Correlates of cervical cancer prevention advocacy and cervical cancer screening in Uganda: Cross-sectional evaluation of a conceptual model

Abstract

An approach to increasing cervical cancer (CC) screening is to empower women who have been screened to act as advocates and encourage other women they know to get screened. We examined correlates of CC screening advocacy and CC screening uptake among constructs in our conceptual model of factors driving engagement in advocacy. A cross-sectional, correlational analysis was conducted with survey data from 40 women (index participants) who had recently screened for CC, and 103 female members of their social network (alter participants) who had not been screened. Variables measured included CC prevention advocacy, as well as internalized CC stigma, sharing of CC screening result, CC knowledge, healthy bodily intake (i.e., diet; alcohol and cigarette use) and self-efficacy related to CC service utilization and CC prevention advocacy, which were hypothesized to be associated with advocacy. Bivariate and multivariable regression analyses, controlling for clustering, were conducted. Among index participants, greater engagement in advocacy was positively correlated with CC knowledge, sharing of CC screening result, and CC service utilization self-efficacy. Women who had screened positive and received treatment for precancerous lesions reported greater CC prevention advocacy, CC knowledge and healthy living, compared to those who screened negative. In multiple regression analyses, CC screening was positively associated with CC prevention advocacy and being age 36 or older, and CC prevention advocacy was also positively associated with CC service utilization self-efficacy. These findings support the validity of our conceptual model regarding factors associated with engagement in CC prevention advocacy among women screened for CC. The strong association between CC prevention advocacy and both CC screening uptake and CC service utilization self-efficacy suggests the potential value of advocacy promotion among women who have been screened, as well as for increasing screening uptake.

1. Introduction

Cervical cancer (CC) is the most common cancer among women in Uganda, with one of the highest incidence rates in the world at 54.8 per 100,000.^[1–3] In Uganda, CC screening via visual inspection of cervix with acetic acid, and thermal therapy for precancerous lesions, are free or low cost, but are scarcely available outside the capital Kampala. Similarly, radiotherapy is used to treat advanced disease, but it is too costly for most people, and only available at one center in Kampala, despite the vast majority of Ugandans living in rural areas. This highlights the importance of screening to prevent onset of cancerous lesions and the emergence of a need for advanced treatment. Nonetheless, most (80%) Ugandan women have advanced CC disease (stage III or higher) at initial presentation for care,^[4] and as little as 5% of women have ever screened for CC.^[4–6]

One approach to increasing CC screening uptake is to empower women who have personal experience with being screened for CC (and treated if applicable) to act as advocates and encourage other women they know to get screened. Peer advocacy interventions have been effective at increasing prevention behaviors and reducing stigma in the context of HIV.^[7–10] However, these interventions are under-studied in the context of CC screening promotion, despite evidence that facilitators of CC screening include encouragement from others to get screened, and knowing someone who has been screened for or diagnosed with CC.^[6,11]

Theories of social diffusion,^[12] cognitive consistency,^[13] and social influence^[14] suggest that behaviors used by a few can be diffused to others through modeling, advocacy, and shifts in social norms. Drawing on these concepts, we developed a conceptual framework (see Fig. 1) for how women who have screened for CC can act as change agents for other women to get screened.^[15] The framework hypothesizes that not feeling internal stigma about one’s level of risk for CC (e.g., not feeling ashamed about getting screened for CC, being at risk for CC, or screening positive for signs of CC risk) facilitates comfort with sharing one’s CC screening experience with others (particularly if her screening resulted in evidence of precancerous or cancerous lesions and need for treatment). Sharing this experience also enhances the credibility of one’s advocacy for CC screening and treatment. To be comfortable and effective in advocating for specific health behaviors, a woman must model the behaviors she is encouraging others to adopt, by prioritizing healthy living (e.g., good nutrition, sexual risk reduction, timely CC screenings). Good knowledge about CC (what causes CC, how it can be prevented, and available treatments) is important for accurate CC-related advocacy. Lastly, learning communication skills and strategies for when, how and who to engage with advocacy helps to bolster confidence and self-efficacy for conducting advocacy. Increased CC prevention advocacy with women in one’s social network should result in increased CC screening uptake in the network and larger community.

Figure 1.

Conceptual framework for promotion of cervical cancer (CC) prevention advocacy among screened women to affect CC screening among social network members.

We enrolled women who had recently screened for CC, as well as unscreened women from within their social networks, to participate in a pilot study to evaluate an intervention to promote CC prevention advocacy. We used baseline data from the screened women to examine correlates of CC prevention advocacy from among constructs in our conceptual model. We hypothesized that engagement in prevention advocacy would be associated with lower internalized CC stigma, greater sharing of personal CC screening experience, and greater CC knowledge, healthy living and self-efficacy related to advocacy. We also examined variables that distinguished the screened women from those who had not screened, to identify factors that may influence uptake of CC screening.

2. Methods

2.1. Study design

We conducted a cross-sectional correlational analysis with baseline data from a randomized controlled trial of an intervention that empowers women who have recently screened for CC (referred to as index participants) to advocate for CC screening and treatment among female members of their social network (referred to as “alters”). Each index participant was asked to refer up to three alters who had not screened for CC within the past three years. Further details of the study protocol are available in a prior publication.^[15] Ethical review and approval was obtained from the Makerere University School of Public Health Research and Ethics Committee (approval number SPH-2021-103), and the Uganda National Council for Science and Technology (approval number HS1633ES). All index and alter participants provided written informed consent.

2.2. Setting

The study took place at Buyinja Health Center in Namayingo, a rural community in the Busogo region of Uganda that is approximately 200 km from Kampala. CC screening and thermal therapy was available at Buyinja Health Center, as well as another nearby health center in the district, and from Rays of Hope Hospice Jinja which conducts periodic mobile CC screening and thermal therapy “day camps.”

2.3. Participants

Recruitment of all participants was conducted in September, 2021. Given the pilot nature of the study, a target sample size was selected based on feasibility rather than statistical power. Women were eligible to enroll as index participants if they were age 18 years or older, had screened for CC within the past year, had stable health status (i.e., judged by the study coordinator to be likely to be healthy enough to complete the 6-month study follow-up, based on physical presentation and reported level of disease), and had shared their CC screening experience with at least one alter who they perceived to not have screened for CC in the past 3 years. Alter participants were eligible if they were at least 18 years of age, were recruited by a woman who was enrolled as an index participant, and self-reported not being screened for CC in the past 3 years. All participants had to speak either Samia or Lusoga, the 2 prominent languages in Namayingo.

Index participants were informed of the study by health care providers either in-person or over the phone, and those who expressed interest were referred to the study coordinator for an in-person screening assessment to confirm eligibility and perform consent procedures. Women who decided to enroll were then administered the baseline survey, which included listing up to 12 women in their social network. To recruit alters, we used data from this social network assessment to randomly select 5 alters who the index participant reported as knowing her CC screening experience (or as many as there were if <5); the index participant was then asked if she was comfortable asking three of these alters to participate. The index participant was asked to call each selected alter at the end of the interview to describe the study in the presence of the coordinator, who then scheduled a study visit if the alter expressed interest in participating. If an alter refused or could not be reached, a replacement was randomly selected from the list of alters who knew the index participant’s CC screening experience (and whom the index participant was comfortable recruiting). When screening the alter, the coordinator confirmed that the individual was not already enrolled (i.e., recruited by another index).

2.4. Measures

The assessment was interviewer-administered using Network Canvas software and conducted in either Samia or Lusoga, depending on the preference of the participant. Measures were translated into Samia and Lusoga using standard translation/backtranslation methodology.^[16] The assessment included a standard survey (index and alter participants) and social network assessment (index participant only). Self-reported CC screening and treatment utilization were verified with abstracted medical chart data. All measures were developed by the study team, except those in which an attribution is cited. For the measures that we developed that include at least 3 items, we cite internal reliability statistics (Cronbach alpha). Participants received 30,000 Uganda Shillings (~$8 United States dollars) after completing the assessment for transportation costs.

2.4.1. Measures administered to all participants.

Demographic and background characteristics included age, level of formal education completed, relationship status and HIV status.

CC knowledge was assessed with 16 statements that reflect the etiology, prevention and treatment of CC; participants responded by indicating whether the statement was “true” or “false” (or “don’t know”) and the sum of correct responses was calculated (Cronbach alpha = .75).

Aspects of healthy living that were assessed included behaviors related to diet and substance use, and ability to notice signs of CC risk and to seek CC-related health services. Healthy bodily intake was assessed with three questions measuring frequency of alcohol use, cigarette smoking, and healthy foods; response options ranged from 0 “never” to 4 “four or more times a week” and a higher mean item score reflects more healthy intake. CC service utilization self-efficacy was assessed with three items indicating confidence to notice a symptom of CC risk, seek health services for a symptom of CC risk, and obtaining treatment if screening revealed signs of CC risk; response options ranged from 0 “you cannot do it at all” to 10 “you are completely certain you can do it,” and a higher mean item score reflects greater self-efficacy (Cronbach alpha = .44).

CC prevention advocacy was assessed with 6 items in which respondents reported the frequency of discussing CC-related topics (e.g., importance of screening, how and where to get screened, importance of getting treatment if signs of CC risk are present) with women they know in the past 6 months; response options ranged from 1 “not at all” to 5 “very much,” and a higher mean item score reflects greater engagement in advocacy (Cronbach alpha = .95).

CC screening and treatment.

Participants reported whether they had ever been screened for CC using visual inspection using acetic acid or pap smear, and if so, when. Those who had been screened were asked if the screening resulted in precancerous lesions or potential cancerous lesions, in separate items; if either type of lesion was reported, receipt of treatment and biopsy was assessed.

2.4.2. Measures administered only to index participants.

Internalized CC stigma was measured with 5 items adapted from a scale of HIV internalized stigma^[17] (e.g., My cervical cancer screening makes me feel ashamed of myself; I think it is safer to keep my cervical cancer screening to myself), that assess the respondent’s feelings about her CC screening and sharing her screening experience with others; response options were 1 “disagree,” 2 “I neither agree nor disagree; I do not have a feeling either way,” and 3 “agree,” and a higher mean score reflects greater stigma (Cronbach alpha = .57).

Sharing of CC screening experience was assessed by asking respondents to what extent they had shared their CC screening result with sexual partners, family, and friends, in separate questions; response options were 0 “none,” 1 “some” and 2 “all,” and a higher mean item score reflects greater disclosure (Cronbach alpha = .74).

CC prevention advocacy self-efficacy was assessed with three items assessing confidence to start a conversation about the need for: CC screening, treatment for signs of CC risk, and telling someone about their CC screening experience; response options ranged from 0 “you cannot do it at all” to 10 “you are completely certain you can do it,” and a higher mean item score reflects greater self-efficacy (Cronbach alpha = .85).

Social network assessment.

Each index participant was asked to list 12 women in their social network (referred to as “alters”) with whom they interact most. For each alter we gathered information to assess network composition (e.g., age, HIV status, relation to and frequency of contact with index participant; level of trust in the alter; perceived history with CC screening and treatment; knowledge of index’s CC screening and treatment). Our prior research shows that alter health seeking behaviors can be accurately reported by index participants.^[18]

Social support from social network members.

Index participants were asked to rate how likely it was that each alter they named in the social network assessment would provide 2 types of support (take you to the doctor if needed; provide suggestions for how to deal with a personal problem); response options were 0 “not likely,” 1 “somewhat likely” and 2 “very likely,” a higher mean item score was calculated for each alter and across all alters named, and higher scores reflect more support.

CC stigma expressed by social network members.

Index participants were asked to indicate whether or not each alter named had expressed CC stigma using 2 items (She has said that a person with cervical cancer must have done something wrong; She looks down on you because of your experience with screening for cervical cancer.) using a yes/no response format; the sum of items reported as “yes” was calculated for each alter and across all alters named, and higher scores reflect more stigma.

2.4.3. Measures administered only to alter participants.

CC screening self-efficacy was assessed only among alter participants, since all index participants had been recently screened and it could be argued they all had good self-efficacy for screening. The measure consisted of 9 items that assessed barriers (e.g., embarrassment to have my private parts looked at; husband/partner would not let me go) to screening. Self-efficacy has been shown to be an important precursor to behavior change,^[19] and therefore may influence the behavior of actively seeking and undergoing CC screening. Response options included 1 “agree,” 2 “disagree” and 3 “don’t know,” with “don’t know” responses not included in the calculation of the mean item score; a higher mean item score reflects greater self-efficacy. Internal reliability was moderate (Cronbach alpha = .64).

2.5. Data analysis

Initial analysis included an assessment of bivariate correlation between CC prevention advocacy and measures drawn from the constructs in the conceptual framework, demographics and CC screening result. We also examined correlations among the constructs in the conceptual framework, in addition to demographics, social network characteristics, and CC screening result. In the above analyses, data from the subgroup of participants who had ever been screened for CC (40 index participants) were used.

Then, to identify correlates of CC screening, we used data from both the 40 index (who had been screened) and 103 alter participants (who had not been screened). The analysis included those measures that were administered to both types of participants, namely demographics, CC knowledge, healthy living, and CC prevention advocacy. We also identified correlates of CC screening self-efficacy among alter participants. Given the statistical power afforded by the larger sample sizes involved with these 2 sets of analyses (compared to the initial analysis with only index participants), we estimated logistic and linear regression models to predict CC screening and CC screening self-efficacy, respectively, in separate analyses. In these regression models, independent variables were significant bivariate correlates (P < .05) of the dependent variable, in addition to age and any secondary education. All analyses that included data from alters, controlled for clustering at the level of each index (i.e., all the alters recruited by a specific index, as well as the index participant themselves, represent a single cluster) by using a SurveyReg or SurveyLogistic routine in statistical analytic software (SAS) 9.2.^[20] All participants had completed the baseline assessment, so there were no missing data to be accounted for in the analysis.

3. Results

3.1. Sample characteristics

Forty women who had screened for CC within the past year were screened for study eligibility, all of whom were eligible and enrolled in the study; 24 (60%) screened positive for signs of precancerous lesions, all of whom received treatment (17 received thermal therapy, 7 received cryotherapy), while the remaining 16 screened negative for any sign of CC risk. From these 40 index participants, 103 female social network members (alters) who had never been screened for CC, enrolled in the study. Table 1 shows the characteristics of the index and alter participants. Most index participants were age 36 or older (58%), compared to just 38% of alter participants (P = .03), but the two subgroups were similar on other demographics; under 30% of both subgroups had any secondary education, and very few women self-reported being HIV-positive.

Table 1.

Baseline sample characteristics by study arm.

Characteristic	Index (n = 40) %/mean (SD)	Alter (n = 103) %/mean (SD)	P
Age 36 years and over	57.5%	37.9%	.03
Any secondary education	22.5%	29.1%	.26
In a committed relationship	85%	73.8%	.08
HIV-positive	7.5%	5.8%	.71
Disclosure of CC screening result	1.5 (0.5)	–	–
CC knowledge	9.4 (3.2)	5.2 (2.8)	.17
CC internalized stigma	1.14 (0.3)	–	–
Engagement in CC prevention advocacy	3.4 (1.1)	1.6 (0.5)	.35
CC advocacy self-efficacy	9.7 (0.5)	–	–
Healthy bodily intake	3.8 (0.3)	3.8 (0.2)	.27
CC service utilization self-efficacy	8.8 (1.3)	7.2 (1.5)	<.001
CC screening self-efficacy	–	2.8 (0.3)	–
Number of social network members (alters) named	7.8 (2.4)	–	–
Mean alter trust level	1.6 (0.4)	–	–
Mean alter support	1.5 (0.5)	–	–
Mean alter expression of CC stigma	0.6 (2.3)	–	–

CC = cervical cancer, SD = standard deviation.

3.2. Correlates of cervical cancer prevention advocacy

Among index participants, greater engagement in advocacy was positively correlated with CC knowledge (r = .66; P < .001), sharing of CC screening result (r = .36; P = .02), and CC service utilization self-efficacy (r = .63; P < .001); it was not associated with internalized CC stigma or prevention advocacy self-efficacy. Also, advocacy was not associated with the demographics; however, the index participants who had screened positive and received treatment for precancerous lesions engaged in more CC prevention advocacy [mean = 3.9 standard deviation (SD = 1.0)] compared to those who screened negative [mean = 2.7 (SD = 0.9); t = 3.9, df = 38, P < .001].

Findings from the correlations among the constructs we hypothesized to be associated with advocacy, as well as with demographics and social network characteristics, indicated that internalized CC stigma was only associated with lower sharing of CC screening result (r= -.40; P = .01). Other correlates of greater sharing of CC screening result were higher level of trust in alters (r = .39; P = .01) and greater perceived support from alters (r = .32; P = .04). CC knowledge was positively correlated with CC service utilization self-efficacy (r = .43; P = .006) while CC prevention advocacy self-efficacy was negatively correlated with level of trust in alters (r= -.36; P = .02). As noted above, CC service utilization self-efficacy was correlated with CC knowledge, but no other construct, and healthy bodily intake behavior was not associated with these constructs.

Those who screened positive and were treated for precancerous lesions had higher levels of CC knowledge (mean = 10.8 vs 7.1; t = 4.3, df = 38, P < .001) and healthy bodily intake behavior (mean = 3.9 vs 3.7; t = 2.1, df = 38, P = .02), but lower CC prevention advocacy self-efficacy (mean = 9.6 vs 10.0; t = 3.6, df = 38, P = .001), compared to those who screened negative.

3.3. Correlates of cervical cancer screening and related self-efficacy

Women who had been screened (index participants) differed from those who had not (alters) in that they had higher levels of CC knowledge [mean = 9.4 vs 5.2; t = 4.8; df = 39; P < .0001; odds ratio, OR (95% confidence interval, CI) = 1.61 (1.32, 1.97)], greater CC service utilization self-efficacy [mean = 8.8 vs 7.2; t = 4.2, df = 39; P < .001; OR (95% CI) = 2.31 (1.54, 3.46)], greater engagement in CC prevention advocacy [mean = 3.4 vs 1.6; t = 5.4, df = 39, P < .0001; OR (95% CI) = 10.55 (4.37, 25.48)], and were older (58% age 36 or older compared to 38%; Rao-Scott chi square = 5.1, df = 1, P = .02). In the multiple logistic regression analysis, engagement in CC prevention advocacy [Adjusted OR (95% CI) = 9.38 (3.28, 26.82)] and age >35 years [Adjusted OR (95% CI) = 4.43 (1.35, 14.52)] remained correlates of CC screening.

CC service utilization self-efficacy is an important proxy for CC screening, as indicated in its association with CC screening noted above. Other correlates of greater CC service utilization self-efficacy included greater CC knowledge [r = .59; P < .0001; beta standard error (SE) = 0.27 (0.04)], engagement in CC prevention advocacy [r = .61; P < .0001; beta (SE) = 0.90 (0.08)], and being in a committed relationship [mean = 7.9 for those in a relationship, vs 6.9 for those who were single; t = 3.4, df = 39, P = .002; beta (SE) = 1.01 (0.30)]; only CC prevention advocacy remained a correlate in the multiple linear regression analysis [beta (SE) = 0.29 (0.03); P < .0001]. CC screening self-efficacy was only measured among alters and it too was positively correlated with CC knowledge [r = .31; P = .001; beta (SE) = 0.03 (0.01)], engagement in CC prevention advocacy [r = .26; P = .008; beta (SE) = 0.13 (0.05)], and CC service utilization self-efficacy [r = .26; P = .03; beta (SE) = 0.05 (0.02)]; in the multiple linear regression analysis, only CC knowledge remained a correlate [beta (SE) = .02 (.01), P = .04].

4. Discussion

In this small sample of women who had recently screened for cervical cancer (CC), several of the variables in our conceptual model were associated with engagement in CC prevention advocacy as hypothesized. Engagement in CC prevention advocacy was positively associated with three of the five hypothesized mechanisms in our model: sharing CC screening results, CC knowledge, and CC-related service utilization self-efficacy. Although advocacy was not associated with internalized CC stigma, our conceptual model posits that low internalized stigma is important for being comfortable sharing CC screening results, and these two mechanisms were indeed negatively correlated in our data. Also, CC knowledge was positively correlated with CC-related service utilization self-efficacy (which was not associated with advocacy). Together, these findings provide preliminary support for the validity of the conceptual framework.

Comfort with sharing one’s CC screening experience establishes a foundation for engagement in CC-related advocacy. Advocacy is more likely to influence behavior change if the advocate is perceived as credible, and sharing one’s personal experience with CC screening (and treatment) helps establish this credibility. Our data showed women are more likely to share their CC screening result with women whom they feel close to, supported by, and who they can trust. This is consistent with other cancer disclosure research,^[21] and supports our intervention’s focus on promoting CC prevention advocacy within one’s personal network.

Women who screened positive and were treated for precancerous lesions reported higher CC prevention advocacy, CC knowledge and healthy bodily intake, compared to the women who screened negative. Women who screen positive and are treated may have more motivation to help others limit their CC risk. Furthermore, receipt of treatment is likely accompanied by more CC information from providers, which could increase engagement in advocacy and healthy behaviors.

CC knowledge appears to play a prominent role in both CC prevention advocacy and CC screening, which were bivariate correlates of knowledge. CC knowledge among Ugandan women in general has been shown to be low^[22,23]; however, our finding that screened women have higher CC knowledge suggests that women are benefiting from the CC-related information provided during the screening (and treatment) process. Knowledge about the importance of CC screening may help women feel more confident about their ability to access screening and services, and better equipped to encourage other women to get screened. These relationships may also be bidirectional –better knowledge and confidence may facilitate engagement in advocacy, but engagement in advocacy may also serve to increase one’s confidence in being able to get screened and access CC-related services in general.

There are several limitations to our analysis. The women in our sample who had screened for CC represent a selection bias, as they had decided to enroll in a study that would train them to engage in CC prevention advocacy; motivation to be such an advocate is likely associated with greater CC knowledge and other constructs we measured, and not representative of screened women in general. Data for some of the measures (e.g., healthy bodily intake, self-efficacy for CC prevention advocacy) were highly positively skewed, causing a potential ceiling effect and limiting the ability to detect associations. This may have contributed to counter intuitive findings, such as the negative correlation between prevention advocacy self-efficacy and trust in one’s social network members. Furthermore, established measures of the constructs assessed were often not available in the literature, requiring the study team to develop new measures, some of which showed low internal reliability; further research is needed to establish reliable, valid measures of the constructs assessed to better examine the associations between constructs. Other limitations include the small sample size, limited statistical power, and the correlational nature of the analysis that precludes any causal inferences.

5. Conclusion

In conclusion, the findings from this cross-sectional analysis provide support for much of our conceptual model regarding factors associated with engagement in CC prevention advocacy among women screened for CC, although not all hypothesized factors were found to be associated with advocacy. These findings also provide support for the composition of the intervention we are currently testing with this sample. The strong association between CC prevention advocacy and both CC screening uptake and CC service utilization self-efficacy suggests the potential value of advocacy promotion among women who have been screened, as well as for increasing screening uptake. The post-intervention follow-up data from this pilot, as well as future trials of the intervention if shown to be effective, will afford opportunities to further evaluate our conceptual framework, including its hypothesis regarding the pathway between CC prevention advocacy and CC screening uptake among social network members.

Acknowledgments

We would like to acknowledge the contribution of our study coordinator, Grace Namisi, who was responsible for participation recruitment and all data collection. We thank the women who agreed to participate in the study. This research is funded by a grant from the Fogarty International Center, National Institutes of Health (R21TW011728; PI: Wanyenze).

Author contributions

Conceptualization: Glenn J. Wagner, Joseph K.B. Matovu, Margrethe Juncker, Eve Namisango, Kathryn Bouskill, Jolly Beyeza-Kashesya, Emmanuel Luyirika, Rhoda K. Wanyenze.

Formal analysis: Rhoda K. Wanyenze.

Funding acquisition: Glenn J. Wagner.

Methodology: Joseph K.B. Matovu, Margrethe Juncker, Eve Namisango, Kathryn Bouskill, Sylvia Nakami, Jolly Beyeza-Kashesya, Emmanuel Luyirika, Rhoda K. Wanyenze.

Project administration: Joseph K.B. Matovu.

Supervision: Joseph K.B. Matovu, Eve Namisango.

Writing – original draft: Glenn Wagner.

Writing – review & editing: Joseph K.B. Matovu, Margrethe Juncker, Eve Namisango, Kathryn Bouskill, Sylvia Nakami, Jolly Beyeza-Kashesya, Emmanuel Luyirika, Rhoda K. Wanyenze.