- Email: [email protected]
Patients’ Barriers to Receipt of Cancer Care, and Factors Associated With Needing More Assistance From a Patient Navigator
o
r
i
g
i
n
a
l
c
o
m
m
u
n
i
c
a
t
i
o
n
Patients’ Barriers to Receipt of Cancer Care, and Factors Associated With Needing More Assistance From a Patient Navigator Samantha Hendren, MD, MPH; Nancy Chin, PhD, MPH; Susan Fisher, PhD; Paul Winters, MS; Jennifer Griggs, MD, MPH; Supriya Mohile, MD; Kevin Fiscella, MD, MPH
Funding/Support: This research was supported by National Cancer Institute grant U01 CA116924-01 awarded to Dr Fiscella, principal investigator. Background: Racial minorities have poorer cancer survival in the United States compared to whites. The purpose of this study was to better understand patients’ barriers to cancer care and to determine which patients have a greater need for assistance from a patient navigator. Methods: Community health workers assisted newly-diagnosed breast and colorectal cancer patients during a randomized trial of patient navigation and collected information about patients’ barriers. Barriers to care were characterized and compared between non-Hispanic white and minority patients. A multivariate model was constructed of factors associated with increased log navigation time, a measure of patients’ need for assistance. Results: Patients’ (n = 103) most commonly identified barriers to care included a lack of social support, insurance/financial concerns, and problems communicating with health care providers. Barriers differed between nonminority and minority patients, and minority patients faced a greater number of barriers (p = .0001). In univariate analysis, log navigation time was associated with race/ethnicity, education, income, employment, insurance type, health literacy, marital status, language, and comorbidity. A multivariate model (R2 = 0.43) for log navigation time was created using stepwise selection, and included the following factors: minority race/ethnicity (p = .032), non–full-time employment (p = .0004), unmarried status (p = .085), university center (p = .0005), and months in study (p < .0001). Conclusions: Newly diagnosed cancer patients’ most common barriers to care include lack of social support, insurance/financial concerns, and problems with health care communications. In this sample of patients, a greater need for assistance was independently associated with minority race/ethnicity and unemployment. These data may help in the design and targeting of interventions to reduce cancer health disparities.
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
Keywords: health disparities n breast cancer n colorectal n cancer n minority health n barriers J Natl Med Assoc. 2011;103:701-710 Author Affiliations: Department of Surgery(Dr Hendren), Department of Medicine, Medical Oncology (Dr Griggs), University of Michigan, Ann Arbor, Michigan; Departments of Community and Preventive Medicine (Dr Chin, Fisher, and Fiscella) and Family Medicine (Mr Winters and Dr Fiscella), Department of Medicine, Medical Oncology (Dr Mohile), and Wilmot Cancer Center (Dr Fiscella), University of Rochester, New York. Correspondence: Samantha Hendren, MD, MPH, University of Michigan, General Surgery, 2124 Taubman Ctr, 1500 E. Medical Center Dr, SPC 5343, Ann Arbor, MI 48109 ([email protected]).
Introduction
I
n the United States, cancer outcome disparities by race and/or socioeconomic status (SES) are well documented for breast cancer, colorectal cancer, prostate cancer, and lung cancer.1,2 Eliminating disparities is a major goal3 of Healthy People 2010. However, understanding the reasons for cancer disparities is important for the design of interventions to address the problem. While later stage at diagnosis due to screening inequities or lack of access to care contributes to disparities, disparities persist even after correcting for stage.4-8 This finding raises the question of disparities in cancer treatment. Using cancer registry data, Gross et al documented that “processes of cancer care” after diagnosis differ between black and white cancer patients.9 Furthermore, these process and outcome disparities have not improved over time.9 Why do traditionally disadvantaged patients receive different care for cancer in the United States? Our group has proposed a theoretical model to explain how “barriers to care” faced by minority or low-SES patients can lead to worse health outcomes (Figure 1). This model shows how patients’ barriers may affect processes and outcomes of cancer care, such as patient adherence, the doctor-patient relationship, and timely receipt of care. However, empirical research studies to prospectively document the range and number of VOL. 103, NO. 8, AUGUST 2011 701
Patients’ Barriers to Cancer Care
barriers faced by minority or poor cancer patients have not been performed. Instead, prior studies have focused on individual barriers. These include financial barriers,10,11 social barriers,12-15 communications barriers,16-18 logistical barriers, and medical comorbidities.19 Patient navigation is one of the few interventions that can potentially address the multiple barriers faced by disadvantaged cancer patients. However, to design patient navigation programs and train navigators, further characterization is needed of the range and frequency of specific barriers faced by cancer patients. There is also a need to identify which patients have a greater need for the types of assistance provided by patient navigation programs. The first aim of this study was to describe the barriers to health care faced by a diverse group of newly diagnosed breast and colorectal cancer patients. These data were prospectively collected by community health workers who had in-depth knowledge of patients. The second aim was to compare these barriers between non-Hispanic white and minority patients. Because race/ethnicity is associated with other demographic features such as income and education, these demographic features were also compared between non-Hispanic white and minority
patients. The final aim of the study was to construct a model to predict which patients have the greatest need for patient navigation. These results may provide information that will assist in the rational design of future interventions to address cancer health disparities.
Methods Data Source The present study utilizes data prospectively collected during a randomized trial of patient navigation conducted through the National Cancer Institute– sponsored Patient Navigation Research Program.20 Patient navigation is an intervention in which trained individuals—in this case, community health workers (CHWs)—assist cancer patients.21,22 CHWs help patients with appointment reminders, coordination of care, insurance paperwork, logistical support, social support, and coaching to promote effective communication with medical providers. Consecutive cancer patients who were assigned to the patient navigation arm of the study were the patient sample for the present study. Patients recruited to participate in the patient
Figure 1. Theoretical Model of Cancer Health Disparities PATIENT CHARACTERISTICS
PATIENT ACCESS BARRIERS
Age*
PROVIDER FACTORS
Competing demands (patient co-morbidity
PRACTICE ORGANIZATION FACTORS
Protocols for follow -up on abnormal cancer screening*
Bias Gender*
Financial*
Race/ethnicity*
Cancer treatment protocols*
Health insurance*
Health literacy* Transportation*
Knowledge
Cultural competency
Use patient reminders*
Childcare* Socioeconomic Status*
Knowledge/beliefs* Language*
Co-morbidity*
*
Activation* (agenda settings, questionasking, assertiveness, clarity of symptoms)
Use of registry for tracking*
PATIENT NAVIGATION ACTIVATION
Systems to promote patient selfmanagement*
Links to community resources*
PATIENT PROCESS
Patient adherence* Timely receipt of care* Quality of cancer treatment*
PATIENT OUTCOMES
Quality of life* Satisfaction* Survival*
Indicates factors potentially affected by patient navigation-activation.
702 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
VOL. 103, NO. 8, AUGUST 2011
Patients’ Barriers to Cancer Care
navigation study were newly diagnosed breast and colorectal cancer patients. Patients were recruited from all Rochester, New York–area cancer centers and from some primary care practices. There were no exclusion criteria based upon socioeconomic status, race, or insurance, but institutionalized patients and those with dementia or prior cancer were excluded. The study wasapproved by all participating institutions’ review boards, and all participants provided written informed consent.
Data Collection Patient information was prospectively collected by research assistants and CHWs during patient navigation activities. Patient demographic information was obtained by patient self-report. Cancer stage and treatment were abstracted from medical charts by cancer physicians. Information on barriers to care was collected by CHWs through semistructured interviews with patients. A standardized form was used, and the types of barriers included on the form are listed in Figure 2. The category
Figure 2. Barriers Faced by Cancer Patients
Insurance includes health insurance problems, being uninsured, being underinsured, or having high copayments. Comoridity includes medical and mental health. System problem means difficulty scheduling care, poor coordination of care, or other obstacles related to providers’ practices. Medical communication refers to communication problems between patients and providers. Perceptions/beliefs refers to beliefs about tests or treatment that may be a barrier to accepting those tests or treatments.
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
VOL. 103, NO. 8, AUGUST 2011 703
Patients’ Barriers to Cancer Care
of “other” was coded when time was spent on behalf of the patients but a specific barrier on the list was not linked to the time (for example, time for the CHW to gather information in response to a patient’s question).
Study Setting The current study took place in Rochester, New York, where medical care is provided by 1 large, academic medical center and 3 smaller hospitals, as well as a
Table 1. Patient Characteristics
Age at time of enrollment, y (mean [SD])
All (n = 103)
White/ NonHispanic (n = 63)
Minority (n = 40)
p Value
55 (12)
56 (12)
53 (13)
.1
56 (89)
37 (93)
.7
52 (83) 11 (18)
35 (88) 5 (13)
.5
1 25 17 11 4 5
(2) (40) (27) (18) (6) (8)
4 (10) 15 (38) 9 (23) 9 (23) 1 (3) 2 (3)
.5
40 15 2 4 2
(64) (24) (3) (6) (3)
14 (35) 9 (23) 12 (30) 4 (10) 1 (3)
.01
0 (0) 2 (3) 15 (24) 17 (27) 8 (13) 14 (22) 7 (11) 20.2 (1)
4 (10) 12 (31) 8 (21) 5 (13) 3 (8) 5 (13) 2 (5) 16.0 (6)
.0001
.0004
63 (100)
35 (88)
.008
3 (5) 9 (14) 7 (11) 6 (10) 2 (3) 24 (38) 12 (19) 46 696 (13,553) 37 (59)
12 (30) 6 (15) 5 (13) 3 (8) 3 (8) 1 (3) 10 (25) 31 864 (12,764) 17 (43)
.02
25 (40) 10 (16) 28 (44) 0.9 (1.0) 33 (52) 7.8 (4)
9 (23) 1 (3) 30 (75) 1.4 (1.2) 23 (58) 7.5 (4)
Gender, No. (%) 93 (90) Female Diagnosis Breast cancer 87 (85) Colorectal cancer 16 (16) Stage at diagnosis 0 5 (5) 1 40 (39) 2 26 (25) 3 20 (19) 4 5 (5) Missing 7 (7) Insurance type Private 54 (52) Medicare 24 (23) Medicaid 14 (14) None 8 (8) Other 3 (3) Education (self-reported) ≤8th grade 4 (4) Some high school 14 (14) High school diploma or equivalent 23 (23) Some college 22 (22) Associate degree 11 (11) College degree 19 (19) Graduate/professional 9 (9) Health Literacy Scale (REALM), mean (SD) 18.7 (4) Language spoken at home, No. (%) English 98 (95) Household income (self-reported) <$10 000 15 (15) $10 000-$19 999 15 (15) $20 000-$29 000 12 (12) $30 000-$39 000 9 (9) $40 000-$49 000 5 (5) ≥$50 000 25 (24) Missing 22 (21) Median household income by zip code (1999), US$ (mean 40 968 (15 054) [SD]) Marital status, No. (%) 54 (52) Married/living as married Employment Full-time 34 (33) Part-time 11 (11) Not employed 58 (56) Charleson Comorbidity Index, mean (SD) 1.1 (1.1) Treatment center, No. (%) at university 56 (54) No. of mo in study, mean (SD) 7.7 (4) No. of treatment modalities, categorical (1-4)
704 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
<.0001 .1 .006 .03 .6 .7 .3
VOL. 103, NO. 8, AUGUST 2011
Patients’ Barriers to Cancer Care
variety of private practices providing outpatient care. The cancer treatment centers in Rochester serve all of Monroe County, and county demographics are similar to the demographics of the US population.23
Variable Definitions Most variables are defined in Table 1; however, detailed variable definitions are given here for selected variables. Race/ethnicity was derived from patient selfreport and was dichotomized as non-Hispanic white (n = 63) vs minority (n = 40). The minority category for race/ ethnicity included: black/non-Hispanic (n = 25), Hispanic (n = 10), and other (n = 5). Health literacy was measured using the Rapid Assessment of Adult Literacy in Medicine (REALM, short version).24,25 The range of scores was 0 to 21, which was the best. Because of a large number of missing values for self-reported income, the 1999 median income for the zip code of patient residence was used in place of self-report income.26 Comorbidity was self-reported based on the Charleson Comorbidity Index.27
Navigation Time CHWs recorded daily the total amount of time spent with each patient and time spent addressing patient barriers. Time spent “just waiting” at medical offices prior to appointments was deducted from time estimates. The total time recorded by CHWs was summed to produce the variable navigation time, a measure of intensity of patient navigation. Because navigation time was not normally distributed, the variable was log transformed (log navigation time), yielding a normal distribution.
Data Analysis Patient sample characteristics and barriers were summarized with descriptive statistics. Associations between continuous variables and race/ethnicity were tested with t tests. Dichotomous factors were tested for association with race/ethnicity using c2 or Fisher exact tests, as appropriate. Categorical variables were compared between categories of race/ethnicity using Wald c2 tests or Cochran-Mantel-Haenszel statistics. A series of bivariate analyses were conducted to test for associations between patient/treatment factors and the dependent variable, log navigation time. The t tests were used for dichotomous variable, Kruskal-Wallis tests were used for ordinal variables, and simple linear regression was used for continuous variables. All p values were 2 sided, and statistical significance was defined as p ≤ .05. To construct a model of factors associated with increased need for patient navigation, log navigation time was used as the dependent variable in multiple linear regression with stepwise selection. Independent variables associated with log navigation time on univariate analysis were included in the stepwise selection. The criteria for selection were: p ≤ .25 for model entry and p ≤ .10 to remain in model. To interpret the results of the multivariate analysis, ß coefficients were exponentiated to return to the outcome navigation time. This gives a ratio of the geometric means of the nonbase value over the base value.28 Mean Navigation Times predicted from the model are shown in Figure 3. These were derived by multiplying the intercept value by the exp(beta) value for each factor to show how much each individual factor would be predicted to
Figure 3. Mean Navigation Timesa From Model 400 350
Navigation Time
300 250 200 150 100 50 0 No Risk Factor a
Nonwhite
Unmarried
Unemployed
Units for navigation time = minutes.
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
VOL. 103, NO. 8, AUGUST 2011 705
Patients’ Barriers to Cancer Care
change the mean navigation time compared to having no risk factors. Software programs used in this study were Excel (Microsoft Corp, Redmond, Washington) and SAS version 9.1 (SAS Institute Inc, Cary, North Carolina).
Results Characteristics of the Patient Sample Characteristics of the patient sample are presented in Table 1. The majority of patients were women with early-stage breast cancer, and 40 of 103 patients in the sample (39%) were minority. The patients in the sample had lower income than the state average—particularly amongst minority patients. While age, gender, cancer type, and cancer stage were similar between white and minority patients, SES factors and comorbidity score differed significantly by race/ethnicity.
Barriers to Cancer Care Figure 2 shows the barriers to cancer care faced by patients. The most common were problems with medical communication, a lack of social support, and medical insurance concerns, each of which affected more than half of the patients. Examples include patients’ reluctance to ask questions or share problems with the medical team, having no one to accompany them to treatments or appointments, and having difficulty completing
or understanding health insurance paperwork. Other common barriers were financial problems, medical or mental health comorbidities, and transportation. In addition, patients’ fears, perceptions or beliefs about tests or treatments, and attitudes toward providers were commonly identified barriers to care. Figure 2 also includes estimates of time spent by CHWs addressing various barriers, indicating that certain barriers were time intensive, such as transportation problems, housing problems, and arranging for interpreters. Of note, these times exclude any face-to-face time between CHW and patient. Table 2 shows the differences in barriers between non-Hispanic white and minority patients in the study. The overall number of barriers was significantly greater for minority patients (mean, 7 vs 5), and certain barriers were also more common. Those included financial problems, medical comorbidities, transportation, perceptions, and beliefs about tests or treatments, language barriers, and childcare issues. The total navigation time is also shown in Table 2 and is significantly greater for minority than for nonminority patients.
Need for Patient Navigation Table 3 gives the results of univariate analysis of patient and study factors that may influence the need for patient navigation as represented by log navigation time. Univariate analysis reveals that log navigation time is
Table 2. Barriers to Care by Race/Ethnicity All (n = 103) No. (%) Communication coaching Social/practical support Insurance (uninsured, underinsured, high copay) Financial problems Medical and mental health comorbidity Fear Communication concerns with medical personnel Transportation System problem with scheduling care Perceptions/belief about tests/treatment Proactive navigation needed Language/interpreter Attitudes towards providers Location of health care facility Patient disability Housing Employment Issues Childcare issues Adult care Literacy No. of barriers per patient, mean (SD) Navigation time (minutes, mean [SD])
65 64 56 41 40 37 36 33 30 21 19 9 9 8 6 5 5 4 3 2 5.8 1489
(63) (62) (54) (40) (39) (36) (35) (32) (29) (20) (18) (9) (9) (8) (6) (5) (5) (4) (3) (2) (3.2) (2036)
White, Non-Hispanic (n = 63) No. (%)
Minority (n = 40) No. (%)
p Valuea
38 (60) 35 (56) 32 (51) 19 (30) 19 (30) 18 (29) 19 (30) 13 (21) 14 (22) 8 (13) 8 (13) 2 (3) 3 (5) 6 (10) 3 (5) 2 (3) 3 (5) 0 (0) 1 (2) 1 (2) 4.9 (2.7) 1084 (1658)
27 (68) 29 (73) 24 (60) 22 (55) 21 (53) 19 (48) 17 (43) 20 (50) 16 (40) 13 (33) 11 (28) 7 (18) 6 (15) 2 (5) 3 (8) 3 (8) 2 (5) 4 (10) 2 (5) 1 (3) 7.2 (3.3) 2127 (2406)
0.5 0.08 0.4 0.01 0.02 0.051 0.2 0.002 0.05 0.02 0.06 0.03 0.09 0.5 0.7 0.4 1.0 0.02 0.6 1.0 0.0001 0.02
* c2 tests, Fisher exact tests, and t tests used as appropriate.
706 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
VOL. 103, NO. 8, AUGUST 2011
Patients’ Barriers to Cancer Care
associated with race/ethnicity, insurance type, education, health literacy, language, income, marital status, employment, and comorbidity score. Duration of time in study, enrollment date, treatment center, and CHW identification number also were associated with log navigation time in univariate analysis. To ensure that the number of months in study did not bias the univariate results, a linear regression of each variable in Table 3 against log navigation time was
performed, correcting for the number of months in study. The distribution of statistically significant and insignificant variables did not change except for the date of enrollment, which became nonsignificant. All other significant variables in Table 3 remained significantly associated with log navigation time, after correcting for months in study (data not shown). The final aim of the present study was to construct a model of factors associated with greater need for patient
Table 3. Univariate Analysis of Factors Potentially Associated With Log Navigation Time Factors Age Gender Cancer type Stage at diagnosis No. of treatment modalities Race/ethnicity White, non-Hispanic Minority Insurance type None Other Medicaid Medicare Private Education ≤8th grade Some high school High school diploma or equivalent Some college Associate degree College degree Graduate/professional Health Literacy Scale Language English Other Household income Marital status Married/living as married Other Employment Not employed Part-time Full-time Comorbidity score Treatment center University Community No. of mo in study Date of enrollment Community health worker identification number 1 2 3 4 5 a
Log Navigation Time (Mean [SD]) or Parameter Estimatea
p Value .3 .9 .9 .1 .5
6.2 (1.2) 7.1 (1.3)
.002
6.8 6.3 7.6 6.8 6.2
(1.4) (2.5) (0.9) (1.2) (1.3)
.004
7.3 (0.2) 7.7 (1.1) 6.2 (1.0) 6.3 (1.5) 6.4 (2.0) 6.4 (0.9) 6.3 (1.5) -0.07
.02
6.5 (1.3) 7.9 (0.5) -0.00003
.02
.02
.001
6.3 (1.4) 6.8 (1.2)
.03
7.0 (1.2) 5.9 (1.1) 5.9 (1.3) +0.4
<.0001
6.9 (1.3) 6.1 (1.3) +0.1 -0.002
.003
7.1 6.7 6.2 7.9 5.7
(1.5) (1.4) (1.0) (0.1) (1.2)
.0009
.0004 .005 .006
For continuous variables.
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
VOL. 103, NO. 8, AUGUST 2011 707
Patients’ Barriers to Cancer Care
navigation. All factors with p ≤ .10 in Table 4 were included in stepwise selection with log navigation time as the dependent variable. The results are given in Table 4. A model including race/ethnicity, marital status, employment, treatment center, and duration of time in the study predicted 43% of the variation in log navigation time in our patient sample. To further interpret these results, the ß coefficients were exponentiated to reflect navigation time. Mean navigation times predicted from the model are shown in Figure 3. This shows that in the model derived from our sample of patients, being unemployed increases navigation time by 157%, minority status increases it by 66%, and being unmarried increases it by 46%.
Discussion
The present study documents in detail the number and types of barriers to care faced by a group of newly diagnosed cancer patients. We found that patients faced a wide variety of social and instrumental barriers. Furthermore, the number and types of barriers differed by race/ethnicity. Certain barriers required more timeintensive interventions to address. Notably, intensity of patient navigation was associated with race/ethnicity and other social factors. The CHWs who prospectively collected these data had in-depth knowledge of the patients; therefore, this study may provide some insights lacking when administrative data or chart review are the primary data sources. In general, difficulty with medical communications, a lack of social support, and problems with health insurance were the most common barriers to newly diagnosed cancer patients. The importance of patientcentered communication is well recognized in cancer care,17 and the present study reinforces that communication difficulties affect patients of diverse social backgrounds. Insurance problems were also common, and Table 4. Multivariate Model for Log Navigation Timea Variables Race/ethnicity Minority Marital status Unmarried Employment Unemployed Part-time Treatment center Not at university No. of mo in study Continuous
Parameter Estimate (ß)b
Exp (ß) p Valueb
0.51
1.659
.0321
0.38
1.461
.085
0.95 0.22
2.573 1.251
.0004
-0.79
0.455
.0005
0.13
1.142
<.0001
Parameter estimate reflects regression equation for logtransformed navigation time.
a
b Multiple linear regression with stepwise selection, n = 95. Model R2 = 0.43.
708 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
these ranged from difficulty understanding paperwork to a complete lack of health care insurance. Published studies confirm the association of insurance status with cancer health disparities.29 However, the existing literature may not adequately emphasize the role of social support in cancer treatment; this was found to be a common barrier in the present study. A lack of social support may affect cancer care at every level, from health care communications (presence of a third person), provision of transportation and assistance with financial pressures, to emotional reassurance and support for treatment adherence. Other common barriers for all patients in the present study included financial problems, comorbidity, transportation, patients’ fears/perceptions/beliefs, attitudes toward providers, and medical system barriers. The finding that patients needed help across a range of barriers underscores the complexity of navigating cancer patients. These results suggest that patient navigators require broad training in communications coaching; providing social support; and instrumental issues, including health insurance, transportation, and others. Another finding of the present study is that minority race/ethnicity was associated with a greater number and different spectrum of barriers to cancer care. These barriers may help to explain the mechanisms of observed treatment disparities. For example, patients facing a multitude of barriers may have financial, medical, or other reasons to miss appointments. They may misinterpret treatment plans or mistrust providers,30,31 leading to further delays in care.32 Patients with complex barriers may even receive different treatment recommendations from providers who perceive that patients cannot “handle” intensive treatment due to their situations.5 Consistent with prior studies, race/ethnicity was also associated in this study with lower income, more comorbid illness, less private insurance coverage, lower educational level, lower health literacy, and unemployment.33,34 Given these findings and the sample size, it is difficult to disentangle the effects of low socioeconomic status from race/ethnicity. Prior research findings vary as to what proportion of observed racial disparities can be explained by socioeconomic factors.19,35 Finally, this study constructed a preliminary model of factors associated with the increased need for patient navigation assistance. The final model derived from stepwise selection included: minority race/ethnicity, being unmarried, unemployment, treatment at the university cancer center, and time in the study (reflecting the duration of active cancer treatment). These factors explain 43% of the variation in log navigation time in this sample of patients. As discussed above, race/ethnicity is a marker for social disadvantage and may represent a constellation of social risk factors.36 Nevertheless, minority status provides a useful tool for targeting interventions to help cancer patients. From a practical perspective, it means that VOL. 103, NO. 8, AUGUST 2011
Patients’ Barriers to Cancer Care
high-need patients can possibly be identified based on a few, baseline sociodemographic factors. While the present study provides in-depth, prospective information about the barriers confronting some newly diagnosed cancer patients, several limitations should be recognized. The outcome variable, navigation time, could theoretically be affected by unmeasured factors, including the personality of the patient or CHW, and the navigator’s level of experience. Nevertheless, the time expenditure of the CHW for each patient was an important indicator of patient navigation intensity and cost in our program. Furthermore, the patient sample for this study was predominantly female, due to the much larger proportion of breast cancer patients than colorectal cancer patients enrolled in the study. Because approximately 90% of the patients were female, the applicability of these findings to male cancer patients is uncertain. Also, the generalizability of these findings to cancer patients across the United States cannot be assumed. It is possible that these cancer patients and/or the local health care system may have unique characteristics that make application of these data to other communities problematic. Despite these limitations, the present study has important health services research implications. Presently, the National Cancer Institute, the American Cancer Society, and cancer centers are devoting considerable resources to the design of interventions to address the problem of cancer health disparities. Patient navigation programs and other interventions seek to level the playing field for traditionally underserved cancer patients, and many of these interventions have already been put in place. However, the design of these programs has relied largely on theoretical models of cancer care and on population-based analyses of risk factors. However, patient-level data on the specific barriers that cancer patients face are lacking. More detailed knowledge of patients’ barriers to care is provided by the present study. Our results highlight the fact that cancer patients’ barriers to optimal care are more complex than insurance coverage problems or transportation to appointments. Social support; medical communications; comorbid illnesses; and fears, attitudes, and beliefs also may affect timeliness and quality of care. Patient navigation programs may be particularly well suited to address these diverse barriers, but practitioners providing patient navigation may require additional training. Also, programs may need to focus more resources on the highest-risk patients because this service is resource intensive. The present study provides a preliminary model to target patient navigation to cancer patients likely to need more help.
Conclusions
The present study shows that newly diagnosed cancer patients’ most common barriers to care include
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
communications barriers, lack of social support, and health insurance concerns. Number and types of barriers differ significantly between non-Hispanic white and minority patients. A multivariate model that predicts high need for patient navigation in our patient sample includes minority race/ethnicity, unemployment, and unmarried status. These data may help in the design and targeting of future interventions to reduce cancer health disparities.
Acknowledgments
The authors would like to thank Starlene Loader, RN, and Sally Rousseau, MSW, of the Rochester Patient Navigation Research Program program for their assistance in obtaining and accurately interpreting the data for this study. We would also like to acknowledge the patient navigators, research assistants, and investigators of the Rochester Patient Navigation Research Program program.
References
1. Jemal A, Clegg LX, Ward E, et al. Annual report to the nation on the status of cancer, 1975-2001, with a special feature regarding survival. Cancer. 2004;101:3-27. 2. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2006. CA Cancer J Clin. 2006;56:106-130. 3. Healthy People 2010: Understanding and Improving Health. Washington, DC: Government Printing Office; 2000. US Department of Health and Human Services. 4. Le H, Ziogas A, Lipkin SM, Zell JA. Effects of socioeconomic status and treatment disparities in colorectal cancer survival. Cancer Epidemiol Biomarkers Prev. 2008;17:1950-1962. 5. Morris AM, Billingsley KG, Hayanga AJ, Matthews B, Baldwin LM, Birkmeyer JD. Residual treatment disparities after oncology referral for rectal cancer. J Natl Cancer Inst. 2008;100:738-744. 6. Morris AM, Billingsley KG, Baxter NN, Baldwin LM. Racial disparities in rectal cancer treatment: a population-based analysis. Arch Surg. 2004;139:151-155. 7. Shavers VL, Brown ML. Racial and ethnic disparities in the receipt of cancer treatment. J Nat Cancer Inst. 2002;94:334-357. 8. Freeman HP, Wasfie TJ. Cancer of the breast in poor black women. Cancer. 1989;63:2562-2569. 9. Gross CP, Smith BD, Wolf E, Andersen M. Racial disparities in cancer therapy: did the gap narrow between 1992 and 2002? Cancer. 2008;112:900-908. 10. Bach PB, Pham HH, Schrag D, Tate RC, Hargraves JL. Primary care physicians who treat blacks and whites. N Engl J Med. 2004;351:575-584. 11. Ward E, Halpern M, Schrag N, et al. Association of insurance with cancer care utilization and outcomes. CA Cancer J Clin. 2008;58:9-31. 12. Allen JD, Shelton RC, Harden E, Goldman RE. Follow-up of abnormal screening mammograms among low-income ethnically diverse women: findings from a qualitative study. Patient Educ Couns. 2008;72:283-292. 13. Garbers S, Jessop DJ, Foti H, Uribelarrea M, Chiasson MA. Barriers to breast cancer screening for low-income Mexican and Dominican women in New York City. J Urban Health. 2003;80:81-91. 14. Hart CL, Smith GD, Blane D. Inequalities in mortality by social class measured at 3 stages of the lifecourse. Am J Public Health. 1998;88:471-474. 15. Mandelblatt JS, Yabroff KR, Kerner JF. Equitable access to cancer services: A review of barriers to quality care. Cancer .1999;86:2378-2390. 16. Ashton CM, Haidet P, Paterniti DA, et al. Racial and ethnic disparities in the use of health services: bias, preferences, or poor communication? J Gen Intern Med. 2003;18:146-152. 17. Epstein RM. Patient-Centered Communication in Cancer Care: Promoting Healing and Reducing Suffering. Bethesda, MD: National Cancer Institute; 2007.
VOL. 103, NO. 8, AUGUST 2011 709
Patients’ Barriers to Cancer Care 18. Hack TF, Degner LF, Parker PA. The communication goals and needs of cancer patients: a review. Psychooncology. 2005;14:831-845. 19. Esnaola NF, Hall BL, Hosokawa PW, et al. Race and surgical outcomes: it is not all black and white. Ann Surg. 2008;248:647-655. 20. Freund KM, Battaglia TA, Calhoun E, et al. National Cancer Institute Patient Navigation Research Program: methods, protocol, and measures. Cancer. 2008;113:3391-3399. 21. Wells KJ, Battaglia TA, Dudley DJ, et al. Patient navigation: state of the art or is it science? Cancer. 2008;113:1999-2010. 22. Freeman HP. Patient navigation: a community based strategy to reduce cancer disparities. J Urban Health. 2006;83:139-141. 23. US Census Bureau. 2006-2008 American Community Survey: Monroe County, NY. http://factfinder.census.gov/servlet/ACSSAFFFacts?_event=Search&geo_id=01000US&_geoContext=&_ street=&_county=monroe+county&_cityTown=monroe+county&_ state=04000US36&_zip=&_lang=en&_sse=on&ActiveGeoDiv=geoSelect&_ useEV=&pctxt=fph&pgsl=010&_submenuId=factsheet_1&ds_ name=ACS_2008_3YR_SAFF&_ci_nbr=null&qr_name=null®=null%3Anull&_ keyword=&_industry=. Accessed October 10, 2010. 24. Davis TC, Long SW, Jackson RH, et al. Rapid estimate of adult literacy in medicine: a shortened screening instrument. Fam Med. 1993;25:391-395. 25. Davis TC, Crouch MA, Long SW, et al. Rapid assessment of literacy levels of adult primary care patients. Fam Med. 1991;23:433-435. 26. Krieger N, Chen JT, Waterman PD, Rehkopf DH, Subramanian SV. Race/ ethnicity, gender, and monitoring socioeconomic gradients in health: a comparison of area-based socioeconomic measures—the public health disparities geocoding project. Am J Public Health. 2003;93:1655-1671. 27. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of clas-
710 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
sifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373-383. 28. UCLA Academic Technology Services. How do I interpret a regression model when some variables are log transformed? www.ats.ucla.edu/stat/ mult_pkg/faq/general/log_transformed_regression.htm. Accessed March 23, 2009. 29. Halpern MT, Ward EM, Pavluck AL, Schrag NM, Bian J, Chen AY. Association of insurance status and ethnicity with cancer stage at diagnosis for 12 cancer sites: a retrospective analysis. Lancet Oncol. 2008;9:222-231. 30. Gordon HS, Street RL Jr, Sharf BF, Kelly PA, Souchek J. Racial differences in trust and lung cancer patients’ perceptions of physician communication. J Clin Oncol . 2006;24:904-909. 31. Bickell NA, Weidmann J, Fei K, Lin JJ, Leventhal H. Underuse of breast cancer adjuvant treatment: patient knowledge, beliefs, and medical mistrust. J Clin Oncol. 2009;27:5160-5167. 32. Adams SA, Smith ER, Hardin J, Prabhu-Das I, Fulton J, Hebert JR. Racial differences in follow-up of abnormal mammography findings among economically disadvantaged women. Cancer. 2009;115:5788-5797. 33. Mead H, Cartwright-Smith L, Jones K, Ramos C, Woods K, Siegel B. Racial and Ethnic Disparities in US Health Care: A Chartbook. New York: The Commonwealth Fund; 2008. 34. National Center for Health Statistics. Health, United States, 2008. Hyattsville, MD: Government Printing Office; 2009. 35. Howard G, Anderson RT, Russell G, Howard VJ, Burke GL. Race, socioeconomic status, and cause-specific mortality. Ann Epidemiol. 2000;10:214-223. 36. Fiscella K, Williams DR. Health disparities based on socioeconomic inequities: implications for urban health care. Acad Med. 2004;79:1139-1147. n
VOL. 103, NO. 8, AUGUST 2011
r
i
g
i
n
a
l
c
o
m
m
u
n
i
c
a
t
i
o
n
Patients’ Barriers to Receipt of Cancer Care, and Factors Associated With Needing More Assistance From a Patient Navigator Samantha Hendren, MD, MPH; Nancy Chin, PhD, MPH; Susan Fisher, PhD; Paul Winters, MS; Jennifer Griggs, MD, MPH; Supriya Mohile, MD; Kevin Fiscella, MD, MPH
Funding/Support: This research was supported by National Cancer Institute grant U01 CA116924-01 awarded to Dr Fiscella, principal investigator. Background: Racial minorities have poorer cancer survival in the United States compared to whites. The purpose of this study was to better understand patients’ barriers to cancer care and to determine which patients have a greater need for assistance from a patient navigator. Methods: Community health workers assisted newly-diagnosed breast and colorectal cancer patients during a randomized trial of patient navigation and collected information about patients’ barriers. Barriers to care were characterized and compared between non-Hispanic white and minority patients. A multivariate model was constructed of factors associated with increased log navigation time, a measure of patients’ need for assistance. Results: Patients’ (n = 103) most commonly identified barriers to care included a lack of social support, insurance/financial concerns, and problems communicating with health care providers. Barriers differed between nonminority and minority patients, and minority patients faced a greater number of barriers (p = .0001). In univariate analysis, log navigation time was associated with race/ethnicity, education, income, employment, insurance type, health literacy, marital status, language, and comorbidity. A multivariate model (R2 = 0.43) for log navigation time was created using stepwise selection, and included the following factors: minority race/ethnicity (p = .032), non–full-time employment (p = .0004), unmarried status (p = .085), university center (p = .0005), and months in study (p < .0001). Conclusions: Newly diagnosed cancer patients’ most common barriers to care include lack of social support, insurance/financial concerns, and problems with health care communications. In this sample of patients, a greater need for assistance was independently associated with minority race/ethnicity and unemployment. These data may help in the design and targeting of interventions to reduce cancer health disparities.
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
Keywords: health disparities n breast cancer n colorectal n cancer n minority health n barriers J Natl Med Assoc. 2011;103:701-710 Author Affiliations: Department of Surgery(Dr Hendren), Department of Medicine, Medical Oncology (Dr Griggs), University of Michigan, Ann Arbor, Michigan; Departments of Community and Preventive Medicine (Dr Chin, Fisher, and Fiscella) and Family Medicine (Mr Winters and Dr Fiscella), Department of Medicine, Medical Oncology (Dr Mohile), and Wilmot Cancer Center (Dr Fiscella), University of Rochester, New York. Correspondence: Samantha Hendren, MD, MPH, University of Michigan, General Surgery, 2124 Taubman Ctr, 1500 E. Medical Center Dr, SPC 5343, Ann Arbor, MI 48109 ([email protected]).
Introduction
I
n the United States, cancer outcome disparities by race and/or socioeconomic status (SES) are well documented for breast cancer, colorectal cancer, prostate cancer, and lung cancer.1,2 Eliminating disparities is a major goal3 of Healthy People 2010. However, understanding the reasons for cancer disparities is important for the design of interventions to address the problem. While later stage at diagnosis due to screening inequities or lack of access to care contributes to disparities, disparities persist even after correcting for stage.4-8 This finding raises the question of disparities in cancer treatment. Using cancer registry data, Gross et al documented that “processes of cancer care” after diagnosis differ between black and white cancer patients.9 Furthermore, these process and outcome disparities have not improved over time.9 Why do traditionally disadvantaged patients receive different care for cancer in the United States? Our group has proposed a theoretical model to explain how “barriers to care” faced by minority or low-SES patients can lead to worse health outcomes (Figure 1). This model shows how patients’ barriers may affect processes and outcomes of cancer care, such as patient adherence, the doctor-patient relationship, and timely receipt of care. However, empirical research studies to prospectively document the range and number of VOL. 103, NO. 8, AUGUST 2011 701
Patients’ Barriers to Cancer Care
barriers faced by minority or poor cancer patients have not been performed. Instead, prior studies have focused on individual barriers. These include financial barriers,10,11 social barriers,12-15 communications barriers,16-18 logistical barriers, and medical comorbidities.19 Patient navigation is one of the few interventions that can potentially address the multiple barriers faced by disadvantaged cancer patients. However, to design patient navigation programs and train navigators, further characterization is needed of the range and frequency of specific barriers faced by cancer patients. There is also a need to identify which patients have a greater need for the types of assistance provided by patient navigation programs. The first aim of this study was to describe the barriers to health care faced by a diverse group of newly diagnosed breast and colorectal cancer patients. These data were prospectively collected by community health workers who had in-depth knowledge of patients. The second aim was to compare these barriers between non-Hispanic white and minority patients. Because race/ethnicity is associated with other demographic features such as income and education, these demographic features were also compared between non-Hispanic white and minority
patients. The final aim of the study was to construct a model to predict which patients have the greatest need for patient navigation. These results may provide information that will assist in the rational design of future interventions to address cancer health disparities.
Methods Data Source The present study utilizes data prospectively collected during a randomized trial of patient navigation conducted through the National Cancer Institute– sponsored Patient Navigation Research Program.20 Patient navigation is an intervention in which trained individuals—in this case, community health workers (CHWs)—assist cancer patients.21,22 CHWs help patients with appointment reminders, coordination of care, insurance paperwork, logistical support, social support, and coaching to promote effective communication with medical providers. Consecutive cancer patients who were assigned to the patient navigation arm of the study were the patient sample for the present study. Patients recruited to participate in the patient
Figure 1. Theoretical Model of Cancer Health Disparities PATIENT CHARACTERISTICS
PATIENT ACCESS BARRIERS
Age*
PROVIDER FACTORS
Competing demands (patient co-morbidity
PRACTICE ORGANIZATION FACTORS
Protocols for follow -up on abnormal cancer screening*
Bias Gender*
Financial*
Race/ethnicity*
Cancer treatment protocols*
Health insurance*
Health literacy* Transportation*
Knowledge
Cultural competency
Use patient reminders*
Childcare* Socioeconomic Status*
Knowledge/beliefs* Language*
Co-morbidity*
*
Activation* (agenda settings, questionasking, assertiveness, clarity of symptoms)
Use of registry for tracking*
PATIENT NAVIGATION ACTIVATION
Systems to promote patient selfmanagement*
Links to community resources*
PATIENT PROCESS
Patient adherence* Timely receipt of care* Quality of cancer treatment*
PATIENT OUTCOMES
Quality of life* Satisfaction* Survival*
Indicates factors potentially affected by patient navigation-activation.
702 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
VOL. 103, NO. 8, AUGUST 2011
Patients’ Barriers to Cancer Care
navigation study were newly diagnosed breast and colorectal cancer patients. Patients were recruited from all Rochester, New York–area cancer centers and from some primary care practices. There were no exclusion criteria based upon socioeconomic status, race, or insurance, but institutionalized patients and those with dementia or prior cancer were excluded. The study wasapproved by all participating institutions’ review boards, and all participants provided written informed consent.
Data Collection Patient information was prospectively collected by research assistants and CHWs during patient navigation activities. Patient demographic information was obtained by patient self-report. Cancer stage and treatment were abstracted from medical charts by cancer physicians. Information on barriers to care was collected by CHWs through semistructured interviews with patients. A standardized form was used, and the types of barriers included on the form are listed in Figure 2. The category
Figure 2. Barriers Faced by Cancer Patients
Insurance includes health insurance problems, being uninsured, being underinsured, or having high copayments. Comoridity includes medical and mental health. System problem means difficulty scheduling care, poor coordination of care, or other obstacles related to providers’ practices. Medical communication refers to communication problems between patients and providers. Perceptions/beliefs refers to beliefs about tests or treatment that may be a barrier to accepting those tests or treatments.
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
VOL. 103, NO. 8, AUGUST 2011 703
Patients’ Barriers to Cancer Care
of “other” was coded when time was spent on behalf of the patients but a specific barrier on the list was not linked to the time (for example, time for the CHW to gather information in response to a patient’s question).
Study Setting The current study took place in Rochester, New York, where medical care is provided by 1 large, academic medical center and 3 smaller hospitals, as well as a
Table 1. Patient Characteristics
Age at time of enrollment, y (mean [SD])
All (n = 103)
White/ NonHispanic (n = 63)
Minority (n = 40)
p Value
55 (12)
56 (12)
53 (13)
.1
56 (89)
37 (93)
.7
52 (83) 11 (18)
35 (88) 5 (13)
.5
1 25 17 11 4 5
(2) (40) (27) (18) (6) (8)
4 (10) 15 (38) 9 (23) 9 (23) 1 (3) 2 (3)
.5
40 15 2 4 2
(64) (24) (3) (6) (3)
14 (35) 9 (23) 12 (30) 4 (10) 1 (3)
.01
0 (0) 2 (3) 15 (24) 17 (27) 8 (13) 14 (22) 7 (11) 20.2 (1)
4 (10) 12 (31) 8 (21) 5 (13) 3 (8) 5 (13) 2 (5) 16.0 (6)
.0001
.0004
63 (100)
35 (88)
.008
3 (5) 9 (14) 7 (11) 6 (10) 2 (3) 24 (38) 12 (19) 46 696 (13,553) 37 (59)
12 (30) 6 (15) 5 (13) 3 (8) 3 (8) 1 (3) 10 (25) 31 864 (12,764) 17 (43)
.02
25 (40) 10 (16) 28 (44) 0.9 (1.0) 33 (52) 7.8 (4)
9 (23) 1 (3) 30 (75) 1.4 (1.2) 23 (58) 7.5 (4)
Gender, No. (%) 93 (90) Female Diagnosis Breast cancer 87 (85) Colorectal cancer 16 (16) Stage at diagnosis 0 5 (5) 1 40 (39) 2 26 (25) 3 20 (19) 4 5 (5) Missing 7 (7) Insurance type Private 54 (52) Medicare 24 (23) Medicaid 14 (14) None 8 (8) Other 3 (3) Education (self-reported) ≤8th grade 4 (4) Some high school 14 (14) High school diploma or equivalent 23 (23) Some college 22 (22) Associate degree 11 (11) College degree 19 (19) Graduate/professional 9 (9) Health Literacy Scale (REALM), mean (SD) 18.7 (4) Language spoken at home, No. (%) English 98 (95) Household income (self-reported) <$10 000 15 (15) $10 000-$19 999 15 (15) $20 000-$29 000 12 (12) $30 000-$39 000 9 (9) $40 000-$49 000 5 (5) ≥$50 000 25 (24) Missing 22 (21) Median household income by zip code (1999), US$ (mean 40 968 (15 054) [SD]) Marital status, No. (%) 54 (52) Married/living as married Employment Full-time 34 (33) Part-time 11 (11) Not employed 58 (56) Charleson Comorbidity Index, mean (SD) 1.1 (1.1) Treatment center, No. (%) at university 56 (54) No. of mo in study, mean (SD) 7.7 (4) No. of treatment modalities, categorical (1-4)
704 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
<.0001 .1 .006 .03 .6 .7 .3
VOL. 103, NO. 8, AUGUST 2011
Patients’ Barriers to Cancer Care
variety of private practices providing outpatient care. The cancer treatment centers in Rochester serve all of Monroe County, and county demographics are similar to the demographics of the US population.23
Variable Definitions Most variables are defined in Table 1; however, detailed variable definitions are given here for selected variables. Race/ethnicity was derived from patient selfreport and was dichotomized as non-Hispanic white (n = 63) vs minority (n = 40). The minority category for race/ ethnicity included: black/non-Hispanic (n = 25), Hispanic (n = 10), and other (n = 5). Health literacy was measured using the Rapid Assessment of Adult Literacy in Medicine (REALM, short version).24,25 The range of scores was 0 to 21, which was the best. Because of a large number of missing values for self-reported income, the 1999 median income for the zip code of patient residence was used in place of self-report income.26 Comorbidity was self-reported based on the Charleson Comorbidity Index.27
Navigation Time CHWs recorded daily the total amount of time spent with each patient and time spent addressing patient barriers. Time spent “just waiting” at medical offices prior to appointments was deducted from time estimates. The total time recorded by CHWs was summed to produce the variable navigation time, a measure of intensity of patient navigation. Because navigation time was not normally distributed, the variable was log transformed (log navigation time), yielding a normal distribution.
Data Analysis Patient sample characteristics and barriers were summarized with descriptive statistics. Associations between continuous variables and race/ethnicity were tested with t tests. Dichotomous factors were tested for association with race/ethnicity using c2 or Fisher exact tests, as appropriate. Categorical variables were compared between categories of race/ethnicity using Wald c2 tests or Cochran-Mantel-Haenszel statistics. A series of bivariate analyses were conducted to test for associations between patient/treatment factors and the dependent variable, log navigation time. The t tests were used for dichotomous variable, Kruskal-Wallis tests were used for ordinal variables, and simple linear regression was used for continuous variables. All p values were 2 sided, and statistical significance was defined as p ≤ .05. To construct a model of factors associated with increased need for patient navigation, log navigation time was used as the dependent variable in multiple linear regression with stepwise selection. Independent variables associated with log navigation time on univariate analysis were included in the stepwise selection. The criteria for selection were: p ≤ .25 for model entry and p ≤ .10 to remain in model. To interpret the results of the multivariate analysis, ß coefficients were exponentiated to return to the outcome navigation time. This gives a ratio of the geometric means of the nonbase value over the base value.28 Mean Navigation Times predicted from the model are shown in Figure 3. These were derived by multiplying the intercept value by the exp(beta) value for each factor to show how much each individual factor would be predicted to
Figure 3. Mean Navigation Timesa From Model 400 350
Navigation Time
300 250 200 150 100 50 0 No Risk Factor a
Nonwhite
Unmarried
Unemployed
Units for navigation time = minutes.
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
VOL. 103, NO. 8, AUGUST 2011 705
Patients’ Barriers to Cancer Care
change the mean navigation time compared to having no risk factors. Software programs used in this study were Excel (Microsoft Corp, Redmond, Washington) and SAS version 9.1 (SAS Institute Inc, Cary, North Carolina).
Results Characteristics of the Patient Sample Characteristics of the patient sample are presented in Table 1. The majority of patients were women with early-stage breast cancer, and 40 of 103 patients in the sample (39%) were minority. The patients in the sample had lower income than the state average—particularly amongst minority patients. While age, gender, cancer type, and cancer stage were similar between white and minority patients, SES factors and comorbidity score differed significantly by race/ethnicity.
Barriers to Cancer Care Figure 2 shows the barriers to cancer care faced by patients. The most common were problems with medical communication, a lack of social support, and medical insurance concerns, each of which affected more than half of the patients. Examples include patients’ reluctance to ask questions or share problems with the medical team, having no one to accompany them to treatments or appointments, and having difficulty completing
or understanding health insurance paperwork. Other common barriers were financial problems, medical or mental health comorbidities, and transportation. In addition, patients’ fears, perceptions or beliefs about tests or treatments, and attitudes toward providers were commonly identified barriers to care. Figure 2 also includes estimates of time spent by CHWs addressing various barriers, indicating that certain barriers were time intensive, such as transportation problems, housing problems, and arranging for interpreters. Of note, these times exclude any face-to-face time between CHW and patient. Table 2 shows the differences in barriers between non-Hispanic white and minority patients in the study. The overall number of barriers was significantly greater for minority patients (mean, 7 vs 5), and certain barriers were also more common. Those included financial problems, medical comorbidities, transportation, perceptions, and beliefs about tests or treatments, language barriers, and childcare issues. The total navigation time is also shown in Table 2 and is significantly greater for minority than for nonminority patients.
Need for Patient Navigation Table 3 gives the results of univariate analysis of patient and study factors that may influence the need for patient navigation as represented by log navigation time. Univariate analysis reveals that log navigation time is
Table 2. Barriers to Care by Race/Ethnicity All (n = 103) No. (%) Communication coaching Social/practical support Insurance (uninsured, underinsured, high copay) Financial problems Medical and mental health comorbidity Fear Communication concerns with medical personnel Transportation System problem with scheduling care Perceptions/belief about tests/treatment Proactive navigation needed Language/interpreter Attitudes towards providers Location of health care facility Patient disability Housing Employment Issues Childcare issues Adult care Literacy No. of barriers per patient, mean (SD) Navigation time (minutes, mean [SD])
65 64 56 41 40 37 36 33 30 21 19 9 9 8 6 5 5 4 3 2 5.8 1489
(63) (62) (54) (40) (39) (36) (35) (32) (29) (20) (18) (9) (9) (8) (6) (5) (5) (4) (3) (2) (3.2) (2036)
White, Non-Hispanic (n = 63) No. (%)
Minority (n = 40) No. (%)
p Valuea
38 (60) 35 (56) 32 (51) 19 (30) 19 (30) 18 (29) 19 (30) 13 (21) 14 (22) 8 (13) 8 (13) 2 (3) 3 (5) 6 (10) 3 (5) 2 (3) 3 (5) 0 (0) 1 (2) 1 (2) 4.9 (2.7) 1084 (1658)
27 (68) 29 (73) 24 (60) 22 (55) 21 (53) 19 (48) 17 (43) 20 (50) 16 (40) 13 (33) 11 (28) 7 (18) 6 (15) 2 (5) 3 (8) 3 (8) 2 (5) 4 (10) 2 (5) 1 (3) 7.2 (3.3) 2127 (2406)
0.5 0.08 0.4 0.01 0.02 0.051 0.2 0.002 0.05 0.02 0.06 0.03 0.09 0.5 0.7 0.4 1.0 0.02 0.6 1.0 0.0001 0.02
* c2 tests, Fisher exact tests, and t tests used as appropriate.
706 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
VOL. 103, NO. 8, AUGUST 2011
Patients’ Barriers to Cancer Care
associated with race/ethnicity, insurance type, education, health literacy, language, income, marital status, employment, and comorbidity score. Duration of time in study, enrollment date, treatment center, and CHW identification number also were associated with log navigation time in univariate analysis. To ensure that the number of months in study did not bias the univariate results, a linear regression of each variable in Table 3 against log navigation time was
performed, correcting for the number of months in study. The distribution of statistically significant and insignificant variables did not change except for the date of enrollment, which became nonsignificant. All other significant variables in Table 3 remained significantly associated with log navigation time, after correcting for months in study (data not shown). The final aim of the present study was to construct a model of factors associated with greater need for patient
Table 3. Univariate Analysis of Factors Potentially Associated With Log Navigation Time Factors Age Gender Cancer type Stage at diagnosis No. of treatment modalities Race/ethnicity White, non-Hispanic Minority Insurance type None Other Medicaid Medicare Private Education ≤8th grade Some high school High school diploma or equivalent Some college Associate degree College degree Graduate/professional Health Literacy Scale Language English Other Household income Marital status Married/living as married Other Employment Not employed Part-time Full-time Comorbidity score Treatment center University Community No. of mo in study Date of enrollment Community health worker identification number 1 2 3 4 5 a
Log Navigation Time (Mean [SD]) or Parameter Estimatea
p Value .3 .9 .9 .1 .5
6.2 (1.2) 7.1 (1.3)
.002
6.8 6.3 7.6 6.8 6.2
(1.4) (2.5) (0.9) (1.2) (1.3)
.004
7.3 (0.2) 7.7 (1.1) 6.2 (1.0) 6.3 (1.5) 6.4 (2.0) 6.4 (0.9) 6.3 (1.5) -0.07
.02
6.5 (1.3) 7.9 (0.5) -0.00003
.02
.02
.001
6.3 (1.4) 6.8 (1.2)
.03
7.0 (1.2) 5.9 (1.1) 5.9 (1.3) +0.4
<.0001
6.9 (1.3) 6.1 (1.3) +0.1 -0.002
.003
7.1 6.7 6.2 7.9 5.7
(1.5) (1.4) (1.0) (0.1) (1.2)
.0009
.0004 .005 .006
For continuous variables.
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
VOL. 103, NO. 8, AUGUST 2011 707
Patients’ Barriers to Cancer Care
navigation. All factors with p ≤ .10 in Table 4 were included in stepwise selection with log navigation time as the dependent variable. The results are given in Table 4. A model including race/ethnicity, marital status, employment, treatment center, and duration of time in the study predicted 43% of the variation in log navigation time in our patient sample. To further interpret these results, the ß coefficients were exponentiated to reflect navigation time. Mean navigation times predicted from the model are shown in Figure 3. This shows that in the model derived from our sample of patients, being unemployed increases navigation time by 157%, minority status increases it by 66%, and being unmarried increases it by 46%.
Discussion
The present study documents in detail the number and types of barriers to care faced by a group of newly diagnosed cancer patients. We found that patients faced a wide variety of social and instrumental barriers. Furthermore, the number and types of barriers differed by race/ethnicity. Certain barriers required more timeintensive interventions to address. Notably, intensity of patient navigation was associated with race/ethnicity and other social factors. The CHWs who prospectively collected these data had in-depth knowledge of the patients; therefore, this study may provide some insights lacking when administrative data or chart review are the primary data sources. In general, difficulty with medical communications, a lack of social support, and problems with health insurance were the most common barriers to newly diagnosed cancer patients. The importance of patientcentered communication is well recognized in cancer care,17 and the present study reinforces that communication difficulties affect patients of diverse social backgrounds. Insurance problems were also common, and Table 4. Multivariate Model for Log Navigation Timea Variables Race/ethnicity Minority Marital status Unmarried Employment Unemployed Part-time Treatment center Not at university No. of mo in study Continuous
Parameter Estimate (ß)b
Exp (ß) p Valueb
0.51
1.659
.0321
0.38
1.461
.085
0.95 0.22
2.573 1.251
.0004
-0.79
0.455
.0005
0.13
1.142
<.0001
Parameter estimate reflects regression equation for logtransformed navigation time.
a
b Multiple linear regression with stepwise selection, n = 95. Model R2 = 0.43.
708 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
these ranged from difficulty understanding paperwork to a complete lack of health care insurance. Published studies confirm the association of insurance status with cancer health disparities.29 However, the existing literature may not adequately emphasize the role of social support in cancer treatment; this was found to be a common barrier in the present study. A lack of social support may affect cancer care at every level, from health care communications (presence of a third person), provision of transportation and assistance with financial pressures, to emotional reassurance and support for treatment adherence. Other common barriers for all patients in the present study included financial problems, comorbidity, transportation, patients’ fears/perceptions/beliefs, attitudes toward providers, and medical system barriers. The finding that patients needed help across a range of barriers underscores the complexity of navigating cancer patients. These results suggest that patient navigators require broad training in communications coaching; providing social support; and instrumental issues, including health insurance, transportation, and others. Another finding of the present study is that minority race/ethnicity was associated with a greater number and different spectrum of barriers to cancer care. These barriers may help to explain the mechanisms of observed treatment disparities. For example, patients facing a multitude of barriers may have financial, medical, or other reasons to miss appointments. They may misinterpret treatment plans or mistrust providers,30,31 leading to further delays in care.32 Patients with complex barriers may even receive different treatment recommendations from providers who perceive that patients cannot “handle” intensive treatment due to their situations.5 Consistent with prior studies, race/ethnicity was also associated in this study with lower income, more comorbid illness, less private insurance coverage, lower educational level, lower health literacy, and unemployment.33,34 Given these findings and the sample size, it is difficult to disentangle the effects of low socioeconomic status from race/ethnicity. Prior research findings vary as to what proportion of observed racial disparities can be explained by socioeconomic factors.19,35 Finally, this study constructed a preliminary model of factors associated with the increased need for patient navigation assistance. The final model derived from stepwise selection included: minority race/ethnicity, being unmarried, unemployment, treatment at the university cancer center, and time in the study (reflecting the duration of active cancer treatment). These factors explain 43% of the variation in log navigation time in this sample of patients. As discussed above, race/ethnicity is a marker for social disadvantage and may represent a constellation of social risk factors.36 Nevertheless, minority status provides a useful tool for targeting interventions to help cancer patients. From a practical perspective, it means that VOL. 103, NO. 8, AUGUST 2011
Patients’ Barriers to Cancer Care
high-need patients can possibly be identified based on a few, baseline sociodemographic factors. While the present study provides in-depth, prospective information about the barriers confronting some newly diagnosed cancer patients, several limitations should be recognized. The outcome variable, navigation time, could theoretically be affected by unmeasured factors, including the personality of the patient or CHW, and the navigator’s level of experience. Nevertheless, the time expenditure of the CHW for each patient was an important indicator of patient navigation intensity and cost in our program. Furthermore, the patient sample for this study was predominantly female, due to the much larger proportion of breast cancer patients than colorectal cancer patients enrolled in the study. Because approximately 90% of the patients were female, the applicability of these findings to male cancer patients is uncertain. Also, the generalizability of these findings to cancer patients across the United States cannot be assumed. It is possible that these cancer patients and/or the local health care system may have unique characteristics that make application of these data to other communities problematic. Despite these limitations, the present study has important health services research implications. Presently, the National Cancer Institute, the American Cancer Society, and cancer centers are devoting considerable resources to the design of interventions to address the problem of cancer health disparities. Patient navigation programs and other interventions seek to level the playing field for traditionally underserved cancer patients, and many of these interventions have already been put in place. However, the design of these programs has relied largely on theoretical models of cancer care and on population-based analyses of risk factors. However, patient-level data on the specific barriers that cancer patients face are lacking. More detailed knowledge of patients’ barriers to care is provided by the present study. Our results highlight the fact that cancer patients’ barriers to optimal care are more complex than insurance coverage problems or transportation to appointments. Social support; medical communications; comorbid illnesses; and fears, attitudes, and beliefs also may affect timeliness and quality of care. Patient navigation programs may be particularly well suited to address these diverse barriers, but practitioners providing patient navigation may require additional training. Also, programs may need to focus more resources on the highest-risk patients because this service is resource intensive. The present study provides a preliminary model to target patient navigation to cancer patients likely to need more help.
Conclusions
The present study shows that newly diagnosed cancer patients’ most common barriers to care include
JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
communications barriers, lack of social support, and health insurance concerns. Number and types of barriers differ significantly between non-Hispanic white and minority patients. A multivariate model that predicts high need for patient navigation in our patient sample includes minority race/ethnicity, unemployment, and unmarried status. These data may help in the design and targeting of future interventions to reduce cancer health disparities.
Acknowledgments
The authors would like to thank Starlene Loader, RN, and Sally Rousseau, MSW, of the Rochester Patient Navigation Research Program program for their assistance in obtaining and accurately interpreting the data for this study. We would also like to acknowledge the patient navigators, research assistants, and investigators of the Rochester Patient Navigation Research Program program.
References
1. Jemal A, Clegg LX, Ward E, et al. Annual report to the nation on the status of cancer, 1975-2001, with a special feature regarding survival. Cancer. 2004;101:3-27. 2. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2006. CA Cancer J Clin. 2006;56:106-130. 3. Healthy People 2010: Understanding and Improving Health. Washington, DC: Government Printing Office; 2000. US Department of Health and Human Services. 4. Le H, Ziogas A, Lipkin SM, Zell JA. Effects of socioeconomic status and treatment disparities in colorectal cancer survival. Cancer Epidemiol Biomarkers Prev. 2008;17:1950-1962. 5. Morris AM, Billingsley KG, Hayanga AJ, Matthews B, Baldwin LM, Birkmeyer JD. Residual treatment disparities after oncology referral for rectal cancer. J Natl Cancer Inst. 2008;100:738-744. 6. Morris AM, Billingsley KG, Baxter NN, Baldwin LM. Racial disparities in rectal cancer treatment: a population-based analysis. Arch Surg. 2004;139:151-155. 7. Shavers VL, Brown ML. Racial and ethnic disparities in the receipt of cancer treatment. J Nat Cancer Inst. 2002;94:334-357. 8. Freeman HP, Wasfie TJ. Cancer of the breast in poor black women. Cancer. 1989;63:2562-2569. 9. Gross CP, Smith BD, Wolf E, Andersen M. Racial disparities in cancer therapy: did the gap narrow between 1992 and 2002? Cancer. 2008;112:900-908. 10. Bach PB, Pham HH, Schrag D, Tate RC, Hargraves JL. Primary care physicians who treat blacks and whites. N Engl J Med. 2004;351:575-584. 11. Ward E, Halpern M, Schrag N, et al. Association of insurance with cancer care utilization and outcomes. CA Cancer J Clin. 2008;58:9-31. 12. Allen JD, Shelton RC, Harden E, Goldman RE. Follow-up of abnormal screening mammograms among low-income ethnically diverse women: findings from a qualitative study. Patient Educ Couns. 2008;72:283-292. 13. Garbers S, Jessop DJ, Foti H, Uribelarrea M, Chiasson MA. Barriers to breast cancer screening for low-income Mexican and Dominican women in New York City. J Urban Health. 2003;80:81-91. 14. Hart CL, Smith GD, Blane D. Inequalities in mortality by social class measured at 3 stages of the lifecourse. Am J Public Health. 1998;88:471-474. 15. Mandelblatt JS, Yabroff KR, Kerner JF. Equitable access to cancer services: A review of barriers to quality care. Cancer .1999;86:2378-2390. 16. Ashton CM, Haidet P, Paterniti DA, et al. Racial and ethnic disparities in the use of health services: bias, preferences, or poor communication? J Gen Intern Med. 2003;18:146-152. 17. Epstein RM. Patient-Centered Communication in Cancer Care: Promoting Healing and Reducing Suffering. Bethesda, MD: National Cancer Institute; 2007.
VOL. 103, NO. 8, AUGUST 2011 709
Patients’ Barriers to Cancer Care 18. Hack TF, Degner LF, Parker PA. The communication goals and needs of cancer patients: a review. Psychooncology. 2005;14:831-845. 19. Esnaola NF, Hall BL, Hosokawa PW, et al. Race and surgical outcomes: it is not all black and white. Ann Surg. 2008;248:647-655. 20. Freund KM, Battaglia TA, Calhoun E, et al. National Cancer Institute Patient Navigation Research Program: methods, protocol, and measures. Cancer. 2008;113:3391-3399. 21. Wells KJ, Battaglia TA, Dudley DJ, et al. Patient navigation: state of the art or is it science? Cancer. 2008;113:1999-2010. 22. Freeman HP. Patient navigation: a community based strategy to reduce cancer disparities. J Urban Health. 2006;83:139-141. 23. US Census Bureau. 2006-2008 American Community Survey: Monroe County, NY. http://factfinder.census.gov/servlet/ACSSAFFFacts?_event=Search&geo_id=01000US&_geoContext=&_ street=&_county=monroe+county&_cityTown=monroe+county&_ state=04000US36&_zip=&_lang=en&_sse=on&ActiveGeoDiv=geoSelect&_ useEV=&pctxt=fph&pgsl=010&_submenuId=factsheet_1&ds_ name=ACS_2008_3YR_SAFF&_ci_nbr=null&qr_name=null®=null%3Anull&_ keyword=&_industry=. Accessed October 10, 2010. 24. Davis TC, Long SW, Jackson RH, et al. Rapid estimate of adult literacy in medicine: a shortened screening instrument. Fam Med. 1993;25:391-395. 25. Davis TC, Crouch MA, Long SW, et al. Rapid assessment of literacy levels of adult primary care patients. Fam Med. 1991;23:433-435. 26. Krieger N, Chen JT, Waterman PD, Rehkopf DH, Subramanian SV. Race/ ethnicity, gender, and monitoring socioeconomic gradients in health: a comparison of area-based socioeconomic measures—the public health disparities geocoding project. Am J Public Health. 2003;93:1655-1671. 27. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of clas-
710 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION
sifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373-383. 28. UCLA Academic Technology Services. How do I interpret a regression model when some variables are log transformed? www.ats.ucla.edu/stat/ mult_pkg/faq/general/log_transformed_regression.htm. Accessed March 23, 2009. 29. Halpern MT, Ward EM, Pavluck AL, Schrag NM, Bian J, Chen AY. Association of insurance status and ethnicity with cancer stage at diagnosis for 12 cancer sites: a retrospective analysis. Lancet Oncol. 2008;9:222-231. 30. Gordon HS, Street RL Jr, Sharf BF, Kelly PA, Souchek J. Racial differences in trust and lung cancer patients’ perceptions of physician communication. J Clin Oncol . 2006;24:904-909. 31. Bickell NA, Weidmann J, Fei K, Lin JJ, Leventhal H. Underuse of breast cancer adjuvant treatment: patient knowledge, beliefs, and medical mistrust. J Clin Oncol. 2009;27:5160-5167. 32. Adams SA, Smith ER, Hardin J, Prabhu-Das I, Fulton J, Hebert JR. Racial differences in follow-up of abnormal mammography findings among economically disadvantaged women. Cancer. 2009;115:5788-5797. 33. Mead H, Cartwright-Smith L, Jones K, Ramos C, Woods K, Siegel B. Racial and Ethnic Disparities in US Health Care: A Chartbook. New York: The Commonwealth Fund; 2008. 34. National Center for Health Statistics. Health, United States, 2008. Hyattsville, MD: Government Printing Office; 2009. 35. Howard G, Anderson RT, Russell G, Howard VJ, Burke GL. Race, socioeconomic status, and cause-specific mortality. Ann Epidemiol. 2000;10:214-223. 36. Fiscella K, Williams DR. Health disparities based on socioeconomic inequities: implications for urban health care. Acad Med. 2004;79:1139-1147. n
VOL. 103, NO. 8, AUGUST 2011