Background: Barriers to timely resolution of abnormal cancer screening tests add to cancer health disparities among low-income, uninsured, and minority populations. We conducted a randomized trial to evaluate the impact of lay patient navigators on time to resolution and completion of follow-up testing among patients with abnormal screening tests in a medically underserved patient population.

Methods: Denver Health, the safety-net health care system serving Denver, is one of 10 performance sites participating in the Patient Navigation Research Program. Of 993 eligible subjects with abnormal screening tests randomized to navigation and no-navigation (control) arms and analyzed, 628 had abnormal breast screens (66 abnormal clinical breast examinations, 304 BIRADS 0, 200 BIRADS 3, 58 BIRADS 4 or 5) whereas 235 had abnormal colorectal and 130 had abnormal prostate screens.

Results: Time to resolution was significantly shorter in the navigated group (stratified log rank test, P < 0.001). Patient navigation improved diagnostic resolution for patients presenting with mammographic BIRADS 3 (P = 0.0003) and BIRADS 0 (P = 0.09), but not BIRADS 4/5 or abnormal breast examinations. Navigation shortened the time for both colorectal (P = 0.0017) and prostate screening resolution (P = 0.06). Participant demographics included 72% minority, 49% with annual household income less than 10,000, and 36% uninsured. Conclusions: Patient navigation positively impacts time to resolution of abnormal screening tests for breast, colorectal, and prostate cancers in a medically underserved population. Impact: By shortening the time to and increasing the proportion of patients with diagnostic resolution patient navigation could reduce disparities in stage at diagnosis and improve cancer outcomes. Cancer Epidemiol Biomarkers Prev; 21(10); 1629–38. ©2012 AACR. Patient navigation programs are being increasingly adopted across the United States, although the evidence for their effectiveness is not well established (1). Patient navigation can potentially impact cancer care across the entire continuum, including screening rates, resolution of screening abnormality, diagnosis to initiation of treatment, treatment adherence and completion, survivorship, and end-of-life care. Most of the reported studies have assessed the impact of patient navigation on screening rates for breast, cervical, and colorectal cancer (2–8). Fewer studies have been published on the effect of patient navigation on diagnostic resolution, treatment outcomes, survivorship, and end-of-life (9). The Patient Navigation Research Program (PNRP) was initiated in 2005 through funding from the National Cancer Institute's Center to Reduce Cancer Health Disparities. PNRP is a cooperative effort of 9 health care institutions across the United States, serving medically underserved populations, including racial and ethnic minorities and those of low socioeconomic status (1, 10). PNRP enrolled subjects with either an abnormal cancer screening finding or an incident diagnosis of breast, cervical, colorectal, and prostate cancer. In this publication the authors describe the results of the individually randomized clinical trial conducted at the Denver PNRP, addressing subjects with abnormal screening findings. ### Setting Denver Health is the public safety net for the City and County of Denver and has a long and distinguished record for serving as the health care provider for Denver's underserved populations. One quarter of Denver residents, or approximately 160,000 adults and children receive their care at Denver Health, of whom 57% are Hispanic, 18% Black, and 20% White. In terms of insurance status, 46% of Denver Health patients are uninsured, 36% have Medicaid, 6% Medicare, 10% commercial, and 2% are covered by the State Children's Health Insurance Program. Denver PNRP conducted a prospective randomized clinical trial, one of 2 individually randomized trials within the National PNRP. The aims of the overall program were to (1) reduce the time from abnormal screening for breast, colorectal, and prostate cancer to diagnostic resolution, (2) reduce the time from breast, colorectal, and prostate cancer diagnosis to start of treatment, (3) increase adherence to cancer treatment, (4) increase patient satisfaction with care, and (5) determine the cost effectiveness of patient navigation. Patient enrollment began at Denver Health in October 2006 and 1,249 screening and treatment patients were enrolled in the study through June 2010. In this manuscript we report our findings for aim 1. ### Study participants Eligibility for Denver PNRP included an abnormal breast, colorectal, or prostate screening test or a new diagnosis of these cancers. Breast screening tests included an abnormal clinical breast examination that resulted in a referral to a specialist or an abnormal mammogram that resulted in further diagnostic examination. Colorectal screening included a positive fecal occult blood test, an abnormal rectal examination, history of bright red blood in the stool among those 40 and over, or an abnormal sigmoidoscopy. Men enrolled with an abnormal prostate screening test had an abnormal prostate finding on digital rectal examination or an elevated prostate-specific antigen (PSA; ≥4 ng/mL). Individuals with a prior history of cancer (excluding nonmelanoma skin cancer) within the past 5 years, prior receipt of patient navigation, pregnancy, or incarceration were not eligible for this study. ### Recruitment, consent, and randomization Potentially eligible patients were identified through weekly electronic lists of patients meeting the criteria from the referral systems and laboratory and radiology results. The recruiter verified eligibility through the electronic medical record. Informational study materials were sent to patients by mail, and a follow-up call was made to obtain verbal consent and baseline information from all patients. Eligibility was confirmed during the initial telephone contact. Patients providing verbal consent and Health Insurance Portability and Accountability Act (HIPAA) authorization were then randomly assigned to either navigation or control groups, using a computer-generated algorithm stratified by cancer type. Navigated patients provided written consent and HIPAA authorization; a waiver for written consent and HIPPA authorization was obtained for the control subjects. Study Institutional Review Board approval was provided by the Colorado Multiple Institution Review Board (COMIRB) before study initiation. Participants assigned to the control group received a letter informing them of their group assignment status and then received usual care with no additional contact until the postintervention survey administration. Those assigned to the navigation group received an introductory call from a patient navigator within 3 days of assignment. ### Charlson comorbidity index The Charlson comorbidity index is widely used and was developed to identify 18 comorbid illnesses that increase the risk of 1-year mortality in a cohort of internal medicine inpatients (11). ### Definition of time to resolution Time to resolution was determined from the date of abnormal screening test to the date of definitive diagnosis. Study subjects without diagnostic resolution were censored at 365 days. Definitive diagnosis of a cancer was defined as the pathologic diagnosis of an invasive cancer on biopsy. ### Navigation intervention The conceptual framework used for this intervention is an expansion of Wagner's Chronic Care Model (CCM) first implemented at Denver Health in 1999 (12). The CCM has 3 main themes: evidence-based, population-based, and patient-centered care. It is operationalized through 6 components: (1) Health System and Organization Change, (2) Delivery System Redesign, (3) Decision Support for Health Care Professionals, (4) Clinical Information Systems, (5) Self-Management Support, and (6) Community Resources. The CCM model has been expanded to the Comprehensive Care Model, and is used to guide care throughout Denver Health's Community Health Centers. This model includes the innovative use of patient navigators to promote diagnostic follow-up and to enhance patient/provider interactions. Within this model, the navigators use a strengths-based approach in navigator/patient interactions to identify assets available to patients while traversing the cancer diagnostic and treatment experience. Patient navigator contact with patients was by phone (63%), email (8%), or in person (29%), depending on patient needs and preferences. A common patient log for navigators was developed to document their work and to capture each patient contact and the activities completed for the patient, along with the duration of each encounter. Patient navigators documented barriers to care from a predefined list, as well as actions taken to address these barriers. During the initial patient contact, the navigator provided an introduction to navigation services and to the navigator's role. Next, the navigator elicited from the patient their potential and current barriers to completing the diagnostic test or treatment. The navigator monitored the patient during the entire diagnostic phase through resolution or, if a cancer diagnosis occurred, continuing through active cancer treatment. The navigator maintained contact with the patient by phone, home visits, and in-person meetings at the medical center. Through ongoing patient contacts, the navigator continued to assess practical barriers, social support, and intention to complete the recommended course of care. Within the health care system, the patient navigator ensured that the required examinations were scheduled and communicated with clinic staff about patient needs and concerns. The navigator accompanied patients to their appointments when fear, social support, and language barriers were identified, or if requested. After each diagnostic test or examination, the patient navigator assessed patient understanding of results and linked the participant to clinic staff as needed for further explanation and to address concerns. If additional examinations were required, the navigator scheduled the appointments and ensured that the appointments were convenient for the patient. If the initial abnormal screening test was resolved with a noncancer diagnosis, the navigator verified patient understanding of surveillance recommendation and terminated navigation services. Denver Health PNRP used 3 patient navigators. Navigator demographic and training information, including any personal experience with cancer, were collected from each navigator for identification of specific characteristics of navigators associated with successful navigation. All navigators attended an annual training conference designed and delivered by the American Cancer Society and the PNRP (13). In addition, frequent institution-based training sessions were conducted on clinical and psychosocial topics. Patient navigators met with the program manager on a weekly basis initially and then continued to meet monthly throughout the research program. Patient navigators also met with clinical staff for training and feedback on medical questions and patient care. A clinical psychologist met regularly with the patient navigators to assist them in working with patients with various behavioral health conditions, as well as in motivating patients and maintaining professional boundaries. ### Statistical analysis Demographic and clinical characteristics were compared using a Cochran–Mantel–Haenszel statistics for ordinal variables, Pearson χ2 test of independence for nominal variables, as well as independent t tests for continuous variables. The effect of navigation on the time to resolution of screening was compared using a stratified log-rank test. The 6 strata were abnormal clinical breast examination, BIRADS 0 (additional imaging required), BIRADS 3 (short interval recommended for follow-up mammogram), BIRADS 4 (suspicious abnormality found requiring biopsy follow-up), or BIRADS 5 (highly suspicious for malignancy), colorectal and prostate. The log-rank test was specified a priori; the strata were later identified based on sample size, location, and typical recommendations (e.g., immediate or 6 month follow-up). Test of effect modification used a Cox regression model with separate baseline hazards for the same 6 strata. Visual inspection of Kaplan–Meier plots suggested that the hazard associated with navigation was not constant over time. Time-varying indicator variables were defined for 0 to 90, 90 to 180, and 180+ days; and testing confirmed the hazard was not constant across these intervals (Wald χ2 (2df) = 21.1, P < 0.001). Effect modification was tested based on interactions of centered covariates with the 3 time-varying indicators in models that included the covariate as a main effect. A total of 3,000 abnormal screening subjects were identified as potentially eligible from laboratory and radiology lists and on initial review. Of these, 1,924 were excluded, 1,162 because of not meeting study inclusion criteria, 327 declined to participate, and 435 were unable to be contacted. A total of 1,076 eligible abnormal screening subjects entered into the Denver PNRP from October 2006 through June 2010 were equally randomized to either the navigation intervention (n = 538) or the control (no navigation) arm (n = 538). Among the subjects in the intervention group, 38 did not receive the intervention because of identification of ineligibility (5 subjects) or refusal to submit written consent and authorization (33 subjects). In the control group, 25 subjects did not receive the intervention because of ineligibility (8 subjects) and refusal to provide consent and authorization (18 subjects). Only one navigated subject revoked consent and authorization. The final analytical sample (993 subjects) included 485 navigated and 508 control subjects. Fourteen navigated and 4 control subjects were excluded in the final group because of issues related to medical records data and data collection. See CONSORT diagram (Fig. 1) for screening subject allocation and follow-up. Figure 1. CONSORT flow diagram of screening patients. Figure 1. CONSORT flow diagram of screening patients. Close modal Results of the time to resolution of abnormal screening tests for breast (n = 628), colorectal (n = 235), and prostate cancer (n = 130) are described in this report. Of the 628 randomized subjects with abnormal breast screens analyzed, 66 had abnormal clinical breast examinations and 304 had BIRADS 0, 200 BIRADS 3, 58 BIRADS 4 or 5 mammographic abnormalities. Results for the impact of patient navigation on newly diagnosed cancer patients will be reported separately. ### Characteristics of study participants The baseline characteristics for the enrolled subjects are shown in Table 1. Table 1. Baseline demographics of study population NavigatedControlCombined VariableLevelCount%Count%Count%P Sex Male 118 24 141 28 259 26 0.22 Female 367 76 367 72 734 74 Age (yrs) <30 16 10 26 30–39 56 12 53 10 109 11 40–49 145 30 160 31 305 31 50–59 141 29 147 29 288 29 0.77 6069 91 19 103 20 194 20 70+ 36 35 71 No answer/missing Race/ethnicity White 122 25 119 23 241 24 Black 96 20 97 19 193 19 Hispanic 250 52 275 54 525 53 0.78 Other 17 16 33 No answer/missing Cancer site Breast 308 64 320 63 628 63 Colorectal 114 24 121 24 235 24 0.99 Prostate 63 13 67 13 130 13 Navigation group Navigated 485 100 485 49 Control 508 100 508 51 Previous navigation No 484 100 501 99 985 99 Yes 0.11 No answer/missing Language English 330 68 335 66 665 67 Spanish 142 29 156 31 298 30 0.71 Other 13 17 30 Education Some high school or less 194 40 201 40 395 40 High school graduate/GED 134 28 125 25 259 26 Some college/AA 106 22 112 22 218 22 0.17 College graduate 34 34 68 Graduate/professional degree 20 27 No answer/missing 10 16 26 Employment None/retired/disabled 312 64 332 65 644 65 Part time 77 16 81 16 158 16 0.56 Full time 95 20 91 18 186 19 No answer/missing Marital status Single/never married 129 27 127 25 256 26 Married/with partner 168 35 194 38 362 36 0.51 Divorced/separated/widowed 188 39 187 37 375 38 No answer/missing Household size 153 32 151 30 304 31 151 31 133 26 284 29 48 10 65 13 113 11 55 11 62 12 117 12 0.20 5+ 78 16 95 19 173 17 No answer/missing Household status Own home 126 26 135 27 261 26 Rent 283 58 279 55 562 57 With friends or family 51 11 62 12 113 11 Other 0.31 Assisted/group home 12 Homeless 14 11 25 No answer/missing 10 12 Household income <10,000 251 52 235 46 486 49
$10,000–$19,999 122 25 149 29 271 27
$20,000–$29,999 47 10 57 11 104 10
$30,000–$39,999 23 29 52 0.53
$40,000–$49,999 13
$50,000+ 26 26 52 No answer/missing 15 Insurance status Private 52 11 60 12 112 11 Public 265 55 263 52 528 53 Uninsured 168 35 185 36 353 36 0.65 No answer/missing Variable Navigated Control Combined P Mean SD Mean SD Mean SD Charlson comorbidity index score 1.1 1.9 1.1 1.7 1.1 1.8 0.78 Number of months in study 6.64 6.59 7.08 6.24 6.87 6.42 0.28 NavigatedControlCombined VariableLevelCount%Count%Count%P Sex Male 118 24 141 28 259 26 0.22 Female 367 76 367 72 734 74 Age (yrs) <30 16 10 26 30–39 56 12 53 10 109 11 40–49 145 30 160 31 305 31 50–59 141 29 147 29 288 29 0.77 6069 91 19 103 20 194 20 70+ 36 35 71 No answer/missing Race/ethnicity White 122 25 119 23 241 24 Black 96 20 97 19 193 19 Hispanic 250 52 275 54 525 53 0.78 Other 17 16 33 No answer/missing Cancer site Breast 308 64 320 63 628 63 Colorectal 114 24 121 24 235 24 0.99 Prostate 63 13 67 13 130 13 Navigation group Navigated 485 100 485 49 Control 508 100 508 51 Previous navigation No 484 100 501 99 985 99 Yes 0.11 No answer/missing Language English 330 68 335 66 665 67 Spanish 142 29 156 31 298 30 0.71 Other 13 17 30 Education Some high school or less 194 40 201 40 395 40 High school graduate/GED 134 28 125 25 259 26 Some college/AA 106 22 112 22 218 22 0.17 College graduate 34 34 68 Graduate/professional degree 20 27 No answer/missing 10 16 26 Employment None/retired/disabled 312 64 332 65 644 65 Part time 77 16 81 16 158 16 0.56 Full time 95 20 91 18 186 19 No answer/missing Marital status Single/never married 129 27 127 25 256 26 Married/with partner 168 35 194 38 362 36 0.51 Divorced/separated/widowed 188 39 187 37 375 38 No answer/missing Household size 153 32 151 30 304 31 151 31 133 26 284 29 48 10 65 13 113 11 55 11 62 12 117 12 0.20 5+ 78 16 95 19 173 17 No answer/missing Household status Own home 126 26 135 27 261 26 Rent 283 58 279 55 562 57 With friends or family 51 11 62 12 113 11 Other 0.31 Assisted/group home 12 Homeless 14 11 25 No answer/missing 10 12 Household income <$10,000 251 52 235 46 486 49
$10,000–$19,999 122 25 149 29 271 27
$20,000–$29,999 47 10 57 11 104 10
$30,000–$39,999 23 29 52 0.53
$40,000–$49,999 13
$50,000+ 26 26 52 No answer/missing 15 Insurance status Private 52 11 60 12 112 11 Public 265 55 263 52 528 53 Uninsured 168 35 185 36 353 36 0.65 No answer/missing Variable Navigated Control Combined P Mean SD Mean SD Mean SD Charlson comorbidity index score 1.1 1.9 1.1 1.7 1.1 1.8 0.78 Number of months in study 6.64 6.59 7.08 6.24 6.87 6.42 0.28 Abbreviations: GED, General Educational Development (equivalent to high school diploma in US); AA, Associate Degree. Overall demographics were notable for high minority representation (53% Hispanic, 24% White, 19% Black), a high percentage of primarily Spanish speakers (30%), low education attainment (40% with some high school or less); low annual household income (less than$10,000 in 49%), high unemployment (65%), and high uninsured rate (36%). These and other baseline subject characteristics were well balanced between the control and intervention arms.

### Time to resolution—overall

Time to resolution for all cancer screening combined was significantly shorter in the navigated group, (stratified log rank test, P < 0.001). From visual inspection of Kaplan–Meier plots (Figs. 2 and 3), navigation especially improves adherence to 6-month follow-up for the breast screening population, as shown most clearly in the BIRADS 0 and BIRADS 3 groups, whereas the impact of navigation in colorectal and prostate screening was most pronounced in the 3- to 6-month period. Details for each stratum are described below.

Figure 2.

Kaplan–Meier estimates for proportion of subjects with resolution by days from initial screening: breast screening subjects—BIRADS 0 and BIRADS 3.

Figure 2.

Kaplan–Meier estimates for proportion of subjects with resolution by days from initial screening: breast screening subjects—BIRADS 0 and BIRADS 3.

Close modal
Figure 3.

Kaplan–Meier estimates for proportion of subjects with resolution by days from initial screening: colorectal and prostate screening subjects.

Figure 3.

Kaplan–Meier estimates for proportion of subjects with resolution by days from initial screening: colorectal and prostate screening subjects.

Close modal

### Time to resolution—breast cancer screening

Figure 2 shows the Kaplan–Meier plots for time to resolution by navigation versus control arms for the mammographic abnormalities BIRADS 0 and BIRADS 3 at time of enrollment. Benefit was found especially in subjects presenting with mammographic BIRADS 3 (P = 0.0003), was of borderline significance for patients with mammographic BIRADS 0 (P = 0.09), but was not significant for those with BIRADS 4/5 or abnormal clinical breast examinations (not shown).

### Time to resolution—colorectal cancer screening

Figure 3 shows the Kaplan–Meier plot for time to resolution of abnormal screening test for colorectal cancer by navigation versus control arms. Navigation significantly shortened the time for colorectal screening resolution (P < 0.002).

### Time to resolution—prostate cancer screening

Figure 3 also shows the Kaplan–Meier plot for time to resolution of abnormal screening for prostate cancer by navigation versus control arms. There is a strong trend for navigation to shorten the time for prostate cancer screening resolution (P < 0.01), although the confidence intervals for point estimates more than the entire range of follow-up displayed in Fig. 3 do overlap.

### Effect modification

Effect modification was tested using interactions of covariates with the 3 time-varying indicators of time in models that included the covariate as a main effect. There were no significant main effects by age, language, marital status, insurance, gender (for colorectal), number of baseline barriers, and comorbidity; nor was there evidence that these patient characteristics modified the effect of navigation.

Patients in the navigated group were more likely to complete diagnostic follow-up and resolution of initial abnormal screening tests as compared with the control patients. As can be seen in Table 2, with all 3 cancer screening sites combined, diagnostic resolution was achieved in 88% versus 70% in the navigated versus control groups, respectively (P < 0.001). For the abnormal breast screening group, 92% of the navigated patients reached diagnostic resolution of the initial abnormal test, as compared with 77% for the control patients (P < 0.001). For the abnormal colorectal screening group, 79% reached diagnostic resolution in the navigated group versus 58% in the control group (P < 0.002), and for the abnormal prostate screening group, diagnostic resolution for the navigated group was 84% as compared with 64% in the control group (P = 0.01).

Table 2.

Definitive tests and diagnostic resolution

NavigatedControlP
SiteTestCount%Count%
All
No cancer 388 80% 324 63% <0.001
No resolution 58 12% 150 30%
Cancers diagnosed 39 8% 34 7% 0.42
Total 485  508
Breast
Excisional biopsy 3% 3%
Stereotactic core biopsy 18 6% 20 6%
US-guided core biopsy 23 7% 23 7%
Core biopsy w/o imaging guidance 1% 1%
Fine needle aspiration biopsy 0% 0%
Breast MRI 0% 0%
Breast US 61 20% 57 18%
Diagnostic mammogram 165 54% 126 39%
Clinical assessment 0% 1%
Other 2% 2%
Total resolved 284 92% 245 77% <0.001
No resolution 24 8% 75 23%
Cancers diagnosed 13 4% 12 4% 0.76
Total 308  320
Colorectal
Colonoscopy w/o biopsy 54 47% 36 30%
Sigmoidoscopy w/o biopsy 0% 1%
Colonoscopy w/o biopsy 29 25% 28 23%
Sigmoidoscopy only 1% 2%
Virtual colonoscopy 0% 0%
Barium enema 0% 0%
Other 5% 2%
Total resolved 90 79% 70 58% <0.002
No resolution 24 21% 51 42%
Cancers diagnosed 3% 4% 0.53
Total 114  121
Prostate
TRUS-guided biopsy 42 67% 34 51%
Ultrasound 0% 0%
Follow-up PSA 5% 1%
Clinical assessment 8% 10%
Other 5% 1%
Total resolved 53 84% 43 64% 0.01
No resolution 10 16% 24 36%
Cancers diagnosed 23 37% 17 25% 0.17
Total 63  67
NavigatedControlP
SiteTestCount%Count%
All
No cancer 388 80% 324 63% <0.001
No resolution 58 12% 150 30%
Cancers diagnosed 39 8% 34 7% 0.42
Total 485  508
Breast
Excisional biopsy 3% 3%
Stereotactic core biopsy 18 6% 20 6%
US-guided core biopsy 23 7% 23 7%
Core biopsy w/o imaging guidance 1% 1%
Fine needle aspiration biopsy 0% 0%
Breast MRI 0% 0%
Breast US 61 20% 57 18%
Diagnostic mammogram 165 54% 126 39%
Clinical assessment 0% 1%
Other 2% 2%
Total resolved 284 92% 245 77% <0.001
No resolution 24 8% 75 23%
Cancers diagnosed 13 4% 12 4% 0.76
Total 308  320
Colorectal
Colonoscopy w/o biopsy 54 47% 36 30%
Sigmoidoscopy w/o biopsy 0% 1%
Colonoscopy w/o biopsy 29 25% 28 23%
Sigmoidoscopy only 1% 2%
Virtual colonoscopy 0% 0%
Barium enema 0% 0%
Other 5% 2%
Total resolved 90 79% 70 58% <0.002
No resolution 24 21% 51 42%
Cancers diagnosed 3% 4% 0.53
Total 114  121
Prostate
TRUS-guided biopsy 42 67% 34 51%
Ultrasound 0% 0%
Follow-up PSA 5% 1%
Clinical assessment 8% 10%
Other 5% 1%
Total resolved 53 84% 43 64% 0.01
No resolution 10 16% 24 36%
Cancers diagnosed 23 37% 17 25% 0.17
Total 63  67

### Number and stage of cancers detected

The number of cancers diagnosed for each cancer screening group was similar (Table 2), and the stage at which these cancers were diagnosed did not differ significantly between navigated and control groups (not shown). Although there were more cancers diagnosed in the navigated prostate screening group (23 patients) as compared with the control group (17 patients), this was not statistically significant due to the lower number of prostate screening patients.

Although study designs differed among the 10 PNRP participating institutions, the major aims and outcomes were the same. For assessing time to resolution of an abnormal screening finding, the Denver PNRP was the only one of the participating institutions with an individually randomized study design, although other institutions used group-randomized or prospective cohort controls. The results of our prospectively randomized controlled trial show that patient navigation positively impacts the time to resolution of abnormal screening tests for breast, colorectal, and prostate cancers in a medically underserved population, and significantly improves the percentage of patients reaching diagnostic resolution.

A number of prior studies have reported that patient navigation resulted in improved resolution of mammographic screening abnormality, ranging from 21% to 29% when compared with control subjects [reviewed by Wells and colleagues, 2008 (1) and Paskett and colleagues, 2011 (14)]. However, only 2 of 8 reported studies randomly assigned subjects to a patient navigation and a no-navigation control group (9, 15). The study by Ell and colleagues randomized women with abnormal mammograms to either the navigation intervention group (96 women) or no-navigation control group (108 women). Their results showed that the intervention group was more likely to be adherent through diagnostic resolution than the control group (90% vs. 66%, OR = 4.48, P < 0.001; 15). In the study by Ferrante and colleagues, 55 subjects were randomized to the navigation intervention and 50 to usual care. Women in the navigation group had shorter times to diagnostic resolution (mean of 25 vs. 43 days, P = 0.001) with 94% of navigated patients achieving diagnostic resolution versus 78% in the control group (9).

These results are of similar magnitude to those reported in our study, namely diagnostic resolution rates of 92% for navigated patients with abnormal mammographic findings as compared with 77% for control patients. In our study, the benefits from navigation were especially striking for patients with initial BIRADS 3 mammographic abnormalities, and to a lesser extent in those with BIRADS 0, where we saw a significant improvement in the adherence rates to the recommended 6-month repeat mammogram in the control group (80% vs. 60% resolution at 240 days). BIRADS 4 and 5 patients, on the other hand, saw rapidly rising and high-diagnostic resolution rates in both navigated and control groups (85% and 80%, respectively, at 90 days), most likely driven by the urgency of the serious mammographic abnormalities reflected by this group of patients.

This is among the first publication, to our knowledge, of a randomized controlled trial evaluating the impact of patient navigation on the diagnostic resolution in patients with abnormal screening tests for colorectal cancer (2, 3) and is the first for prostate cancer navigation. As with our patients with abnormal mammograms, we found that patient navigation shortens the time to diagnostic resolution and significantly increased the proportion of patients adherent through diagnostic resolution for colorectal screening abnormalities (positive fecal occult blood, rectal bleeding, abnormal digital rectal examination, or sigmoidoscopy). For patients with prostate cancer screening abnormalities (elevated PSA, abnormal digital rectal examination), navigation led to improvement in time to and in percentage completing diagnostic resolution, although to a somewhat lesser degree than for patients with colorectal cancer screening abnormalities. This is most likely because of the smaller numbers enrolled and a lower sense of urgency in evaluating these abnormalities by both patients and providers.

The populations studied by Ell and colleagues and Ferrante and colleagues included a large proportion of minority patients, as did our study population (9, 15). All 3 populations are notable for low-annual household income, low-education attainment, high unemployment, high uninsured rate, and high comorbidity. These populations encounter numerous barriers to accessing and maintaining health care, including the timely resolution of abnormal cancer screening findings (16, 17). Patient navigation is designed to assist patients to overcome these barriers, provide a safe, nonthreatening forum for deliberation, and lend emotional support to patient decision making and dealing with an adverse screening outcome.

Study strengths include the use of a prospective individually randomized controlled clinical trial design, large number of patients enrolled compared with prior studies, a high-enrollment rate of 77% among subjects able to be contacted, and utilization of a centralized electronic medical record system to identify potential participants and to document study outcomes. We enrolled significant numbers into the breast (628) and colorectal (235) cancer abnormal screening components of the study, with fewer patients with abnormal prostate screening (130). Our results are especially applicable to other safety net health care settings in light of our largely minority, low-income and uninsured study population, with multiple challenges in their ability to adhere to demanding diagnostic evaluations.

An additional strength included consistent interpretation of screening mammograms by 2 mammographers over the 4 years of participant enrollment. The percentage of BIRADS 0 and 3 remained stable over that period, although the percentages of readings falling into these 2 categories were consistently higher than at other participating institutions. We also experienced little staff turnover and were able to retain a cadre of experienced navigators throughout.

Limitations of this study include the delay between identifying eligible patients, contacting them for consent, enrolling them into the study, and initiating the intervention. Our electronic medical record greatly assisted in the identification of potentially eligible patients; however, up-to-date contact information was not always available. In addition, the intervention could not be initiated until consent had been obtained. This delay to enrollment and initiation of the navigation or control intervention was similar for both groups; mean of 36.6 days (SD ± 28.3) for the navigated group versus mean of 34.1 days (SD ± 26.4) for the control group. It is possible that this delay blunted the impact of navigation during the early part of patients' dealing with the abnormal screen results. In future studies, methods need to be explored to shorten this delay, allowing for earlier intervention to further improve outcomes.

A second limitation is the relatively small number of patients entered into the study with abnormal prostate cancer screening findings. Screening practice has undergone changes over the past decade, resulting in fewer men being screened for prostate cancer, especially in the older age groups (18). Although our study still suggests a benefit for the abnormal prostate cancer screening patients, the smaller numbers resulted in borderline statistical significance.

Patient navigation is an effective strategy for improving adherence to diagnostic evaluation and resolution, regardless of ethnicity, insurance status, and education level. Our study provides additional evidence of its effectiveness in patients with abnormal mammographic findings and provides new information in patients with abnormal screening results for colorectal and prostate cancer in an underserved patient population.

No potential conflicts of interest were disclosed.

Conception and design: P.C. Raich, E. Whitley, D. Fairclough

Development of methodology: P.C. Raich, E. Whitley

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P.C. Raich, E. Whitley, P. Valverde

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P.C. Raich, E. Whitley, W. Thorland, D. Fairclough

Writing, review, and/or revision of the manuscript: P.C. Raich, E. Whitley, P. Valverde, D. Fairclough

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P.C. Raich, E. Whitley, W. Thorland, P. Valverde

Study supervision: P.C. Raich, E. Whitley, P. Valverde

The authors thank Denver Health patients and patient navigators, as well as to colleagues Rachel Everhart, MS, Kristin Kilbourn, PhD, MPH, Betsy Risendal, PhD, Al Marcus, PhD, and Sandra Vannice, RN, CNS, AOCN.

Supported by grant 5U01CA6903 from the Center to Reduce Health Disparities of the National Cancer Institute, NIH.

1.
Wells
KJ
,
Battaglia
TA
,
Dudley
DJ
,
Garcia
R
,
Greene
A
,
Calhoun
E
, et al
Patient Navigation: State of the art or is it science?
Cancer
2008
;
113
:
1999
2010
.
2.
Lasser
KE
,
Murillo
J
,
Lisboa
BA
,
Casmir
AN
,
Valley-Shah
L
,
Emmons
KM
, et al
Colorectal cancer screening among ethnically diverse, low-income patients
.
Arch Int Med
2011
;
171
:
906
12
.
3.
Percac-Lima
S
,
Grant
RW
,
Green
AR
,
Ashburner
JM
,
Gamba
G
,
OO
S
, et al
A culturally tailored navigator program for colorectal cancer screening in a community health center: a randomized controlled trial
.
J Gen Int Med
2009
;
24
:
211
7
.
4.
Taylor
VM
,
Jackson
JC
,
Yasui
Y
,
Nguyen
TT
,
Woodall
E
,
Acorda
E
, et al
Evaluation of a cervical cancer control intervention using lay health workers for Vietnamese American women
.
Am J Public Health
2010
;
100
:
1924
9
.
5.
Nguyen
TT
,
McPhee
SJ
,
Gildengorin
G
,
Nguyen
T
,
Wong
C
,
Lai
KQ
, et al
Papanicolaou testing among Vietnamese Americans: results of a multifaceted intervention
.
Am J Prev Med
2006
;
31
:
1
9
.
6.
Ahmed
NU
,
Haber
G
,
Semenya
KA
,
Hargreaves
MK
.
Randomized controlled trial of mammography intervention in insured very low-income women
.
Cancer Epidemiol Biomarkers Prev
2010
;
19
:
1790
8
.
7.
Christie
J
,
Itzkowitz
S
,
Lihau-Nkanza
I
,
Castillo
A
,
Redd
W
,
Jandorf
L
.
A randomized controlled trial using patient navigation to increase colonoscopy screening among low income minorities
.
J Natl Med Assoc
2008
;
100
:
278
84
.
8.
Jandorf
L
,
Gutierrez
Y
,
Lopez
J
,
Christie
J
,
Itzkowitz
SH
.
Use of a patient navigator to increase colorectal screening in an urban neighborhood clinic
.
J Urban Health
2005
;
82
:
216
24
.
9.
Ferrante
JM
,
Chen
P-H
,
Kim
S
.
The effect of patient navigation on time to diagnosis, anxiety, and satisfaction in urban minority women with abnormal mammograms: a randomized controlled trial
.
J Urban Health
2008
;
85
:
114
24
.
10.
Freund
KM
,
Battaglia
TA
,
Calhoun
E
,
Dudley
DJ
,
Fiscella
K
,
E
, et al
National Cancer Institute Patient Navigation Research Program: methods, protocol and measures
.
Cancer
2008
;
113
:
3391
9
.
11.
Charlson
ME
,
Pompei
P
,
Ales
KL
,
MacKenzie
CR
.
A new method of classifying prognostic comorbidity in longitudinal studies: development and validation
.
J Chronic Dis
1987
;
40
:
373
83
.
12.
Wagner
EH
.
Chronic disease management: what will it take to improve care for chronic illness?
Eff Clin Pract
1998
;
1
:
2
4
.
13.
Calhoun
EA
,
Whitley
EM
,
Esparza
A
,
Ness
E
,
Valverde
PA
,
Greene
A
, et al
A national patient navigator training program
.
Health Promot Pract
2010
;
11
:
205
15
.
14.
ED
,
Harrop
P
,
Wells
KJ
.
Patient navigation: an update on the state of the science
.
Cancer
2011
;
61
:
237
49
.
15.
Ell
K
,
Vourlekis
B
,
Lee
PJ
,
Xie
B
.
Patient navigation and case management following an abnormal mammogram: a randomized clinical trial
.
Prev Med
2007
;
44
:
26
33
.
16.
Raich
PC
,
Valverde
P
,
Thorland
W
,
Kilbourn
K
,
Risendal
B
,
Borrayo
E
, et al
The relationship between patient co-morbidities and perceived barriers in a group of underserved patients with cancer or at risk for cancer
. In:
Poster presentation at the 30th Annual Meeting of the Society of Behavioral Medicine; 2009 Apr 25
;
:
Springer
;
2009
(
Poster #D-28
).
17.
Valverde
P
,
Thorland
W
,
Risendal
B
,
Whitley
E
,
Raich
P
.
Gender differences in barriers experienced by low-income participants of a patient navigator clinical trial
. In:
Poster presentation at the Annual Meeting of the American Association of Cancer Research; 2009 February 3–6
;
Phoenix, AZ
:
AACR
;
2009
.
18.
Chou
R
,
Croswell
JM
,
Dana
T
,
Bougatsos
C
,
Blazina
I
,
Rongwei
F
, et al
Screening for prostate cancer: a review of the evidence for the U.S. Preventive Services Task Force
.
Ann Int Med
2011
;
155
:
762
71
.