Author Affiliations: Division of Geriatrics, San Francisco Medical Center and the University of California, San Francisco.
Screening guidelines recommend that mammography be targeted to women likely to live longer than 5 years. Because women undergoing dialysis have a reduced but variable life expectancy, their appropriate use of screening is controversial. Therefore, we conducted this study to describe national mammography rates among women undergoing dialysis with differing prognostic factors and to determine whether screening is targeted to healthier women who live longer.
Using the US Renal Data System, we identified 17 090 women aged 50 years or older who started dialysis in 1997. We tracked women for 5 years to ascertain their use of screening mammography or death.
The 5-year survival rate was 25%. The biennial screening mammography rate was 25%, ranging from 12% for women aged 80 years or older to 69% for women who were ever on the transplant list. Women who were screened in the past year had a lower death rate than those who were not (hazards ratio, 0.55; 95% confidence interval, 0.51-0.59). Yet, 2198 women (13%) who died within 5 years underwent screening, and 2004 women (12%) who lived more than 5 years while receiving dialysis did not undergo screening.
Screening mammography rates are appropriately low among women undergoing dialysis because the 5-year survival rate is low. Screening is being targeted to women who are healthier and live longer. However, targeting could be improved by increasing screening in the few women undergoing dialysis with substantial life expectancies while decreasing screening in most women undergoing dialysis who live less than 5 years.
Guideline panels and reviews generally agree that screening mammography is unlikely to be beneficial for women with significant comorbidity or with a life expectancy of less than 5 years, and may actually lead to workups in these women that result only in harm.1- 5 However, as is true for many chronic conditions, there are no screening mammography guidelines specific to patients with end-stage renal disease undergoing dialysis, who are growing in number and predicted to exceed 600 000 in the United States by 2010.6,7 Because patients undergoing dialysis have a significantly reduced but variable life expectancy, the appropriate use of cancer screening is controversial in this population.8 Some clinicians argue that cancer screening is underused in the dialysis population and that quality improvement efforts should be directed at increasing screening rates, whereas others advise against routine cancer screening in dialysis patients.9- 12
Cost-effectiveness analyses have shown that breast cancer screening is likely to have minimal effect on life expectancy in the dialysis population as a whole because of the high competing risks of death in these women.13- 15 However, a subset of these women have substantial life expectancies. For example, younger patients and those who are African American survive longer while undergoing dialysis, and it has been recommended that screening be targeted to them.13- 16 However, it is unclear to what extent screening mammography is actually targeted to women undergoing dialysis with good prognostic factors who might reasonably benefit and is avoided in women with life expectancies of less than 5 years who will probably experience only the harms of screening, such as additional procedures owing to false-positive results, identification and treatment of clinically insignificant disease, and psychological distress.1,5 In the general elderly population (≥70 years), screening mammography is not avoided by women in poor health.17 Therefore, we conducted this study to describe national rates of screening mammography use among women undergoing dialysis with differing prognostic factors and to determine whether screening is being appropriately targeted to healthier women who live longer.
Data used in this longitudinal cohort study came from the US Renal Data System (USRDS), which contains demographic and clinical information on all patients with end-stage renal disease who undergo chronic dialysis or kidney transplantation. The USRDS contains data from various sources, including Medicare billing records, United Network for Organ Services transplant records, death notification reports, and the Centers for Medicare and Medicaid Services (CMS) Medical Evidence form 2728, which must be completed by renal providers for all new patients with end-stage renal disease.18
The potential analytic cohort for this study included the 29 955 women aged 50 years or older who began dialysis between January 1 and December 31, 1997. We excluded 1440 women (5%) who had cancer within the past 10 years designated on the CMS Medical Evidence form and 2536 women (8%) who died within 90 days after starting dialysis, because this exclusion is used by the USRDS to define stable chronic dialysis.18 We also excluded 7048 women (24%) who did not have fee-for-service Medicare or Medicaid as their primary payer because mammography services provided to them are not reliably reported in the USRDS claims data. In addition, we excluded 1841 women (6%) who had only diagnostic mammograms during the 5 years of follow-up because these women presumably had breast symptoms and were not eligible for screening. Diagnostic mammograms were defined as any mammogram other than a bilateral screening mammogram identified by the Health Plan Employer Data and Information Set 2002 algorithm.19 This left a final cohort of 17 090 women.
We identified potential prognostic characteristics in women undergoing dialysis, including age, race/ethnicity, comorbid conditions, primary cause of renal failure, and body mass index, which were obtained from the CMS Medical Evidence form at the start of dialysis.20 Treatment modality at 90 days (hemodialysis or peritoneal dialysis) was obtained from the USRDS treatment history file. We also identified women who were listed on the transplant waiting list or who underwent kidney transplantation at any point during the 5 years after starting dialysis. In addition, the 18 dialysis networks were grouped into 5 regions (west and mountain, midwest, northeast, southeast, and Texas and Oklahoma).18
Receipt of mammography during the 5 years after starting dialysis was determined using USRDS Medicare payment data (institutional and physician/supplier files) containing inpatient and outpatient claims for each patient between January 1, 1997, and December 31, 2002. Screening mammograms were identified by the Current Procedure Terminology code 76092 (bilateral screening mammogram).21,22 Inpatient and outpatient claims were also examined for the diagnosis of breast cancer (codes 174.0-174.9 of the International Classification of Diseases, Ninth Revision, Clinical Modification).23,24 Because none of the women included in our cohort had a history of cancer, breast cancers identified after the start of dialysis were considered incident cancers. In addition, women were considered to have died of breast cancer if they were diagnosed as having breast cancer and the USRDS Death Notification form (CMS-2746-U3) indicated the primary cause of death was malignant disease. Additional causes of death and the date of death were also based on the Death Notification form. Loss to follow-up was defined as having no Medicare claims in the last year among those women not known to be dead.18 However, this occurred in less than 0.4% of our cohort.
The Committee on Human Research at the University of California, San Francisco, the USRDS Coordinating Center, and the Committee for Research and Development at the San Francisco VA Medical Center approved the study.
Because of the high rate of death, screening mammography rates were calculated as the number of first screening mammograms performed after the start of dialysis divided by total time (in person-years) eligible for screening. Women were considered eligible for screening from the time they started dialysis until the occurrence of a screening mammogram, a breast cancer diagnosis, death, loss to follow-up, or the end of 5 years. Mammography rates are expressed as biennial rates because the recommended screening interval is every 2 years.4
We conducted 3 analyses to examine the extent to which screening mammography was targeted to women with better prognoses. First, we evaluated whether prognostic characteristics that have higher 5-year survival rates were associated with higher screening mammography rates. We used Cox proportional hazards models to measure the bivariate relationship between each prognostic characteristic and the time to first screening mammogram. Second, we evaluated whether screening mammography was associated with survival using a Cox proportional hazards model. The outcome was survival time among women who lived at least 1 year after starting dialysis, and the predictor variable was screening mammography within the past year entered as a time-dependent covariate to account for changes in screening status during the 5 years of follow-up.25 Finally, to identify the number of women who in hindsight were mistargeted for screening (eg, underwent screening but died within 5 years or never had a screening mammogram despite living more than 5 years), we described screening mammography use among 5-year survivors and nonsurvivors. We used SAS software, version 9.1 (SAS Institute Inc, Cary, NC), for all analyses.
The mean (SD) age of the 17 090 women in this cohort was 70 (9) years. Thirty-three percent were African American. Diabetes mellitus or hypertension was the cause of renal failure in 78%. Ninety-one percent were undergoing hemodialysis (Table 1). Five years after starting dialysis therapy, only 4320 women (25%) remained alive. This survival rate is lower than the 5-year survival rate of 39% reported by US Life Tables for 90-year-old women in the general US population (Figure 1).26 Five-year survival rates for women undergoing dialysis ranged from 10% for those 80 years or older to 74% for women who were ever on the kidney transplant waiting list, which includes women awaiting transplant and those who received a transplant. Survival rates were highest among women who were younger, who did not have comorbid disease, or who were on the kidney transplant waiting list (Table 1). However, even women who started dialysis at ages 50 to 59 years without diabetes or cardiovascular disease (n = 696) had a 5-year survival rate of only 59%.
Five-year survival rates for women 50 years or older undergoing dialysis compared with those of elderly women (aged 70, 80, or 90 years) in the general US population not undergoing dialysis. Survival data are from the US Life Tables 1997.26
A total of 4514 women had at least 1 screening mammogram during the total follow-up time of 36 717 person-years, for a biennial screening rate of 25% (95% confidence interval [CI], 24%-25%). Biennial screening mammography rates ranged from 12% for women aged 80 years or older to 69% for women who were ever on the kidney transplant waiting list (Table 2). Characteristics that had higher 5-year survival rates were generally associated with higher screening rates. In addition, 37% of the 4514 women who were screened had 2 or more screening mammograms after starting dialysis, for a total of 7226 screening mammograms performed during follow-up.
Among women who survived at least 1 year on dialysis (n = 13 355), those who had a screening mammogram within the past year were less likely to die than those who did not (hazards ratio, 0.55; 95% CI, 0.51-0.59). To rule out the possibility that screening caused this survival benefit, we repeated the analysis in women who were never diagnosed as having breast cancer because a survival benefit due to screening can only occur in women who develop breast cancer.1 The association between screening mammography and survival was the same, suggesting that screening appears to have been targeted to healthier women who lived longer.
However, screening was not only received by women who lived longer than 5 years after starting dialysis. Among the 12 770 women who died, 2845 screening mammograms were performed in 2198 women. Therefore, 2845 (39%) of all 7226 screening mammograms were performed in women who died within 5 years. In addition, 1266 (58%) of these 2198 women who were screened and died had at least 1 unfavorable prognostic characteristic at the start of dialysis (Figure 2). Figure 2 also illustrates that 2004 women who lived longer than 5 years were never screened and 54% had at least 1 unfavorable prognostic characteristic. On the other hand, clinicians appropriately avoided screening in 10 572 women who died within 5 years, while they provided screening to 2316 women who lived longer than 5 years.
The number of women undergoing dialysis who were mistargeted or appropriately targeted for screening mammography, stratified by the presence or absence of unfavorable prognostic characteristics at the start of dialysis. Mistargeting was defined in hindsight as screening after the start of dialysis in a woman who died within 5 years or never screening after the start of dialysis in a woman who lived more than 5 years. BMI indicates body mass index (calculated as weight in kilograms divided by the square of height in meters); CAD, coronary artery disease; CHF, congestive heart failure; CVD, cerebrovascular disease; and PVD, peripheral vascular disease. Women with at least 1 unfavorable prognostic characteristic (n = 10 961) had only a 19% 5-year survival rate.
In addition, breast cancer was a rare cause of death during the first 5 years of dialysis. Including the 1841 women who only had diagnostic mammograms after starting dialysis, 743 (4%) of 18 931 women were diagnosed as having breast cancer, including 152 (20%) of these diagnosed within 1 year after a screening mammogram. Of the 743 women diagnosed as having breast cancer, 44 (6%) died of malignancy and 449 (60%) died of other causes. Therefore, the overall risk of dying of a newly detected breast cancer during the 5 years after starting dialysis was 0.2% (44/18 931). In comparison, the 5-year risk of dying of cardiovascular disease was 32% (5975/18 931), and the risk of dying of infection was 11% (2101/18 931).
This study examined survival and screening mammography utilization by women who started dialysis at age 50 years or older and were followed up for an extended 5-year period that began 6 years after the introduction of the Medicare mammography benefit. Five-year survival was only 25% among women in this chronic stable dialysis cohort, defined as those living more than 90 days after starting dialysis.6 Therefore, the biennial screening mammography rate of 25% is appropriately low. In addition, although screening mammography was not perfectly targeted to only those women who lived more than 5 years, screening appears to have been targeted to healthier women undergoing dialysis who lived longer.
Although there are no specific guidelines for cancer screening in patients undergoing dialysis, many screening guidelines and cost-effectiveness analyses recommend that screening mammography be targeted to women who are likely to live longer than 5 years.1- 3,13,27 As few women aged 50 years or older lived longer than 5 years after starting dialysis, the low screening mammography rates in this cohort are consistent with general screening mammography guidelines. The low screening rates in this study are also consistent with those of a previous study in the mid-1990s in New Jersey, which found that 26% of women aged 52 to 69 years underwent screening or diagnostic mammography during the first year of dialysis.9
In addition to being less likely to benefit from screening mammography because of low 5-year survival rates, women undergoing dialysis have several reasons why they may be more likely than women in the general population to experience harm from screening mammography. First, women undergoing dialysis often have secondary hyperparathyroidism, which increases breast calcifications on mammography and may increase the likelihood of false-positive results.11,28 Second, women undergoing dialysis may have more complications caused by invasive diagnostic and therapeutic procedures (eg, biopsy, surgery, chemotherapy) owing to their higher prevalence of cardiovascular disease and their greater risk for infection and delayed wound healing.6,29 Third, we found that even women diagnosed as having breast cancer were 10 times more likely to die of causes other than cancer, suggesting that screening for women undergoing dialysis may frequently cause harm by finding clinically insignificant disease for which treatment causes burden but does not lengthen life.1
In line with mammography guidelines and concerns about potential harms, we found that screening mammography appears to be targeted to women undergoing dialysis who are expected to live longer, such as younger women and those without cardiovascular disease. We also found that at any time, women who were screened in the past year had half the risk of dying compared with women not recently screened, regardless of whether they were diagnosed as having breast cancer, which further supports that screening is being targeted to healthier women.
However, there is room for improvement in how mammography is targeted using prognostic characteristics. For example, 2198 women (13%) in our dialysis cohort underwent screening mammography but died within 5 years and therefore were subjected only to the potential harms of screening. More than half of these women had at least 1 characteristic associated with a greater than 80% 5-year mortality (eg, age ≥80 years, congestive heart failure, vascular disease, low body mass index), in whom screening in general should not be recommended. Although 2101 women (12%) with an unfavorable characteristic lived more than 5 years, these women are still sufficiently different from women in randomized controlled trials such that the benefits of screening them are uncertain and at best small.30,31 Breast cancer is a rare cause of death among women undergoing dialysis.32 Therefore, the major emphasis of screening mammography decisions in women undergoing dialysis should be on “doing no harm” by minimizing screening in women who have limited life expectancies and by allowing women to spend their limited free time while not receiving dialysis away from the health care setting. This would also allow for greater emphasis to be placed on treating cardiovascular risk factors, constructing permanent vascular access, and optimizing nutrition, which have been shown to improve the lives of women undergoing dialysis.33
However, some clinicians have decried low screening mammography rates among women receiving dialysis as poor-quality care and have called for quality improvement measures, such as formal reporting procedures and incentive systems, to increase screening rates in this population.9,10 However, any policy that seeks only to increase overall screening rates in the dialysis population will not improve preventive care but will increase the number of women with limited life expectancies who experience the harms of screening without benefit. Instead, tools are needed to help clinicians identify women undergoing dialysis who have substantial life expectancies and for whom there is some chance that the potential benefits of finding an early-stage cancer outweigh the potential harms (eg, women on the kidney transplant list). Such tools would promote informed screening decisions in which accurate estimates of benefits and harms are weighed according to patient preferences.1,34 Although women who choose dialysis may have a more interventional approach to medical care and decide to continue mammography even when it offers little chance of benefit, there is evidence that patients will often withdraw requests for unhelpful treatments when potential harms are discussed with them.35
Our study has several limitations. First, because we relied on Medicare claims for the receipt of screening mammography, we excluded women who did not have Medicare as their primary payer. Including these women in survival calculations did not change 5-year survival rates, but women with health maintenance organization or private insurance may have higher screening rates.36 Second, even among women with Medicare coverage, claims will miss mammograms paid for by other sources (eg, out-of-pocket, free clinics), and there may be errors in coding screening mammograms as diagnostic, such that claims probably underestimate use of screening mammography.21,22 However, several studies have shown that when our study was conducted, use of the screening mammography Current Procedural Terminology code was widespread and a valid measure of the percentage of beneficiaries provided screening through Medicare.22,37 Third, although we did not have tumor registry data to confirm diagnoses of breast cancer, incidence rates for breast cancer calculated from inpatient and outpatient Medicare claims have been shown to closely approximate Surveillance, Epidemiology, and End Results program data.23,38 We also had the advantage of using USRDS data to exclude women who had breast cancer within 10 years of starting dialysis to better differentiate incident from prevalent breast cancer cases. Finally, although this study includes several important prognostic characteristics from the CMS Medical Evidence form completed for all persons starting dialysis, it does not reliably include all factors that may determine a person's life expectancy or cancer risk, such as family history, physical functioning, severity of comorbid conditions, or laboratory values.20
In conclusion, low screening mammography rates in the dialysis population do not mean that women receiving dialysis are receiving poor-quality care. Only 25% of the women in our chronic dialysis cohort lived longer than 5 years, and survival has not significantly improved since the late 1990s.6 Therefore, screening mammography rates should not exceed 25% in this population and probably should be much lower given the uncertain benefit of mammography in women undergoing dialysis and the increased likelihood of harm. Rather than developing policies or performance measures to increase overall screening mammography rates, tools should be designed to better target screening to the small number of women who have substantial life expectancies while avoiding screening in most of the women undergoing dialysis who are more likely to be harmed than to benefit.
Correspondence: Louise C. Walter, MD, Division of Geriatrics, San Francisco VA Medical Center, 4150 Clement St, Mailcode 181G, San Francisco, CA 94121 (Louise.Walter@ucsf.edu).
Accepted for Publication: February 19, 2006.
Author Contributions: Dr Walter had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Financial Disclosure: None.
Funding/Support: This study was supported by a Veterans Affairs Career Development Award in Health Services Research and Development (Drs Walter and O’Hare) and grant DK-56182 from the National Institute of Diabetes and Digestive and Kidney Diseases (Dr Johansen).
Disclaimer: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, and approval of the manuscript. The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs.
Thank you for submitting a comment on this article. It will be reviewed by JAMA Internal Medicine editors. You will be notified when your comment has been published. Comments should not exceed 500 words of text and 10 references.
Do not submit personal medical questions or information that could identify a specific patient, questions about a particular case, or general inquiries to an author. Only content that has not been published, posted, or submitted elsewhere should be submitted. By submitting this Comment, you and any coauthors transfer copyright to the journal if your Comment is posted.
* = Required Field
Disclosure of Any Conflicts of Interest*
Indicate all relevant conflicts of interest of each author below, including all relevant financial interests, activities, and relationships within the past 3 years including, but not limited to, employment, affiliation, grants or funding, consultancies, honoraria or payment, speakers’ bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued. If all authors have none, check "No potential conflicts or relevant financial interests" in the box below. Please also indicate any funding received in support of this work. The information will be posted with your response.
Some tools below are only available to our subscribers or users with an online account.
Download citation file:
Web of Science® Times Cited: 11
Customize your page view by dragging & repositioning the boxes below.
Enter your username and email address. We'll send you a link to reset your password.
Enter your username and email address. We'll send instructions on how to reset your password to the email address we have on record.
Athens and Shibboleth are access management services that provide single sign-on to protected resources. They replace the multiple user names and passwords necessary to access subscription-based content with a single user name and password that can be entered once per session. It operates independently of a user's location or IP address. If your institution uses Athens or Shibboleth authentication, please contact your site administrator to receive your user name and password.