initiating Chronic Dialysis
1175
28.5%
41.9%
Under Care of a Nephrologist for > 12 Months Prior to Initiating Chronic Dialysis
1389
33.6%
49.5%
Under Care of a Nephrologist Prior to Initiating Chronic Dialysis
2805
67.9%
100.0%
Not under Care of a Nephrologist Prior to Initiating Chronic Dialysis
1144
27.7%
Don't Know
180
4.4%
4129
100.0%
AVF Placement
A patient who had been under a nephrologist’s care prior to beginning chronic dialysis was 3.1 times as likely to begin dialysis with an AVF in place. Of all incident Network hemodialysis patients, 40.8% (95% confidence interval 39.9% to 43.0%) began chronic dialysis with an AVF in place. The Network goal for incident-patient AVF-in-place rate matches the corresponding KDOQI Guideline at 50%.
Among patients who had been under the care of a nephrologist prior to initiating chronic dialysis, 51.6% (49.7% to 53.4%) began dialysis with an AVF in place, compared to 17.9% (14.5% to 18.9%) of patients who had not been under a nephrologist’s care. This difference is highly significant statistically (p << 0.001), as the following graph shows. The AVF-in-use rate for early-referral patients was 22.0%, and for others 2.3%.
The longer a patient had been under a nephrologist’s care prior to initiating chronic dialysis, the greater the likelihood that they began dialysis with an AVF in place, as the graph and table below show. These four groups are all highly statistically significantly different from one another (p << 0.001).
Percent of Patients with AVF in Place
95% Confidence Interval
Length of Time Under Care of a Nephrologist Prior to Initiating Chronic Dialysis
None
17.9%
15.8% to 20.0%
0-6 Months
32.4%
26.5% to 38.3%
6-12 Months
46.7%
43.9% to 49.6%
>12 Months
59.0%
56.4% to 61.6%
AVF-in-use rates were 15.4% for patients under a nephrologist’s care for less than six months, 24.0% for 6-12 months, and 40.3% for > 12 months.
Trend
These differences have not changed significantly over the course of the eighteen months of data that we have. The following graphs show the early-referral and AVF-in-place rates for each of the three six-month periods.
The average increase from one six-month period to the next was 2.1 percentage points for all incident patients and 2.2 percentage points for early-referral patients.
By Gender
Males were 2% more likely than females to have been under care of a nephrologist prior to initiating chronic dialysis, but the difference was not statistically significant, as this graph shows. 71.7% (95% confidence interval 69.9% to 73.5%) of males and 70.1% (67.8% to 72.3%) of females were under a nephrologist’s care before their chronic dialysis began.
However, males were 17% (p < 0.001) more likely to begin dialysis with an AVF in place, as the following graph illustrates. 43.4% (95% confidence interval 41.4% to 45.3%) of males began dialysis with an AVF in place, compared to 36.9% (34.6% to 39.2%) of females.
Among patients who had been under a nephrologist’s care prior to initiating chronic dialysis, males were only 13% more likely to begin dialysis with an AVF in place (p < 0.005), implying perhaps that early referral was able to some extent to correct a pre-existing bias against females receiving the best care available. 54.0% (95% confidence interval 51.6% to 56.3%) of early-referral males and 48.0% (45.0% to 50.9%) of early-referral females began dialysis with an AVF in place. This graph shows the difference.
By Age
There were no statistically significant differences among age groups in early-referral rates, although 0-to-19-year-olds were 11% more likely to have been under a nephrologist’s care prior to beginning chronic dialysis than 40-to-59-year-olds, and 20-to-39-year-olds were 11% less likely to, as the following graph and table illustrate.
Age
Early-Referral Rate
95% Confidence Interval
AVF-in-Place Rate
95% Confidence Interval
0-19
80.6%
66.7% to 94.6%
29.0%
13.1% to 45.0%
20-39
64.9%
59.7% to 70.1%
38.1%
32.9% to 43.4%
40-59
72.5%
69.9% to 75.1%
44.3%
41.5% to 47.1%
60-79
70.9%
68.8% to 72.9%
41.4%
39.1% to 43.6%
>79
71.5%
67.9% to 75.0%
34.3%
30.7% to 37.9%
There were significant differences among age groups in AVF-in-place rates (p < 0.005). Incident patients older than 79 were 17% less likely to begin dialysis with an AVF than 60-to-79-year-olds (p < 0.01). Patients under 19 received AVFs at the lowest rate, but their standard error was too large to indicate significant difference, because of their small number (here 31). These differences are illustrated below.
By Race
Race produced significant variation in both early-referral and AVF-in-place rates, both at p < 0.025. However, among only patients who had been under the care of a nephrologist prior to initiating chronic dialysis, there were no race differences in AVF-in-place rates, implying that race was a significant determinant in early referral more than it was in AVF placement. In early-referral rates, white patients were significantly lower (10% lower) than black patients (p < 0.05). Native American patients were least likely to have been under a nephrologist’s care prior to initiating dialysis (6% less likely than whites), but because of the relatively small number of ESRD Native Americans (here 118), the difference was not statistically significant. This graph compares prior-nepthrologist rates.
Race
Early-Referral Rate
95% Confidence Interval
AVF-in-Place Rate
95% Confidence Interval
Native American
66.1%
57.6% to 74.6%
38.1%
29.6% to 46.6%
Asian
75.9%
69.9% to 82.0%
49.7%
42.8% to 56.7%
Black
78.2%
73.1% to 83.2%
46.4%
40.4% to 52.3%
White
70.3%
68.7% to 71.8%
39.9%
38.3% to 41.5%
Pacific Islander
76.7%
66.0% to 87.4%
35.9%
24.2% to 47.7%
AVF rates by race, tabled above and graphed below, reveal that Asian incident patients received significantly more (25% more) AVFs than white incident patients (p < 0.05).
AVF-in-place rates by race for only those patients who were under the care of a nephrologist prior to initiation of chronic dialysis are graphed below. No differences in this graph are statistically significant.
By State
Incident hemodialysis patients residing in Idaho or Montana had been under a nephrologist’s care prior to beginning chronic dialysis significantly less frequently than patients in Oregon or Washington (p < 0.005). As this graph and table show, Montana patients were 14% less likely and Idaho patients 16% less likely than Washington patients to have been under a nephrologist’s care prior to starting chronic dialysis.
State of Residence
Early-Referral Rate
95% Confidence Interval
AVF-in-Place Rate
95% Confidence Interval
Alaska
72.5%
63.4% to 81.7%
33.9%
25.5% to 42.2%
Idaho
61.0%
55.9% to 66.1%
27.8%
23.1% to 32.4%
Montana
62.6%
56.5% to 68.8%
45.0%
38.7% to 51.3%
Oregon
73.0%
70.5% to 75.6%
40.3%
37.6% to 43.1%
Washington
72.5%
70.6% to 74.4%
43.1%
41.0% to 45.2%
However, Montana residents began dialysis with an AVF in place more often than residents of any other Network state (p << 0.005) – Montana residents were 62% more likely to begin dialysis with an AVF in place than Idaho patients. While the differences were not significant, Alaska residents were less likely to receive an AVF than Washington or Oregon residents, even though they were as likely to have been under a nephrologist’s care prior to beginning dialysis.
The graph of AVF-in-place rates for early-referral patients only (p < 0.005) confirms that these biases are nephrologist differences (or community-resource differences) more than differences in early referral patterns. Incident early-referral hemodialysis patients in Montana were 55% more likely than Idaho patients and 58% more likely than Alaska patients to begin chronic dialysis with an AVF in place.
By Insurance Type
Early-referral rates vary significantly with the type of insurance an incident patient had (p << 0.05). Patients with only group insurance had been under the care of a nephrologist prior to beginning chronic dialysis 14% more often than patients with only Medicare insurance. Patients with no insurance were least likely to have been early-referral patients – 42% less likely than only-Medicare patients. VA patients were most likely to have been under a nephrologist’s care prior to beginning chronic dialysis. These analyses and the graphs below exclude all patients with more than one type of insurance.
Those with no insurance were significantly less likely (by about 30%) than those with Medicaid, group, or “other” insurance to begin chronic dialysis with an AVF in place (p < 0.025), as the following graph and table illustrate. In the table, note the differences in age. Medicare Advantage patients have the highest AVF rates in spite of having the highest average age, and uninsured patients have the lowest AVF rates in spite of having the lowest average age.
Type of Insurance
Average Patient Age
Early-Referral Rate
95% Confidence Interval
AVF-in-Place Rate
95% Confidence Interval
Medicaid
49
70.2%
65.6% to 74.8%
44.5%
39.7% to 49.4%
VA
58
80.6%
67.6% to 93.5%
36.6%
21.8% to 51.3%
Medicare
70
66.8%
62.8% to 70.8%
39.1%
35.1% to 43.2%
Medicare Advantage
75
71.3%
63.0% to 79.6%
45.4%
36.4% to 54.3%
Group
50
76.1%
72.3% to 79.8%
43.0%
38.8% to 47.3%
Other
54
72.2%
66.5% to 77.9%
44.8%
38.7% to 51.0%
None
46
38.8%
32.4% to 45.3%
30.8%
24.9% to 36.7%
Even among early-referral patients, those with no insurance were still significantly less likely (about 30% less likely) to have an AVF in place at the beginning of dialysis than those with Medicaid or group insurance (p < 0.05).
By Nephrologist
When this Medical Evidence data is examined by the nephrologist who signed the form, there is a broad range of AVF-in-place rates, with two primary modes, one below the Network average at 25-30% AVF-in-place and the other just above the KDOQI Guideline at 50-55% AVF-in-place. This graph is a histogram of nephrologists AVF-in-place rates. While rates range from 0% to 100%, the tails are reasonably long. Most nephrologists fall between 20% and 70% AVF-in-place. The bimodality implies that some nephrologists may not have “gotten the message,” or that their community lacks the resources for them to receive early referrals or place many AVFs.
Sixty-three percent of Network nephrologists are below the Network goal and KDOQI Guideline of 50% AVF in place for incident hemodialysis patients. The percentage of nephrologists whose incident patients fall below 50% ranges from 93% in Idaho to 43% in Western Washington, as this table shows.
State or Region
Percent of Nephrologists with AVF-in-Place Rate Lower than 50%
Idaho
93%
Alaska
86%
Eastern Washington
76%
Montana
71%
Oregon
64%
Western Washington
43%
Network
63%
