Exploring health literacy and the correlates of Pap testing among African immigrant women: Findings from the AfroPap study

Abstract

Background:

African immigrant (AI) women have low rates of Pap testing. Health literacy plays a pivotal role in health behaviors. Sources and types of health information could shape health literacy and inform the Pap testing behaviors of AI women. However, the influences of health literacy, sources and types of health information along with cultural and psychosocial correlates on the Pap testing behaviors of AI women are poorly understood.

Objective:

To examine how sources and types of health information impact health literacy, and in turn, how health literacy, cultural and psychosocial factors influence the Pap testing behaviors of AI women.

Methods:

An adapted Health Literacy Skills Framework guided the selection of variables for this cross-sectional study. Convenience sampling was used to recruit 167 AI women, 21–65 years. Multivariate logistic regression was used to assess correlates of Pap testing after adjusting for covariates (age, education, English proficiency, employment, income, health insurance, access to primary care, marital status, and healthcare provider recommendation).

Results:

Most participants (71%) had received a Pap test in the past and used multiple (two or more) sources (65%) and types (57%) of health information. Using multiple sources of health information (aOR: 0.11, p<0.01) but not types of health information was associated with Pap testing. Having negative cultural beliefs (aOR:0.17, p=0.01) and high self-efficacy (aOR: 9.38, p<0.01) were significantly associated with Pap testing after adjusting for covariates. High health literacy (OR: 3.23, p<0.05) and high decisional balance (OR: 5.28, p<0.001) were associated with Pap testing in bivariate models but did not remain significant after controlling for covariates.

Conclusions:

Cultural beliefs was a significant correlate of AI women’s Pap testing behaviors regardless of other known social determinants of health (education, English proficiency, age, access to primary care). Disseminating health information through various sources have the potential to promote Pap testing among AI women. Larger studies which utilize a robust sampling strategy and include a diverse group of AI women are needed in order to optimize health interventions aimed at improving Pap test screening behaviors among AI women.

Cervical cancer is preventable, and among those who develop cervical cancer, early detection and treatment results in high survival rates and better health outcomes.¹ In particular, since the Papanicolaou (Pap) test was introduced as the ideal method for early detection of cervical cancer, there has been a substantial decrease in the number of women who die from cervical cancer in the United States (U.S.) each year.² Nevertheless, the American Cancer Society estimates that in 2019, about 13,170 women in the U.S will be diagnosed with cervical cancer and about 4,250 of these women will die from the disease.² Disparities in cervical cancer outcomes persist, particularly among racial/ethnic and immigrant populations. For example, Black women have higher cervical cancer death rates compared to other racial/ethnic groups due, in large part, to lower Pap testing rates.³ Within the Black ethnicity, African immigrant (AI) women (nationally not specified) represent the most vulnerable subgroup of women, having 2–3 times less likelihood of reporting a lifetime receipt of a Pap test in comparison to African American women.^4,5

Health literacy is a multi-dimensional concept that addresses the various skillsets a person needs to successfully “obtain, process, understand and use basic health information and services to make appropriate health decisions.”⁶ Health literacy has been reported as a strong determinant of Pap testing in middle-age white women,^7,8 Asian immigrant women,⁹ and Hispanic immigrant women.^10,11 None of the studies included AI women, however. Further, prior research assessed health literacy using single domain instruments (i.e., print literacy) such as the Test of Functional Literacy in Adults (TOFHLA) and Rapid Estimate of Adult Literacy in Medicine (REALM), without addressing other skillsets such as navigational literacy (operate effectively in a healthcare setting), familiarity (knowledge and understanding of health concepts), oral literacy (speaking and listening effectively) and numeracy (appropriate use of quantitative information) which are essential to inform a health behavior such as cancer screening.^8,12,13

Sources of health information such as mass media (i.e. television, radio, Internet), social media (i.e. Facebook and WhatsApp) and interpersonal relationships made through friends, family, religious organizations and with healthcare providers are known to influence health behaviors.^14,15 Within the African cultural context, female friends/family serve as a central portal for the dissemination of women’s health information, and hence, may play a crucial role in shaping health literacy and the adoption of health behavior such as cancer screening.^16–19 The use of different types (verbal, written text, pictures) of health information exchange is also suggested as a possible mechanism through which health literacy is shaped and health behaviors are adopted.^6,20 For example, a significant number of adults with low health literacy prefer verbal communication to written text.^21,22

In addition to health literacy, a number of sociodemographic, cultural, and psychosocial factors have been associated with Pap testing behaviors. For instance, low English proficiency, limited cancer knowledge, low perceived susceptibility and embarrassment were negatively associated with Pap testing behaviors among Asian immigrant women,^23,24 so were low income, no health insurance coverage, low cancer knowledge, fatalism and embarrassment among African American women.^25,26 AI women have unique cultural experiences and it is likely that the correlates of Pap testing among this population may differ from those of women belonging to other ethnic/racial and immigrant groups. For example, AI women report higher levels of English proficiency in comparison to certain subgroups of Asian immigrant women (i.e. Chinese and Korean) who often report having difficulty communicating in English.^27–29 AIs also report higher perceived incidents of discrimination by healthcare practitioners due to their accents, immigration status and cultural mode of dressing whereas African Americans report age and poverty as the main reasons for higher perceived healthcare-related discrimination.^30–33

Taken together, there is limited research addressing health literacy and other correlates in relation to Pap testing behaviors among AI women. Considering that high mortality with low rates of Pap testing are reported among AI women,^3–5 examining relevant correlates of their Pap testing behaviors is imperative. In addition, since health literacy can be modified through culturally appropriate interventions,⁸ it is important to examine and understand how health literacy is shaped among AI women. Therefore, the purpose of this study is to examine how sources and types of health information impact health literacy, and in turn, how health literacy and other factors influence the Pap testing behaviors of AI women.

THEORETICAL FRAMEWORK

An adapted Health Literacy Skills (HLS) conceptual framework guided the selection of study variables.³⁴ The original premise of the HLS framework is to address the multidimensionality of health literacy and illustrate “the full continuum of relations among predictors, mediators and outcomes of health literacy.”³⁴ In African communities, female friends/family also serve as portals for the dissemination of women’s health information,^16,17,19,35 and they play a crucial role in African women’s adaptation of health behaviors such as cancer screening.¹⁸ Based on these findings, we adapted the HLS framework to include antecedents of health literacy (i.e., sources and types of health information), health literacy, and mediators (i.e. self-efficacy, decisional balance, cultural beliefs/attitudes and cancer knowledge) of the relationship between health literacy and health behavior.³⁴ Sources of health information was defined as the interpersonal relationships made through social ties or with healthcare providers that influence health behaviors and health beliefs,^36,37 and included family/relatives, friends, ethnic church, TV/Radio, Internet, social media and physician/healthcare provider. Types of health information was defined as the communication strategies for disseminating health information with the aim of influencing health beliefs and behaviors,³⁶ and included verbal, written text and pictures. For the purpose of this study, health literacy was conceptualized as a multi-dimensional concept to include individuals’ “ability to use medical terminology (familiarity) and apply relevant medical terms throughout the cancer screening trajectory (navigation).”^12,38 Finally, the following psychosocial and cultural variables were assessed as correlates of health behavior: Cancer knowledge, conceptually defined as what a person knows about cervical cancer;³⁹ self-efficacy, defined as how confident a person is in carrying out tasks such as receiving a Pap test;⁴⁰ decisional balance, conceptually defined as the perceived pros and cons of a health behavior,⁸ and cultural beliefs/attitudes, defined as the principles, values and customs that inform a person’s health behavior such as receiving a Pap test.⁴¹

METHODS

Study Design and Sampling

The African immigrant women Pap testing behavior (AfroPap) study used a cross-sectional design. A convenience sample of 167 women were recruited using two main sampling methods: in-person and online surveys via Qualtrics, an encrypted, data management and web-based survey tool. Participants were eligible to enroll in this study if there were females, 21–65 years old, self-identified as AI, had no history of hysterectomy, could read and write English, and resided in the U.S.

Procedures

All study procedures were approved by an academic institutional review board (IRB). Potential participants were approached at various African churches and community organizations where the study was announced, and flyers were disseminated. Trained bilingual research assistants obtained oral informed consent from in-person participants. The online survey included a copy of the IRB-approved oral consent script and an eligibility self-screener. Participants who completed the in-person surveys received a weblink to the online survey and were encouraged to share it with other AI women within their social networks. The weblink to the online survey was also disseminated using WhatsApp, a social media and communication platform commonly used within the African immigrant community. The online survey was limited to individuals who could complete the survey on their own with no assistance from the research team. Data collection occurred between November 2017 and December 2018. Study participants received a $5 gift card after survey completion. Individuals who completed the survey online received the gift card via mail.

Measures

Individual characteristics and screening-related variables:

Individual characteristics and screening-related variables were collected using a questionnaire developed for the purpose of this study. The information collected included age (21–65 years), health insurance coverage (yes/no), educational attainment (less than high school, high school, college, more than college), marital status (married, never married, separated/divorced), employment (employed full- or part-time/unemployed), primary health care access (yes/no). Income was assessed with the question, “Can you give an estimate of your annual household income?” and participants were categorized as low or high based on the sample median. English proficiency was measured with the questions, “(1) How well do you speak in English? (2) Can you converse in English? (3) Can you speak over the phone in English? (4) Can you read English newspapers? (5) How well can you interact at the hospital without the assistance of translators? and (6) Do you need help reading instructions or pamphlets you receive from the doctor or pharmacist?” The first five items were scored on a 4-point (1–4) Likert scale and a 5-point (1–5) Likert scale for the last item. Higher scores indicated higher English proficiency. Physician recommendation was measured with the question, “Has a healthcare provider recommended that you get a Pap test during the past 3 years?” (yes/no). Finally, history of Pap testing was measured with the question, “Have you ever had a Pap test?” (yes/no), “When was your last Pap test (month/year/place).

Antecedents of health literacy:

Types of health information was measured with the question, “How is women’s health information often presented to you?” (verbal/written text/pictures/other), with participants given the option to mark all that apply. Sources of health information was measured with the question, “From whom do you most likely get advice about women’s health? (female family/relatives, female friends, pastor/church, TV/radio, internet, social media, physician/healthcare practitioner, and other), with participants given the option to mark all that apply.

Health literacy:

The Assessment of Health Literacy-Cancer (AHL-C) is a validated instrument to assess health literacy in the context of women’s cancer screening with evidence of reliability and validity.¹² Based on Baker’s¹³ conceptualization of health literacy, the AHL-C addresses multiple dimensions of health literacy: print literacy, comprehension, familiarity, numeracy and navigation. A recent study revealed that the familiarity and navigational literacy sub-scales of the AHL-C were strongly associated with an increased likelihood of Pap testing.³⁸ Hence these two sub-scales were used to assess health literacy in our study. Familiarity scale includes 12 items which assess individuals’ ability to use relevant medical terminologies. Navigational literacy scale includes 12 clause items that assess individuals’ ability to apply relevant medical terms throughout the cancer screening trajectory) with questions such as, “Please sit down and roll up your sleeve. I will measure your ___;” “Please tell me whether you have abnormal symptoms such as __ in your breast.”¹² Higher scores on the AHL-C scales indicated higher health literacy levels. The internal consistency (Cronbach’s alpha) for the navigation and familiarity sub-scales were 0.92 and 0.96, respectively, in the original validation sample, and 0.86 and 0.98 in the current study sample.

Cultural beliefs/attitudes:

The Tang, Solomon and McCraken instrument⁴¹ consist of a 5-point Likert scale with 3 items measuring the use of home remedies with questions such as, “ I use home remedies as treatment for health problems” and a 5-point Likert scale with 6 items measuring perceived cultural barriers to Pap testing with questions such as, (1) “I would feel embarrassed with a doctor examining my cervix as part of a medical exam,” and (2) “I would feel uncomfortable talking about my body with a doctor.” Higher scores indicated negative culture beliefs and attitudes. In the original validation sample, the Cronbach’s alpha coefficients for the use of home remedies and perceived cultural barriers sections were 0.72 and 0.68 respectively. In our study sample, the internal consistency (Cronbach’s alpha) for the combined sections was 0.78.

Cervical cancer knowledge:

The modified Cervical Cancer Knowledge Test consists of a binary (true/false) scale with 21-items. The first 10 items (Cronbach’s alpha: 0.80–0.89)⁴² assess knowledge of cervical cancer risk factors and symptoms with questions such as, “If a woman gets cervical cancer, it can be detected early.” The remaining 11 items (Cronbach’s alpha: not reported)⁴³ assess knowledge on HPV infection and prevention such as, “HPV can be prevented by vaccination.” Higher scores indicated higher cervical cancer knowledge. The modified 21-item scale yielded an internal consistency (Cronbach’s alpha) of 0.87 in our study sample.

Self-efficacy:

The 4-item self-efficacy scale⁴⁴ measures how confident a woman is in carrying out tasks in relation to Pap testing using a 4-point (1=not at all confident to 4=very confident) Likert scale. The scale includes questions such as, “Do you feel confident that you can have a Pap test on a regular schedule?” Higher scores indicate higher self-efficacy. In our study sample, the self-efficacy scale yielded an internal consistency (Cronbach’s alpha) of 0.93.

Decisional balance:

Decisional balance refers to women’s perception about pros and cons of cervical cancer screening. The decisional balance scale has 12 items and uses a 5-point Likert scale (1=strongly disagree to 5=strongly agree). The scale includes questions such as (1) “A Pap test can be done so quickly that it is not a bother to have one;” (2) “A Pap test is not as important as people say it is.” Negatively worded items were recoded, and total scores were calculated. Higher scores indicate higher decisional balance. The decisional balance scale had an internal consistency (Cronbach’s alpha) of 0.76–0.86 in the original validation sample.⁴⁵ The Cronbach’s alpha for our study sample was 0.81.

Sample Size

Previous community-based, cancer prevention studies in the U.S included AI sample sizes ranging from 63 to 100, and estimated the prevalence of Pap testing among AI women to be between 19% to 70%.^18,46 Using a sample size of 157 and alpha of 0.05, our study would have 90% power to detect an OR of 3.43 or higher to be statistically significant.

Data Analysis

Descriptive statistics such as frequencies and percentages were calculated for all categorical variables. For continuous variables, means and standard deviations were reported. Two participants had missing data for the entire decisional balance scale and their values were replaced with the sample mean for the scale. To examine the association between selected study variables and past Pap testing, we used a series of logistic regression models. First, in order to examine the relationship between antecedents (i.e., types and sources of health information) and health literacy, dummy variables were created for participants who used 0–1 and 2+ sources and types of health information. For health literacy, participants were categorized as high and low by using 75^th percentile scores. Second, bivariate logistic regression was used to test the relationship between theoretically selected study variables and Pap testing. Multivariable logistic regression was then used to evaluate factors associated with Pap testing after adjusting for covariates. For psychosocial factors (i.e., cancer knowledge, decisional balance, and self-efficacy), participants were categorized as high and low by using 75^th percentile scores for cancer knowledge and mean for other variables. Based on a literature review of Pap test screening among immigrant women, the following variables were controlled for as covariates: age, education, English proficiency, employment, income, insurance, access to primary care, marital status, and healthcare provider recommendation. Multicollinearity was assessed and the variance inflation factor (VIF) was reasonable (VIF<5). Finally, we tested a possible mediation of theoretically selected psychosocial variables in the relationship between health literacy and Pap testing using the Baron and Kenny⁴⁷ four step approach. Tests of mediation using regression analyses would not be warranted if no significant association is found between health literacy and Pap testing. The statistical significance was considered when p < 0.05.

RESULTS

Sample Characteristics

The description of sample characteristics is reported in Table 1. A total of 167 AI women participated in the AfroPap study, with a majority of participants (n=91) recruited in-person and the remaining (n=76) recruited online. Participants’ ages ranged from 22–65 years and the mean (SD) age was 40.90 (12.25) years. Most participants were college educated (68%), married (57%), employed (80%), and reported having health insurance coverage (79%) and a primary care physician (74%). The average length of stay in the U.S. was 15 (7.76) years, and more than half (54%) of the participants had lived in the U.S. for 15 years or more. The average (SD) English proficiency score was 22.69 (3.86), indicative of high overall English proficiency among study participants. Participants recruited online were significantly younger (mean age: 33.96 years) and had higher average English proficiency scores (23.71) than those recruited in person (mean age: 46.70 years, English proficiency score: 21.84). No significant differences were found between the online and in-person groups in relation to employment status, income, health insurance coverage, length of stay and history of Pap testing.

Table 1:

Sample Characteristics

Characteristics	Overall (N=167)	In-person surveys (n=91)	Online surveys (n=76)	p-value
Age, y Mean (SD)	40.9 (12.25)	46.7 (11.27)	34.0 (9.48)	<0.001
Range, y	22–65	22–65	22–61
Education
<HS	19 (11)	17 (19)	2 (3)
High school	34 (20)	31 (23)	13 (17)	0.001
College	114 (68)	53 (58)	61 (80)
English proficiency score Mean (SD)	22.7 (3.86)	21.8 (4.56)	23.71 (2.46)	0.001
Employment (n, %)
Employed	134 (80)	71 (78)	63 (83)	0.215
Unemployed	23 (14)	16 (18)	7 (9)
Other	10 (6)	4 (4)	6 (8)
Income (n, %)
Low	84 (50)	45 (49)	38 (50)	0.944
High	83 (50)	46 (51)	38 (50)
Health insurance (n, %)
Yes	132 (79)	68 (75)	64 (84)	0.134
No	35 (21)	23 (25)	12 (16)
Primary care
Yes	123 (74)	68 (75)	55 (72)	0.731
No	44 (26)	23 (25)	21 (28)
Marital status (n, %)
Married	95 (57)	62 (68)	33 (43)	<0.001
Never married	45 (27)	10 (11)	35 (46)
Separated/Divorced	27 (16)	19 (21)	8 (11)
Physician recommendation (n, %)
Yes	97 (58)	54 (59)	43 (57)	0.719
No	70 (42)	37 (41)	33 (43)
Length of stay Mean (SD)	15.12 (7.76)	15.91 (8.29)	14.14 (6.99)	0.14
Sources of health information (n, %)
0–1	59 (35)	31 (34)	28 (34)	0.709
2 or more	108 (65)	60 (66)	48 (66)
Types of health information (n, %)
0–1	72 (43)	45 (49)	27 (36)	0.070
2 or more	95 (57)	46 (51)	49 (64)
Cervical cancer knowledge (n, %)
High	24 (14)	8 (9)	16 (21)
Low	143 (86)	83 (91)	60 (79)	0.024
Self-efficacy (n, %)
Low	71 (43)	26 (29)	45 (59)	<0.001
High	96 (57)	65 (71)	31 (41)
Decisional balance (n, %)
Low	63 (38)	36 (40)	27 (36)	0.592
High	104 (62)	55 (60)	49 (64)
Cultural beliefs/attitudes
Low (positive)	79 (47)	45 (49)	34 (45)	0.543
High (negative)	88 (53)	46 (51)	42 (55)
Health Literacy (n, %)
Low	79 (47)	51 (56)	28 (37)	0.013
High	88 (53)	40 (44)	48 (63)
History of Pap testing (n, %)
Yes	118 (71)	70 (77)	48 (63)	0.052
No	49 (29)	21 (23)	28 (37)

Overall, AI women’s knowledge on cervical cancer risk factors and symptoms were low; more than two thirds of the participants (86%) scored lower than the 75^th percentile on the Cervical Cancer Knowledge Test. Most participants (57%) reported high self-efficacy, and in relation to their perceived pros and cons of Pap testing, 62% of participants had a high decisional balance. Most participants (53%) reported negative cultural beliefs and attitudes towards Pap testing. More than half of the study participants reported using multiple sources of health information (65%), and often having health information presented in multiple forms (57%). Seventy-one percent of the participants (n=118) reported that they had received a Pap test in the past. Among AI women who reported no history of Pap testing (n=49), 75% reported that no healthcare provider had recommended that they receive a Pap test in the past 3 years. On the AHL-C scale, 53% (n=88) of AI women had high health literacy levels and 47% (n=79) had low health literacy levels.

Multiple logistic regression analyses

Antecedents of health literacy

As shown in Table 2, there were no significant associations between using multiple sources and types of health information and health literacy after controlling for all covariates. As indicated in Table 3, there was an association between AI women who used multiple sources of health information and Pap testing, and this relationship remained significant (aOR: 0.11, p<0.01) even after controlling for covariates. No significant association was found between multiple types of health information and Pap testing, however.

Table 2:

Logistic regression model showing the association between sources and types of health information and health literacy

Characteristics	Unadjusted OR (95% CI)	p-value	*Adjusted OR (95% CI)	p-value
Sources of health information
0–1	Ref		Ref
2 or more	1.38 (0.73–2.62)	0.317	0.97 (0.37–2.58)	0.958
Types of health information
0–1	Ref		Ref
2 or more	1.98 (1.07–3.69)	0.031	2.56 (0.97–6.72)	0.056

^*Adjusted for age, education, English proficiency, employment, income, insurance, access to medical care, marital status, physician recommendation, length of stay

Table 3:

Logistic regression showing factors associated with Pap testing

Characteristics	Unadjusted OR (95% CI)	p-value	Adjusted OR (95% CI)	p-value
Age, y Mean (SD)	1.03 (1.00–1.06)	0.028	1.11 (1.01–1.22)	0.031
Education
<HS	Ref	Ref	Ref
High school	1.58 (0.51–4.91)	0.423	0.23 (0.01–3.66)	0.299
College	1.47 (0.36–6.00)	0.007	0.43 (0.03–6.69)	0.546
English proficiency score	1.22 (1.11–1.34)	0.000	1.50 (1.05–2.13)	0.023
Employment
Unemployed	Ref	Ref	Ref	Ref
Employed	2.81 (1.13–6.95)	0.026	1.34 (0.16–11.22)	0.789
Other	0.92 (0.20–4.05)	0.909	0.19 (0.03–1.34)	0.095
Income
Low	Ref	Ref	Ref	0.501
High	2.81 (1.39–5.65)	0.004	0.60 (0.13–2.70)
Health insurance
No	Ref	Ref	Ref	0.739
Yes	4.04 (1.85–8.81)	0.000	0.74 (0.12–4.44)
Primary care
No	Ref	Ref	Ref	Ref
Yes	7.29 (3.40–15.62)	0.000	9.14 (1.90–44.07)	0.006
Marital status
Married	Ref	Ref	Ref	Ref
Never married	0.24 (0.11–0.52)	0.000	0.66 (0.12–3.47)	0.619
Separated/Divorced	0.71 (0.26–1.94)	0.508	1.81 (0.13–26.08)	0.662
Physician recommendation
No	Ref	0.000	Ref	0.006
Yes	7.94 (3.69–17.07)		8.31 (1.83–37.67)
Length of stay	1.08 (1.03–1.13)	0.003	1.08 (0.97–1.20)	0.172
Sources of health information
0–1	Ref	Ref	Ref	Ref
2 or more	0.31 (0.14–0.69)	0.004	0.11 (0.02–0.53)	0.006
Types of health information
0–1	Ref	Ref	Ref	Ref
2 or more	0.99 (0.50–1.93)	0.966	1.57 (0.32–7.66)	0.575
Cervical cancer knowledge
Low	Ref	0.327	Ref	0.528
High	1.69 (0.59–4.81)		0.54 (0.08–3.67)
Self-efficacy
Low	Ref	Ref	Ref	Ref
High	12.47 (5.43–28.64)	0.000	9.38 (2.10–41.93)	0.003
Decisional balance
Low	Ref	Ref	Ref	Ref
High	5.28 (2.58–10.82)	0.000	1.70 (0.33–8.67)	0.525
Cultural beliefs/attitudes
Low (positive)	Ref		Ref
High (negative)	0.13 (0.06–0.30)	0.000	0.17 (0.04–0.71)	0.015
Health literacy
Low	Ref		Ref
High	3.23 (1.60–6.51)	0.001	0.55 (0.10–3.17)	0.504

Full model presented | OR: Odds Ratio CI: Confidence interval

Health literacy, cultural and psychosocial correlates of Pap testing

As indicated in Table 3, in the bivariate model, high self-efficacy (OR: 12.47, p<0.001), high decisional balance (OR: 5.28, p<0.001) and negative cultural beliefs/attitudes (OR: 0.13, p<0.001) were associated with Pap testing. After adjusting for all covariates, only the relationship between high self-efficacy (aOR: 9.38, p<0.01), negative cultural beliefs/attitudes (aOR: 0.17, p=0.01) and Pap testing remained significant. High health literacy (OR: 3.23, p<0.01) was associated with Pap testing in the bivariate model, but the relationship was no longer significant after adjusting for main study variables (cervical cancer knowledge, self-efficacy, decisional balance, cultural beliefs/attitude) and other covariates. Since after controlling for covariates, we found no significant association between health literacy and Pap testing, the first step of the Baron and Kenny⁴⁷ test of mediation was not met. Therefore, no further tests of mediation were warranted.

DISCUSSION

The purpose of this study was to examine theoretically selected variables (i.e., antecedents of health literacy such as types and sources of health information, health literacy, and other psychosocial and cultural factors in relation to Pap testing among AI women living in the United States. A large proportion of our study participants (71%) reported receipt of a Pap test. This result is contrary to findings from previous studies which reported lower rates of Pap testing among AI women whose screening rates ranged from 20% to 70%, in comparison to 35%−85% in other immigrant samples (Asian, Hispanic, Eastern European).^46,48–50 Compared to previous studies, our study participants also reported higher rates of health insurance coverage (74%) and access to primary care (79%), two factors known to be associated with higher rates of Pap testing among immigrants.^16,51 This may have resulted in the higher prevalence of Pap testing reported in our study sample. Secondly, previous studies have shown that self-reported Pap testing rates are overestimated and often vary from Pap test rates that are based on medical records.⁵² The prevalence of Pap tests reported in the AfroPap study were based on women’s self-reports and could be an overestimation of women’s actual Pap testing history.

Using television, print media and the internet to convey health information with the goal of encouraging behavior modification such as healthy diets and exercise has been studied. What is lacking, however, are quantitative studies to examine the strength of association between sources and types of health information and cancer screening behaviors.^15,53 One study which included predominantly non-Hispanic white women showed an association between the use of television as a source for health information and an increased odds of mammography.¹⁵ Previous studies suggest that AI women health behaviors are potentially influenced by the health information they receive and share within their social circles.^18,54 Based on these findings, we also performed an exploratory analysis to examine the impact of sources (female relatives, female friends, pastor/church, TV/Radio, Internet, social media, physician/healthcare provider) and types (verbal, written text, pictures) of health information on AI women’s Pap testing behaviors. Our findings indicate that using multiple (2 or more) sources of health information is associated with an 89% lower odds of Pap testing. We also found a positive correlation between women’s educational attainment and use of the internet and healthcare providers as sources of health information. An inverse relationship was found among educational attainment and the use of family, friends and ethnic churches as health information sources (data not shown). This finding is similar to previous studies on Haitian immigrants⁵⁵ and Iraqi adults¹⁴ which showed that highly educated individuals are likely to report using single sources (i.e. the internet or healthcare providers) of health related information compared to lower educated individuals who use multiple sources (i.e. friends, family, religious organizations) of health information to inform their health behaviors. We found no association between types of health information and Pap testing among AI women. The lack of variability in the types of health information study participants received regardless of their history of Pap testing could explain our non-significant findings. Our results highlight the need for further studies to explore AI women’s preferred sources/types of health information, and how their perceived usefulness influence AI women’s health behaviors.

AI women with higher cultural beliefs/attitudes scores, indicative of fatalistic beliefs about cervical cancer and negative attitudes towards Pap testing, were 83% less likely to report the receipt of a Pap test in the past. This finding shows that among AI women, culture plays a critical role in informing and shaping cancer screening behaviors regardless of other known social determinants of health such as education, healthcare access and English proficiency. This result is new from previous studies which showed that cultural beliefs and attitudes played a less critical role in the cancer screening behaviors of immigrant women with high English proficiency and educational levels.^9,56 Our finding also shows that when addressing the cancer health needs of AI women, the development and use of culturally tailored programs to address the cancer health needs of AI women could be an effective strategy.

Previous studies have reported that immigrants who have stayed longer in a host country are more likely to be well acquainted with the host country’s healthcare system and are likely to utilize the preventive care services which are provided.^57,58 In the AfroPap study, we found no significant association between years of stay in the U.S and AI women’s receipt of a Pap test. Among immigrant groups, acculturation is shown to be a stronger predictor of health behaviors than individuals’ length of stay in a host country. Previous studies that used length of stay as a proxy measure for acculturation reported mixed findings.^16,59,60 This could explain the nonsignificant findings in our study. Future studies that use psychometrically tested instruments to assess the influence of acculturation on Pap testing behaviors among AI women are needed. Higher odds of Pap testing were seen among AI women who reported having a healthcare provider compared to women who reported no primary care access. In addition, AI women with higher English proficiency scores reported significantly higher odds of Pap testing compared to those with lower English proficiency scores. This finding is consistent with several studies which reported access to a primary healthcare as known facilitator of Pap testing, and low English proficiency as a barrier to women’s receipt of a Pap test.^51,56,57 The majority of our participants have health insurance coverage (79%) so they are likely to have access to primary care. Also, participants migrated mainly from African countries where English is spoken as an official language, so participants are likely to be more English proficient than other groups migrating from countries where English is not the official language.

Low health literacy is a strong predictor of the utilization of preventative health services among immigrants, with previous studies indicating the important role that improving the health literacy of immigrants could help in addressing health disparities, particularly in relation to cancer outcomes.^8,9,57 Our findings indicated that AI women with high health literacy levels were 3 times more likely to report receipt of a Pap test compared to those with lower health literacy levels at bivariate level. Individuals’ health literacy levels are directly impacted by their educational attainment.⁶ Our findings show that 47% of AI women have low health literacy levels. This proportion is drastically lower than what was reported (71%) in a Norwegian study on low educated (70% less than high school education) Somali immigrants.⁵⁷ A high proportion of our study participants are English proficient (77%) and college educated (68%), and this may have potentially influenced the higher health literacy levels reported.

Recruiting and enrolling immigrants and ethnic/minority groups in research studies can be particularly challenging.^61–63 In the U.S., the recruitment of AIs for cancer prevention studies are not well documented, and previous studies have mostly included African immigrant participants residing in a specific region (e.g. Minnesota, Washington D.C) or originating from a specific country (e.g. Ghana, Nigeria and Somalia).⁶⁴ In order to reach and recruit a diverse sample of AI women across the United States for the AfroPap study, we recruited participants using both online and in-person strategies. Within a one-year period, we met our study goal of recruiting a total of 167 AI women, and online recruitment resulted in the highest yield (data reported in another study). Details of the recruitment strategies are recorded elsewhere.⁶⁴ Similar to previous studies, participants recruited online were younger than those recruited in person.^65,66 The two groups, however, did not differ in employment status, income, health insurance coverage, length of stay and history of Pap testing.

Strengths and limitations

Study participants’ history of Pap testing was self-reported, which limits the generalizability of study findings due to potential recall bias. We used a cross-sectional design which precludes causality. Our convenience sample of 167 also delimits the study’s generalizability because participants recruited online were significantly younger and had higher English proficiency scores compared to AI women recruited in-person who were older and reported lower English proficiency scores. However, it is worth noting that the demographic characteristics of our study participants are similar to national data on African immigrants currently living in the U.S.²⁹ We found no significant association between health literacy and Pap testing after controlling for covariates so no further tests of mediation were warranted. Larger longitudinal studies are needed to test for potential psychosocial and cultural mediators of the relationship between health literacy and Pap testing among AI women. We observed wide confidence intervals on the variables, “self-efficacy “and “primary care.” A large proportion of participants reported access to a primary care provider (74%) and high self-efficacy (57%). It is possible that both variables were underpowered when we conducted multiple adjusted analyses. This study had several strengths which outweigh the limitations we identified. First, this is one of the first studies to quantitatively analyze the health literacy levels of AI women in the U.S. We successfully recruited a potentially “hard-to-reach” immigrant population using both in-person and online recruitment strategies. In order to increase study participants’ willingness to answer gender-specific and sensitive reproductive health questions, the research staff, including the principal investigator were female, African immigrants. The research team members spoke various African dialects (Twi, Ga, Pidgin, Igbo), thus enabling effective communication with potential study participants who required some assistance in completing the study surveys.

IMPLICATIONS

Physician recommendations, access to primary health care, sources of health information and negative, fatalistic cultural beliefs/attitudes may explain why AI women decide to utilize cancer screening services. To further understand how these factors impact AI women’s cancer screening behaviors calls for larger studies which utilize a robust sampling strategy and include a diverse group of African immigrants. To gain an in-depth understanding of how cultural beliefs impact health behaviors among AI women calls for methodologically rigorous studies that utilize both quantitative and qualitative research designs. In terms of recruitment strategies, using the internet appears to be a promising strategy to consider for future studies that include African immigrants.