The authors evaluated the effectiveness and cost effectiveness of 2 interventions designed to promote colorectal cancer (CRC) screening in safety-net settings.
The authors evaluated the effectiveness and cost effectiveness of 2 interventions designed to promote colorectal cancer (CRC) screening in safety-net settings.
A 3-arm, quasi-experimental evaluation was conducted among 8 clinics in Louisiana. Screening efforts included: 1) enhanced usual care, 2) literacy-informed education of patients, and 3) education plus nurse support. Overall, 961 average-risk patients ages 50 to 85 years were eligible for routine CRC screening and were recruited. Outcomes included CRC screening completion and incremental cost effectiveness using literacy-informed education of patients and education plus nurse support versus enhanced usual care.
The baseline screening rate was <3%. After the interventions, the screening rate was 38.6% with enhanced usual care, 57.1% with education, and 60.6% with education that included additional nurse support. After adjusting for age, race, sex, and literacy, patients who received education alone were not more likely to complete screening than those who received enhanced usual care; and those who received additional nurse support were 1.60-fold more likely to complete screening than those who received enhanced usual care (95% confidence interval, 1.06-2.42; P = .024). The incremental cost per additional individual screened was $1337 for education plus nurse support over enhanced usual care.
Fecal occult blood test rates were increased beyond enhanced usual care by providing brief education and nurse support but not by providing education alone. More cost-effective alternatives to nurse support need to be investigated. Cancer 2013;119:3879–3886. © 2013 American Cancer Society.
Of all cancer screening initiatives, colorectal cancer (CRC) screening rates are the lowest. Although CRC screening rates have improved in the general population, disparities persist among adults who have lower socioeconomic status, limited health literacy, are members of racial/ethnic minority groups, and/or live in rural locations.[1-5] Reducing these disparities is a primary public health objective.
Numerous studies have examined barriers and facilitators to CRC screening among vulnerable populations.[2, 7-17] Limited health literacy has been linked specifically to less CRC knowledge, negative attitudes toward screening, lower self-efficacy, and less likelihood of screening completion.[7, 18-23] Because evidence clearly demonstrates an association between limited health literacy and poor health outcomes, the US Department of Health and Human Services has called for the dissemination of health information that is accessible, understandable, and actionable. Despite this, few low-literacy initiatives have been specifically developed to promote CRC screening.[20, 25, 26] Even fewer are suitable for use in rural and urban community clinics. The Colorectal Cancer Roundtable recently recommended focusing future initiatives on Federally Qualified Health Centers (FQHCs) to address national screening challenges.
In response, our team designed, implemented, and evaluated a multifaceted intervention to increase CRC screening rates among low-income, uninsured patients receiving care from FQHCs. FQHCs provide primary care to more than 20 million individuals in the United States. Our study was designed to test 2 strategies to promote CRC screening: 1) the use of literacy-informed educational materials with accompanying “teach back” to confirm comprehension and 2) the use of this educational strategy with support and follow-up by a nurse manager. These strategies promoted the use of fecal occult blood tests (FOBTs), the most feasible, cost-effective screening option for low-income and uninsured patients.[30, 31] Each intervention arm was compared with an enhanced usual care arm. Given the resource-constrained FQHC environment, we evaluated not only the effectiveness of our 2 interventions but also their cost effectiveness.
This study took place between May 2008 and August 2011 in North Louisiana. Two intervention strategies were tested in a 3-arm study: 1) enhanced usual care, in which patients were given a recommendation for CRC screening and an FOBT kit; 2) an educational strategy, in which patients received enhanced usual care plus brief education that included pamphlet, video, and simplified FOBT instructions; and 3) the nurse support arm, in which patients received enhanced usual care, the educational strategy, and additional nurse support and follow-up to encourage the completion of CRC screening.
The target population was the 5 FQHCs in predominantly rural Louisiana. Three FQHCs participated in this study, and the other 2 were involved in cancer screening programs at the time. Each FQHC was randomly assigned to 1 of the 3 study arms as part of this 3-arm, quasiexperimental evaluation. Each study FQHC was affiliated with multiple clinics, which were assigned to the same arm as their parent FQHC. At the time of randomization, there were a total of 7 clinics associated with the 3 FQHCs, including 2 in enhanced usual care arm, 2 in the educational strategy arm, and 3 in the nurse support arm. After the first year of the study, 1 additional clinic was enrolled in the enhanced usual care arm because of limited patient recruitment in that arm. The 8 participating clinics were located in 8 towns within 7 parishes across the state. Six clinics were located in rural towns, with populations ranging from 450 to 13,000; and 2 clinics were located in low-income areas of cities with populations of 63,000 and 199,000, respectively. Baseline rates of CRC screening at each of the 8 study clinics ranged from 1% to 2%.
Patients were recruited through a multistep process. First, while taking patients' vital signs, a medical assistant at each clinic identified potentially eligible participants by the age listed in their chart (range 50-85 years). Medical assistants were trained to ask patients ages 50 to 85 years whether they would be willing to talk to an onsite research assistant (RA) about participating in a CRC screening study before their physician encounter. Those who were interested met with the RA, who screened them for further eligibility: 1) English-speaking, 2) current clinic patient, 3) not requiring screening at an earlier age according to American Cancer Society (ACS) guidelines, 4) not up-to-date with US Preventives Services Task Force (USPSTS) CRC screening recommendations (ie, a FOBT annually, flexible sigmoidoscopy every 5 years, or colonoscopy every 10 years), and 5) not having an acute medical concern. The Louisiana State University Health Sciences Center-Shreveport Institutional Review Board approved the study. Each patient received $10 for their participation in the baseline survey.
The study interview included demographic and CRC screening items from validated questionnaires used previously by the authors.[11, 33, 34] A detailed description of the survey, which was written on a fourth-grade level and was administered orally, has been reported previously. Literacy was assessed using the Rapid Estimate of Adult Literacy in Medicine (REALM). Raw REALM scores (range, 0-66) can be converted into reading grade levels that correlate with literacy skills.
The intervention components were designed by following health literacy best practices and the theory of health learning capacity to simplify the complexity of independently completing an FOBT.[29, 37] Health belief model and social cognitive theories guided the framing of intervention content to address the salience of CRC screening and the need to take action.[38, 39] The educational strategy and nurse support arms were designed to overcome key patient barriers to CRC screening such as access to tests, limited knowledge, negative beliefs, poor self-efficacy and lack of motivation. The education materials were developed and pilot tested using focus groups and cognitive interviews with FQHC patients, providers, and our community advisory boards to help ensure they were useful, understandable, appealing, and culturally appropriate. The nurse support arm was included as an intervention strategy to determine the added benefit of more in-depth counseling and telephone follow-up support to encourage FOBT completion.
All staff and providers in each clinic attended a 2-hour in-service on CRC screening and an orientation to the study during a quarterly clinic meeting. RA training in the enhanced usual care arm included practice interviewing patients and administrating the survey and literacy test. RAs were given a script for recommending CRC screening and introducing the FOBT kit. For the educational strategy arm, RAs were given additional training in using health literacy techniques. For the nurse support arm, the nurse manager training also included motivational interviewing techniques, use of a tracking system, and a protocol for contacting patients and assisting them with navigation if a test was positive.
At enrollment, after completing the structured interview, the clinic-based RA gave patients a recommendation to complete CRC screening, the FOBT kit, and a suggestion to talk with their primary care provider about screening during their visit that day. Patients returned FOBTs to the clinic by mail using a preaddressed, stamped envelope. Regular clinic protocol was followed for positive test results and if diagnostic testing was needed.
At enrollment, after completing the structured interview, the clinic RA gave patients the same FOBT kit and recommendation that were given in the enhanced usual care arm, but simplified FOBT instructions were provided. In addition, the RA gave a brief literacy-informed educational intervention that included a colorful, illustrated CRC pamphlet written on a fifth-grade level that provided actionable information organized from a patient's perspective as well as a short video that was developed by the authors that captured FQHC patients discussing barriers and facilitators to screening and a physician making a recommendation while showing key steps in FOBT completion. The education also included the RA giving a concrete demonstration of FOBT instructions. The RA used techniques like “teach back” to confirm the accuracy and completeness of patients' understanding. Patients returned FOBTs to the clinic by mail. Tracking and follow-up were done in the same manner as the enhanced usual care arm.
At enrollment, after the structured interview, the nurse manager provided the same materials and FOBT instructions that were provided in the education arm before the patient's physician visit. The nurses used motivational interviewing techniques to identify and problem-solve barriers and motivate patients to complete FOBTs. To promote comprehension and confidence, the nurses often showed the patents how to complete the FOBTs and called within 1 week to ask whether they had questions and, if necessary, to review the instructions. If patients did not return their FOBT, the nurses followed up by telephone within 2 weeks and again in 1 month. If the results were positive, the nurse manager called patients to discuss the results, facilitate appointments with their primary care provider, and, if indicated, schedule patients for a diagnostic colonoscopy at the appropriate treatment center.
The EZ Detect FOBT kit (Biomerica, Inc., Irvine, Calif) was chosen based on the simplicity of the test from a patient perspective and recommendations from FQHC patients, providers, CEOs, and our community advisory board. The test requires 3 separate bowel movements but no stool handling. Sensitivity and specificity of the kit have been reported previously. Previous research also indicated that patients overwhelmingly preferred the EZ Detect many other hemoccult cards (92% vs 8%). Eligible patient FOBT completion at 3 months postenrollment was the primary outcome measure, as documented by the clinic nurse (enhanced usual care and educational strategy arms) or nurse manager (nurse support arm).
FOBT completion rates were defined as the percentage of FOBT cards returned to the clinic. A barrier scale was calculated using 4 questions concerning confusion, embarrassment, trouble, and messiness related to the FOBT test. This scale ranged from 4 to 20, and high values indicated that the participant thought these were barriers. To examine whether patients in the study arms differed on continuous baseline characteristics of age and the barrier index, analysis of variance was used. Chi-square tests were performed for categorical factors, including literacy level. Screening ratios were defined as the ratio of FOBT completion rates between 2 groups. Both screening ratios and pairwise tests for FOBT completion were calculated using logistic regression, which accounted for clustering by clinic. Multivariate analyses were adjusted for age, race, sex, and literacy level. Tests for interaction between literacy level and study arm were assessed.
Cost data were collected from purchase orders, receipts, and questioning research staff. Total incremental costs and additional numbers of individuals screened were calculated for the education arm and the nurse case manager arm (comparison arms) over the enhanced usual care arm (reference arm). Incremental costs for the education arm were writing, producing, and editing a DVD ($10,000); brochures ($2000); and RA ($1036). Costs for the nurse arm were postcards ($60) and 40% of salaries for 2 nurses for 1 year ($106,280). Comparison arm costs and numbers screened were normalized to the reference arm to account for differences in sample size. The incremental cost-effectiveness ratio was calculated as the total incremental cost of a comparison arm relative to the reference arm divided by the total number of additional individuals screened.
Overall, 1055 patients were identified who met the age criteria, and, of these, 33 patients (3.1%) refused to participate and 61 (5.8%) were ineligible because they were up to date on CRC screening. In total, 961 patients were consented and enrolled with a determined cooperation rate of 91.1%. Baseline characteristics are compared among groups in Table 1. Participants ranged in age from 50 years to 85 years, 77% were women. The majority (67%) were and the majority (67%) were African American. Over half (56%) had limited literacy (ie, less than ninth-grade level). There were significant differences across groups for age, race/ethnicity, marital status, literacy, prior recommendation, previous FOBT, wanting to know whether they had CRC, and positive beliefs concerning FOBT efficacy.
|No. of Patients (%)|
|Characteristic||All Patients, n = 961||Enhanced Usual Care, n = 275||Education, n = 282||Nurse, n = 404||P|
|Age: Mean ± SD, y||58.4 ± 7.3||57.7 ± 7.5||57.8 ± 6.5||59.2 ± 7.5||.014|
|Age category, y|
|50–59||611 (64)||190 (69)||181 (64)||240 (59)||.11|
|60–69||265 (28)||63 (23)||80 (28)||122 (30)|
|70–85||85 (9)||22 (8)||21 (7)||42 (10)|
|Women||740 (77)||207 (75)||224 (79)||309 (77)||.50|
|≤High school||313 (33)||98 (36)||92 (33)||123 (31)||.26|
|High school graduate||435 (45)||109 (40)||139 (49)||187 (47)|
|Some college||157 (16)||50 (18)||40 (14)||67 (17)|
|≥College graduate||53 (6)||18 (7)||11 (4)||24 (6)|
|African American||645 (67)||199 (72)||114 (40)||332 (83)||< .0001|
|Caucasian/Hispanic||313 (33)||76 (28)||168 (60)||69 (17)|
|Single||276 (29)||62 (23)||55 (20)||159 (40)||< .0001|
|Married||330 (34)||101 (37)||142 (50)||87 (22)|
|Separated||66 (7)||22 (8)||14 (5)||30 (7)|
|Divorced||155 (16)||47 (17)||38 (13)||70 (17)|
|Widowed||131 (14)||43 (16)||33 (12)||55 (14)|
|Literacy level: REALM score|
|Limited: 0–60||537 (56)||188 (68)||98 (35)||251 (62)||<.0001|
|Adequate: 61–66||424 (44)||87 (32)||184 (65)||153 (38)|
|Seen physician in past 12 mo||849 (89)||236 (86)||258 (91)||355 (89)||.09|
|Prior recommendation||357 (39)||96 (35)||83 (29)||178 (48)||< .0001|
|Ever completed an FOBT||262 (28)||62 (23)||26 (9)||174 (47)||< .0001|
|Would want to know if have CRC|
|Yes||837 (90)||242 (90)||261 (93)||334 (89)||.03|
|No||56 (6)||12 (4)||15 (5)||29 (8)|
|Don't know||34 (4)||17 (6)||6 (2)||11 (3)|
|FOBT finds CRC early|
|Strongly agree/agree||889 (96)||255 (94)||276 (98)||358 (96)||.24|
|Disagree/strongly disagree||9 (1)||3 (1)||2 (1)||4 (1)|
|Don't know||29 (3)||13 (5)||4 (1)||12 (3)|
|FOBT decreases chances of dying from CRC|
|Strongly agree/agree||742 (80)||194 (72)||243 (86)||305 (82)||.0001|
|Disagree/strongly disagree||109 (12)||43 (16)||29 (10)||37 (10)|
|Don't know||76 (8)||34 (13)||10 (4)||32 (9)|
|Barrier index, Mean ± SD||8.98 ± 2.29||9.42 ± 2.24||8.04 ± 2.11||9.38 ± 2.26||< .0001|
The FOBT return rate was 38.6% in the enhanced usual care arm, 57.1% in the education arm and 60.6% in the nurse support arm (Table 2). Adjusting for age, race, sex, and literacy, there was a significant difference in the screening ratio across arms (P = .012). Participants in the nurse support arm were 1.60 times more likely to be screened (95% confidence interval, 1.06-2.42; P = .024) compared with those in the enhanced usual care arm, but they were no more likely to be screened than those in the educational arm (P = .09). Those in the educational arm were not more likely to be screened compared with those in the enhanced usual care arm (P = .20).
|No. of Patients (%)|
|Outcome Measure||All Patients, n = 961||Enhanced Usual Care, n = 275||Education, n = 282||Nurse, n = 404||P|
|FOBT returned: Screened||512 (53)||106 (38.6)||161 (57.1)||245 (60.6)||< .0001|
|FOBT not returned||449 (47)||169 (61.4)||121 (42.9)||159 (39.4)|
|Screening ratio [95% CI]||1.00||1.39 [0.86–2.22]||1.52 [1.00–2.31]||.11|
|Screening ratio [95% CI]||1.00||1.10 [0.86–1.40]|
|Adjusted screening ratio [95% CI]||1.00||1.36 [0.85–2.18]||1.60 [1.06–2.42]||.012|
|Adjusted screening ratio [95% CI]||1.00||1.18 [0.97–1.42]|
Table 3 indicates there were significant differences across study arms in FOBT completion among patients with limited literacy (P = .006) but not among those with adequate literacy (P = .064). The interaction term for study arm and literacy level was not statistically significant (P = .80), indicating that the screening ratios did not differ significantly between literacy groups.
|No. of Patients (%)|
|Outcome Measure||All Patients||Enhanced Usual Care||Education||Nurse||P|
|Limited literacy||N = 537||N = 188||N = 98||N = 251|
|FOBT returned: Screened||264 (49)||69 (36.7)||51 (52)||144 (57.4)||.0006|
|FOBT not returned||273 (51)||119 (63.3)||47 (48)||107 (42.6)|
|Screening ratio [95% CI]||1.00||1.39 [0.97–1.98]||1.60 [1.19–2.16]|
|Screening ratio [95% CI]||1.00||1.15 [0.99–1.34]|
|Adequate literacy||N = 424||N = 87||N = 184||N = 153|
|FOBT returned: Screened||248 (58)||37 (42.5)||110 (59.8)||101 (66)||.064|
|FOBT not returned||176 (42)||53 (57.5)||74 (40.2)||52 (34)|
|Screening ratio [95% CI]||1.00||1.27 [0.74–2.20]||1.58 [0.99–2.54]|
|Screening ratio [95% CI]||1.00||1.24 [0.92–1.67]|
Table 4 presents the incremental cost-effectiveness ratio of the educational arm relative to the enhanced usual care arm and for the nurse support arm relative to the educational arm. The incremental cost of the educational intervention per additional individual screened was $250 over enhanced usual care, whereas the incremental cost of the nurse intervention per additional individual screened was $1337 over enhanced usual care.
|Row ID||Variable||Education (Comparison Arm) vs EUC (Reference Arm)||Nurse (Comparison Arm) vs EUC (Reference Arm)|
|Additional no. of individuals screened in comparison arm|
|A||Sample size in reference arm||275||275|
|B||No. screened in reference arm||106||106|
|C||Sample size in comparison arm||282||404|
|D||No. screened in comparison arm||161||245|
|E||No. screened in comparison arm normalized to size of reference arm||157||167|
|F||Additional no. screened in comparison arm normalized to size of reference arm = row E − row B||51||61|
|Incremental costs of comparison arm, $|
|I||Total incremental costs||12,731||119,376|
|J||Total incremental costs normalized to size of reference arm, $||12,415||81,258|
|Incremental cost-effectiveness ratio = row J/row F, $||250||1337|
Among urban and rural southern FQHC patients, our study documented extremely low baseline CRC screening rates. Sizable gains over baseline were detected in screening completion in our enhanced usual care arm and literacy informed education arm; further benefit of our educational strategy combined with the support of a nurse manager was substantial. FOBT completion rates in our education with nurse support strategy were among the most successful reported in CRC screening interventions to date among lower income populations that have ranged from 43% to 70%.[30, 43-46]
Previous studies in safety-net clinics cited low rates of physician recommendation and inadequate insurance as causes of low CRC screening rates.[12, 28] Overcoming this barrier may be a reason for the substantial improvement in our enhanced usual care arm, in which nonmedical staff gave patients an FOBT kit, a CRC recommendation, and a suggestion to talk with their provider during their routine office visit. Recent studies demonstrate that some of the strongest interventions include the use of nonphysician staff to communicate with patients and offer FOBT kits[30, 43]
The patients who received nursing support were most likely to complete CRC screening. When nurses called the patients, the most common barrier for not returning FOBTs was confusion about instructions or losing the kit, rather than additional decision-making or cancer-related anxiety. Given the cost of the nurse, other strategies to prompt patients, such as letters, texting, automated calls, or personal calls by a less costly medical assistant, should be investigated. If a less expensive clinic staff were used, then estimates could fall to $389 per additional individual completing screening, representing approximately 33% of the cost of the nurse. However, this still may be cost prohibitive among FQHCs. With the increasing presence of electronic health records (EHRs), especially among community health centers, the amount of staff time dedicated to identifying and tracking patients also might be substantially reduced. None of the FQHCs in this investigation had an integrated EHR system at the time the study. Future studies that aim to evaluate other low-cost interventions also need to use currently recommended FOBTs, such as the fecal immunochemical test.
Our study had limitations. Differences were noted between arms in sociodemographic characteristics, perceived barriers, individuals wanting to know whether they had CRC, and belief that FOBTs would decrease chances of dying of CRC, but not for the primary outcome of screening rates. Therefore, adjustments for key variables were made in statistical analyses. Other limitations relate to the generalizability of our results; we included predominantly African American and female patients receiving care from FQHCs in 1 state. However, this is generally representative of FQHC populations in the southern United States.
It is noteworthy that our choice of the EZ Detect FOBT kit was determined both by currently available and USPSTF-recommended options at the time of the study onset as well as the perceived simplicity of the test from a patient's perspective. The EZ Detect kit demonstrated sensitivity and specificity comparable to those of the Hemoccult II FOBT (Beckman Coulter, Inc., Pasadena, Calif). However, USPSTF and ACS recommendations have changed, and currently both recommend immunochemical tests, which have superior sensitivity, and no longer recommend tests with less sensitivity, such as Hemoccult II or EZ Detect. By the time we were aware of this change, patient screening had been completed in the current study. On the basis of ACS and USPSTF guidelines, tests like EZ Detect and Hemoccult II should no longer be used for screening or research. Future efforts should consider how strategies similar to those we studied may incorporate the newer, recommended FOBTs among similar vulnerable populations.
Achieving the Healthy People 2020 goal of having >70% of eligible adults up to date with CRC screening guidelines may require a variety of approaches.[5, 30] Because the menu of screening options in FQHCs is often limited to annual FOBT, strategies are needed to overcome the challenges of limited resources for patient education and continued outreach. Given highly constrained resources, decision makers require information on the cost effectiveness of interventions. Future studies should use currently recommended tests that have superior sensitivity, employ strategies to make annual FOBT completion easier for patients to access and understand, as well as offer community clinics an effective, low-cost, and potentially sustainable means for screening.
Funding for this project was supported by the National Institutes of Health and National Cancer Institute (R01CA115869).
Ms. Reynolds, Dr. Arnold, Mr. Carias, and Ms. Platt have received grant support and support for travel to meetings from the National Cancer Institute (NCI), National Institutes of Health (NIH). Dr. Davis has received grant support and support for travel to meetings from the NCI, NIH; has received compensation as a consultant to McNeil and for lectures from Ochsner Health Systems and the NCI-supported Health Literacy Consortium of San Antonio; owns stock/stock options in Johnson & Johnson, McNeil, and Abbott Laboratories; and has received reimbursement from the State of Louisiana for travel to the National Colon Cancer Round Table. Dr. Rademaker has received grant support from the NIH; has received compensation as a consultant to the NIH, Georgetown University, and the American Association for Cancer Research; has various grants pending from the NIH; and has received reimbursement for travel, accommodations, and meeting expenses from the American Association for Cancer Research, Georgetown University, and the NIH.