Colorectal cancer knowledge and screening adherence among low-income Hispanic employees

Abstract

Hispanics have the lowest colorectal cancer (CRC) screening rates of all racial/ethnic groups and comprise the largest proportion of low-income manual laborers in the nation. We partnered with businesses to implement a community health worker (CHW)-led intervention among Hispanic workers in service-related and manual labor occupations, which often pay low wages and do not provide health insurance. CHWs measured knowledge, screening adherence and perceptions of CRC risk before and after educational interventions via interview. CHWs provided fecal immunochemical tests (FITs) to participants aged ≥50 years. Chi-square tests and logistic regression identified pre-intervention predictors of CRC knowledge of all participants and adherence among eligible participants. Adherence among participants increased from 40% (n = 307) pre-intervention to 66% post-intervention. Knowledge about CRC was associated with age ≥50 years (OR = 8.90 [95% CI = 2.61–30.35]; ref = 18–30) and perceived personal risk for CRC (Likely, OR = 3.06 [95% CI = 1.40–6.67]; ref = Not likely). Insurance status was associated with screening adherence pre-intervention (OR = 3.00 [95% CI 1.10–8.12]; ref = No insurance). Improvement in adherence post-intervention was associated with income between $25 000 and ≥$55 000 (OR = 8.49 [95% CI 1.49–48.32]; ref = $5000–<$10 000). Community-based health programs can improve CRC screening adherence among Hispanic workers in service-related and manual labor positions, but lowest-income workers may need additional support.

Introduction

Colorectal cancer (CRC) is the third leading cause of cancer-related death in the United States [1, 2]. CRC screenings reduce CRC mortality because they detect precancerous colon polyps needing removal. Screenings include colonoscopies, fecal occult blood tests (FOBTs) and fecal immunochemical tests (FITs) [3]. The US Preventive Services Task Force and other leading medical organizations recommend that screenings begin at age 45 years [4, 5]. Despite these recommendations nearly 40% of US adults aged ≥50 years, an estimated 20 million, have never been screened for CRC [3, 6, 7].

CRC is the second most common cancer among Hispanic men and women [8]. Hispanics have the lowest CRC screening rates of any racial/ethnic group [8–11]. In 2010, 47% of Hispanic adults aged ≥50 years had a CRC screening compared with 62% of non-Hispanic Whites and 56% of African Americans [3]. Because of this disparity Hispanics have the highest risk for advanced stage CRC diagnosis compared with other racial/ethnic groups [8, 12, 13]. The low CRC screening rate among Hispanics is attributed to low knowledge regarding CRC and screenings [14], lack of physician recommendation [15], inadequate linguistically appropriate educational materials [16], limited access to health insurance [17, 18] and a limited understanding of the connection between family history and personal cancer risk [19].

Multicomponent interventions designed for Hispanic populations with show great promise in increasing awareness of CRC screening [20, 21]. Previous US-based interventions among Hispanics that utilize community health educators [22], telephone care management with linguistic support [16, 23] and culturally appropriate storytelling techniques [24] improve knowledge about CRC and participation in CRC screenings.

Similar to national statistics, CRC screening adherence among Hispanics in Utah is lower than other racial/ethnic groups [25]. Hispanics are the largest and most rapidly growing minority group in Utah due to high birth rates and immigration [26]. They comprise 13.7% of the population [27] and 36% of Hispanics residing in Utah were born outside the United States. While Hispanic immigrants work in a variety of professional occupations nationwide, a large percent of workers in occupations related to service and manual labor are Hispanic [28, 29]. These occupations include jobs in construction (32.3% Hispanic), building cleaning and maintenance (36.7% Hispanic) and food preparation and service (24.9% Hispanic) [28, 29]. Job in these occupations typically does not provide health insurance or paid time off for medical care including CRC screenings.

We partnered with companies in Salt Lake County who employ a large proportion of Hispanic workers to implement a community health worker (CHW)-led intervention. We designed the intervention to increase knowledge of and adherence to breast, cervical and CRC screenings. This paper focuses on CRC. We identify participant characteristics associated with pre-intervention knowledge of CRC and CRC screenings. Among participants age ≥50 years, the recommended age of CRC screening at the time of this intervention, we identify characteristics associated with adherence to screenings pre-intervention and characteristics predictive of improvement in adherence post-intervention.

Methods

This study is part of a larger intervention examining the effects of a CHW-led educational intervention among Hispanic workers that occurred between January 2015 and February 2016 [30]. We adapted methods from the Prevention Care Management program from the National Cancer Institute Research-Tested Intervention Programs to tailor the intervention to the needs of the Hispanic working community [16, 23]. This intervention was a result of a partnership between researchers at Huntsman Cancer Institute; two nonprofit community health organizations that serve the Hispanic population in Salt Lake City, Utah and employ CHWs; and managers of professional home cleaning, hotel cleaning, construction, transportation and culinary/restaurant service companies in Salt Lake City. Community leaders and health educators associated with these nonprofits provided input on survey design and recruitment methods. Before the intervention, CHWs participated in two trainings conducted by a bilingual and bicultural community outreach specialist employed by Huntsman Cancer Institute. These trainings taught CHWs how to deliver a cancer education curriculum that could be presented via telephone or in-person using a PowerPoint slide deck as a guide. The CHWS also received instruction about the research protocol and how to interpret the results of FITs. The Institutional Review Board of the University of Utah approved this study.

Participant recruitment and data collection

Workplace managers advertised the study to current and former employees using flyers, postcards and sign-up sheets. Eligible employees were aged ≥18 years, spoke Spanish or English, and were employed at one of job sites in a manual or service labor job. Employees provided their phone numbers if interested. Sign-up sheets were given directly to either the CHWs or the research team who provided names and phone numbers to the CHWs. CHWs also recruited current and former service/manual workers from their own social networks.

We recruited Hispanics working in service-related and manual labor occupations because the low wages, inadequate health insurance benefits and lack of paid time off common in these positions typically do not facilitate CRC screenings.

CHWs contacted eligible participants to schedule an educational session held via telephone, in-person at a participant-selected location, or at a business partner office according to participant preference. CHWs administered the pretest interview (116 open and close-ended questions), conducted the educational intervention and distributed FIT tests to eligible participants. CHWS also helped eligible participants schedule a screening if needed, including FIT. Participants typically completed the pre-intervention interview in 30–60 min. The pre-intervention interview measured attitudes, knowledge and perceptions about CRC, screening guidelines and current screening adherence.

Post-intervention interviews (98 questions) began as early as a week after the pre-intervention interview and ended 1 year after the intervention. CHWs conducted the post-intervention interviews via phone or in-person. CHWs contacted participants a maximum of eight times to complete the post-intervention interview. The post-intervention interview determined retention of knowledge about CRC screenings and completion of FIT or other CRC screenings. CHWs assessed the results of FIT over the phone if they were completed. Participants with positive FIT results were referred for no-cost colonoscopies to the Huntsman Cancer Institute.

Intervention

CHWs presented educational information from the US Preventive services task force about breast, cervical, and CRC and the screening test(s) available for each specific cancer. They explained at what ages the screenings are recommended and to whom they are recommended for. CHW also presented general guidance on healthy behaviors such as nutrition, exercise, stress management, and emotional/mental health.

We provided participants with incentives to complete the pre- and post-intervention interviews and FIT test. Participants received $15 for completing the pre-intervention interview, $20 for the post-intervention interview, and $25 for the FIT.

Demographic variables

During the pre-intervention interview participants reported their age at time of interview, sex, ethnicity, birthplace (Inside United States versus Outside United States), language(s) primarily spoken, educational attainment (less than high school, high school graduate or more), current occupation, work status, health insurance, and annual household income.

The majority of participants (88.3%) spoke only Spanish, Portuguese or Spanish in combination with a non-English language. Only 3% of participants reported English as their sole primary language and 6.5% reported their primary languages were English in combination with Spanish. We categorized participants into two language groups to reflect these self-reported language patterns: (i) English only or English as one of the reported languages (English); and (ii) Spanish only, Portuguese only, or Spanish and other non-English language(s) (Spanish). We developed this dichotomization scheme because English is the language spoken primarily in the workplace and by healthcare personnel in Utah.

Participants reported employment in service, cleaning, gardening, restaurant, or other manual labor jobs. Participants could report more than one job and any managerial role in their organization. We categorized participants as service/manual labor if they were managers for any service/manual labor job. We considered participants office workers if they noted a managerial or clerical role in a nonmanual position (e.g. accountant). Participants also indicated work status (Full-time, Part-time, Temp work/hourly/per diem, Not working for wages/missing), health insurance (Insured/No insurance), and household income ($5000 to <$10 000; $10 000 to <$25 000; $25 000 or more).

Family history and perception of risk

Before the intervention CHWs asked questions about personal cancer history (Yes, No), cancer history of any immediate family members (parent, sibling, child, grandparents; Yes, No), and cancer history of other relatives (Yes, No). Participants reported perceptions of personal risk for CRC (Likely, Somewhat likely, Not likely), and rated their perceived importance of CRC screening (Important, Somewhat important, Not important).

Outcomes

We measured three general knowledge outcomes: (i) Heard of CRC (e.g. Before today, had you heard of CRC?), (ii) Recognized at least one CRC test (e.g. Do you know what a FOBT is?) and (iii) Identified at least one CRC guideline (e.g. At what age should CRC screenings begin?). Questions and answers that measured each outcome are listed in the Supplementary Appendix.

These outcomes were dichotomized into Yes or No groups. For the one question measuring ‘Heard of CRC’ we marked responses as ‘Yes’ if participants chose ‘Yes.’ Multiple questions assessed ‘Recognized at least one CRC test’ so we marked participants ‘Yes’ if participants answered ‘Yes’ to one or more of the FOBT/sigmoidoscopy/colonoscopy questions. Multiple questions also assessed ‘Identified at least one CRC guideline.’ If participants answered one of the four questions correctly we included them in the ‘Yes’ category. For all outcomes we categorized participants as ‘No’ if they answered ‘No’ to all questions or if they left all responses blank. We counted blank responses as ‘No’ answers because Hispanics may feel discomfort in selecting ‘No’ as a response in health interviews [31].

Predictors of adherence to CRC screening among participants aged ≥50 before and after the intervention included: (i) Current adherence to CRC screening: Participants reported if they had a: (a) FOBT/FIT within the past year; (b) a sigmoidoscopy in the past 5 years and (c) a colonoscopy in the past 10 years. If participants indicated one positive response for any of the questions, they were considered adherent; (ii) Previous screening: (FOBT/FIT, sigmoidoscopy, colonoscopy) if participants indicated ‘Yes’ for any screening tests they were marked as having a previous screening, regardless of when the screening happened; (iii) Previous colonoscopy was ascertained by asking if participants ever had a colonoscopy in any given time frame (Yes/No). Age ≥50 years was the cut off for screening per guidelines relevant at the time of the study [7].

We asked participants the same questions about their screening practices after the intervention. Participants who previously were not adherent, but were adherent after the intervention were coded as ‘Improved with any screening.’ We noted improvements in screening adherence for FIT which was provided to participants aged ≥50 years as part of the study.

Statistical methods

We excluded participants who did not report Hispanic/Latino ethnicity from the analysis (n = 5, <2%). Chi-square tests and Fisher’s exact tests (for cells sized <5) indicated differences in the outcomes stratified by the variables of interest, including missing data. We restricted analyses for current CRC screening adherence, previous CRC screenings and previous colonoscopy to participants aged ≥50 years per relevant guidelines [7].

We calculated prevalence odds ratios (ORs) and 95% confidence intervals (95% CI) using logistic models with a binary distribution and logit link to approximate relative risks. We calculated the crude OR first, then adjusted each model for age and sex where appropriate. After examining the effect estimates for those models, we placed all variables with non-null sex- and age-adjusted effect estimates in a single model to identify significant predictors of CRC knowledge and adherence, presented in the multivariable tables below. Analyses were conducted using PROC GENMOD in SAS version 9.4. Values were significant at P ≤ 0.05.

Results

Our population was largely female, born outside the United States and under 49 years of age (Table I). The majority of participants (88.3%) reported Spanish as their primary language and nearly half (42.0%) did not graduate from high school. Most participants worked full-time (44.3%) and had service/manual labor positions (75.7%). Participants who were not currently employed in service/manual labor positions previously worked in those positions or were office workers in the businesses. Most participants did not have health insurance (70.2%) and reported an annual household income below $25 000 (64.9%).

Table I.

Demographic, economic, cancer history and perceived risk characteristics of Hispanic service industry and manual workers (N = 307)

	N	%
Demographic characteristics
Age at survey (years)
18–30	22	7.2
31–49	162	52.8
≥50	122	39.7
Missing	1	0.3
Sex
Female	245	79.8
Male	61	19.9
Missing	1	0.3
Birthplace
Outside United States	290	94.5
Inside United States	13	4.2
Missing	4	1.3
Primary language
Spanish	271	88.3
English	30	9.8
Missing	6	2.0
Education
<High school	129	42.0
≥High school	171	55.7
Missing	7	2.3
Economic characteristics
Current occupation
Office worker	17	5.6
Service/manual worker	231	75.7
Not working/other	52	17.1
Missing	5	1.6
Self-reported work status
Full-time	135	44.3
Part-time	100	14.4
Temp work/hourly/per diem	26	32.8
Not working for wages/missing	44	8.5
Insurance status
Insured	80	26.2
No insurance	214	70.2
Missing	11	3.6
Annual household income
$5000 to <$10 000	62	20.3
$10 000 to <$25 000	136	44.6
≥$25 000	88	28.9
Missing	19	6.2
Cancer history
Previous diagnosis of cancer
No	288	93.8
Yes	19	6.2
Immediate family member with diagnosis
No family member with diagnosis	198	64.5
Has family member(s) with diagnosis	94	30.6
Missing	15	4.9
Other relative with diagnosis
No relative with diagnosis	168	54.7
Has relative(s) with diagnosis	115	37.5
Missing	24	7.8
Perceived CRC risk
Personal risk of CRC
Likely	123	40.1
Somewhat likely	110	35.8
Not likely	57	18.6
Missing	17	5.5
Importance of CRC screening
Important	273	88.9
Somewhat important	6	2.0
Not important	3	1.0
Missing	25	8.1

Few participants had a personal cancer history (6.2%) or an immediate family member with a cancer history (30.6%). Most viewed their personal risk for CRC as ‘Likely’ or ‘Somewhat likely,’ and rated CRC screenings as ‘Important.’ Most participants heard of CRC before the intervention, could recognize at least one CRC test and could identify at least one CRC screening guideline (Table II). Before the intervention 40% of adults aged ≥50 years were adherent to screening guidelines. After the intervention, 66% of eligible adults were adherent with the majority of improvement from the provided FITs (not shown).

Table II.

Differences in pre-intervention knowledge of CRC and screening guidelines among Hispanic service industry and manual workers (all ages) by demographic, economic, family history and personal risk characteristics

	Row totala	Heard of CRC					Recognized at least one CRC screening test					Identified at least one guideline correctly
		No (n = 72)		Yes (n = 233)		P c	No (n = 120)		Yes (n = 186)		P c	No (n = 44)		Yes (n = 254)		P c
	N = 305	n	%b	n	%b		n	%b	n	%b		n	%b	n	%b
Demographic characteristics
Age at survey (years)
18–30	21	12	57.1	9	42.9	<0.01*	9	42.9	12	57.1	<0.01*	2	9.5	19	90.5	0.09
31–49	151	39	25.8	112	74.2		72	47.7	79	52.3		28	18.5	123	81.5
≥50 years	121	18	14.9	103	85.1		33	27.3	88	72.7		12	9.9	109	90.1
Primary language
English	28	12	42.9	16	57.1	0.03*	9	32.1	19	67.9	0.76	5	16.7	25	83.3	0.75
Spanish	261	57	21.8	204	78.2		104	39.8	157	60.2		38	14.6	223	85.4
Education
<High school	125	36	28.8	89	71.2	0.12	71	56.8	54	43.2	<0.01*	26	20.8	99	79.2	0.02*
≥High school	163	32	19.6	131	80.4		41	25.2	122	74.8		17	10.4	146	89.6
Economic characteristics
Self-reported work status
Full-time	135	29	21.5	106	78.5	0.55	50	37	85	63	0.61	21	16.4	107	83.6	0.72
Part-time	100	26	26	74	74		38	37.6	63	62.4		13	13	87	87
Temp work/hourly/per diem	26	5	19.2	21	80.8		11	42.3	15	57.7		5	19.2	21	80.8
Not working for wages/missing	44	12	27.3	32	72.7		21	47.7	23	52.3		5	11.4	39	88.6
Insurance status
Insured	83	17	20.5	66	79.5	0.24	23	28.4	58	71.6	0.01*	7	8.4	76	91.6	0.03*
No insurance	215	54	25.1	161	74.9		95	44.4	119	55.6		36	17.4	171	82.6
Annual household income
$5000 to <$10 000	62	19	30.6	43	69.4	0.35	24	38.7	38	61.3	0.88	10	16.4	51	83.6	0.64
$10 000 to <$25 000	136	28	20.6	108	79.4		56	41.2	80	58.8		16	12.3	114	87.7
≥$25 000	88	19	21.6	69	78.4		32	36	57	64		16	18	73	82
Family history and personal risk
Immediate family with previous diagnosis
No family member with diagnosis	193	52	26.9	141	73.1	0.07	77	39.9	116	60.1	0.02*	28	14.5	165	85.5	0.07
Has family member(s) with diagnosis	90	14	15.6	76	84.4		28	31.1	62	68.9		10	11.1	80	88.9
Other relative with previous diagnosis
No relative with diagnosis	163	40	24.5	123	75.5	0.23	72	44.2	91	55.8	<0.01*	23	14.1	140	85.9	0.56
Has relative(s) with diagnosis	111	22	19.8	89	80.2		30	27	81	73.0		18	16.2	93	83.8
Personal risk for CRC
Likely	122	25	20.5	97	79.5	<0.01*	49	40.2	73	59.8	0.02*	19	15.6	103	84.4	0.52
Somewhat likely	106	17	16	89	84.0		33	31.1	73	68.7		12	11.3	94	88.7
Not likely	55	24	43.6	31	56.4		24	43.6	31	56.4		11	20	44	80

^aTotal not equivalent due to missing data.

^bRow percent.

^cChi-square statistics used, Fishers exact test used if cells contained <25% of data.

^*Significant at P ≤ 0.05.

Descriptive analyses

Table II summarizes population characteristics according to their CRC knowledge, screening recognition and screening adherence. We found a significantly higher percent of participants heard about CRC if they were age ≥50 years compared with younger age groups, spoke Spanish as their primary language and rated their personal risk for CRC as ‘Likely’ or ‘Somewhat likely.’ A significantly higher percent of participants who were age ≥50 years, had a high school degree, and had health insurance recognized at least one CRC screening. A larger percent of participants who had immediate family or other relatives with a cancer history, or who rated their personal risk for CRC as ‘Likely’ or ‘Somewhat Likely’ also recognized at least one CRC screening. A significantly larger percent of participants who graduated from high school and participants who had health insurance correctly identified at least one CRC screening guideline.

For participants aged ≥50 years (Table III), current adherence to screening was more frequent among participants with insurance and among women compared with men. Participants with insurance reported a significantly higher prevalence of previous CRC screening and of previous colonoscopy.

Table III.

Differences in current CRC screening adherence, previous screenings and previous colonoscopy among Hispanic service industry and manual workers aged ≥50 years pre-intervention

	Row totala	Currently adherent to CRC screeningb					Previous CRC screeningc					Previous colonoscopy onlyd
		No (n = 80)		Yes (n = 45)		P f	No (n = 75)		Yes (n = 50)		P f	No (n = 89)		Yes (n = 34)		P f
	N = 125	n	%e	n	%e		n	%e	n	%5		n	%e	n	%e
Demographic characteristics
Sex
Female	94	55	58.5	39	41.5	0.04*	52	55.3	42	44.7	0.12	66	70.2	28	29.8	0.29
Male	26	21	80.8	5	19.2		19	73.1	7	26.9		21	80.8	5	19.2
Primary language
English	8	5	62.5	3	37.5	0.87	5	62.5	3	37.5	0.74	5	62.5	3	37.5	0.49
Spanish	110	70	63.6	40	36.4		65	59.1	45	40.9		81	73.6	29	26.4
Education
<High school	48	31	64.6	17	35.4	0.94	29	60.4	19	39.6	0.94	37	77.1	11	22.9	0.4
≥High school	70	44	62.9	26	37.1		41	58.6	29	41.4		49	70.0	21	30.0
Economic characteristics
Self-reported work status
Full-time	50	31	62.0	19	38.0	0.69	29	58.0	21	42.0	0.67	39	78.0	11	22.0	0.57
Part-time	39	24	61.5	15	38.5		24	61.5	15	38.5		26	66.7	13	33.3
Temp work/hourly/per diem	14	11	78.6	3	21.4		10	71.4	4	28.6		11	78.6	3	21.4
Not working for wages/missing	18	11	61.1	7	38.9		9	50.0	9	50.0		12	66.7	6	33.3
Insurance status
Insured	33	15	45.5	18	54.5	0.01*	13	39.4	20	60.6	0.01*	17	51.5	16	48.5	<0.01*
No insurance	80	58	72.5	22	27.5		55	68.8	25	31.3		67	83.8	13	16.3
Annual household income
$5000 to <$10 000	27	20	74.1	7	25.9	0.54	19	70.4	8	29.6	0.44	22	81.5	5	18.5	0.53
$10 000 to <$25 000	50	30	60.0	20	40.0		27	54.0	23	46.0		35	70.0	15	30.0
≥$25 000	35	23	65.7	12	34.3		22	62.9	13	37.1		25	71.4	10	28.6
Family history and personal risk
Immediate family with previous diagnosis
No family member with diagnosis	63	44	69.8	19	30.2	0.11	40	63.5	23	36.5	0.27	50	79.4	13	20.6	0.08
Has family member(s) with diagnosis	51	28	54.9	23	45.1		27	52.9	24	47.04		33	64.7	18	35.3
Other relative with previous diagnosis
No relative with diagnosis	67	44	65.7	23	34.3	0.87	41	61.2	26	38.8	0.94	50	74.6	17	25.4	0.93
Has relative(s) with diagnosis	46	30	65.2	16	34.8		29	63.0	17	37.0		34	73.9	12	26.1
Personal risk for CRC
Likely	53	31	58.5	22	41.5	0.37	28	52.8	25	47.2	0.2	37	69.8	16	30.2	0.57
Somewhat likely	42	28	66.7	14	33.3		26	61.9	16	38.1		32	76.2	10	23.8
Not likely	21	16	76.2	5	23.8		16	76.2	5	23.8		17	81.0	4	19.0

^aTotal not equivalent due to missing data.

^bCurrently adherent to colonoscopy, sigmoidoscopy and/or FOBT/FIT guidelines.

^cDefined as ever having had an FOBT/FIT, sigmoidoscopy and/or colonoscopy.

^dFOBT/FIT and sigmoidoscopy not counted in analysis.

^eRow percent.

^fChi-square statistics used, Fishers exact test used if cells contained <25% of data.

^*Significant at P ≤ 0.05.

Multivariable analyses

The results of multivariable analysis for predictors of knowledge about CRC and CRC screening and guidelines are summarized in Table IV. Participant characteristics associated with having heard of CRC included older age at interview: 31–49 years (OR = 4.53, 95% CI = 1.46–14.01), and ≥50 years (OR = 8.90, 95% CI = 2.61–30.35) compared with participants aged 18–30 years. Participants who rated their risk for CRC as ‘Somewhat likely’ (OR = 4.48, 95% CI = 1.94–10.34) and ‘Likely’ (OR = 3.06, 95% CI = 1.40–6.67) had a significantly higher odds of hearing of CRC than participants who reported risk as ‘Not likely.’ Only those with a high school degree or higher had a significant, positive association for recognizing at least one CRC test, (OR = 3.21, 95% CI = 1.78–5.80, ref = Less than high school). Participants aged ≥50 years had a higher odds of recognizing at least one guideline (OR = 2.34, 95% CI = 0.70–7.81) than adults aged 18–30 years, although this association was not significant. We found no differences in the recognition of CRC screenings between participants aged 31–49 years and 18–30 years. Health insurance had a significant, positive association with the correct identification of at least one screening guideline (OR = 2.88, 95% CI = 1.05–7.90, ref = No insurance).

Table IV.

Multivariable ORs and 95% CIs (95% CI) for pre-intervention knowledge of CRC and screening guidelines among Hispanic service industry and manual workers (all ages)

	Heard of CRCa (N = 305)		Recognized at least one CRC testb (N = 306)		Identified at least one guideline correctlyc (N = 298)
	OR	95% CI	OR	95% CI	OR	95% CI
Demographic characteristics
Age at survey (years)
18–30	Ref		Ref		Ref
31–49	4.53*	1.46–14.01	1.09	0.34–3.45	0.50	0.07–1.62
≥50	8.90*	2.61–30.35	2.34	0.70–7.81	0.67	0.13–3.49
Language
English	Ref		Ref		Ref
Spanish	1.13	0.39–3.33	0.73	0.24–2.17	1.56	0.50–4.88
Educational attainment
<High school	Ref		Ref		.
≥High school	1.54	0.79–3.00	3.21*	1.78–5.80	.	.
Economic characteristics
Self-reported work status
Not working for wages/missing	Ref		Ref
Full-time	1.92	0.71–5.23	1.85	0.75–4.60	.
Part-time	1.52	0.54–4.27	1.94	0.75–5.05	.
Temp work/hourly/per diem	1.50	0.38–5.87	1.60	0.45–5.72	.
Insurance status
No insurance	Ref		Ref		Ref
Insured	1.14	0.52–2.44	1.42	0.70–2.87	2.88*	1.05–7.90
Family history and personal risk
Other relative with previous cancer diagnosis
No relative with diagnosis	.		Ref		.
Has relative(s) with diagnosis	.		1.63	0.88–3.03	.
Personal risk for CRC
Not likely	Ref		Ref		Ref
Somewhat likely	4.48*	1.94–10.34	1.78	0.80–3.98	1.81	0.70–4.67
Likely	3.06*	1.40–6.67	1.23	0.57–2.64	1.61	0.66–3.93

^aHeard of CRC pre-intervention.

^bRecognized sigmoidoscopy, colonoscopy and/or FOBT/FIT as a CRC screening.

^cCorrectly identified age and/or frequency of CRC guideline for sigmoidoscopy, colonoscopy and/or FOBT/FIT. ‘.’ Indicates the variable was not included in the model.

^*Significant at P ≤ 0.05.

Multivariable models examined how participant characteristics were associated with pre-intervention adherence to CRC screening guidelines and history of CRC screenings among participants aged ≥50 years, shown in Table V. Following adjustment for age and sex, insurance status was the only significant predictor for current adherence to CRC screening guidelines (OR= 3.00, 95% CI = 1.10–8.12), history of any CRC screening (OR= 3.40, 95% CI = 1.25–9.20) and previous colonoscopy (OR= 5.00, 95% CI = 1.72–14.53). Although not statistically significant, rating personal risk for CRC as ‘Likely’ had a 2-fold association with current adherence (OR= 2.34, 95% CI = 0.63–8.69, ref = Not likely), and a 3-fold association with previous screening (OR= 3.01, 95% CI = 0.81–11.13, ref = Not likely).

Table V.

Multivariable ORs and 95% CIs for differences in current CRC screening adherence, screening history and previous colonoscopy among Hispanic service industry and manual workers aged ≥50 years pre-intervention

	Currently adherent to screening guidelinesa (N = 122)		Any previous CRC screeningb (N = 122)		Previous colonoscopyc (N = 120)
	OR	95% CI	OR	95% CI	OR	95% CI
Demographic characteristics
Sex
Female	Ref		Ref		Ref
Male	0.27	0.07–1.10	0.47	0.14–1.56	0.50	0.12–2.11
Education
<High school	Ref		Ref		Ref
≥High school	0.69	0.28–1.72	0.71	0.29–1.73	0.91	0.33–2.52
Economic characteristics
Insurance status
No insurance	Ref		Ref		Ref
Insured	3.00*	1.10–8.12	3.40*	1.25–9.20	5.00*	1.72–14.53
Annual household income
$5000 to <$10 000	Ref		Ref		Ref
$10 000 to <$25 000	2.16	0.69–6.77	1.28	0.75–6.94	2.24	0.62–8.09
≥$25 000	1.07	0.28–4.03	0.92	0.25–3.33	0.93	0.21–4.24
Personal risk
Personal risk for CRC
Not likely	Ref		Ref		Ref
Somewhat likely	1.39	0.34–5.58	1.76	0.44–6.95	1.07	0.22–5.24
Likely	2.34	0.63–8.69	3.01	0.81–11.13	1.99	0.45–8.71

All variables in table included in model.

^aCurrently adherent to colonoscopy, sigmoidoscopy and/or FOBT/FIT guidelines.

^bDefined as ever having an FOBT/FIT, sigmoidoscopy and/or colonoscopy.

^cFOBT/FIT and sigmoidoscopy not counted in analysis.

^*Significant at P ≤ 0.05.

Characteristics influencing post-intervention improvements in adherence to CRC screening guidelines are summarized in Table VI. The majority of improvement was due to FIT provided to all eligible participants aged ≥50 years. Only six participants received colonoscopies after the intervention while 36 participants completed the FIT post-intervention (data not shown). Annual income ≥$25 000 had a significant association with improvement in any screening after the intervention (OR= 8.60, 95% CI = 1.52–48.80, ref = $5000 to <$10 000). Improvement in FIT was also significantly associated with a higher income (OR= 8.49, 95% CI = 1.49–48.32, ref = $5000 to <$10 000).

Table VI.

Multivariable ORs and 95% CIs for post-intervention improvements in any CRC screening and FIT tests among Hispanic service industry and manual workers aged ≥ 50 years

	Improvement in any CRC Screeninga (N = 59)		Improvement in FIT (N = 59)
	OR	95% CI	OR	95% CI
Demographic characteristics
Sex
Female	Ref		Ref
Male	0.61	0.14–2.73	0.72	0.16–3.22
Education
<High school	Ref		Ref
≥High school	1.26	0.38–4.12	1.03	0.31–3.38
Economic characteristics
Insurance status
No insurance	Ref		Ref
Insured	2.00	0.46–8.65	2.19	0.51–9.51
Annual household income
$5000 to <$10 000	Ref		Ref
$10 000 to <$25 000	2.47	0.60–10.06	2.23	0.54–9.13
≥$25 000	8.60*	1.52–48.80	8.49*	1.49–48.32
Personal risk
Personal risk for CRC
Not likely	Ref		Ref
Somewhat likely	2.47	0.53–9.36	2.50	0.53–11.67
Likely	2.02	0.44–9.36	1.61	0.35–7.41

^aColonoscopy, sigmoidoscopy, FOBT and/or FIT.

^*Significant at P ≤ 0.05.

Discussion

Improving adherence to CRC screenings can reduce the racial/ethnic disparities in the colorectal mortality rates among Hispanics. In Utah racial/ethnic minorities will comprise 40% of the Salt Lake County population by 2050 [26], doubling the size of the current racial/ethnic minority population. The rate of CRC screening adherence among Hispanics in our study (40%) was lower than the national averages for both the general US population (60%) and Hispanics nationwide (47%) [3]. Understanding the complex factors influencing CRC screening in this fast-growing population is needed to reduce current CRC disparities in Utah and inform future interventions in this population [32]. This intervention provides support for FIT as a method to bridge the gap in CRC screening in certain Hispanic populations. FIT is considered a simple and easy method for CRC screening by Hispanics [33] but provision of FIT and education from a CHW in this study did not improve adherence among Hispanics with the lowest incomes ($5000 to <$10 000) and who are uninsured. These participants may have refused to take a test home, not used them due to a lack of resources for treatment or had miscommunication about the cost of the test. Uninsured and low-income participants may harbor other non-measured feelings, beliefs and attitudes that prevented completing the FIT [34]. Understanding motives behind completion and non-completion of FIT in the lowest income and uninsured Hispanic populations is an important next step.

Provision of the FITs to eligible participants who were overdue for CRC screening increased CRC screening adherence from 40% to 66%. Hispanic men, individuals with insurance, and those with incomes ≥$25 000 showed the most improvement in screening adherence. The increased use of FITs among males, although not significant, is especially encouraging as Hispanic men report the lowest screening rates among all Hispanic subgroups [21, 35].

A lack of health insurance, lower incomes and language barriers are likely significant contributors to the low level of CRC screening seen in our study population before the intervention [8, 11, 17, 35, 36]. Income-related gaps in insurance coverage, noted in previous assessments [37, 38], are particularly responsible for placing Hispanics at a disadvantage for screening adherence. In our study population of low-wage workers, 70% are uninsured and employers are not mandated to provide health insurance to all employees in Utah. Despite the small number that is insured, we still report that insurance is the only significant predictor of pre-intervention adherence to CRC screening guidelines. In a sub-analysis we report no difference between participants with insurance provided by the Affordable Care Act Exchange and other forms of insurance. As Utah recently voted to expand Medicaid in April 2019, the number of uninsured Hispanics and thus nonadherence to CRC screenings may change in the future [39].

Completing high school was the only significant predictor of recognizing at least one CRC screening method. This education gap is consistent with previous studies [11, 40]. Interventions tailored to Spanish-speaking populations appear to effectively bridge gaps in CRC awareness for populations with less formal education [24, 32]. However, increased awareness and knowledge does not necessarily translate to adherence. Much of the improvement in adherence we report was attributed to distribution of FIT paired with appropriate education. Thus education paired with access to resources is necessary to reduce CRC disparities.

Interventions targeting Hispanic populations should consider educating adults aged <50 years about CRC screenings so they can make informed decisions later in life. In our sample, we found no differences in the recognition of at least one CRC screening between Hispanic adults aged 31–49 years and those aged 18–30 years. Early education may help familiarize individuals with CRC screening procedures and alleviate potential embarrassment or fear [14]. Providing information about screenings to younger adults could also aid in identifying early onset CRC. The incidence of CRC in Hispanics aged <50 years increased nationally despite a decrease in the overall incidence rate [8].

Hispanics are at higher risk for genetic mutations related to CRC [41] and presence of a family member with CRC has been linked to a 20-fold increased risk for CRC diagnosis in Hispanics [42]. Individuals with first-degree relatives who were diagnosed with CRC are highly encouraged adhere to screening guidelines beginning in early adulthood [3]. However Hispanic adults with family history of CRC are less likely to adhere to screening guidelines than non-Hispanics with no family history of CRC [19]. Because of these inherited risks interventions should consider incorporating family history into CRC prevention education.

Knowing about family history of cancer had a positive but nonsignificant association with recognizing one CRC test and knowing about one guideline. In our study population knowing about family history of cancer was correlated with an increase in a participants’ perception of personal risk. However, personal risk does not translate to improved screening adherence as the effect estimates we found for personal risk were not significant. Education about family risk, physician recommendation and access to care may clear a path to improving screening adherence among Hispanics with a family history of CRC [43, 44].

This intervention demonstrates the feasibility of developing partnerships with businesses to implement workplace-based health interventions. Interventions like these can give community health organizations access to underserved populations that are otherwise difficult to reach.

Limitations and strengths

We focused on recruiting Hispanics who work in low-income service-related and manual labor jobs to target the most vulnerable population possible. Thus, our results are not generalizable to the entire Hispanic population in Utah. We only recruited participants from Salt Lake County which limits the geographic reach of our study and potentially its external validity. Thus, our results may not be generalizable to the entire Hispanic population in Utah. The target population limited our sample size, potentially affecting the width of our CIs.

Participation in the intervention was not mandatory for employees or businesses. Some businesses were enthusiastic about the partnership while others were less so. The more enthusiastic businesses may have created a culture and structural changes, such as flexible scheduled, that aided in the improvement of CRC screenings among their employees. Employees who chose to participate may have also significant differences from those that did not. The confounding from these unmeasured differences may influence the results in this study. However, the number of employees from businesses that participated in the intervention varied greatly and the enthusiasm from the different businesses also varied greatly. Because of this random variation, we do not suspect that our study's results are greatly obscured by confounding. Future studies could examine the influence of workplace culture and flexible scheduling on adherence with cancer screenings.

Although participants were asked which language(s) they speak, we did not determine English or Spanish language proficiency which may influence the effect of language on the effect estimates. We do not have an exact range for the upper-income bracket, but the incomes should be relatively similar across all jobs included in the study. CHWs conducted the post-intervention assessment at different time points ranging from a few days to a few months post-intervention due to participant availability and preference. These time frames may not have provided participants with sufficient time to schedule or complete CRC screenings. Despite this variation in the post-intervention assessments, we saw a significant improvement in CRC screening adherence among those who received FITs.

Our study has several strengths, including identification of the role of health insurance and demographic characteristics on CRC knowledge, and the roles of family history and personal risk on CRC adherence. We trained CHWs to provide education about CRC screenings to reach low-income Hispanic workers through workplace partnerships. This has never before been done in Utah, other organizations have partnered with healthcare businesses [45]. Our results demonstrate that these types of interventions could be successfully disseminated among low-income workers in Utah. To our knowledge this is also the first assessment of CRC knowledge and screening adherence in a majority immigrant Hispanic working population in Utah. We also assessed CRC screening adherence before and after the intervention in our high-risk Hispanic population which is rare among CRC interventions that primarily focus on education [32].

Conclusion

Reducing racial/ethnic disparities in screenings within the rapidly growing Hispanic population can reduce CRC mortality. This intervention demonstrated the feasibility of forming partnerships among businesses, community health organizations and a university to implement an intervention targeted at low-income, under insured Hispanic workers. Low-income and uninsured Hispanics may require more support than other populations to improve adherence. Interventions should provide culturally and linguistically appropriate education paired with access to low cost screenings such as FITs. Future studies should examine the feasibility of partnering with larger businesses to conduct similar interventions on a bigger scale.