Barriers to healthcare contribute to delays in follow-up among women with abnormal cancer screening: data from the Patient Navigation Research Program

Abstract

Background

There is limited understanding of the association between barriers to care and clinical outcomes within patient navigation programs.

Methods

Secondary analyses of data from the intervention arms of the Patient Navigation Research Program (PNRP), including navigated participants with abnormal breast and cervical cancer screening tests from 2007 to 2010. Independent variables were (a) number of unique barriers to care (0, 1, 2, or 3+) documented during patient navigation encounters and (b) presence of socio-legal barriers originating from social policy (yes/no). Median time to diagnostic resolution of index screening abnormalities was estimated using Kaplan-Meier cumulative incidence curves. Multivariable Cox proportional hazards regression examined the impact of barriers on time to resolution, controlling for socio-demographics and stratifying by study center.

Results

Among 2600 breast participants, three-quarters had barriers to care (25% 1 barrier, 16% 2 barriers and 34% 3+ barriers). Among 1387 cervical participants, more than half had barriers (31% 1 barrier, 11% 2 barriers, and 13% 3+ barriers). Among breast participants, the presence of barriers was associated with less timely resolution for any number of barriers compared to no barriers. Among cervical participants, only the presence of 2 or more barriers was associated with less timely resolution. Both types of barriers, socio-legal and other barriers, were associated with delay among breast and cervical participants.

Conclusions

Navigated women with barriers resolve cancer screening abnormalities at a slower rate compared to navigated women with no barriers. Further innovations in navigation care are necessary to maximize the impact of patient navigation programs nationwide.

INTRODUCTION

Despite remarkable advances in cancer care, low-income communities and racial/ethnic minorities continue to bear an unequal burden of cancer-related illness such as higher mortality ¹ and lower stage-specific survival when compared with more affluent communities.^2,3 These disparities in cancer outcomes have been attributed to differences in care at each point in the care continuum. Recent immigrants and the uninsured are less likely to complete cancer screening⁴ and Hispanic women have lower rates of Papanicolau testing.^5,6 African-American and Hispanic women experience longer time until follow-up of an abnormal mammogram than White women.^7–9 Such delays between detection and initiation of treatment of breast cancer have been associated with later stage of diagnosis and lower survival.¹⁰

Evidence suggests that these delays may be explained by barriers or obstacles to healthcare disproportionately faced by low-income and racial minority patients that impede timely receipt of healthcare services. Patient navigation has emerged as one partial solution to address disparities in cancer care delivery. Patient navigation has been defined as “the support and guidance offered to persons with abnormal cancer screening or a new cancer diagnosis in accessing the cancer care system; overcoming barriers; and facilitating timely, quality care provided in a culturally sensitive manner”.¹¹ The central purpose of patient navigation is the identification and elimination of patient-level barriers to care.¹²

In 2005, the National Cancer Institute's (NCI) Center to Reduce Cancer Health Disparities initiated the multicenter Patient Navigation Research Program (PNRP) to examine the benefits of patient navigation across a diverse population.^11,12 Growing evidence from this scientifically rigorous study has demonstrated the efficacy of patient navigation to decrease the time from an abnormal cancer screening test to completion of follow-up testing^13–18 and the time from diagnosis of cancer to initiation of treatment^19,20 among participants who received navigation compared to usual care.

However, even as patient navigation expands in popularity and penetration,^21,22 delays in care still exist within navigation programs. At some centers participating in the PNRP, up to 15% of participants who had received navigation for abnormal breast cancer screening tests still had not completed follow-up testing one year after their abnormal screening.²³ We hypothesized that these continued delays may be due to the number or type of barriers faced by navigation recipients. We examine barriers through the lens of social determinants of health. Certain unique barriers originate from the conditions and structure of society, such as lack of affordable housing, impaired access to education and employment, or inadequate income supports; these barriers are the focus of medical-legal partnerships (MLPs). The MLP model recognizes that many social determinants of health have legal solutions.²⁴ Legal advocates have the expertise to identify violations of rights and take necessary steps to hold the appropriate parties accountable.²⁵ Preliminary studies suggest that the presence of barriers in general, and these socio-legal barriers in particular, contribute to persistent delays in care despite navigation.²⁶ We define socio-legal barriers as those social problems related to meeting life's most basic needs that are supported by public policy, regulation, and programming and thus potentially remedied through legal advocacy or action.

In this analysis, our first aim was to examine the relationship between the number of barriers to care and timeliness of care. We hypothesized that participants with one or more barriers to care would have longer time to diagnostic resolution of abnormal breast and cervical cancer screening tests compared to participants with no barriers to care, and that as the number of barriers identified for a participant increased, the time to diagnostic resolution would increase.

Our second aim was to examine the relationship between the type of barriers (socio-legal barriers and non- socio-legal barriers) and timeliness of care. We hypothesized that participants with any socio-legal barriers to care would have longer time to diagnostic resolution of abnormal breast and cervical cancer screening tests compared to participants with non- socio-legal barriers to care or participants with no barriers to care. Better understanding of the association between barriers to care and clinical outcomes could inform the design of future navigation programs, potentially identifying vulnerable subgroups of participants within the navigation cohort who need additional assistance based on their barrier profile.

MATERIALS AND METHODS

Design Overview

We conducted a secondary analysis of data from the intervention arm of the PNRP, a multicenter study of patient navigation for cancer care among underserved populations.¹¹ The PNRP compared patient navigation with usual care on time to diagnosis or treatment for participants with breast, cervical, colorectal, or prostate screening abnormalities and/or cancers from ten study centers between 2007 and 2010. Our analysis focused on participants who were navigated for abnormal breast and cervical cancer screening tests. We examined the relationship between the number and types of barriers to care identified by patient navigators and time to diagnostic follow-up of abnormal cancer screening. Barrier information was only collected on participants who received navigation, therefore this analysis is restricted to navigated participants and does not include non-navigated control participants. All PNRP centers received approvals from their local Institutional Review Boards for their study designs.¹¹

Study Participants

For this analysis, eligible participants included women 18 years of age or older enrolled with a breast or cervical cancer screening abnormality, as previously described.¹¹ Participants were excluded if they had cognitive impairment that prevented interactions with navigators, had previously received navigation for cancer, had a history of cancer treated within the last five years, or were pregnant.

Study Measures

Predictor variable: barriers to care

Based on review of the literature and expert consensus, the PNRP investigators developed a list of potential barriers for investigation in the PNRP (Table 1).¹¹ Navigators recorded their encounters with each enrolled participant in patient-specific logs. This log included the type of encounter with the participant (e.g. in-person, by telephone call, etc.), the length of time with the participant during the encounter, the total time spent on navigation activities on behalf of the participant, barriers to care identified during that encounter, and actions taken to address those barriers.

Table 1.

Barriers to care in the Patient Navigation Research Program

		Abnormal breast screening N=2600		Abnormal cervical screening N=1387
Number of barriers		N	%	N	%
0		676	26	631	45
1		645	25	429	31
2		403	16	147	11
3+		876	34	180	13
Types of barriers
None		676	26	631	45
Non socio-legal barriers only		937	36	433	31
At least one socio-legal barrier		987	38	323	23
Barrier	Description
Socio-Legal Barriersa
Health Insurance	Paying for all aspects of health care is a problem	702	27.0	239	17.2
Financial Problems	Dealing with financial problems is interfering with receiving health care	358	13.8	40	2.9
Employment Issues	Work demands make getting health care difficult	220	8.5	74	5.3
Childcare Issues	Not having childcare when the patient needs medical care	84	3.2	14	1.0
Adult Care	Difficulty finding support for other family when the patient needs medical care	61	2.4	5	0.4
Housing	Worrying about where the patient lives during her health care	50	1.9	12	0.9
Non Socio-Legal Barriers
System Problems with Scheduling Care	Care provided to patient is not convenient/efficient to patient's needs	695	26.7	215	15.5
Language/Interpreter	Health care personnel and patient do not share a common language for communication	575	22.1	125	9.0
Fear	Fear about any aspect of medical care or their health	512	19.7	199	14.4
Communication Concerns with Medical Staff	Barriers to understanding the information given to a patient by medical personnel	433	16.7	84	6.1
Transportation	Difficulty getting from home to where the patient obtains her health care	424	16.3	73	5.3
Social/Practical Support	Lacks a person/community to help them through the care	402	15.5	73	5.3
Perceptions/Beliefs about Tests/Treatment	Personal or cultural beliefs that affect receiving health care	355	13.7	86	6.2
Other	Barrier other than one defined in the PNRP framework	339	13.0	34	2.5
Location of Health Care Facility	Distance from health care facility is a barrier even if patient has transportation	325	12.5	22	1.6
Medical/Mental Health Comorbidity	Medical health problems or mental health problems that make getting health care difficult	224	8.6	63	4.5
Literacy	Difficulty understanding written communication from the health care setting	204	7.9	30	2.2
Attitudes toward Providers	Perceptions and beliefs about the health care providers that impact receiving care	72	2.8	14	1.0
Out of town/country	Patient known to be out of area during their care	70	2.7	29	2.1
Patient Disability	Disability that makes getting health care difficult	57	2.1	24	1.7
Total number of barriers		6162		1455

^aThese six barriers reflected the domains served by MLP services and summarized in the acronym I-HELP [Income supports and insurance; Housing and utilities; Education and employment; Legal (immigration) status; and Personal and family stability and safety].²⁶

With each encounter, the navigator was trained to ask open-ended questions and elicit from the participant which of these barriers, if any, were operating to affect her receipt of follow-up diagnostic testing. The navigator was expected to document the barrier or barriers and what actions s/he took in response. If no barriers were found, then the navigator documented “No barriers identified”. If participants had more than one barrier type, then each unique barrier was noted in the navigator log. For this analysis, if a participant had multiple navigation contacts and the same barrier to care was noted at subsequent contacts, then that unique barrier was counted only once.

The predictor variable, barriers to care, was characterized in four different ways for our analysis: (1) barriers: yes/no; (2) number of barriers: zero, one, two, or three or more; (3) socio-legal barriers: yes/no; and (4) number and type of barrier combined. The six barriers defined as socio-legal barriers in our analysis were Insurance, Financial problems, Housing, Employment issues, Childcare, and Adult care (Table 1). These barriers reflected the domains served by MLP services.²⁷

Participants were classified into one of three mutually exclusive categories based on socio-legal barriers: (1) no barriers to care, (2) non- socio-legal barriers to care only (barriers present but none of the six socio-legal barriers highlighted in our study), and (3) at least one socio-legal barriers to care (at least one of the six socio-legal barriers documented at some point during the navigation experience). In order to better observe the effect of the number of barriers to care among participants with different types of barriers (socio-legal barriers or non- socio-legal barriers), we also divided participants into five categories: (1) no barriers to care, (2) exactly one non- socio-legal barriers, (3) exactly one socio-legal barriers, (4) two or more non- socio-legal barriers, and (5) two or more barriers of which at least one is a socio-legal barrier. This five-category mutually exclusive classification allowed us to compare participants with similar quantity of barriers but with qualitatively different barriers

Outcome variable: time to diagnostic resolution

Clinical data on type and dates of tests ordered, tests completed, and test results were manually abstracted from electronic health records, physical charts, registration, and appointment databases depending on the research center.¹¹ The outcome variable was time to diagnostic resolution of a screening abnormality, defined as a continuous variable as the number of days from the index screening abnormality until the date of completion of definitive diagnostic care (e.g. breast biopsy, colposcopy).

Covariates

Socio-demographic information was collected from registration data or self-report at the respective research centers. Covariates included demographic variables that the PNRP required all centers to collect and that have been shown to be associated with delays in completing recommended follow-up.^28,29

Statistical Analysis

Analyses were done separately for breast and cervical participants since the clinical follow-up testing and respective time for diagnostic test results differs between breast and cervical cancer and since preliminary examination of the study data showed that the distribution of barriers was different between breast and cervical participants. For each analysis, we first calculated frequencies of the number of barriers to care and the types of barriers to care among all breast and cervical participants. We conducted descriptive statistics to compare the socio-demographic characteristics of study participants with and without barriers to care using analysis of variance (ANOVA) for means of continuous variables and chi-square tests for categorical variables.

Next, we conducted unadjusted bivariate analyses to explore associations between the number and type of barriers present and time to diagnostic resolution. The median number of days to diagnostic resolution was estimated using Kaplan-Meier survival curves according to (1) the presence or absence of barriers (2) the number of barriers (3) the three-category designation of socio-legal barriers (4) the five-category designation of socio-legal barriers. Median days to resolution were compared using the Kruskall-Wallis test. Results from unadjusted analyses are available in online supporting information (Supporting Table S1).

We created several Cox proportional hazards regression models with likelihood of diagnostic resolution as the outcome. The first model included number of barriers as the main predictor and did not include a term for socio-legal barriers. The second model included the three-category socio-legal barrier variable and did not include the number of barriers. The third model included the five-category socio-legal barrier variable, which takes into account both the number of barriers and whether the barriers are socio-legal barrier or non- socio-legal barriers. All models were adjusted for continuous age, race/ethnicity, insurance, and language, and stratified by research center to account for baseline differences among centers. Adjusted hazard ratio (aHR) greater than 1.0 indicates more timely diagnostic resolution compared to the reference group. There was no violation of the proportional hazards assumption. The complete results for all three models are available in online supporting information (Supporting Tables S2 and S3).

RESULTS

Characteristics of participants

Table 2 shows the socio-demographic characteristics of breast and cervical screening participants according to the presence of barriers to care. Nearly three-fourths of breast participants and over half of cervical participants had one or more barriers to care. Compared to participants with no barriers, those with barriers were more often Hispanic/Latina, uninsured, and non-English language speakers (p<0.0002).

Table 2.

Socio-demographic characteristics of women with abnormal breast or cervical screening, by presence of barriers

	Abnormal breast screening			Abnormal cervical screening
Characteristic	Total N=2600	No barriers N=675 (26%)	At least one barrier N=1918 (74%)	Total N=1387	No barriers N=631 (45%)	At least one barrier N=756 (55%)
Age
Mean (SD)		48.7 (12.5)	46.7 (12.1)		30.1 (9.9)	31.0 (10.5)
Race/Ethnicity
White	774	53%	22%	241	20%	15%
Black/African-American	492	22%	18%	437	38%	26%
Hispanic/Latina	1163	19%	54%	692	40%	58%
Other	170	5%	7%	16	1%	1%
Insurance Status
Uninsured	1019	16%	48%	527	28%	47%
Public	911	37%	35%	561	46%	37%
Private	634	46%	17%	291	27%	17%
Primary Language
English	1481	80%	50%	878	70%	58%
Non-English	1079	20%	50%	509	30%	42%

p-value < 0.0002 for all comparisons of characteristics of participants with no barriers to care versus participants with at least one barrier to care except for mean age of cervical screening participants

Barrier prevalence

The distribution of the number and specific types of barriers for both breast and cervical participants are displayed in Table 1. About one-fourth of breast screening participants had one barrier to care documented during their navigation experience, while 16% had two barriers and 34% had three or more barriers. There were a total of 6162 barriers identified among 2600 participants navigated for abnormal breast cancer screening. The most common barriers were health insurance (27% of breast participants) and system problems scheduling care (27%).

Less than one-third of cervical screening participants had exactly one barrier to care, while 11% and 13% had two and three or more barriers to care, respectively (Table 1). There were 1455 barriers identified among 1387 participants navigated for abnormal cervical cancer screening. The most common barriers were similar to those among breast participants.

Adjusted association of barriers and time to diagnostic resolution

Table 3 presents the three Cox proportional hazards regression models for breast screening participants, stratified by research center. After adjustment for race, insurance, language, and age, the presence of any number of barriers was associated with less timely resolution compared to no barriers (Model 1). The magnitude of the effect was overall similar regardless of the number of barriers (aHR 0.76 for one barrier, 0.69 for two barriers, and 0.74 for three or more barriers).

Table 3.

Cox proportional hazards regression models for time to diagnostic resolution among women with abnormal breast (N=2600) or cervical (N=1387) cancer screening according to number and type of barriers

Variable	Level	Model 1: number of barriers		Model 2: SLBs (3-category)		Model 3: SLBs (5-category)
		aHR	95% CI	aHR	95% CI	aHR	95% CI
BREAST
Number of barriers	0	1.0 (ref)
	1	0.76	0.67, 0.86
	2	0.69	0.6, 0.8
	3+	0.74	0.65, 0.84
SLBs – 3 category	None			1.0 (ref)
	Non-SLBs only			0.74	0.66, 0.82
	At least 1 SLB			0.74	0.65, 0.84
SLBs – 5 category	None					1.0 (ref)
	1 Non-SLB only					0.77	0.68, 0.87
	1 SLB only					0.71	0.55, 0.91
	2+ Non-SLBs only					0.69	0.6, 0.79
	2+ Barriers including at least 1 SLB					0.74	0.65, 0.84
CERVICAL
Number of barriers	0	1.0 (ref)
	1	0.89	0.78, 1.01
	2	0.6	0.49, 0.74
	3+	0.8	0.66, 0.96
SLBs – 3 category	None			1.0 (ref)
	Non-SLBs only			0.87	0.76, 0.99
	At least 1 SLB			0.71	0.61, 0.83
SLBs – 5 category	None					1.0 (ref)
	1 Non-SLB only					0.91	0.79, 1.05
	1 SLB only					0.81	0.65, 1.01
	2+ Non-SLBs only					0.75	0.6, 0.94
	2+ Barriers including at least 1 SLB					0.67	0.55, 0.8

Models adjusted for race/ethnicity (White, Black/African-American, Hispanic/Latina, Other), insurance (private, public, uninsured), language (English, Non-English), and continuous age, and stratified by research center; SLB: socio-legal barriers

Adjusted hazard ratio (aHR) greater than 1.0 indicates more timely diagnostic resolution compared to the reference group.

In Model 2, non-socio-legal barriers and socio-legal barriers were associated with significantly less timely follow-up care (aHR 0.74 for both) compared to no barriers. Using the five-category classification of barriers (Model 3), non- socio-legal barriers and socio-legal barriers of any number were associated with less timely follow-up care compared to no barriers with no clear pattern to the impact based on type of barrier.

Having either two barriers or three or more barriers was significantly associated with less timely resolution in Model 1 for cervical screening participants (Table 3). Non-socio-legal barriers and socio-legal barriers were both associated with less timely resolution (Model 2), with a slightly greater impact from socio-legal barriers. In the five-category model (Model 3), presence of two or more barriers, whether exclusively non-socio-legal barriers (aHR 0.75, 95% CI 0.6, 0.94) or including at least one socio-legal barrier (aHR 0.67, 95% CI 0.55, 0.8), was associated with delay in achieving diagnostic resolution.

DISCUSSION

In this secondary analysis of a large, multicenter patient navigation program, we found that women with barriers to care experienced delays in follow-up for abnormal breast or cervical cancer screening tests compared to women without barriers. Among those navigated for breast screening abnormalities, the negative effect on timeliness of care was observed regardless of the given number of barriers or the type of barrier. For women with cervical screening abnormalities, the impact was most prominent for those with multiple barriers, but also did not depend on the type of barrier. Within a patient navigation program for cancer care that has demonstrated the benefits of navigation overall compared to usual care,²³ barriers were still associated with delays.

This work is among the few studies that have examined the relationship between barriers and clinical outcomes in a patient navigation program, but the results of our analyses were unexpected compared to prior work. We hypothesized that, as the number of barriers increased, delays would also increase, that is, a possible dose-response relationship would exist similar to one observed in comparable analyses performed using data from the Boston PNRP site.³⁰ Instead, in the adjusted model for breast participants, we found that the impact on timeliness of care was relatively similar whether there were one, two, or three or more barriers. In the case of cervical participants, barriers significantly affected time to resolution when there were at least two barriers.

Using the same socio-legal barrier framework and five-category classification, Primeau and colleagues in the Boston PNRP found that participants with multiple barriers experienced less timely diagnostic resolution compared to those with no barriers or only one barrier, but that participants with socio-legal barriers experienced even greater delays than those with a similar number of non- socio-legal barriers.²⁶ In our multicenter replication of this work, we did not find that the type of barriers had an appreciable differential impact on timeliness of care even when accounting for number of barriers. These dissimilar findings may be due to the larger sample of participants with socio-legal barriers used in our study: 38% of breast and 23% of cervical navigated participants nationally had socio-legal barriers compared to only 6% of the navigated participants in Boston. Alternatively, the quality of the socio-legal barriers in Boston may differ from other centers in different states.

The current study also combined data from multiple research centers with varied patient populations. These various centers likely had different resources available to patients and navigators to address prevalent barriers and different distributions of barriers. Whereas health insurance was an uncommon barrier in the Boston PNRP due to Massachusetts healthcare reform initiated in 2006, it was the most frequently identified barrier in both breast and cervical participants across the other PNRP centers. Language/Interpreter barriers may have been more commonly identified in research centers that served large patient populations with limited English proficiency and magnified the impact of non- socio-legal barriers on the outcome of interest.

These findings have important implications for the design and implementation of future navigation programs. While all of the PNRP research centers recruited participants from underserved, low-income, and/or minority communities, our analyses demonstrate that those with barriers were especially vulnerable. They were more often uninsured, non-English language speaking, Hispanic/Latina, and, in the case of breast participants, younger. These measurable markers of disadvantage are part of a collection of social risk factors including ones not directly measured in this study (e.g. low income, low educational attainment). Such socio-demographic characteristics have been suggested as one way to identify those most likely to have barriers and therefore most at risk of experiencing delays. Hendren and colleagues examined barriers to care collected on newly-diagnosed cancer patients receiving navigation services and found that minority race/ethnicity was associated with greater time spent with navigators.²⁸ Together with unemployment, unmarried status, and cancer treatment variables, race/ethnicity explained much of the variation in navigation need. The authors suggested using minority status as a tool for directing navigation interventions to vulnerable cancer patients. While this recommendation may not be pertinent to all communities,³¹ it highlights the broader importance of tailoring navigation programs to local needs.

Notably, one-fourth of breast subjects and nearly half of cervical subjects did not have any barriers to care identified. These differences in barrier prevalence may reflect the different challenges faced by older versus younger women in accessing care or the complexities of diagnostic breast cancer care that often requires multiple disciplines and modalities. Our data showed that as the number of barriers increased, the degree of navigation support also increased, as measured by the number of navigator encounters and the time spent during the encounters. In the era of accountable care, one could argue that a resource-intense service like patient navigation should be directed to those patients who need it most – patients with barriers and delays. On the other hand, in a population where barriers are widespread, as they were among the breast subjects, it may be more efficient to provide navigation uniformly without attempting to differentiate those with barriers from those without barriers. The more likely scenario, as navigation spreads to more settings and populations, will be that a proportion of patients will not have barriers and thus not need navigation. The challenge for future navigation programs therefore will be to effectively identify their target patient population. One of the strengths of this study is that it included participants who represent the most disadvantaged members of the population and therefore the very individuals for whom patient navigation is intended. We improved on the generalizability of prior work by including participants from diverse communities across the country, and our large sample size allowed us to examine differences between several groups using the five-category socio-legal barrier classification. Barriers in the PNRP were collected prospectively, and navigators received standardized training on how to identify and document barriers in a common navigator tracking log which allowed us to combine data across centers.

Still, we recognize certain limitations in this secondary analysis. Barrier information was not collected on non-navigated control participants, therefore we cannot assess the impact of the navigation intervention on subgroups by number or type of barriers. The various research centers participating in the PNRP differed in their study designs, patient populations, and distribution of barriers, and barrier information was most likely collected differently in ways that reflect the local burden of barriers, availability of resources to address the barriers, and individual knowledge and skills of the navigators. Other factors besides barriers to care, including provider factors, such as cultural competency, and navigator factors, such as personality and racial concordance, also influence patient outcomes.³² We attempted to account for differences between research centers by stratifying by center in adjusted analyses. Although we cannot assess fidelity of implementation of the navigation intervention, navigators at all PNRP centers did receive uniform training and competency evaluation on the identification and documentation of barriers in order to standardize their practices.

We created the novel category of socio-legal barriers using the six barriers from the PNRP framework that best reflected obstacles for which legal regulations and protections exist, however, the original barriers in the PNRP were not defined with this classification in mind. Our approach is therefore exploratory in nature.

In this analysis, there is a risk of confounding by indication since delays in care provide more opportunities for encounters and therefore more opportunities to identify barriers. We sought to minimize this potential risk by only counting barriers once even if the same barrier was documented on subsequent encounters. One weakness of this approach is that it does not allow investigation of the effect of persistent barriers over time. Such barriers that are repeatedly identified may represent the most challenging ones for navigators to address and the most deleterious to clinical outcomes.³¹ Still, we found that the majority of barriers were found within the first three encounters, suggesting that a limited proportion of barriers were identified with additional encounters.

This analysis used data from women needing follow-up for abnormal breast and cervical cancer screening, therefore the findings cannot be generalized to other cancer screening tests or patients diagnosed with cancer and receiving navigation. While it was not our intention to compare breast and cervical participants, we observed that the distribution and impact of barriers differed depending on the cancer site.

In conclusion, in a multicenter patient navigation program, we found that patients with documented barriers to care are among the most vulnerable members of the population and that barriers to care, whether socio-legal barriers or non socio-legal barriers, are associated with less timely follow-up of abnormal breast and cervical cancer screening tests. In order to equitably eliminate delays in care, patient navigation programs should refine and strengthen processes of recognizing patients who have barriers to care and target navigation services accordingly. The particular barriers that are most common and most important will vary by cancer screening site and patient population, therefore navigation programs should characterize the needs of their local communities in order to optimize the efficiency and effectiveness of their interventions. Future research should collect information on barriers in patients not receiving navigation for comparison to navigated patients in order to understand the differential benefits of navigation according to barrier number and type.

Condensed abstract.

Among women who received navigation for abnormal breast or cervical cancer screening in the Patient Navigation Research Program, women with barriers resolve cancer screening abnormalities at a slower rate compared to those with no barriers. The negative effect on timeliness of care was observed regardless of the given number of barriers or the type of barrier.