Do Community Health Worker Interventions Improve Rates of Screening Mammography in the United States? A Systematic Review

Abstract

INTRODUCTION

In 2010, an estimated 207,090 women were diagnosed with breast cancer (BC) in the United States.(1) The U.S. Preventive Services Task Force recommends screening mammography every 2 years for women age 50 to 74 years, and recommends screening for women age 40 to 49 years be based on an individual’s risk factors.(2) The American Cancer Society recommends yearly mammography beginning at age 40.(3) Early detection of BC is associated with reductions in mortality and improvements in survival rates.(4, 5)

There are significant racial and socioeconomic disparities in BC mortality, survival rates, and cancer stage at diagnosis in the United States.(6-10) Women who are less likely to adhere to screening mammography guidelines include those who are: ethnic or racial minorities; lacking comprehensive health insurance or a usual source of medical care; non-English speakers; immigrants; living in rural areas; or socioeconomically disadvantaged.(11-19)

One model of BC screening promotion that has been implemented and evaluated frequently is the community health worker (CHW) model. CHWs are lay individuals trained to serve as liaisons between members of their communities and health care providers and services.(20) Historically, CHWs serve low-income, medically underserved, racial/ethnic minority, and hard-to-reach populations.(21)

Previous systematic reviews on the effectiveness of CHWs on increasing screening mammography have several limitations. These reviews combined mammography with other health behaviors,(22) combined CHW interventions with other mammography-enhancing intervention strategies,(23-25) and combined between-group studies (comparing CHW interventions to comparison group) and within-group studies (comparing mammography rates over time in a CHW intervention group),(26) making it difficult to draw conclusions about effectiveness of CHW interventions in improving screening mammography. To conclusively assess effect of CHW interventions in increasing screening mammography rates, we performed a systematic review. The objective was to synthesize and critically appraise available evidence on effectiveness of CHW interventions in increasing screening mammography rates compared to a control group in any population.

MATERIALS AND METHODS

A study protocol was developed clearly outlining a priori all stages of the systematic review process.

RESULTS

Study Identification

The initial database search yielded 265 articles (Figure 1), of which 24 met inclusion criteria.(41-64) Of the 24 articles included in the systematic review (Table 1), 18 (75%)(41-47, 50, 52, 55-63) had extractable data. Three articles (41, 42, 62) provided data enabling calculation of multiple effect sizes. One study provided data on multiple interventions,(42) one study reported results in two different strata based on age,(41) and one study reported results separately by race and ethnicity.(62) Of the 24 included studies (Table 1), 14 were (58%) RCTs(41, 42, 44, 45, 50, 51, 53, 54, 56, 58, 60, 61, 63, 64) and 10 were quasi-experimental studies.(43, 46-49, 52, 55, 57, 59, 62)

Open in a separate window

Figure 1

Flow Diagram of Literature Review Process for Identifying Studies Evaluating Use of Community Health Workers to Improve Screening Mammography Rates.

Table 1

Summary of Community Health Worker Interventions for Breast Cancer Screening by Ethnic/Target Group

First Author/Year	Target Population	CHW Number & Characteristics	Length; Type of Training	Intervention Description	Control Description
African American
Earp,2002 46	Rural, African American women 50+ years	-170 volunteer lay health advisors -4 paid community outreach specialists (indigenous community leaders)	-3–5 sessions, 10-12 hours instruction; role playing, BC and screening practices.	-2 community outreach activities per month -Lay health advisors conducted 2 weekly individual sessions	UC in comparison counties
Erwin, 1999 47	Rural, African American women, Mississippi River Delta, Arkansas.	7 African American BC survivors	Training provided; interviewing strategies, breast health education	-Church-based group about early detection and screening -Free mammogram vouchers	UC in comparison counties
Paskett, 1999 55	Predominately African- American women 40+ years, low-income housing, Winston-Salem, North Carolina (intervention) and Greensboro, North Carolina (control)	ND	ND	-Educational sessions by lay health educators -Church program conducted by lay health educator	UC in comparison city
Sung,1997 60	Low-income, inner city African American women	Lay health workers recruited from National Black Women’s Health Project	10 weeks; interviewing and health education	-Two home visits, 1 booster session. -Education about breast and cervical cancer, Pap test and breast exam video -Reproductive health education	Cancer screening educational materials
West,2004 63	Rural, low-income African American women 50-80	Indigenous African American female health care workers.	Training provided; semi-structured interview’s content, counseling style	-Personalized reminder letters followed by intensive phone counseling -Phone call included semi-structured interview eliciting barriers, facilitators to mammography -Problem solving to overcome barriers -Provided phone numbers for scheduling mammography	Eligible for a no- cost mammogram; tailored letter describing risk for BC
Zhu,2002 64	Single African American women 65+, living in public housing	African American women from same housing complex	-Four 3-hour sessions; BC, screening practices, possible barriers	-Home-based BC screening education -Taught participants to overcome barriers -Brochures	Public housing complexes, no intervention
Hispanic/Latina
Fernandez, 2005 48	Hispanic farm workers	Lay health workers	Lay health workers given training curriculum including teaching guide (12 lessons)	-Individual and group intervention sessions with lay health worker -Education sessions included videos, flip charts, discussions of breast and cervical screening	Comparison clinics, no intervention
Navarro, 1995 53	Low-income Latinas, San Diego, California	36 consejeras	Training provided; conducting small group sessions on health topics	-12 week, 90 minute group session program -Consejeras taught cancer prevention with culturally-appropriate materials.	12-week Community Living Skills program
Navarro, 1998 54	Latinas, San Diego, California	36 consejeras	Training provided; following consejera manual to conduct weekly educational group sessions	12 week, group sessions on breast and cervical cancer early detection	12-week Community Living Skills program
Navarro, 2000 51	Latinas, San Diego, California	36 consejeras	Training provided; conducting small group sessions on breast and cervical cancer screening	12 week group cancer screening intervention following educational materials	12-week Community Living Skills program
Navarro, 2007 52	Low-income, low acculturated Latinas, San Diego, California	17 consejeras	Five 2-hour sessions; 14 program sessions, recruitment strategies, role playing to lead sessions.	12 90 minute weekly group sessions and 2 monthly sessions focusing on breast and cervical cancer	Friends and/or family (learning partners) who received information from class participants
Sauaia, 2007 57	Latinas, Colorado	“Peer counselors”, 4 Catholic churches, Denver, Colorado (Promotoras)	Training provided	-Bimonthly meetings after mass and other church events -Respected leader delivered homilies addressing breast health at least twice at each church -1 to 3 home-based health groups per church -Newsletter	209 Catholic churches sent: --Letter describing project -Bilingual printed materials - display unit -short messages for delivery at pulpit and church bulletin
Suarez, 1997 59	Low-income Mexican American women 40+, Texas	Mexican-American volunteers, Spanish speakers	ND	- Verbal and newsletter cancer- screening education	Comparison community, no intervention
Welsh, 2005 62	Latinas, Colorado	“Peer counselors”, 4 Catholic churches, Denver, Colorado	Training provided; standardized curriculum	-Monthly visits to each church -Promotoras approached peers after Sunday masses and during church fairs, other church related activities -Promotoras facilitated home-based platicas (health groups) about breast health	209 churches mailed: -introductory letter -NCI BC screening educational materials -display unit -messages to be delivered at pulpit or church bulletins
Asian American/Pacific Islander
Bird,1998 43	Vietnamese-American women 18+, California	-16 indigenous lay health worker neighborhood leaders -68 neighborhood assistants	-Training provided; delivery of prevention education -Provided training manual	-Small-group home-based educational sessions - Focused on general prevention, routine checkups, breast and cervical cancer screening -Flip chart and facilitated discussion	Comparison city, no intervention
Gotay, 2000 49	Native Hawaiian women, 18+, Oahu, Hawaii	Native-Hawaiian lay health educators	-Training provided; breast, cervical cancer screening.	-Small-group, traditional Hawaiian “talk-story” methods -Free mammogram and Pap test vouchers -Audiovisual aids used	Comparison area, no intervention
Other Populations
Allen, 2001 41	Women who work in sites with Service Employees International Union representation	80 female employee peer health advisors	-16 hours; cancer screening guidelines, epidemiology, early detection methods, community resources	- 16-month program, 6 small group discussions, role modeling individual outreach, counseling, social support focused on screening. -2 worksite-wide campaigns	Workshop at study conclusion
Andersen, 2000 42	Women 50-80, rural Washington	Women from participating communities	Training provided	-Telephone intervention using barrier- specific counseling to promote mammography - Brochures	Control communities, no intervention
Calle,1994 44	African American and White women 40+	80 ACS volunteer peer educators, Jacksonville, Orlando, Florida	-Half-day; mammography and mammography facilities, breast health guidelines, intervention process, intervention practice sessions, BC fact sheet, resource guide for mammography centers	-Phone intervention emphasized importance of regular mammograms -Set and confirmed date appointment would be scheduled and completed.	No intervention
Duan, 2000 45	African-American, Latina, and White women 50-80, churches in Los Angeles County, California	Part-time peer counselors	ND	-Peer counselors provided telephone barrier-focused mammography counseling annually for 2 years - Women provided information about risk status, BC prevalence rates, encouraged to ask physician for referral	Churches, no intervention
Margolis, 1998 50	African American, Native American, and White women 40+, non- primary-care outpatient clinics	Low-income, lay female senior aides (lay health advisers)	-One month; peer education techniques, breast health, communication skills, cultural diversity, principles of randomization, adherence to research protocol	-Appointments scheduled with female nurse practitioner for those due for screening -Women who declined screening encouraged to follow up with health care provider -Women who were up-to-date were encouraged to get regular screening and offered mailed reminder	UC
Paskett, 2006 56	African-American, Native American, and White women from rural areas and low-income background.	2 Native American and African American lay health advisors	1 week plus follow-up sessions; breast health, practice on breast models, resource manual, practice intervention sessions	- 3 in-person individual visits, follow- up phone calls, mailings over 9-12 months. -Educational materials about cancer risk, overcoming barriers to mammography -Discussion of mammography, BC, breast self-exam, scheduling mammography -2 postcard mailings addressing women’s readiness to change	Letter and NCI brochure about cervical cancer screening; after follow up, letter and NCI brochure about mammography
Slater, 1998 58	Low-income women, public housing	ACS and resident volunteers	-Training provided; volunteers provided intervention scripts and protocols.	60 minute “Friend to Friend” program with health professional speaker, small group discussions, assistance in obtaining mammogram, mammogram reminder -Free mammograms -Assistance with appointment scheduling -Free transportation	Delayed intervention
Weber, 1997 61	African-American, Latina, and White low- income, urban women 52-77, 6 primary care facilities Rochester, New York	6 women from local community (4 African American, 1 Hispanic, 1 White)	ND	-1 personalized mammography reminder letter from physician -1 low literacy mammography reminder letter from community health educator -Patient education and reminders -Identification and removal of barriers	1 personalized mammography reminder letter from physician plus usual care

Open in a separate window

ND=not described; BC=breast cancer; UC=usual care

Outcomes of CHW Intervention on Receipt of Screening Mammography

All Trials

Most studies (75%) collected data on screening mammography use via self-report interviews or surveys.(41-47, 49, 51-55, 58-60, 63, 64) Other studies collected mammography data through chart reviews,(56, 61) health insurance claims data,(57, 62) or through a combination of self report and chart reviews.(50) The most common time frame in which screening mammography was collected was one year(45, 49-51, 54-56, 64) and two years,(41, 46, 49, 50, 55, 57, 59, 61, 62, 64) with two studies reporting time frames in between one and two years.(43, 58) Several studies evaluated the CHW intervention’s effect on lifetime receipt of screening mammography.(43, 47, 49, 52, 53)

Of 18 studies with sufficient data to evaluate receipt of screening mammography in the meta-analysis, 10 (56%) were RCTs,(41, 42, 44, 45, 50, 56, 58, 60, 61, 63) and 8 were quasi-experimental.(43, 46, 47, 52, 55, 59, 62) The pooled RR of obtaining screening mammography based on 28,836 mammography events (9,342 intervention;19,494 control) was 1.06 (95% CI:1.02,1.11,p=0.003), indicating a statistically significant effect of CHW interventions on improving rate of screening mammography (Figure 2). However, there was a statistically significant heterogeneity among included studies (I²=80%;p<0.00001).

Open in a separate window

Figure 2

Forest plot of comparison: 1 Intervention versus no Intervention, outcome: 1.1 Receipt of mammography.

Quasi-Experimental Studies

The pooled RR of obtaining screening mammography based on 14,677 mammography events (2,235 intervention; 12,442 control) in 8 quasi-experimental studies (9 comparisons)(43, 46, 47, 52, 55, 57, 59, 62) was 1.03 (95% CI:0.89,1.18,p=0.71, Figure 2), indicating no effect of CHW interventions on rate of screening mammography. There was a statistically significant heterogeneity among included studies (I²= 84%;p<0.00001).

Randomized Controlled Trials

The pooled RR of obtaining screening mammography based on 14,159 mammography events (7,107 intervention;7,052 control) in 10 RCTs (14 comparisons)(41, 42, 44, 45, 50, 56, 58, 60, 61, 63) was 1.07 (95% CI:1.03,1.12,p=0.0005, Figure 2) indicating a statistically significant effect of CHW interventions on improving rate of screening mammography. There was a statistically significant heterogeneity among included trials (I²=78%;p<0.00001).

Sensitivity Analyses

To assess robustness of our findings and account for observed heterogeneity among included studies, we performed additional analyses. Results from quasi-experimental studies varied both in magnitude and direction of effect. Therefore, quasi-experimental studies were excluded from further sensitivity analysis. Sensitivity analysis focused on the following factors that may be associated with success of CHW interventions in the 10 RCTs: methodological quality (cluster versus individual randomization, intent-to-treat analysis), choice of control intervention, method of measuring study outcome (self-report versus chart review), setting of participant recruitment (medical versus community and rural versus urban), number of CHW intervention components, and characteristics of CHWs.

Methodological Quality

A critical appraisal of methodological quality of all studies was conducted, including information about study design and analyses (Table 2).

Table 2

Randomized Controlled Trial Characteristics and Study Quality

First Author, Year	Number of Participa nts, Baseline	Mean Age CHW Intervention Participants (years)	Mean Age Control Particip ants (years)	Length of Interventi on (months)	Time from Baseline to Follow Up (months)	Generation of Randomization Sequence	Power Analysis	Described Drop Outs	Participants Selected Due to Lack of Adherence	Definition and Measurement of Mammography Outcome	Intent to Treat Analysis Conducted
Allen 2001 41	2943	ND	ND	16	30	Work sites blocked on size of worksite and type of agency to create 4 blocks. Within each block random-number generator assigned work sites to each condition.	ND	Yes	No	Past 12-24 months;self- report	No
Anderson 2000 42	8907	ND	ND	24	36	40 communities assigned to 10 blocks of 4 communities for block randomization. Each community within one of ten blocks randomly assigned to a study arm using one of 24 permutation patterns of 4 for each block.	ND	Yes	No	Within 24 months of follow up interview; self- report	No
Calle 1994 44	738	ND	ND	6	8	American Cancer Society volunteers each generated list of 10 women. Lists collected and women randomized to either intervention or control groups.	ND	Yes	No	Since start of intervention; self-report	No
Duan 2000 45	1113	ND	ND	24	12	30 churches matched to 12 blocks and randomized half the churches in each block to peer counseling and other half to control. Matching variables included race/ethnicity, membership size, resources, and denomination.	ND	Yes	No	Past 12 months;self- report	No
Margolis 1998 50	1658	54.5	55.9	1 day plus mailed reminders	12+	Patients assigned to intervention or usual care according to whether medical record number was odd or even	ND	Yes	No	12 or 24 months (Depending on age);self-report plus chart review	Yes
Paskett 2006 56	897	54.5	55.7	12-14	12-14	Stratified selection, blocked randomization by race and clinic, performed by study statistician	Yes	Yes	Yes	Past 12 months; chart review	No
Slater 1998 58	427	68.9	67.4	24	12	41 public housing high rise buildings randomized to treatment or control groups. Buildings stratified based on number of women per building who were 40 years and older.	ND	Yes	No	Within 15 months of intervention; self-report	No
Sung 1997 60	321	ND	ND	11	17	Participants randomly assigned to intervention or control groups	ND	Yes	No	Depended on age (within 4 years if 35-39; within three years if 40-49; yearly if older than 49); self- report	No
Weber 1997 61	376	63	63	4	7	Patients randomized to one of two groups. Randomization stratified by practice site.	Yes	Yes	Yes	During intervention period; chart review	Yes
West 2004 63	320	ND	ND	1 day	6	Envelope containing randomization group opened by CHW after completing telephone interview.	ND	Yes	Yes	Past 6 months; self-report	Yes

Open in a separate window

ND =Not described

Unit of Randomization

Six RCTs (60%) randomized individual participants to either CHW intervention or comparison group(44, 50, 56, 60, 61, 63) whereas in the other 4, unit of randomization was a cluster or group to which participants belonged, such as a work site or church.(41, 42, 45, 58) None of the CRTs reported power calculations or whether they were adjusted for clustering effects. Three trials(41, 42, 45) incorporated clustering effects in the main statistical analysis, but did not report estimated ICC values. The fourth trial(58) reported the value of ICC (−0.0015), which was appropriately assumed to indicate no design effect for the primary outcome measure, and data were analyzed at individual rather than cluster level.

In RCTs that randomized individual participants, CHW intervention was associated with a statistically significant increase in screening mammography (RR:1.39,95% CI:1.13, 1.70,p=0.002). In RCTs that randomized groups to treatment condition, CHW interventions were not associated with increases in screening mammography (RR:1.02, 95% CI:1.00,1.04,p=0.05). There was statistically significant heterogeneity among included trials that used individuals as unit of randomization (I²=80%; p=0.0002). RCTs that used groups as unit of randomization did not demonstrate heterogeneity (I²=27%;p=0.22). Compared with unadjusted estimates, OR estimates adjusted for ICC and imbalance were statistically significant, indicating that CHWs improve rates of screening mammography. However, adjusted ORs were closer to the point of no effect (Figure 3).

Open in a separate window

Figure 3

Effect of Intra-Class Correlation Adjustment on Odds Ratio Estimates.

Intent-to-Treat Analysis

Pooled data from 3 RCTs which performed ITT analysis(50, 61, 63) indicate that a CHW intervention was not associated with increases in screening mammography (RR:1.48, 95% CI:0.85,2.59,p=0.17). In 7 studies (11 comparisons) that did not perform ITT analysis,(41, 42, 44, 45, 56, 58, 60) a CHW intervention was associated with increases in screening mammography (RR:1.06, 95% CI:1.02,1.10,p=0.004). Both studies that conducted ITT analysis (I²=82%;p=0.004) and studies that did not conduct ITT analysis (I²=77%;p<0.00001) demonstrated significant heterogeneity.

Choice of Control Intervention

RCTs comparing CHW interventions to routine care (7 trials involving 11 comparisons due to multiple intervention groups or different comparisons by age)(41, 42, 44, 45, 50, 58, 60) were associated with a statistically significant increase in screening mammography (RR:1.04,95% CI:1.01,1.07,p=0.007). The only RCT employing health education as control(56) compared with CHW also showed a statistically significant benefit with use of a CHW intervention (RR:1.56;95% CI:1.29,1.89;p<0.00001). The pooled RR from 2 RCTs that used mammography reminders as control(61, 63) showed a statistically non-significant difference compared with CHW (RR:1.82,95% CI:0.83,4.01,p=0.14). However, there was statistically significant heterogeneity among included trials that used routine care (I²=65%;p=0.002) or mammography reminders (I²=73%;p=0.05) as control.

Method of Measuring Study Outcome

There was significant variation in measurement of study outcome, receipt of mammography, which limited sensitivity analyses that could be conducted on length of time in which mammography was measured (Table 2). The interval of measurement reported in the RCTs ranged from receipt of mammography in the past 6 months to receipt of mammography in the past 3 years. Seven of 10 RCTs measured screening mammography outcome using participant self-report.(41, 42, 44, 45, 58, 60, 63) Two studies measured screening mammography with chart review,(56, 61) and one combined self-report and a review of patients’ medical records.(50) CHW interventions were effective at increasing screening mammography when mammography was measured by chart review alone (RR:1.93,95% CI:1.15,3.23,p=0.01) or by chart review plus self report (RR:1.10,95% CI:1.02,1.19,p=0.009). The pooled RR for the 7 RCTs (11 comparisons) which measured mammography using self report indicated CHW interventions were not associated with increases in screening mammography (RR:1.03,95% CI:1.00,1.06,p=0.02). There was statistically significant heterogeneity among included trials that used chart review alone (I²=73%;p=0.06) and self-report alone (I²=59%;p=0.007) to measure the mammography outcome.

Study Setting

Medical versus Community Setting

Five studies recruited participants from a medical setting,(50, 56, 60, 61, 63) and 4 (7 comparisons) recruited participants from a community setting (e.g., church, neighborhood).(42, 44, 45, 58) In one study, participants were recruited from a work setting(41) which was not coded as either a community or medical setting, and was therefore excluded from sensitivity analysis. CHWs were effective in increasing rates of screening mammography in RCTs recruiting participants from a medical setting (RR:1.41,95% CI:1.09,1.82,p=0.008) or community setting (RR:1.05,95% CI:1.01,1.10,p=0.02). There was statistically significant heterogeneity among included trials for both subgroups (medical setting:I²=81%;p=0.0003; community setting:I²=70%;p=0.003).

Urban versus Rural Setting

Six RCTs were conducted in an urban setting,(44, 45, 50, 58, 60, 61) 3 (7 comparisons) in a rural setting,(42, 56, 63) and one(41) (2 comparisons) did not report information on setting. Studies performed in an urban area showed (RR:1.23,95% CI:1.09,1.39,p=0.001) a statistically significant benefit associated with CHW compared with control. Pooled results for studies performed in a rural setting (RR:1.05,95% CI:1.00,1.11,p=0.06) or unknown setting (RR:1.01,95% CI:0.98,1.03,p=0.69) did not demonstrate a benefit associated with CHW compared with control. There was statistically significant heterogeneity among the subgroup of studies performed in an urban (I²=74%;p=0.002) or rural setting (I²=78%;p=0.0003), but not the two included comparisons for the study that did not report setting (I²=0%;p=0.75).

Number of Intervention Components

The CHWs were described using several different terms, such as indigenous community leaders,(46) consejeras,(51-54) promotoras,(57) lay health advisors,(46, 50, 56) lay health educators,(49, 55, 64) community health educators,(61) lay health workers,(43, 60) peer health advisors,(41) peer counselors,(45) community health care workers,(63) and volunteers.(42, 44, 58, 59) The most commonly reported components of CHW interventions included health education,(41-43, 45-49, 51-54, 56, 57, 59-64) referrals to health care,(42, 43, 45, 47, 48, 50, 58, 59, 63) appointment scheduling, (48, 50, 56, 61) and vouchers, free mammograms, or lower cost mammograms.(46-49, 63) Less common intervention components included emotional or social support,(41, 61) financial paperwork,(48, 61) communication with the health care team,(61) mailed reminders,(50, 61) and child care.(61) Sensitivity analysis showed a statistically significant increase in receipt of screening mammography associated with increase in number of interventions. RCTs where CHW interventions included 1 intervention component(44, 45, 60) (RR:1.23,95% CI:1.00,1.52,p=0.05) or 2 intervention components(41, 42, 56, 58) (RR:1.03,95% CI:1.00,1.07;p=0.06) were associated with a statistically non-significant increase in rates of screening mammography. Interventions that included 3 intervention components(50, 63) (RR:1.10,95% CI:1.03,1.19,p=0.008) or more than 3 intervention components (61) (RR:2.67,95% CI:1.59,4.48,p=0.0002) were associated with a statistically significant increase in rates of mammography. There was statistically significant heterogeneity among included RCTs in subgroups where CHW used 1 (I²=75%;p=0.02) or 2 intervention components (I²=74%;p=0.0004). There was a statistically non-significant heterogeneity among included RCTs in the subgroup with 3 CHW intervention components (I²=0%;p=0.81).

Racial or Ethnic Concordance of CHW and Target Population

The studies reviewed targeted a number of different populations. Ninety-two percent of the studies included ethnic or racial minority participants or focused on low-income populations. Pooled results from RCTs that reported matching CHW and target population by race or ethnicity(44, 56, 61, 63) (RR:1.58,95% CI:1.29,1.93,p<0.0001) and that did not report matching CHW and target population by race or ethnicity(41, 42, 45, 50, 58, 60) (RR:1.03,95% CI:1.01,1.05,p=0.02) showed a statistically significant improvement in adherence to screening mammography, however, the effect was stronger for the RCTs that matched CHWs to participants by race and ethnicity. There was not statistically significant heterogeneity among either subgroup of studies (concordant:I²=45%;p=0.14; not concordant:I²=45%;p=0.06).

DISCUSSION

The result from this meta-analysis of 18 studies enrolling a total of 26,660 participants indicates CHW interventions are associated with a statistically significant increase in rates of screening mammography, but tend to have stronger effects in specific settings and study designs and when participants and CHWs were similar ethnically or racially. When RCTs were compared to quasi-experimental studies, the significant increase in screening mammography rates due to CHW interventions was observed in RCTs, but not in quasi-experimental studies. In RCTs, all studies demonstrated either a neutral (intervention neither increased nor decreased mammography rates) or positive intervention effect (intervention increased mammography rates), indicating CHWs were more effective than control in improving screening mammography rates. In quasi-experimental studies, direction of intervention effects was both positive and negative. These findings point out to inherent biases associated with observational study designs. Overall, there was significant heterogeneity in studies, with varying populations, varying lengths of intervention, varying lengths of follow-up after intervention, and varying time frames for the mammography outcome. Imputing different values of ICC had minimal effect on the distribution of point estimates of ORs or credibility intervals. So, there was little impact of similarity of different clusters (e.g., clinics) as a result of using a CRT design on overall effectiveness of CHW interventions on screening mammography.

Overall, our findings are similar to results of previous studies.(21, 65) Results of the present study are also similar to findings of a recent systematic review (26) which found a subset of CHW interventions was associated with significantly greater screening mammography utilization rates when compared to controls or other interventions (mail, print, minimal CHW). Our study improves on this recent review by providing a meta-analysis of all studies evaluating impact of CHWs on screening mammography. In addition, through sensitivity analyses of data from RCTs, the present study provides information regarding specific situations where CHW interventions are most likely to be beneficial (interventions conducted in urban settings, recruitment of participants from medical settings, measuring mammography outcome using a chart review, CHW interventions with 3 or more components, and interventions where CHW and patients are racially or ethnically concordant). The findings of our study differ from those of Lewin,(22) who found no beneficial effect (RR:1.05, 95% CI: 0.99,1.12,p=0.10) of CHW interventions on screening mammography based on a meta-analysis of 4 RCTs conducted internationally. While our analysis includes 3 of the 4 studies reported by Lewin,(22) our analysis also includes twice as many RCTs as well as quasi-experimental studies. Thus, our results came from a larger pool of data that may not be as strongly influenced by results of one individual study.

A lack of information reported in articles did not allow for detailed sensitivity analysis of methodological study quality. Sensitivity analysis comparing unit of randomization indicates CRTs of CHW interventions did not lead to increased mammography rates, whereas studies that randomized individuals were associated with a significant increase in mammography. Additionally, there was no significant increase in mammography screening in 3 studies that utilized ITT analysis to evaluate a CHW intervention,(50, 61, 63) but there was a significant increase in studies without ITT analysis. This lack of intervention effect may be related to smaller pooled sample size of the 3 studies (n=2019) or may suggest possible biases favoring intervention effect in studies that used per-protocol analyses, rather than ITT.(66)

CHW interventions were associated with increases in rates of screening mammography in studies with routine care or health education, but not in studies with mammography reminders. Screening reminders may exert a strong effect on mammography rates, as noted by others.(67, 68) Most RCTs reviewed included mammography referrals, reminders, or appointment scheduling as one CHW intervention component.(42, 44, 45, 50, 56, 58, 61, 63) This finding points to the need to determine components of CHW interventions most strongly associated with increases in screening mammography in order to improve efficiency of CHW programs.

Both community and medical settings of recruitment were associated with increases in screening mammography related to a CHW intervention, but the effect was stronger in participants recruited from a medical setting. Connecting patients to a medical setting is important in designing a CHW program as establishing a usual source of care is a known predictor of mammography screening.(16, 69) In addition, CHW interventions that took place in an urban setting were associated with screening mammography increases, whereas CHW interventions conducted in rural settings were not. These differences may reflect difficulties in obtaining a mammogram in a rural area.(19)

RCTs that delivered at least 3 types of interventions were associated with increases in screening mammography. These findings concur with previous findings(24) indicating interventions with multiple intervention components were associated with stronger increases in screening mammography rates. While the studies reviewed tended to focus or include historically or medically underserved populations, the RCTs that reported concordance between participants and CHWs on race or ethnicity indicated a stronger CHW intervention effect on screening mammography than studies where race and ethnic concordance was not described or performed. In contrast, a previous systematic review found concordance between physicians’ and patients’ race or ethnicity was not associated with improved health outcomes or patients’ utilization of health care.(70) Sensitivity analyses could not be conducted to evaluate differences in effectiveness of CHW interventions by different population groups because the majority of the RCTs targeted multiple underserved populations. Thus, there was no way to categorize them in a meaningful way for comparisons.

There are several limitations to this systematic review and meta-analysis. The study may be biased as it included only published peer-reviewed articles. Although unpublished works were eligible for inclusion in the review, none met inclusion criteria. This review was limited to studies published in English and implemented in the United States. Thus, meta-analysis results may not generalize to countries providing better or worse access to mammography screening services. In addition, sensitivity analyses were based on a small number of RCTs and should be interpreted with caution. The systematic review is also limited by reporting of data in original articles. Several sensitivity analyses could not be conducted due to significant variability in populations, interventions, and study designs reported. This variability is expected in the context of conducting research on interventions designed to meet local needs of various populations and under different breast screening guidelines in place throughout the years, but variability makes it difficult to combine data into meaningful categories.

The findings have significant implications for public health practice by indicating CHW interventions are associated with improvements in rate of one time screening mammography, especially in medical settings, urban settings, and in participants who are racially or ethnically concordant with the CHW. However, as mammography is a behavior that must be repeated multiple times, there is still much that is unknown regarding efficacy of CHW interventions in increasing repeated BC screening. A recent systematic review and meta-analysis of the effectiveness of various types of interventions designed to promote repeat BC screening found studies that utilized screening reminders only and studies that used more intensive interventions (including CHWs) were both associated with increases in repeat mammography.(68) Future research should evaluate whether CHW interventions are associated with repeat mammography screening and initiation of mammography screening. Future systematic reviews are necessary to compare CHW interventions to other mammography promoting interventions, such as media interventions.(71) In addition, future systematic reviews should evaluate whether CHW interventions are associated with other cancer screening behaviors. Finally, future research should be conducted to definitively determine which participants and populations benefit most from CHW interventions and why.

ACKNOWLEDGEMENTS

Footnotes

REFERENCES