2. 98 Sabatino et al / Am J Prev Med 2012;43(1):97–118
Context ● reducing client out-of-pocket costs and reducing struc-
tural barriers (enhancing access);
C
ancer is the second-leading cause of death in the
● provider reminders, provider assessment and feed-
U.S.1 According to U.S. Cancer Statistics,2 more
back, and provider incentives (increasing provider
than 560,000 people died from cancer in 2007.
delivery).
Screening reduces cancer mortality, and in some cases,
incidence from breast, cervical, and colorectal cancers.3–5 Findings from these reviews led to Task Force recom-
The U.S. Preventive Services Task Force recommends mendations for seven interventions to increase use of one
age-appropriate screening for breast cancer with mam- or more of these recommended cancer screening tests.
mography; cervical cancer with Pap tests; and colorectal There was insuffıcient evidence to determine the effec-
cancers with fecal occult blood test (FOBT), flexible sig- tiveness for remaining intervention categories.17–20
moidoscopy, or colonoscopy.3–5 Given the number of intervention categories for which
Although screening use has improved over time for effectiveness was not established for one or more cancer
several screening tests,6 – 8 rates are still suboptimal. This screening sites, the relative lack of evidence across re-
is particularly true for colorectal cancer screening; ap- views for colorectal cancer screening, and the particular
proximately 35%–50% of the population has not been need to increase uptake of colorectal cancer screening in
screened at recommended intervals.8 –11 For breast can- appropriate populations, the Community Guide team,
cer screening, 25%–30% of age-eligible women report not Task Force, and the Cancer Prevention and Control
having had recent mammograms6,11; for cervical cancer Research Network (CPCRN) sought to update these
screening, approximately 20% of women aged 18 – 44 systematic reviews. This article presents results from
years have not had Pap tests within the prior 3 years.6,11 the updated reviews of effectiveness for group educa-
Rates of regular screening use are even lower,12,13 and tion, one-on-one education, client incentive, client re-
screening rates have not risen in recent years.14 Further, minder, mass media, reducing out-of-pocket costs, re-
for many cancers, there are disparities in screening use ducing structural barriers, provider assessment and
for underserved groups, such as those with low income, feedback, and provider incentive interventions to in-
no insurance, or no usual source of care.8,11,14 –16 Inter- crease screening for breast, cervical, and colorectal
ventions to increase appropriate screening use can help cancers. Summary Task Force fındings from the origi-
achieve national screening objectives (www.healthy- nal reviews and from these updates are presented in
people.gov/2020) and save lives, and may reduce dispar- Table 1.
ities in screening. These updated reviews sought to address three
The Guide to Community Preventive Services (Commu- questions:
nity Guide), under the guidance of the independent, non- ● whether interventions for which there was insuffıcient
federal Community Preventive Services Task Force (the evidence to determine effectiveness in the previous
Task Force), previously conducted systematic re- reviews17–20 now had suffıcient evidence to determine
views17–20 on effectiveness of interventions to increase effectiveness;
screening for breast, cervical, and colorectal cancers. Ev- ● whether additional evidence would lead to a change in
idence for reviews was based on studies published be- fındings for interventions found to have suffıcient or
tween 1966 and 2004, and provided the basis for Task strong evidence of effectiveness on previous review;
Force recommendations for intervention use. Interven- ● what important research gaps remain.
tions selected for these reviews were included in one of
An updated review for small media interventions is
three strategies conceptualized to increase screening: in-
underway. A review17 of provider reminders was pub-
creasing community demand for screening, reducing
lished recently.
barriers to access, and increasing screening service deliv-
ery by healthcare providers. The fırst two strategies in- Evidence Acquisition
cluded client-directed approaches; the third strategy in-
Methods for conducting the original Community Guide systematic
cluded provider-directed approaches to promote use of
reviews of interventions to increase breast, cervical, and colorectal
appropriate screening. cancer screening are described elsewhere.21 These methods were
Eleven intervention categories were defıned and followed for the current updates with the exception of adaptations
grouped within these three strategies, which are as described in this section.
follows: Analytic frameworks for the three primary strategies assessed
through updated reviews are shown in Figures 1–3. These frame-
● client reminders, client incentives, one-on-one educa- works are unchanged from those used in the original reviews with
tion, group education, mass media, and small media the exception that they were revised to incorporate healthcare
(increasing community demand); system factors. Updated reviews used the same primary strategies,
www.ajpmonline.org
3. Sabatino et al / Am J Prev Med 2012;43(1):97–118 99
a
Table 1. Original and updated Community Preventive Services Task Force findings for cancer screening interventions
Intervention Original review findingsb Updated review findings
INCREASING COMMUNITY DEMAND FOR SCREENING
Group education
Breast cancer screening Insufficient evidence Recommended: sufficient evidence
Cervical cancer screening Insufficient evidence Insufficient evidence
Colorectal cancer screening Insufficient evidence Insufficient evidence
One-on-one education
Breast cancer screening Recommended: strong evidence Recommended: strong evidence
Cervical cancer screening Recommended: strong evidence Recommended: strong evidence
Colorectal cancer screening Insufficient evidence Recommended: sufficient evidencec
Client remindersd
Breast cancer screening Recommended: strong evidence Recommended: strong evidence
Cervical cancer screening Recommended: strong evidence Recommended: strong evidence
c
Colorectal cancer screening Recommended: sufficient evidence Recommended: strong evidencec
Client incentives
Breast cancer screening Insufficient evidence Insufficient evidence
Cervical cancer screening Insufficient evidence Insufficient evidence
Colorectal cancer screening Insufficient evidence Insufficient evidence
Mass media
Breast cancer screening Insufficient evidence Insufficient evidence
Cervical cancer screening Insufficient evidence Insufficient evidence
Colorectal cancer screening Insufficient evidence Insufficient evidence
INCREASING COMMUNITY ACCESS TO SCREENING
Reducing structural barriers
Breast cancer screening Recommended: strong evidence Recommended: strong evidence
Cervical cancer screening Insufficient evidence Insufficient evidence
c
Colorectal cancer screening Recommended: strong evidence Recommended: strong evidencec
Reducing out-of-pocket costs
Breast cancer screening Recommended: sufficient evidence Recommended: sufficient evidence
Cervical cancer screening Insufficient evidence Insufficient evidence
Colorectal cancer screening Insufficient evidence Insufficient evidence
INCREASING PROVIDER DELIVERY OR PROMOTION OF SCREENING
Provider assessment and feedback Recommended: sufficient evidencec Recommended: sufficient evidencec
Provider incentives Insufficient evidence Insufficient evidence
a
Strength of evidence based on the number of available studies, the suitability of study design for evaluating effectiveness, the quality of
execution of studies, the consistency of results across studies, and the magnitude of effect.21,23
b
Findings published in Baron et al.18,19 and Sabatino et al.20
c
Insufficient evidence to determine effectiveness for colorectal cancer screening with tests other than FOBT
d
For client reminders, the original review was limited to studies with greatest design suitability (e.g., RCTs) because of the large number of such
studies identified. All update studies for client reminder interventions had greatest design suitability except for one. That study was not
included in the assessment of absolute change in screening use for cervical or colorectal cancer, and exclusion of that study did not change
overall conclusions for any of the three cancer screening sites.
July 2012
4. 100 Sabatino et al / Am J Prev Med 2012;43(1):97–118
Other positive or negative
effects on preventive care
and service Efficacy established
Reminders
Incentives Change Follow-up Decrease
Increase completed
Mass media Knowledge screening Diagnosis Morbidity
Small media Attitudes (early detection) Treatment
Intentions Mortality
One-on-one educ.
Group educ.
Healthcare
Intervention
system factors
Mediators or intermediate outcomes
Intermediate outcome measuring
intervention effectiveness
Ultimate/desired health outcomes
Figure 1. Analytic framework: client-directed interventions to increase community demand for cancer screening services
educ, education
intervention categories, and defınitions as the original reviews. because these interventions previously were determined to have
This report includes updates for nine of these reviews. strong evidence of effectiveness.
For interval updates, evidence from update studies also was
Reaffirmation Updates, Interval Updates, and Full compared with that from the original review. For these updates,
Updates studies were scored for quality of execution.21 For full updates,
Three types of update approaches were possible. The approach evidence from the update was combined with evidence from the
selected by the team depended on strength of evidence in the original review and synthesized using standard Community Guide
original review. Where evidence of effectiveness was strong or methods.21 In some instances where reaffırmation or interval up-
suffıcient (Table 1), the team pursued reaffırmation and interval dates were undertaken, evidence from both reviews was combined
updates, respectively. Where evidence was insuffıcient to deter- to address specifıc research questions of interest identifıed by the
mine effectiveness, full updates were undertaken. For reaffırmation team.
updates, evidence from studies identifıed during update was com- Updated reviews were based on evidence from literature pub-
pared with evidence from the original review for consistency. In the lished between January 2004 and October 2008. Although some
interest of effıciency, scoring studies for quality of execution, studies in the original review were published in 2004, the original
whereby the internal validity of included studies was assessed using reviews did not include the entire calendar year of 2004. Thus, 2004
a standardized Community Guide process,21 was not required, was included in the search strategy for updated reviews. Studies
Other positive or negative
effects on preventive care
services received
Efficacy established
Reduce barriers
Structural Increase access Increase completed Decrease
Follow-up
barriers Physical screening Diagnosis Morbidity
Out-of-pocket Economic (early detection) Treatment Mortality
costs
Change client intent
Attitude Intervention
Perception Other healthcare Mediators or intermediate outcomes
system factors Intermediate outcome measuring
intervention effectiveness
Ultimate/desired health outcomes
Figure 2. Analytic framework: client-directed interventions to increase community access to cancer screening services
www.ajpmonline.org
5. Sabatino et al / Am J Prev Med 2012;43(1):97–118 101
Other positive or negative Change client
effects on client behavior or Knowledge
preventive services received Attitudes
Intentions
Increase
Discussion of test
with clients Increase test Increase
Provider assessment Recommendation, Completed screening
and feedback offer, order (early detection)
Provider incentives
Provider reminders Change provider
Attitudes
Other healthcare
Intentions
system factors
Follow-up
Diagnosis
Intervention Treatment
(Efficacy established)
Mediators or intermediate outcomes
Intermediate outcome measuring
intervention effectiveness Decrease
Morbidity
Ultimate/desired health outcomes
Mortality
Figure 3. Analytic framework: provider-directed interventions to increase screening for breast, cervical, and colorectal
cancers
from 2004 included in the original reviews were excluded from the use in the intervention group attributable to the intervention and
body of evidence for updates. concurrent change in the comparison group. When this was not
The team searched fıve computerized databases for potentially possible, effectiveness was evaluated either by examining the dif-
eligible studies (MEDLINE; the Cumulative Index to Nursing and ference in post-intervention screening use between groups or
Allied Health database [CINAHL]; the Chronic Disease Prevention change from pre-intervention to post-intervention in the same
database [CDP, Cancer Prevention and Control subfıeld]; Psy- group, depending on the data available. Interpretation of fındings
cINFO; and the Cochrane Library databases). (Search terms are and conclusions followed Community Guide rules, with evidence
available at www.thecommunityguide.org/cancer/screening/ about intervention effectiveness categorized as strong evidence of
provider-oriented/supportingmaterials/SSclient_provider.html.) effectiveness, suffıcient evidence of effectiveness, or insuffıcient
The team also reviewed citations received from team members and evidence to determine effectiveness, based on number of available
reference lists from articles, as appropriate. Conference abstracts studies, suitability of study design for evaluating effectiveness,
were not included. quality of execution of studies, consistency of results across studies,
The search identifıed 18,906 citations for which titles and ab- and magnitude of effect.21,23
stracts were screened for potential relevance to the interventions Conclusions of insuffıcient evidence to determine effectiveness
and outcomes of interest. Full-text review was undertaken for 319 do not indicate that interventions are ineffective, but rather that
of these articles. As in the original reviews, studies had to (1) be a more information is needed to determine whether or not interven-
primary investigation of at least one of the defıned intervention tions are effective. The number of studies required to determine
categories; (2) be conducted in a country with a high-income effectiveness varied depending on the quality of execution and
economy21 to increase applicability to the U.S.22; (3) provide infor- design suitability of studies included.23 Information about popula-
mation on one or more cancer screening outcomes of interest tions and settings for which recommendations are relevant is pro-
(breast, cervical, and/or colorectal cancer screening); and (4) in- vided in the Applicability sections. According to Community Guide
clude a comparison group that either reflected screening use prior rules,21 where evidence of intervention effectiveness was suffıcient
to intervention implementation or a concurrent group unexposed or strong, information about effectiveness, applicability, additional
to the intervention category of interest. A total of 45 studies quali- benefıts and potential harms, barriers to implementation, and re-
fıed for these reviews. search gaps was summarized. Where evidence was insuffıcient to
Qualifying studies were abstracted independently by two ab- determine effectiveness, remaining questions about effectiveness
stractors using a standardized abstraction form. Following Com- were summarized.
munity Guide methods, information about study design suitability, For client reminders, the original review was limited to studies
quality of execution, sample, intervention and comparison groups, with greatest design suitability (e.g., RCTs), because of the large
outcomes, and effect was abstracted. When necessary, conflicts number of such studies identifıed. However, the updated reviews
were resolved by review by a third team member. Design suitability were expanded to include all designs to maximize the potential to
categories included greatest, moderate, and least suitable according address additional research questions, including examining incre-
to Community Guide rules.23 Quality of execution is used to assess mental effects of client reminder interventions when added to
biases and limitations in study execution. Quality was categorized other interventions. All update studies had greatest design suitabil-
as good, fair, or limited.23 Studies of limited quality were excluded ity except for one study. This study was not included in the assess-
from analyses, consistent with Community Guide rules. ment of absolute change in screening use for cervical or colorectal
Consistent with previous reviews, intervention effectiveness was cancer. Exclusion of this study did not change overall conclusions
evaluated by examining the difference between change in screening for any of the three cancer screening sites.
July 2012
6. 102 Sabatino et al / Am J Prev Med 2012;43(1):97–118
As in the original reviews of provider-directed interventions, (1.31, 95% CIϭ0.99, 1.74) levels. Group-level analysis
studies that reported only screening tests recommended or of- was adjusted for baseline measures and screening behav-
fered but not completed were not included in the determination iors. Individual-level analysis was adjusted additionally
of intervention effectiveness. These studies could be used to
for intervention group, level of acculturation, age, educa-
provide information about applicability, implementation, and
other effects. Also consistent with the original reviews,17,20 tion, and insurance status.
effectiveness of provider-directed interventions was deter- Cervical cancer screening promotion (full up-
mined by considering evidence across all three cancer screening
date). Five studies qualifıed for the combined body of
sites combined, as long as there were not differences in effec-
evidence; one study37 identifıed during update was ex-
tiveness by screening test. Additional information about Com-
munity Guide methods is available at www.thecommunityguide.org/ cluded because of limited quality of execution. Three
about/methods.html. qualifying studies26,28,32 had greatest design suitability
and fair quality of execution and two36,38 had least-
Evidence Synthesis suitable designs and fair quality of execution. Four stud-
ies26,28,36,38 examined self-reported Pap test use. For one
Increasing Community Demand for study,36 the type of education (interactive or didactic)
Screening: Group Education could not be determined; the four remaining studies in-
Definition. Group education conveys information cluded an interactive format. Education sessions were
about indications for, benefıts of, and ways to overcome delivered by lay health workers or peer facilitators in
barriers to screening with goals of informing, encourag- three studies28,32,36 and by health professionals in the
ing, and motivating participants to seek recommended others. Where specifıed, interventions were conducted in
screening. Group education usually is conducted by the U.S., among African Americans, Latin Americans,
health professionals or by trained lay people who use Filipino Americans, and whites, and in populations of
presentations or other teaching aids in lectures or inter- low- to mixed- or middle-class SES. Most programs were
active formats; they often incorporate role modeling or delivered in churches or homes in the community.
other methods. Group education can be given to a variety Data from four studies could be converted to a com-
of groups, in different settings, and by different types of mon metric and yielded a post-intervention absolute me-
educators with different backgrounds and styles. dian percentage point change in screening completed of
10.6 (range of values: 0 to 59.1). The remaining study32
Breast cancer screening promotion (full update). Of reported an AOR of receiving a Pap test at group (0.69
13 qualifying studies of group education interventions to [95% CIϭ0.41, 1.19]) and individual (1.12 [95%
increase breast cancer screening, ten24 –33 had greatest CIϭ0.91, 1.37]) levels. ORs were adjusted for factors de-
design suitability with good to fair quality of execution, scribed in Breast Cancer Screening Promotion, above.
and three34 –36 had least suitable designs with fair qual-
ity of execution. Twelve studies24 –34,36 examined post- Colorectal cancer screening promotion (full up-
intervention completion of mammography screening, as date). Two studies36,39 were included in the combined
determined by self-report. One study35 examined body of evidence. One study39 with three intervention
county-level mammography rates. Most studies24,26 –35 arms had greatest study design suitability and good qual-
used interactive education programs with one or more ity of execution. The other36 had least-suitable design and
sessions intended to improve participants’ screening fair quality of execution. Both studies examined post-
awareness, knowledge, and attitudes. Eight studies fo- intervention changes in colorectal cancer screening by
cused specifıcally on breast cancer, and four others ad- FOBT as determined by the proportion of returned
dressed multiple cancers. Where specifıed, interventions FOBT kits. One study39 offered interactive group sessions
were conducted in the U.S. and specifıcally targeted mi- delivered by peer facilitators, and the other36 offered ses-
nority26 –28,32–36 and elderly25,26,29,31,35 populations. An sions delivered by promotoras with an in-class format.
additional study30 targeted self-identifıed gay, lesbian, For this study, the team was unable to determine if dis-
and transgendered participants. Most programs were de- cussions were delivered in an interactive or didactic for-
livered in churches or homes within communities. mat. Interventions were interactive education programs
Of these studies, 12, with 13 intervention arms (one report24 delivered in churches or homes in the community. Pop-
ulations included Latinas, African Americans, and white
included two study arms), yielded a post-intervention me-
Americans. The two studies included four intervention
dian absolute percentage point change of 11.5 (interquar-
arms and yielded a median absolute percentage point
tile interval [IQI]ϭ5.5, 24.0). Results from the remaining
change of 4.4 (range of values: Ϫ13 to 37).
study32 could not be expressed as an absolute percentage
point change. It reported AORs of receiving a mammo- Conclusion. According to Community Guide rules of evi-
gram at group (0.82, 95% CIϭ0.44, 1.56) and individual dence, there is now suffıcient evidence that group education
www.ajpmonline.org
7. Sabatino et al / Am J Prev Med 2012;43(1):97–118 103
is effective in increasing screening for breast cancer point increases in mammography use. Of fıve stud-
(Table 1). There still is insuffıcient evidence to determine ies40,42,43,45,48 of absolute change not specifıc to partici-
the effectiveness of group education in increasing screening pants at increased risk, the median increase for seven
for cervical cancer and colorectal cancer because of small intervention arms was 11.9 percentage points (range of
numbers of studies, methodologic limitations of identifıed values: 6.5 to 15.2).
studies, and inconsistent fındings. To compare effects of tailored interventions with those
not explicitly tailored (referred to as “untailored”), the
Applicability. Based on populations and settings included
in these studies, group education interventions to increase team examined evidence from both reviews. Among the
breast cancer screening should be applicable across a range 30 studies measuring absolute change, 23 stud-
of settings and populations, provided they are adapted to ies40,42,43,45– 47,49 – 65 evaluated 30 tailored intervention
target populations and delivery context. Results from studies arms, and demonstrated a median effect of 9.7 percentage
targeting specialized populations may not be generalizable points (IQIϭ6.5, 15.2). For the nine studies48,56,64,66 –71
to interventions directed at the general population. evaluating nine untailored intervention arms, the median
effect was 6.3 percentage points (IQIϭ2.0, 11.4). Findings
Increasing Community Demand for from the three studies56,64,72 in the original review pro-
Screening: One-on-One Education viding intra-study comparisons of tailored versus untai-
lored intervention arms were consistent with the larger
Definition. One-on-one education conveys informa- effect seen for tailored interventions. No update studies
tion to individuals by telephone or in person about indi- provided information about both tailored and untailored
cations for, benefıts of, and ways to overcome barriers to interventions.
screening with the goal of informing, encouraging, and To examine the effect of one-on-one education inter-
motivating people to seek recommended screening. ventions among underserved populations, the team iden-
These messages are delivered by healthcare workers or tifıed nine studies40,43,45,48 –50,52,65,70 from both reviews
other health professionals, lay health advisors, or volun- that described their samples as including predominantly
teers, and are conducted in medical, community, work- low-income women, or that reported that Ͼ30% of their
site, or household settings. Interventions can be untai- sample had income less than $15,000 –$20,000. The me-
lored to address the overall target population or tailored dian effect across 13 effect estimates from these nine
according to individual assessments to address the recip- studies was 10.4 percentage points (IQIϭ9.4, 15.1), com-
ient’s individual characteristics, beliefs, or perceived bar- pared with a median of 8.8 percentage points (IQIϭ2.0,
riers to screening. As defıned for this review, one-on-one 14.4) for the remaining 21 studies (nϭ26 effect
education may be accompanied by a small media or a estimates).
client reminder component. The team also sought to evaluate the incremental effect
Breast cancer screening promotion (reaffirmation up- of one-on-one education interventions beyond the effect
date). The original review18 found strong evidence of of other intervention components common to two or
effectiveness based on a median increase in mammogra- more study arms. Five studies40,41,44 – 46 were identifıed
phy use across 23 studies of 9.2 percentage points from the update, seven studies49 –51,55,60,63,69 from the
(IQIϭ4.9, 14.4), and ORs from four additional study original review, and one study73 from the review of mul-
arms in the favorable direction. Nine studies40 – 48 were ticomponent interventions that allowed this type of com-
included in the update. All had greatest design suitability. parison. Two of these studies44,63 provided information
As in the original review,18 outcomes were assessed by about three comparisons that included a different form of
self-report40,41,46,47 or medical record review.42– 45,48 In- one-on-one education in comparison groups (e.g., the
terventions were delivered in the home40 – 48 or clinic,44 incremental effect of phone education beyond the effect
by medical40,44 and nonmedical professionals,41– 43,45– 48 of in-person education combined with other intervention
by telephone,40,42,44 – 48 or in person.41,43,44 Most studies components44). Across all 13 studies (nϭ15 effect esti-
included tailored components.40,42– 47 Studies were con- mates), the overall median incremental effect was 6.1
ducted in the U.S. and included urban40,41,44,45,48 and percentage points (IQIϭ2.0, 11.0). Effects for the three
rural populations.42,43 Studies included participants who comparisons that included forms of one-on-one educa-
were African-American, Hispanic,40,41,43,45– 48 Asian- tion in comparison groups were Ϫ17.4, Ϫ3.0, and 11.0
American,46,47 and Native American43; had low SES percentage points.
had increased risk for breast cancer46,47; and were non-
adherent with recent screening. Cervical cancer screening promotion (reaffirmation
Results from two studies46,47 of participants with in- update). The original review18 found strong evidence of
creased breast cancer risk ranged from 1 to 18 percentage effectiveness based on a median increase of 8.1 percent-
July 2012
8. 104 Sabatino et al / Am J Prev Med 2012;43(1):97–118
age points (IQIϭ5.7, 17.3), with three studies evaluating cluded. No clear differences emerged, although the num-
fıve tailored intervention arms, and two studies evaluat- ber of effect estimates in some strata was small.
ing three untailored arms. No additional studies were
Conclusion. According to Community Guide rules of
identifıed during update.
evidence, there is strong evidence that one-on-one edu-
cation is effective in increasing screening for breast and
Colorectal cancer screening promotion (full up-
cervical cancers, and suffıcient evidence of its effective-
date). All seven qualifying studies in the combined
ness in increasing colorectal cancer screening with FOBT
body of evidence had greatest design suitability, four74 –77 (Table 1). However, evidence remains insuffıcient to de-
with good and three78 – 80 with fair quality of execution. termine the effectiveness of one-on-one education in in-
Most studies75,76,78,80 ascertained screening use via med- creasing colorectal cancer screening with other modali-
ical record review, although others reviewed appoint- ties, because too few studies were identifıed, and results
ment attendance,77 screening program records,79 or as- for those studies were inconsistent (colonoscopy).
certained use by self-report confırmed by physician
survey.74 Interventions were delivered in both Applicability. In the original review, the Task Force
home74,77–79 and clinic settings,75,76,80 by phone,77–79 in concluded that fındings for breast and cervical cancer
person,74 –76,80 or by medical professionals74,75,80 or oth- screening should apply both to tailored and untailored
ers.74,76 –79 Three studies74,78,79 evaluated tailored inter- interventions across a range of populations, provided
ventions. Most studies included participants aged Ն50 intervention programs were adapted to target population
years, although two included participants in their and delivery context. Studies included in the update sup-
40s.74,80 In addition to white participants, studies in- port these conclusions. Recommendations for colorectal
cluded African-American,75,77 Hispanic,75 and Asian- cancer screening with FOBT also should apply across a
American74,76 participants; participants with low range of populations. Although no studies explicitly
SES76,80; and urban populations.75–77 No studies speci- noted the inclusion of rural populations, fındings are not
fıed inclusion of rural populations. One study74 included expected to differ substantially from those of urban and
unspecifıed populations.
participants at increased risk due to a fırst-degree family
history of colorectal cancer. All studies were conducted in
the U.S. Increasing Community Demand for
These seven studies evaluated 15 intervention arms Screening: Client Reminders
(one study80 included six intervention arms) and re-
Definition. Client reminders or recalls are textual (let-
ported outcomes for FOBT (nϭ10 effect esti-
ter, postcard, e-mail) or telephone messages advising
mates)75,76,78 – 80; flexible sigmoidoscopy (nϭ1 effect es-
people that their screening is due (reminder) or overdue
timate)78; colonoscopy (nϭ2 effect estimates)78,80; and (recall). Client reminders may be enhanced by one or
with any test (nϭ2 effect estimates).74,78 The median more of the following: follow-up printed or telephone
effect for FOBT was 19.1 percentage points (IQIϭ12.9, reminders; additional text or discussion with information
25.1). Effects for any cancer screening test ranged from 1 about indications for, benefıts of, and ways to overcome
to 11 percentage points, and for colonoscopy ranged from barriers to screening; and assistance in scheduling ap-
0 to 11 percentage points. The one study reporting flexi- pointments. Interventions can be untailored to address
ble sigmoidoscopy outcomes reported no effect. the overall target population or tailored with the intent to
Among the fıve studies of FOBT screening, two evalu- reach one specifıc person, based on characteristics unique
ated tailored interventions78,79 and three studies,75,76,80 to that person, related to the outcome of interest, and
with evaluable data from eight intervention arms, did not. derived from an individual assessment.
Effects for tailored interventions ranged from 1 to 20.7
percentage points. The median for untailored interven- Breast cancer screening promotion (reaffirmation up-
tions was 20.7 percentage points (IQIϭ13.8, 25.8). No date). The original review18 of client reminders found
studies included within-study comparisons of tailored strong evidence of effectiveness based on a median in-
and untailored interventions. The few studies of tailored crease of 14.0 percentage points in recent mammography
and of untailored arms along with overlapping fındings (19 studies; IQIϭ2.0, 24.0) and three additional studies
by tailored status, makes drawing conclusions for FOBT demonstrating an increase in repeat mammography. In
based on tailoring diffıcult. The team also stratifıed anal- the update, six additional studies81– 86 were included. All
yses by whether interventions were delivered by phone or had greatest design suitability except for one83 with least-
in person, by medical professionals or others, and suitable design. Exclusion of this study83 did not change
whether small media and/or client reminders were in- overall conclusions.
www.ajpmonline.org
9. Sabatino et al / Am J Prev Med 2012;43(1):97–118 105
Outcomes for update studies of breast cancer screen- terventions enabled this type of comparison. Across all
ing promotion were ascertained via self-report,86 medical nine studies (nϭ12 effect estimates), the overall median
record review,85 administrative records,81,82,84 or screen- incremental effect was 5.0 percentage points (IQIϭ1.6,
ing program attendance.83 Interventions included both 6.7).
textual83– 86 and telephone reminders,81,82 which in- One study82 in the update provided information about
cluded automated interactive voice response reminders the effect of a telephone client reminder in increasing
(AIVR) by phone81 as well as tailored interventions86 and screening use by either clinical breast exam or mammog-
enhanced interventions82,84 – 86 (as in the original re- raphy. Because of the different outcome, it was not in-
view,18 defıned as including follow-up reminders, addi- cluded in analyses of absolute change in mammography
tional text, discussion, or appointment scheduling assis- use. This study82 showed an absolute increase in screen-
tance). Studies included reminders delivered by clinical ing of 8 percentage points.
practices or organizations,81,85 screening programs or regis-
tries,82– 84 or other sources.86 Where specifıed, interventions Cervical cancer screening promotion (reaffirmation
were conducted in the U.S.82,84 – 86 and Norway.83 Studies update). The original review18 found strong evidence of
included white,82,84 – 86 African-American,82,86 and His- effectiveness based on a median increase in Pap test use
panic participants.86 No studies specifıed inclusion of across 14 intervention arms of 10.2 percentage points
other racial or ethnic groups, although several included (IQIϭ6.3, 17.9). In the update, six additional qualifying
groups of unspecifıed race. Others did not report race or studies81,85,95–98 were identifıed. All had greatest design
ethnicity. Individuals with low SES82,84 and urban or suitability except for one95 with least-suitable design.
mixed urban/rural populations82,84,85 also were included. This study95 was not included in the assessment of abso-
Several studies did not report this information. lute change in screening use; exclusion of this study did
Of four update studies81,83,84,86 providing information not change overall conclusions.
about absolute change in mammography use, two81,83 Outcomes for update studies of cervical cancer screen-
provided information about recent screening only, de- ing promotion were ascertained via medical record re-
fıned as completion of the most recent mammogram view,85 administrative records,81 or screening registry re-
within a specifıed interval; one84 provided information cords.96 –98 Method of ascertainment was not reported in
about repeat mammography only, defıned as examining one study.95 Interventions included printed reminders
two or more consecutive, on-time mammograms; and only,85,96,98 telephone reminders only,81 and printed re-
one86 provided information about both. The only phone minders with telephone follow-up reminders.95,97 Re-
intervention among these four studies was the AIVR minders were delivered by clinical practices or organiza-
study.81 When studies from both reviews were combined tions81,85,95 and screening programs or registries.96 –98
to examine differences by recent versus repeat screening No studies included tailored interventions, and
use, the median increase for recent use was 12.3 percent- four85,95,97,98 included enhanced interventions. Where
age points (IQIϭ3.0, 18.9; nϭ30 effect estimates) and for specifıed, interventions were conducted in the U.S., Swe-
repeat mammography was 6.0 percentage points den,97 Belgium,96 and Australia.95,98 One study85 re-
(IQIϭ3.0, 19.1; nϭ8 effect estimates). ported including nonwhite participants but did not pro-
Findings from the original review also suggested that vide more-specifıc information. The remaining studies
unenhanced, printed reminders have smaller effects than did not report race/ethnicity. The one study97 that re-
enhanced or telephone reminders (median 3.6 percent- ported SES included low-SES participants. Three stud-
age points across 12 studies vs 18.5 percentage points ies85,95,98 included urban or mixed urban/rural popula-
across 13 studies, respectively). This conclusion was sup- tions. The other three studies81,96,97 did not report urban/
ported by all nine intra-study comparisons. When the rural status.
team incorporated update studies,81,83,84,86 including one Four studies81,96 –98 evaluating fıve intervention arms
study84 with separate arms for unenhanced and enhanced provided information about absolute changes in screen-
client reminders, fındings reaffırmed that enhanced or ing use. One97 provided information about both fol-
telephone reminders may have a greater effect (15.5 per- low-up printed reminders and follow-up telephone re-
centage points [IQIϭ7.0, 29.0] vs 4.5 percentage points minders. The median increase was 2.8 percentage points
[IQIϭ1.9, 14.0]). (range of values: 1.6 to 31.4). Although the increase in the
The team also examined the incremental effect of client update was smaller than in the original review,18 effect
reminders beyond the effect of other intervention com- estimates from the update fell within the range of effects
ponents common to two or more study arms. One study85 in the original review.
in the update, six studies87–92 in the original review, and As for breast cancer screening, fındings from the
two studies93,94 from the review of multicomponent in- original review18 suggested that unenhanced printed
July 2012
10. 106 Sabatino et al / Am J Prev Med 2012;43(1):97–118
reminders may have a smaller effect than enhanced or No update studies provided information about abso-
telephone reminders (median increase 9.8 percentage lute changes in screening. The two studies85,100 with
points vs 15.5 percentage points, respectively). This greatest suitability provided information about incre-
conclusion was supported by one intra-study compar- mental effects of client reminders on FOBT screening.
ison.18 Among three update studies81,97,98 evaluating One study80 from the original review and two studies93,94
four intervention arms that included telephone and/or from the review of multicomponent interventions also
enhanced reminders, the range of effects was 1.6 to allowed this type of comparison for FOBT screening.
31.4 percentage points. The one update study of Across all fıve studies80,85,93,94,100 evaluating nine inter-
printed unenhanced reminders reported a 1.8 percent- vention arms, the median incremental effect for FOBT
age point increase.96 No differences were noted ac- use was 10.9 percentage points (IQIϭ6.0, 13.5).
cording to other study characteristics, although there No studies in the original review provided information
were few update studies, which limited the authors’ about colorectal cancer screening with tests other than
ability to detect differences. FOBT. Two update studies85,100 evaluating fıve interven-
The team also examined the incremental effect of client tion arms provided information about incremental ef-
reminders beyond the effect of other intervention com- fects of client reminders on use of flexible sigmoidoscopy,
ponents common to two or more study arms. One study85 colonoscopy, or barium enema. The median increase
in the update interval, one study99 in the original review, across these fıve effect estimates was 0.5 percentage
and two studies93,94 from the original review of multi- points (range of values: 0.0 – 6.0). One study100 reported
component interventions enabled this type of compari- the incremental effect on completion of any colorectal
son. These studies evaluated fıve intervention arms and cancer screening test (FOBT, flexible sigmoidoscopy,
provided information about the incremental effect of cli- colonoscopy, or barium enema) to be 1.0 percentage
ent reminders in addition to provider-directed interven- point.
tions. The overall median incremental effect was 3.7 per- One update study101 provided information about a
centage points (range of values: Ϫ3.5 to 25.2). One update printed follow-up reminder to participants randomized
study95 reported the relative increase in number of Pap to one of fıve screening-test regimens who did not re-
tests performed over 2 years to be 6.3%. Because of the spond. The outcome was FOBT or flexible sigmoidos-
different outcome (i.e., number of tests), it was not in- copy completion. Because of the different nature and
cluded in analyses of absolute change. outcome of this study,101 it was not included in analyses
of absolute or incremental change. Absolute increases
Colorectal cancer screening promotion (interval up- associated with reminders were reported to be 9.2% and
date for fecal occult blood testing). The original re- 11.1% for participants invited to complete mailed FOBT
view found suffıcient evidence of effectiveness for client kits and FOBT delivered by general practitioners, respec-
reminders to increase colorectal cancer screening with tively, and 3.3% for participants invited to complete one-
FOBT based on a median increase across four studies time sigmoidoscopy and 3.2% for flexible sigmoidoscopy
(nϭ8 effect estimates) of 11.5 percentage points followed by FOBT.
(IQIϭ8.9, 20.3). The update included three additional
Conclusion. According to Community Guide rules of
studies.85,100,101 All had greatest design suitability except
evidence, there is strong evidence that client reminders
for one101 with least-suitable design. That study was not
are effective in increasing screening for breast and cervi-
included in the assessment of absolute change in screen-
cal cancers and for colorectal cancer with fecal occult
ing and its exclusion did not change overall conclusions.
blood testing (Table 1). However, evidence remains in-
All three studies had fair quality of execution.
suffıcient to determine its effectiveness in increasing
Outcomes for update studies were ascertained via sur-
colorectal cancer screening with other tests (colonos-
vey100 and medical record review.85,101 All interventions
copy, flexible sigmoidoscopy) because evidence from the
were printed, none were tailored, and two85,100 were en-
two additional studies identifıed produced inconclusive
hanced. Reminders were delivered by clinical practices or
results.
organizations85,101 or screening programs.100 Where
specifıed, interventions were conducted in the U.S. and Applicability. The original review18 concluded that rec-
Italy.101 One85 study reported including nonwhite partic- ommendations for client reminder interventions to in-
ipants, although it did not provide more-specifıc infor- crease screening for breast, cervical, and colorectal cancer
mation. The remaining studies did not report race/eth- (FOBT only) should be applicable across a range of set-
nicity. Two studies85,100 included mixed urban/rural or tings and populations, provided they are adapted to the
non-urban populations; the third did not report urban/ target populations and delivery context. Studies included
rural status. during the update support these conclusions.
www.ajpmonline.org
11. Sabatino et al / Am J Prev Med 2012;43(1):97–118 107
Increasing Community Demand for attempt to capitalize on existing interventions and infra-
Screening: Client Incentives structure. This review evaluated the effectiveness of mass
media used alone or its individual contribution to the effec-
Definition. Client incentives are small, noncoercive re-
tiveness of multicomponent interventions.
wards (e.g., cash or coupons) to motivate people to seek
cancer screening for themselves or to encourage others Breast cancer screening promotion (full update). Two
(e.g., family members, close friends) to seek screening. studies103,104 qualifıed for review. Both had fair quality of
Incentives are distinct from interventions designed to execution, one103 with greatest and one104 with moderate
improve access to services (e.g., transportation, child design suitability. One study104 evaluated the effect of a
care, reducing client out-of-pocket costs). radio and newspaper advertisement campaign compared
with usual care among urban, Italian-speaking women in
Breast cancer screening promotion (full up- Australia. The outcome was the number of mammo-
date). One study102 qualifıed for review and had great- grams performed per month, ascertained through review
est design suitability and fair quality of execution. This of screening program records. For women in their 50s,
study evaluated the effect of a $10 incentive for women the relative percentage change in number of mammo-
who completed mammography screening through a pre- grams completed was Ϫ16.1% for initial screens and
existing program that provided free mammograms to Ϫ4.2% for subsequent screens. Among women in their
low-income, under-, or uninsured women. The inter- 60s, the relative percentage changes were Ϫ10.8% and
vention was sent to all women in a commercial data- 9.0% for initial and subsequent screens, respectively.
base who were aged 40 – 63 years and from census The second study103 compared a multicomponent in-
blocks having household size and income characteris- tervention including a higher-intensity mass media com-
tics consistent with program guidelines. However, ponent (messages on city buses, newspaper ads and/or
only program-eligible women were included in assess- articles, radio and/or TV programs, and public service
ing mammography completion. announcements) with a multicomponent intervention
The study provided information about the incremental including a lower-intensity mass media component
effect of adding client incentives to screening availability (campus newspapers and yard signs [reported seldom to
information and appointment scheduling assistance. The be employed]). Other components in both arms included
incremental effect was 0.52 percentage points (95% group education, small media, and health fairs. The sam-
CIϭ0.32, 0.72). Results restricted to women eligible for ple included African-American women living in census
the free screening program, rather than all women in the tracts with a high proportion of African-American resi-
identifıed census blocks, yielded an incremental effect of dents. The outcome was self-reported completion of a
Ͻ2.0 percentage points. mammogram within 2 years; clinical breast exam within
Cervical cancer screening promotion (full up- 2 years also was reported. The absolute change in screen-
date). No qualifying studies evaluating the effect of client ing was Ϫ2.4 percentage points (95% CIϭ Ϫ9.0, 4.2) for
incentives on cervical cancer screening were identifıed. mammography and 4.2 percentage points (95% CIϭ
Ϫ1.1, 9.5) for clinical breast exam, respectively. No stud-
Colorectal cancer screening promotion (full up- ies included information provided through other modes,
date). No qualifying studies evaluating the effect of client such as magazines or the Internet.
incentives on colorectal cancer screening were identifıed.
Cervical cancer screening promotion (full up-
Conclusion. According to Community Guide rules of date). Three studies qualifıed for review, of which
evidence, there is insuffıcient evidence to determine the two103,105 had greatest and one106 had least-suitable de-
effectiveness of using client incentives to increase screen- signs. All had fair quality of execution. The two studies from
ing for breast, cervical, or colorectal cancers, because only the original review included three intervention arms. Rela-
one study102 for breast cancer and no studies for cervical tive percentage increases in Pap test completion ascertained
and colorectal cancers qualifıed for review (Table 1). by record review were reported to be 20.4% and 47.6% in
one study and 21.3% in the other. The update study assessed
Increasing Community Demand for the effect of higher- versus lower-intensity mass media as
Screening: Mass Media part of a multicomponent intervention (described further
above). The absolute change in women screened within
Definition. Mass media—including TV, radio, newspa-
2 years was 4.7 percentage points.
pers, magazines, and billboards—are used to communicate
educational and motivational information in community or Colorectal cancer screening promotion (full up-
larger-scale intervention campaigns. Mass media interven- date). One study103 qualifıed for review. This study ex-
tions, however, almost always include other components or amined the effect of higher- versus lower-intensity mass
July 2012
12. 108 Sabatino et al / Am J Prev Med 2012;43(1):97–118
media as part of a multicomponent intervention (de- Other health issues also were discussed (e.g., prostate and
scribed further above). Outcomes included ever having colorectal cancer screening). Women residing on
had FOBT and ever having had proctoscopy. (Like Molokai Island who were aged Ն40 years were eligible for
colonoscopy or sigmoidoscopy, proctoscopy involves in- mammography screening. The post-intervention in-
sertion of a tube into the rectum to look for signs of crease in mammography screening was 18 percentage
cancer or other problems, although proctoscopy is an points (95% CIϭ Ϫ1.0, 37.0). The secondary outcome,
older test that used a rigid tube.107) The absolute change clinical breast examinations, increased by 34 percentage
in screening was Ϫ4.7 percentage points (95% CIϭ points (95% CIϭ19.0, 49.0).
Ϫ12.3, 2.9) for FOBT, and Ϫ8.0 percentage points (95%
CIϭ Ϫ15.2, Ϫ0.8) for proctoscopy. Cervical cancer screening promotion (full up-
date). Three studies qualifıed for review. Two stud-
Conclusion. According to Community Guide rules of ies109,110 were of greatest design suitability with fair qual-
evidence, there is insuffıcient evidence to determine the ity of execution, and the remaining study111 had a least-
effectiveness of mass media interventions in increasing suitable design with fair quality of execution. All three
screening for breast, cervical, and colorectal cancers be- studies of reducing structural barriers examined self-
cause too few studies qualifıed for review (Table 1). reported, post-intervention Pap test use. Two of three
studies109 –111 investigated effectiveness of alternative
Increasing Community Access to Screening: screening sites. One study111 examined a nurse-led clinic
Reducing Structural Barriers within a correctional facility, and another110 offered on-
Definition. Structural barriers are non-economic bur- site screening to residents at a high-rise apartment build-
dens or obstacles that impede access to screening. Inter- ing. The fınal study109 invited participants to receive
ventions designed to reduce these barriers may facilitate screening during extended hours. Studies were con-
access by reducing time or distance between service de- ducted in the U.S., Canada, and Australia. Participants
livery settings and target populations; modifying hours of included low-income female residents of a high-rise
service to meet client needs; offering services in alterna- apartment building,110 incarcerated women,111 and fe-
tive or nonclinical settings (e.g., mobile mammography male patients of a university-based general practice who
vans at worksites or in residential communities); and were due or overdue for screening.109 For the overall
eliminating or simplifying administrative procedures and body of evidence, the median increase in Pap screening
other obstacles (e.g., scheduling assistance or patient nav- was 13.6 percentage points (range of values: 5.9 –17.8).
igators, transportation, dependent care, translation ser-
Colorectal cancer screening promotion (reaffirmation
vices, limiting the number of clinic visits). Such interven-
update). The original review19 found strong evidence of
tions often include one or more secondary supporting
effectiveness of interventions to reduce structural barri-
measures, such as printed or telephone reminders; edu-
ers to colorectal cancer screening with FOBT. The me-
cation about cancer screening; information about screen-
dian increase was 16.1 percentage points (IQIϭ12.1, 22.9;
ing availability (e.g., group education, pamphlets, or bro-
nϭ11 effect estimates). Five additional studies were in-
chures); or measures to reduce client out-of-pocket costs.
cluded in the update. Four studies108,112–114 had least-
Interventions principally designed to reduce client costs
suitable study designs and fair quality of execution. One
are considered a separate class of approaches (discussed
study115 had greatest suitability of study design, with
below).
good quality of execution.
Breast cancer screening promotion (reaffirmation up- Outcomes in update studies included completion of
date). The original review19 found strong evidence of FOBT alone112,115; colonoscopy or FOBT (including fecal
effectiveness for reducing structural barriers to breast immunochemical tests)113; any of the three testing mo-
cancer screening, based on a median overall increase in dalities (FOBT, sigmoidoscopy, or colonoscopy)108; and
mammography use across seven studies of 17.7 percent- the mean number of colonoscopies per month.114 Out-
age points (IQIϭ11.5, 30.5). The update included one comes were ascertained by proportion of returned
additional study108 with a least-suitable study design. kits,112,115 self-report,113 and medical record108 or hospi-
That study examined self-reported, post-intervention tal record review.114 Most evidence focused on ap-
completion of mammography, and clinical breast exam proaches to reduce time and distance to completing
screening. The intervention was a 1-day community cel- screening (e.g., mailing FOBT cards to clients). Studies
ebration in Hawaii with personalized recruitment, one- were conducted in the U.S. and France and in medical
on-one talk story education sessions, and culturally rele- care and community settings. All studies enrolled men
vant education brochures. Subjects met with physicians and women aged Ն50 years. One study112 enrolled par-
of the same gender, who were flown in for the event. ticipants who were due or overdue for screening. Another
www.ajpmonline.org
13. Sabatino et al / Am J Prev Med 2012;43(1):97–118 109
115
study enrolled participants who had not received Increasing Community Access to Screening:
screening in the previous year. The remaining studies did Reducing Out-of-Pocket Costs
not specify screening histories. Specifıed racial/ethnic
Definition. These interventions attempt to minimize or
groups included whites, Hispanics/Latinos, African
remove economic barriers that impede client access to
Americans, and Native Hawaiians. Included populations
cancer screening services. Costs can be reduced through a
also varied, from residents of urban communities113 to variety of approaches, including vouchers, reimburse-
residents of a remote Hawaiian Island.108 ments, reduction in copays, or adjustments in federal or
Based on four effect estimates in the update studies, state insurance coverage. Efforts to reduce client costs
there was a median 36.9 percentage point increase across may be combined with measures to provide client educa-
colorectal cancer screening tests (range of values: 16.3 to tion, information about program availability, or mea-
41.1). Additional evidence showed a 9.5% relative in- sures to reduce structural barriers.
crease in the mean number of colonoscopies per
month.114 Breast cancer screening promotion (interval up-
Conclusion. According to Community Guide rules of date). The original review19 found suffıcient evidence
evidence, there is strong evidence that reducing struc- of effectiveness to recommend interventions that reduce
tural barriers is effective in increasing screening for out-of-pocket costs to promote breast cancer screening,
breast and colorectal cancers (by mammography and based on a median increase in completed mammography
FOBT, respectively; Table 1). Evidence is insuffıcient, across eight intervention arms of 11.5 percentage points
however, to determine whether reducing structural (IQIϭ6.0, 28.5). No additional studies were identifıed
barriers is effective in increasing colorectal cancer during the update.
screening by flexible sigmoidoscopy or colonoscopy
because only one study114 using these screening proce- Cervical cancer screening promotion (full up-
dures was identifıed. Evidence was also insuffıcient to date). One study116 qualifıed for review and had least-
determine the effectiveness of reducing structural bar- suitable design and fair quality of execution. This study
riers in increasing screening for cervical cancer be- reported an increase in completed Paps tests of 17 per-
cause only three relevant studies were identifıed, and centage points.
these had methodologic limitations.
Colorectal cancer screening promotion (full up-
Applicability. The original review concluded that the date). No qualifying studies of reducing client out-of-
evidence for reducing structural barriers interventions to pocket costs interventions to increase colorectal cancer
increase breast cancer screening should be applicable screening were identifıed.
across a range of settings for target populations with
limited access to mammography. That review placed Conclusion. According to Community Guide rules of
heavy emphasis on strategies to reduce time and distance evidence, there is suffıcient evidence that reducing client
or create alternative screening locations. In addition to out-of-pocket costs is effective in increasing screening for
including populations similar to the original review, the breast cancer (Table 1). There is insuffıcient evidence to
updated body of evidence included a study that focused determine its effectiveness in increasing screening for
on rural populations. Hence, fındings from the original cervical or colorectal cancer because too few (cervical
cancer) or no (colorectal cancer) studies were identifıed.
review are supported, such that recommendations should
Nonetheless, the consistently favorable results for inter-
apply across a range of populations and settings, provided
ventions that reduce costs for breast cancer screening and
that programs are adapted to target populations and de-
several other preventive services suggest that such inter-
livery contexts.
ventions are likely to be effective for increasing cervical
For colorectal cancer screening, original review fınd-
and colorectal cancer screening as well.
ings were limited to FOBT screening and applicable
across a range of settings where target populations may Applicability. The original review19 concluded that rec-
have limited physical access to FOBT. Included studies ommendations for use of interventions that reduce out-
generally represented white and African-American pop- of-pocket costs to increase screening for breast cancer
ulations but no other racial ethnic groups. For the up- should be applicable across a range of settings and popu-
dated review, applicability may be expanded, given the lations where target populations may have limited fınan-
addition of studies from another high-income econ- cial resources for mammography. Because no additional
omy112 and studies whose samples included other popu- studies were identifıed during the update, conclusions
lations (e.g., Native Hawaiians, Hispanics). about applicability remain unchanged.
July 2012
14. 110 Sabatino et al / Am J Prev Med 2012;43(1):97–118
Increasing Provider Delivery or Promotion of sigmoidoscopy, or colonoscopy was a 45 percentage
Screening: Provider Assessment and point increase.
Feedback Conclusion. According to Community Guide rules of
Definition. Provider assessment and feedback interven- evidence, there is suffıcient evidence that provider assess-
tions both evaluate provider performance in offering ment and feedback interventions are effective in increas-
and/or delivering screening to clients (assessment) and ing screening for breast cancer (mammography); cervical
present providers with information about their perfor- cancer (Pap test); and colorectal cancer (FOBT; Table 1).
mance in providing screening services (feedback). Feed- Evidence was insuffıcient, however, to determine effec-
back may describe the performance of a group of provid- tiveness of this intervention in increasing colorectal can-
ers (e.g., mean performance for a practice) or individual cer screening using methods other than FOBT.
providers and may be compared with a goal or standard. Applicability. The original review concluded that rec-
ommendations to increase screening for breast, cervical,
Breast, cervical, and colorectal cancer screening pro- and colorectal cancer (FOBT only) should be applicable
motion (full update). Nine qualifying studies117–125 across settings and populations described, with the caveat
were included in the review. Four studies117,122–124 had that provider training status potentially was related to
greatest, two119,125 had moderate, and three118,120,121 had magnitude of effect. Considering additional information
least suitable designs. Quality of execution was fair for all from the update, conclusions about applicability remain
except two,121,122 for which it was good. unchanged.
Seven studies118,120 –125 reported completed screening,
and four studies117,119,121,123 reported screening ordered by
Increasing Provider Delivery or Promotion of
providers. No studies of ordered screening were included in
Screening: Provider Incentives
the update. Completed screening outcomes were ascer-
tained through medical record review.118,120 –123,125 Assess- Definition. Provider incentives are direct or indirect
ment of provider screening performance was conducted rewards intended to motivate providers to perform
by providers auditing charts of their own patients118 or cancer screening or make appropriate referral for their
another provider’s patients,120 via computer patients to receive these services. Rewards are often
search123,124 or chart review by researchers121,122 or monetary but can include nonmonetary incentives also
others.125 Feedback was provided concerning individ- (e.g., continuing medical education credit). Because
ual provider performance,120,124 group perfor- some form of assessment is needed to determine
mance,121,125 or both.118,122,123 Feedback received by whether providers receive rewards, an assessment
providers varied from a single occurrence118,121 to regu- component may be included in the intervention.
lar intervals.120,122–125 Studies of completed screening Breast, cervical, and colorectal cancer screening pro-
were conducted in the U.S.120 –125 and the United King- motion (full update). Five studies qualifıed for review.
dom,118 and included both trainee120,122–124 and non- Of these, three126 –128 had greatest and two129,130 had
trainee physicians.118,121 Two studies specifıed patient least-suitable designs. All had fair quality of execution
race/ethnicity, including African-American, Hispanic, except for one128 with good quality of execution.
and Asian participants,122,123 and several specifıed the Of these fıve studies, three128 –130 reported completed
inclusion of urban120,122,124 and rural groups.121 screening, one127 reported recommended or offered
For completed screening, four effect estimates121–124 screening, and one126 reported both. The four studies of
were included for mammography, four118,120,122,124 for completed screening ascertained outcomes from medical
Pap test, and three120,122,124 for colorectal cancer screen- records,126,129 self-report,126 performance reports,128 or
ing with FOBT, with one study122 also providing an esti- claims data130 from health plans. Interventions included
mate for flexible sigmoidoscopy. One study125 evaluated provider incentives alone128 –130 or with provider assess-
change in use of FOBT, flexible sigmoidoscopy, or ment and feedback and reminders.126
colonoscopy. Findings across all screening sites led to a The nature of and details provided about incentives
median increase in screening use of 13.0 percentage varied across studies. Interventions included a quarterly
points (IQIϭ5.5, 21.8). Findings for mammography var- practice bonus of approximately $0.23 per member per
ied from 3.4 to 20.6 percentage points, for Pap from 4.0 to month for each performance target met, with bonus po-
29.5 percentage points, and for FOBT screening from tential representing approximately 5% of capitation128;
12.3 to 23.0 percentage points. The one estimate for flex- quarterly practice bonuses with the amount related to
ible sigmoidoscopy showed essentially no effect. The es- whether higher or lower screening thresholds were
timate from the update study examining FOBT, flexible met129; a physician bonus based on the percentage re-
www.ajpmonline.org
15. Sabatino et al / Am J Prev Med 2012;43(1):97–118 111
ferred for screening during each audit period (i.e., $50 for For client reminders, barriers may include limited in-
a 50% referral rate)126; and a year-end physician bonus frastructure and staffıng and/or computer support to
program with specifıcs of the bonus unavailable.130 identify patients due for screening and deliver reminders
Studies of completed screening took place in the effıciently.18 Further, costs of generating and delivering
U.S.126,128,130 and Scotland.129 Physician settings ranged reminders may be a substantial barrier, and barriers re-
from large, multispecialty organizations128 to individual lated to tailoring may apply.18 (When done on a large
practice associations or physician practices.126,129,130 Pa- scale, such interventions may cost little per person.) As
tient populations included commercially insured health noted in the original review,19 potential barriers for re-
plan members130 and patients of selected practices.126,129 ducing structural barriers interventions may include lim-
The four studies of completed screening evaluated ited resources to provide mobile mammography services,
seven intervention arms: two for mammography, two for diffıculty identifying alternative screening sites, ade-
Pap tests, one for FOBT, one for endoscopic screening, quately staffıng facilities at alternative sites or during
and one for double-contrast barium enema. The median alternative hours, and ensuring follow-up of abnormal
change in screening use across studies was 1.7 percentage tests for clients lacking access to care. Barriers to imple-
points (IQIϭ Ϫ0.1, 3.6). Findings for mammography mentation were not addressed for client incentives, mass
varied from Ϫ2.0 to 1.7 percentage points, for Pap from media, and provider incentives, because effectiveness was
3.6 to 8.0 percentage points, and for colorectal screening not established for any cancer screening site.
from Ϫ0.1 to 2.8 percentage points.
Conclusion. According to Community Guide rules of Research Gaps
evidence, there is insuffıcient evidence to determine the The team found suffıcient to strong evidence that inter-
effectiveness of provider incentives in increasing screen- ventions using one-on-one education, client reminders,
ing for breast, cervical, or colorectal cancers (Table 1). provider assessment and feedback, and reducing struc-
Evidence is insuffıcient because results were inconsistent tural barriers are effective in promoting colorectal cancer
and generally small. screening with FOBT. However, more information is
needed to determine whether interventions are effective
for other forms of colorectal cancer screening. Effective-
Additional Benefits and Potential Harms of ness for these other tests has not been established for any
Interventions intervention.
No reports of other positive or negative effects of inter- Further, as new screening tests emerge (e.g., fecal im-
ventions on use of other healthcare services, health be- munochemical tests), information will be needed about
haviors, or informed decision making were found while whether effects differ for these tests. It is also unknown
updating reviews in all intervention categories. For client whether interventions to promote colorectal cancer
incentives, no other positive or negative effects of incen- screening are equally effective when specifıc to one type
tives with small monetary value were identifıed in the of test, or when addressing colorectal cancer screening
literature reviewed or by the review team. However, the more generally. Because there is more than one recom-
team noted that, at some point, as the monetary value of mended screening test for colorectal cancer, focusing in-
incentives increases, they have the potential to become terventions on only one test may limit client choices,
coercive. disregard client preferences, or fail to consider provider
preferences. More information also is needed about effec-
tiveness of interventions using incentives, both client-
Potential Barriers to Implementing and provider-directed, and mass media. Where informa-
Interventions tion about these interventions may be available, or where
In general, limited resources and infrastructure appear to plans to employ such interventions may already be in
be the most important barriers to implementing inter- place, the publication of such data or evaluation to exam-
ventions.18 For one-on-one education interventions, re- ine effectiveness of these interventions would help bridge
cruitment and training of educators, quality-control these gaps.
measures, duration of educational sessions, travel for in- Effectiveness of group education and reducing out-of-
person education, and professional backgrounds of edu- pocket cost interventions was established for breast can-
cators may influence costs and feasibility of implementa- cer screening although not for cervical or colorectal can-
tion. In addition to costs, these interventions may require cer screening. However, given consistently favorable
special skills or tools to develop messages, including tai- results for interventions that reduce costs for breast can-
lored messages, which also may pose implementation cer screening and other preventive services, there is no
barriers.18 reason to conclude a priori that results for breast can-
July 2012