Testing a Prostate Cancer Educational Intervention in High-Burden Neighborhoods

Abstract

To better capitalize on our enhanced understanding of prostate cancer (PCa) risk factors it is important to better understand how knowledge and attitudes contribute to ethnic disparities in PCa outcomes. The goal of this study was to test the impact of a targeted PCa educational intervention vs. a healthy lifestyle educational control intervention on levels of knowledge, concern, and intention to screen for PCa. We recruited 239 men from neighborhoods with the highest PCa burden in Philadelphia. We assigned 118 men from two of the neighborhoods to the control group 121 men from 2 other neighborhoods to the intervention group. Repeated outcome assessment measures were obtained by administering the survey at baseline, post-session, 1 month post-session, and 4 months post-session. We conducted descriptive statistics to characterize the study sample and linear mixed effect regression models to analyze the intervention’s effect on the outcomes. At baseline, we observed no differences in the outcomes between the PCa-targeted intervention and healthy lifestyle control groups. We found that knowledge of PCa and intention to screen increased significantly over time for both the control and intervention groups (p≤0.01 at the 4-month follow-up). In contrast, change in the level of PCa concern was only significant for the intervention group immediately post-session and at 1-month follow-up (p=0.04 and p=0.01, respectively). This study showed that gathering at-risk men for discussions about PCa or other health concerns may increase their PCa knowledge and intention to talk to a doctor about PCa screening.

Introduction

Prostate cancer is a leading cause of morbidity and mortality in the US. Of the top ten leading types of cancer in US males, prostate cancer (PCa) was number one in 2020 for estimated new cases (191,930) and number two for estimated deaths (33,330) [1]. When compared to non-Hispanic Whites, the burden of PCa in African American males is substantially greater with an 2012–2017 incidence rate (per 100,000) of 173.0 (vs. 104.1) and death rate of 38.7 (vs. 19.1) [1]. The basis for this racial disparity is thought to be multifactorial involving an interplay of biological, socioeconomic (SES), cultural, and behavioral factors. [2] This high degree of disease complexity has made it extremely difficult to quantify the extent to which each of these factors contributes to overall outcomes in PCa. [2]

To better capitalize on our enhanced understanding of PCa risk factors it is important to better understand how knowledge and attitudes contribute to ethnic disparities in PCa outcomes. In a recent literature review [3], possible explanations for poorer PCa outcomes in African Americans included lower access to healthcare and multiple differences in other social and behavioral factors. These included mistrust of the healthcare system, a fatalistic viewpoint on cancer (one will die regardless of medical intervention), and the misperception that African American men are not at a greater risk for PCa. All of these potential factors lend themselves to the possibility that enhanced education with regard to general healthcare as well as PCa-specific knowledge would be beneficial.

The goal of this study was to test the impact of a targeted PCa educational intervention vs. a healthy lifestyle educational control intervention on levels of knowledge, concern, and intention to screen for PCa. We hypothesized that men in the PCa-focused intervention group would have higher levels of knowledge, lower concern, and increased intention to screen for PCa compared to the healthy lifestyle educational control group. We recruited male participants from four predominately African American Philadelphia communities to test the hypothesis that a single PCa educational session would positively impact overall PCa risk awareness and the likelihood for increased screening events. Furthermore, we determined whether a PCa-specific educational session provided greater benefit when compared to an educational session that focused predominantly on general healthcare.

Methods

Study Sample

We assessed 325 men in Philadelphia, PA from 2017 for participation in the Empowering Men about Prostate Cancer Together (EMPaCT) Study. The men were residents from four Philadelphia neighborhoods that we determined to be high burden areas for prostate cancer based upon cancer registry data. [4] They were recruited using fliers distributed in the community, social media, and word-of mouth through peer health workers and community stakeholders. After determining eligibility and ability to attend a scheduled educational session, the final enrolled sample consisted of 273 men of whom 239 completed baseline surveys. (Figure 1) We used 2010 census data to compare the demographics of our final sample to those of the general population in our selected neighborhoods.

Figure 1.

Box plot of PCa knowledge test scores by study group over time.

Study eligibility included living in one of the high risk neighborhoods with a main (but not exclusive) focus on men ages 40–69, an age range for whom screening recommendations would be most relevant. Men who did not live in one of the four selected high-burden neighborhoods, or who self-reported that they have previously been diagnosed with PCa were excluded.

Study Design

The study evaluated an educational intervention using a non-randomized comparison group design. Men residing in two of the Philadelphia neighborhoods received the targeted PCa educational intervention group and two similar neighborhoods (with similar sociodemographics) received the healthy lifestyle educational control group.

The PCa targeted intervention included an educational session to inform participants about the prostate health, prostate cancer and prostate cancer screening. The control group included a lifestyle educational session to inform participants about men’s health issues (e.g., health screenings, nutrition, tobacco use, visiting the doctor.) Trained Peer Educators delivered the educational sessions in selected neighborhoods. Sessions were held at city recreation centers, churches and synagogues, transitional housing sites, public libraries, and civic neighborhood organizations. Both types of sessions were conducted by the same group of Peer Educators and were held for the same amount of time using the same structure.

Variables

The independent variable indicated study group assignment (i.e., intervention vs. control). The questions used to assess the primary outcomes are listed in Table 1. The primary outcomes were change in PCa knowledge score, concern about PCa, and intent to screen for PCa. PCa knowledge was evaluated using a survey comprised of true-false questions. The final set of questions were approved by members of the research team (prostate cancer researchers, consultants, and Peer Educators). Our evaluation after preliminary data analysis showed that 3 knowledge questions (bolded in Table 1) showed little variation among participants, with greater than 90% of respondents answering correctly at baseline. These questions were removed from final analysis for the longitudinal repeated measures analysis of this study. The knowledge scoring metric was true/false coded as correct/incorrect responses, summed for a score of 0–7. The survey item assessing concern with a Likert scale rating concern about prostate cancer from not concerned (1 point) to very concerned (4 points). A question to assess the interest in talking to a doctor about PCa screening used a Likert scale from very unlikely (1 point) to very likely (5 points).

Table 1.

Survey Questions to Assess Prostate Cancer Outcomes

Prostate Cancer Knowledge Questions*		True	False
A	The prostate is a walnut-sized organ that only men have.	□	□
B	Older men are more likely to get prostate cancer than younger men.	□	□
C	African American men are less likely to get prostate cancer and die from their disease than white men.	□	□
D	Prostate cancer always has symptoms.	□	□
E	It is important for every man to talk to a doctor about whether prostate cancer screening is right for him.	□	□
F	Screening can include a digital rectal exam (DRE) and a blood test called the prostate specific antigen (PSA).	□	□
G	Medical experts agree on routine screening for prostate cancer.	□	□
H	Prostate cancer is easier to treat if it is detected early.	□	□
I	All prostate cancer grows at the same rate.	□	□
J	If no one in your family has had prostate cancer, you are not at risk.	□	□

^*Bolded items were removed in final analysis because >90% of respondents answered correctly

Prostate Cancer Concern

How concerned are you about getting prostate cancer?

□ Not concerned □ A little bit concerned □ Neutral □ Moderately concerned □ Very concerned

Intent to Screen for Prostate Cancer

How likely is it that you will do each of the following in the next 6 months?

□ Very unlikely □ Somewhat likely □ Neutral □ Somewhat likely □ Very likely

Data Collection

Repeated outcome assessment measures were obtained by administering the survey at multiple time points: baseline (pre-educational session with collection of demographics), post-session, 1 month post-session, and 4 months post-session. Participants completed the baseline and post-session surveys in person (self-administered.) One-month and 4-month follow-up surveys were administered by phone.

Data Analysis

Descriptive statistics

Descriptive statistics were used to analyze demographic factors. Median and range values were provided for continuous variables and point estimates of central tendency were analyzed by Kruskal-Wallis tests to evaluate the statistical significance of differences between study groups. Categorical variables were analyzed by Chi-square tests.

Box plots

Box plots were used to summarize the distributions of outcomes data between study groups over time on participants having complete baseline knowledge test scores. These plots show the mean (represented by the circle in the box); the median (represented by the line inside the box); the first & third quartiles (represented by the lower & upper bounds of the box); the smallest and largest observations within 1.5 interquartile ranges from the lower or upper bound of the box, respectively (represented by the “whiskers”); and observations beyond 1.5 interquartile ranges from the lower or upper bound of the box (potential outliers represented by circles).

Longitudinal linear mixed effect regression models

Longitudinal linear mixed effect regression models were used to model differences in outcome means between groups at different times over the course of the study. Random intercept effects were included in the model to account for the correlation in observations from the same neighborhood and applied an autoregressive covariance matrix structure to account for the serial correlation in outcome observations within participants over time. Fixed effects for time (baseline, post-educational session, 1-month follow-up, or 4-month follow-up), study group indicator (intervention vs. control), and their interactions were included and adjusted for participant age, body mass index (BMI under 25, 25-<30, or ≥30 kg/m²), current smoking status, marital status (married/partnered or other), and education (no high school diploma, diploma/college, or post-college/graduate-level) as covariates. We compared means over time within groups and between groups using linear contrasts of parameters from these adjusted models and presented results with 95% confidence intervals (CIs) and p-values.

The significance level for all tests was set a priori at 0.05 and all statistical analyses were conducted using SAS v9.4 (SAS Institute, Cary, NC, USA).

Results

Of the 239 men enrolled in the study, 118 (49%) men were assigned to the control group and attended general health educational sessions, and 121 men (51%) were assigned to the intervention group and attended PCa-specific educational sessions. Of the 239 study participants, 234 completed the post-educational session survey. A subset of participants also completed the later follow-up surveys (1 month surveys, n=190 and 4 month surveys, n=161).

As shown in Table 2, with the exception of two variables (education and smoking status), baseline characteristics were not significantly different between the two groups. The median age of the control group was 53 (range 29 to 69) and the median age of the intervention group was 54 (range 27 to 83) (p=0.466). Consistent with the racial composition of their community, 95% percent of the control group and 89% of the intervention group were African-American (p=0.156). The time of residence in the current neighborhood ranged from 1 month to 63 years and was not significantly different between the two groups (median of 10 years in both groups, p=0.277). Educational attainment was higher in the intervention group (p=0.005). Marital and employment status were similar between the two groups (p=0.136 and p=0.120, respectively). Rates were similar for both groups regarding having health insurance (86% control, 89% intervention; p=0.456) and Likert Scale health ratings (median=4 control, median=4 intervention; p=0.454). The intervention group had significantly higher rates of smoking (p<0.001) than the control group. The number of clinically-defined obese individuals was somewhat disparate but differences between the control and intervention groups did not reach statistical significance (32% control, 44% intervention; p=0.065). Baseline characteristics related to PCa-specific health practices including screening (historical or planned), as well as fundamental knowledge of the disease and inherent concerns were all similar between the two groups.

Table 2.

Demographics of Participants in the EMPaCT Study (n=239)

Variables		All Participants N=239	Participants from Control Neighborhoods N=118	Participants from Intervention Neighborhoods N=121	p-value
Median age (range)		53 (27–83)	53 (29–69)	54 (27–83)	0.466
Median years resided in current neighborhood (range)		10 (0.1–63)	10 (0.1–61)	10 (0.2–63)	0.277
Race	African American	212 (92%)	106 (95%)	106 (89%)	0.334
	White	8 (4%)	2 (2%)	6 (5%)
	Other	11(5%)	4 (4%)	7 (6%)
Education	Less than a HS diploma	40 (17%)	25 (23%)	15 (13%)	0.005
	HS diploma	102 (44%)	50 (45%)	52 (44%)
	Some college	49 (21%)	22 (20%)	27 (23%)
	4 yr. college degree	21 (9%)	12 (11%)	9 (8%)
	Post-college/graduate degree	18 (8%)	2 (2%)	16 (14%)
Married	Married	53 (23%)	21 (19%)	32 (27%)	0.136
	Single	141 (61%)	73 (65%)	68 (57%)
	Separated/Divorced	31 (13%)	13 (12%)	18 (15%)
	Widowed	6 (3%)	5 (5%)	1 (1%)
Work Status	Full time	57 (25%)	30 (27%)	27 (23%)	0.120
	Part time	29 (13%)	12 (11%)	17 (14%)
	Retired	22 (10%)	6 (5%)	16 (14%)
	On medical leave	58 (25%)	26 (23%)	32 (27%)
	Seasonal work	14 (6%)	10 (9%)	4 (3%)
	Unemployed	50 (22%)	27 (24%)	23 (19%)
General Health Status, Median Score		4	4	4	0.454
Currently Have Some Form of Health Insurance	Yes	203 (88%)	97 (86%)	106 (89%)	0.456
	No	29 (13%)	16 (14%)	13 (11%)
Current Smoker	Yes	136 (59%)	83 (74%)	53 (45%)	<0.001
	No	95 (41%)	29 (26%)	66 (56%)
Obese	Yes	91 (38%)	38 (32%)	53 (44%)	0.065
	No	148 (62%)	80 (68%)	68 (56%)
Concerned about PCa	Not concerned	17 (8%)	8 (8%)	9 (8%)	0.968
	Little Bit Concerned	36 (17%)	16 (15%)	20 (17%)
	Moderately concerned	70 (36%)	38 (37%)	41 (36%)
	Very Concerned	85 (39%)	42 (40%)	43 (38%)
Ever screened for PCa		113 (51%)	56 (51%)	57 (51%)	0.997
Last screened > 1 yr ago for PCa		55 (63%)	29 (67%)	26 (59%)	0.419
Intention to Screen for PCa in next 6 months, Median Score		4	4	5	0.479
KNOWLEDGE questions		% Correct
Only men have a prostate		190 (80%)	94 (80%)	96 (79%)	0.951
Older men more likely to get PCa		158 (66%)	80 (68%)	78 (64%)	0.586
African Americans less likely to get PCa		180 (75%)	86 (73%)	94 (78%)	0.389
PCa always has symptoms		116 (49%)	49 (42%)	67 (55%)	0.032
Important to talk to Dr about screening		230 (96%)	111 (94%)	119 (98%)	0.099
Screening can include DRE and PSA		221 (94%)	107 (92%)	114 (95%)	0.385
Medical experts agree on screening		25 (11%)	105 (11%)	107 (10%)	0.815
PCa easier to treat if detected early		223 (93%)	108 (92%)	115 (95%)	0.277
All PCa grows at the same rate		199 (83%)	96 (81%)	103 (85%)	0.435
If no family history, not at risk		196 (82%)	97 (82%)	99 (81%)	0.938
Knowledge SCORE (range)		80 (40–100)	70 (40–100)	80 (40–100)	0.204

Figures 1, 2, and 3 show box plots characterizing the distributions of the PCa outcomes data on knowledge, concern, and likelihood of screening in the next 6 months, respectively, between the control and intervention groups over the course of the study. The impact of the health interventions, when measured as estimated mean changes from baseline, revealed key similarities and differences between the two groups (Table 3). Knowledge of PCa increased for both the control and intervention groups and was statistically significant for both groups at the 4-month follow-up (p≤0.01). Likewise, changes in plans for PCa screening were statistically significant for both groups immediately following the educational session (p≤0.01). The magnitude and significance of this effect was sustained at the 1- and 4-month follow-ups. In contrast, the changes in the level of PCa concern was only significant for the intervention group immediately post-session and at 1-month follow-up (p=0.04 and p=0.01, respectively). This effect appeared to diminish over time as the level of PCa concern was no longer significantly different from baseline at the 4-month follow-up (p=0.40). Differences in mean scores between the two study groups were fairly unremarkable (Table 4). Differences in PCa knowledge were similar between groups, though, at 1-month follow-up, the intervention group had significantly higher knowledge than controls (difference in means = 0.53, 95% CI: 0.10, 0.97, p=0.02). Similar to the knowledge scores, the intervention group had significantly higher concern for PCa, but this too was observed only for the 1-month follow-up, (difference in means = 0.41, 95% CI: 0.03, 0.80, p=0.04). No significant differences were found between groups for their likelihood to get PCa screening within the next 6 months.

Figure 2.

Box plot of levels of concern for getting PCa by study group over time.

Figure 3.

Box plot of intention for PCa screening in the next 6 months responses by study group over time.

Table 3.

Estimated mean changes from baseline assessment in knowledge scores, concerns, and likelihood of screening by study group over time.^*

	Estimated Changes from Baseline: Mean (95% CI), p-value
	Immediately Post-Session	At 1-month Follow-up	At 4-month Follow-up
Knowledge
Control	0.29 (−0.02, 0.61), p=0.07	0.26 (−0.06, 0.57), p=0.11	0.36 (0.08, 0.64), p=0.01
Intervention	0.39 (0.08, 0.69), p=0.01	0.73 (0.43, 1.04), p<.01	0.64 (0.36, 0.92), p<.01
Concern
Control	0.22 (−0.05, 0.49), p=0.11	−0.11 (−0.37, 0.15), p=0.40	0.21 (−0.03, 0.45), p=0.08
Intervention	0.26 (0.01, 0.51), p=0.04	0.32 (0.07, 0.57), p=0.01	0.10 (−0.13, 0.33), p=0.40
Intent to Screen
Control	0.44 (0.10, 0.78), p=0.01	0.47 (0.11, 0.84), p=0.01	0.78 (0.40, 1.16), p<.01
Intervention	0.64 (0.31, 0.97), p=0.01	0.54 (0.11, 0.97), p=0.01	0.91 (0.49, 1.32), p<.01

^*based on longitudinal mixed effects regression modeling adjusted for age, BMI, smoking, marital status, and education.

Table 4.

Differences in mean knowledge scores, concerns, and likelihood of screening between study groups over time.^*

	Estimated Group Mean Differences: Mean (95% CI), p-value
Outcomes (range of scores)	Baseline	Post-Session	1-month Follow-up	4-month Follow-up
Knowledge (0–7 points)	0.06 (−0.35, 0.47), p=0.78	0.15 (−0.26, 0.56), p=0.47	0.53 (0.10, 0.97), p=0.02	0.34 (−0.11, 0.79), p=0.14
Concern (1–4 points)	−0.02 (−0.39, 0.35), p=0.92	0.02 (−0.36, 0.40), p=0.92	0.41 (0.03, 0.80), p=0.04	−0.13 (−0.52, 0.27), p=0.53
Intent to Screen (1–5 points)	0.14 (−0.34, 0.63), p=0.57	0.34 (−0.13, 0.82), p=0.15	0.21 (−0.36, 0.77), p=0.47	0.27 (−0.31, 0.85), p=0.37

^*the difference was computed as intervention mean minus control mean and based on longitudinal mixed effects regression modeling adjusted for age, BMI, smoking, marital status, and education.

Discussion

The goal of this PCa intervention study was to decrease concern, increase PCa knowledge, and increase intent to screen for PCa among high risk populations in Philadelphia, PA. At baseline, we observed no differences in the outcomes between the PCa-targeted intervention and healthy lifestyle control groups. At 1-month post-educational session, levels of PCa knowledge and concern were significantly higher for the intervention group, but that difference dissipated by the 4-month point. Interestingly, tracking within group changes over time, we observed that knowledge and concern generally increased for those in the intervention group. Intent to screen increased significantly over time for both the intervention and control groups.

Ten years ago, the USPSTF first advised against PSA testing for all men in the general population. However, other scientific organizations were not in total agreement with the recommendation and suggested that patients and doctors engage in shared decision making before PSA screening. [6] In a survey of > 1000 men shortly after the USPSTF announcement, 62% agreed with the original recommendation. However, only 13% intended to follow the recommendation and not be screened for PCa in the future. More than half of those surveyed planned to not follow the recommendation. Another 33% were uncertain about their personal future screening. Men who disagreed with the recommendation were more likely to be older and to experience more worry about PCa. [7]

More recently, the USPSTF has suggested that PCa screening for men ages 55–69 should be decided by the patient in consultation with his physician. That consultation should include a discussion of the potential harms and benefits of PCa screening. Additionally, physicians are advised not to screen men against their preference. However, patients need accessible and adequate health education related to early detection and diagnosis of PCa to assist in the decision-making process, as in some situations the patient with an interest may have to initiate the discussion. It is also notable that there are no definitive recommendations for screening of African American men, one of the most high risk populations for prostate cancer development and related mortality. [8]

A cross-sectional study found that among US-born and Caribbean-born men, men of African descent had higher levels of fear and worry about PCa compared to White men. The highest fear of screening scores were observed among US-born African American men. Caribbean-born men of African descent had highest levels of worry about PCa. [9] Once men received one PCa screening (Digital Rectal Exam), they were as likely to have annual screening as those with low fear of PCa. [9] Having a personal physician, annual health exams, and family history of PCa increase the likelihood of receiving PCa screening. High perceived level of health care system barriers may reduce the likelihood of PCa screening. [9]

Previous studies have shown that educational interventions have been successful in promoting prostate cancer awareness and healthcare. [10] Interventions have included different vehicles (educational pamphlets, in-person presentations) and different venues (community settings, barbershops, work places, health fairs, and places of worship.) [10, 11] Community-based educational interventions have been shown to increase cancer knowledge and decisional self-efficacy (important components for engaging in shared decision making) and intention to screen. Such interventions also decrease perceived barriers to screening, suggesting that community educators can effectively address concerns related to logistics and costs of screening.[11] Culturally tailored decision support instruments for African American men are effective tools for increasing patient’s certainty during the decision-making process. [8] More than 80% of African American men surveyed in one community-based educational intervention study indicated that they planned to talk to a physician about screening, despite exit interviews that suggested that some men feel devalued and racially discriminated against in the healthcare system. [8]

The strengths of this study include the focus on at-risk men in pre-determined high-risk areas of the city. We focused on an age group (40–69 years) for which screening guidelines are most relevant. The literature suggests that, this age group might be most in need of prostate health information, because they may not understand that they are susceptible to prostate cancer. [5, 6] We also determined that the demographics of our final sample were similar to those of the local population. Census data showed that the 4 neighborhoods that we selected were 86–95% African American, 11–22% of residents had less than a high school diploma, and 15–24% were unemployed. These findings suggest we may have drawn a representative sample. Our research team developed a culturally-tailored intervention (features and content) with input from focus groups (conducted in the areas of interest), PCa survivors (project consultants), and Peer Educators (African American men that resided in the targeted neighborhoods.) We were able to follow most participants over 4 months to observe how well information was retained and how perspectives on PCa changed over time. Additionally, the intervention and healthy lifestyle educational sessions were delivered by peer educators in sites located in targeted neighborhoods. Based on our focus groups, we believe that African American men respond to educational information best when they are in comfortable settings with educators who they identify with and who identify with them.

The limitations of this study include that some of the participants were lost to follow-up, so we were unable to contact them by phone to follow them to study completion. Although the sample appears representative of the general population from these neighborhoods, the participants still could represent a select group of men that were willing to participate in educational health sessions. As a result, we may have a sample of men that is more motivated to pursue PCa health and screening than some other men in the community. Though the study demonstrated sufficient statistical power to detect even modest improvements in outcomes measures, the generalizability of the findings is uncertain since sampling was limited to four Philadelphia communities.

In conclusion, this study showed that gathering at-risk men for discussions about PCa or other health concerns may increase their willingness to talk to a doctor about PCa screening. The next steps is to replicate our findings. We are currently planning a similar study to engage African American men in PCa-specific discussions to determine if our methods and materials increase PCa awareness and intent to screen among targeted high risk populations. Future multi-institutional studies may link these participants to their electronic health records to examine how the targeted intervention may impact clinical outcomes.

Availability of data and material:

Interested parties may contact the authors for access to anonymized data and study materials.