1. Epidemiology and Global Health
Download icon

Cancer risk perception and physician communication behaviors on cervical cancer and colorectal cancer screening

  1. Diane M Harper  Is a corresponding author
  2. Madiha Tariq
  3. Asraa Alhawli
  4. Nadia Syed
  5. Minal Patel
  6. Ken Resnicow
  1. Department of Family Medicine, University of Michigan School of Medicine, United States
  2. Department of Obstetrics & Gynecology, University of Michigan School of Medicine, United States
  3. Department of Women’s Studies, University of Michigan, College of Literature, Science and the Arts, United States
  4. Arab Community Center for Economic and Social Services (ACCESS), United States
  5. Department of Health Behavior & Health Education, University of Michigan School of Public, United States
  6. Outreach and Health Disparities Research, University of Michigan Rogel Cancer Center, United States
  7. Center for Health Communications Research, University of Michigan, School of Public Health, United States
Research Article
  • Cited 0
  • Views 240
  • Annotations
Cite this article as: eLife 2021;10:e70003 doi: 10.7554/eLife.70003

Abstract

Background:

Women 50–65 years of age have the lowest cervical and colorectal cancer (CRC) screening rates among ages recommended for screening. The primary aim of this work is to determine how cancer risk perceptions and provider communication behaviors, in addition to known demographic factors, influence the uptake of both cervical and CRC screening or a single screen among women in southeast Michigan.

Methods:

Fourteen health services and communication behavior questions were adapted from the Health Information National Trends Survey (HINTS) and administered to a multiethnic sample of adults in southeast Michigan. The outcome variable was self-reported up-to-date cervical cancer and/or CRC screening as defined by the United States Preventive Services Task Force (USPSTF). Demographic and cancer risk/communication behavior responses of the four screening populations (both tests, one test, no tests) were analyzed with multinomial regression for all comparisons.

Results:

Of the 394 respondents, 54% were up to date for both cervical and CRC screening, 21% were up to date with only cervical cancer screening and 12% were up to date for only CRC screening. Of the 14 risk perception and communication behavior questions, only ‘Did your primary care physician (PCP) involve you in the decisions about your health care as much as you wanted?’ was significantly associated with women having both screens compared to only cervical cancer screening (aOR 1.67; 95% CI: 1.08, 2.57). The multivariate model showed age, and Middle East and North African (MENA) ethnicity and Black race, in addition to PCP-patient dyad decision-making to be associated with the cancer screenings women completed.

Conclusions:

Optimizing PCP-patient decision-making in health care may increase opportunities for both cervical cancer and CRC screening either in the office or by self-sampling. Understanding the effects of age and the different interventional strategies needed for MENA women compared to Black women will inform future intervention trials aimed to increase both cancer screenings.

Funding:

This work was supported by NIH through the Michigan Institute for Clinical and Health Research UL1TR002240 and by NCI through The University of Michigan Rogel Cancer Center P30CA046592-29-S4 grants.

Introduction

Women 50–65 years of age have the lowest cervical cancer screening rate among those 21–65 years of age, as reported in three national surveillance databases: Behavioral Risk Factor Surveillance System (BRFSS), Health Information National Trends Survey (HINTS), and Health Center Patient Survey (HCPS) (Harper et al., 2020). Likewise, women 50–65 years of age also have the lowest colorectal cancer (CRC) screening rates among those 50–75 years of age (American Cancer Society, 2020). Only half of the women in this age group are up to date for both screens, while 25% have cervical only, 12% CRC only, and 12% have neither screen (Harper et al., 2021a). As we have previously reported in the evaluation of women in southeast Michigan, those more likely to be up to date for only cervical cancer screening compared to both CRC and cervical cancer screens are younger women and women of Middle East and North African (MENA) ethnicity (Harper et al., 2021a).

MENA women in southeastern Michigan have become an at-risk hidden population for health care as there is no US Census category that represents their ethnicity on national surveillance health surveys. The Arab immigration to the Detroit area (known as the ‘Arab capital of North America’ [Crowder, 2015] happened over the past century in four waves [Semaan, 2015]) in part due to homeland wars and economic poverty, with an opportunity to relocate for better employment opportunities in the auto-industry in southeastern Michigan. Early Arab immigrants were Christian and primarily from Lebanon, Israel, and Syria (Cumoletti and Batalova, 2018) with later increased immigration by Muslim Arabs from primarily Palestine, Iraq, and Yemen. While Arab immigration to the United States has been increasing year over year since the 1950s, health care indicators such as cervical cancer screening, incidence, or mortality have only been reported for Arab Americans after the September 11, 2001, attacks when an Arab American identity was forged (Seeff and McKenna, 2003; Schwartz et al., 2004).

Over the past two decades, cervical cancer screening rates have improved among MENA women who were significantly less likely to ever receive cervical cancer screening in their lifetime than White women (Dallo and Kindratt, 2015; Yassine et al., 2010) to the most recent report showing near equivalency with White women (Harper et al., 2021aHarper et al., 2021b). However, while the cervical cancer screening prevalence for MENA women is similar to White women, participation in both CRC and cervical cancer screening differs for MENA women. Instead of half obtaining both cancer screenings, only 18% of MENA women are up to date for both, 50% have cervical cancer screening only, 7% CRC screening alone, and 20% report no screening at all (Harper et al., 2021a).

Personal risk perceptions, and cancer perceptions and knowledge as well as physician communication behaviors, are powerful influencers of screening uptake. In particular, fatalism about cancer leads to less participation in cancer screening programs among White, African American, and Hispanic race/ethnicities (Niederdeppe and Levy, 2007; Chavez et al., 1997). Greater perceptions of self-cancer risk relative to other people have been linked with an increased uptake of CRC screening (Zhao and Nan, 2016; Blalock et al., 1990, Atkinson et al., 2015).

In addition, physician communication behaviors may be a key modifiable determinate of cancer screening behaviors (Peterson et al., 2016) so much so that increasing communication behavior is included as a major objective in the Healthy People 2030 guidelines (HP 2030 Guidelines, 2020). Simple physician recommendations for a prevention intervention have been successful in increasing human papillomavirus (HPV) vaccine uptake (Kempe et al., 2019), CRC screening (Hudson et al., 2012; Atkinson et al., 2015), and cervical cancer screening (Luque et al., 2018). But more specific communication patterns that examine involvement in health care decisions and dealing with women’s feelings of uncertainty might better deconstruct the association with the cancer type and its related screening behaviors.

The primary aim of our study was to determine if CRC and/or cervical cancer screening uptake among the multiethnic sample of White, Black, and MENA women 50–65 years of age from southeast Michigan was associated with personal risk perceptions, cancer risk perceptions, and physician communication behavior.

Materials and methods

Survey measures

Request a detailed protocol

A cross-sectional health survey that included screening determinants questions from the NCI HINTS 5 study was developed and piloted for southeast Michigan (Harper et al., 2021a). Only women 50–65 years of age who answered questions about cervical cancer and CRC screening were included in this study. Questions about hysterectomy and colectomy were not included in the survey. Prior demographic descriptors significant to this screening population were included in the analyses (Harper et al., 2021aHP 2030 Guidelines, 2020).

The respondent’s perceptions of the communication behaviors of her primary care physician (PCP) were assessed with seven items adapted from the Consumer Assessment of Health Plans Study (Marshall et al., 2001). These include questions such as ‘Did your primary care physician give you the chance to ask all the health-related questions you had?’ on a four-point Likert scale (1-4): always, usually, sometimes, and never. The risk perception and knowledge section of HINTS were developed from the Health Belief Model, the Precaution Adoption Model, the Transactional Model of Stress and Coping, the Self-Regulation Model of Health Behavior, and the Protection Motivation Theory (Vernon, 1999Table 1). The seven questions included items such as ‘When I think about cancer, I automatically think about death’ which were also scored on a four-point Likert scale (1-4): strongly disagree, somewhat disagree, somewhat agree, strongly agree. While these two sets of seven questions are often aggregated for a summary score, we found the detailed questions provided more insight about the screening behaviors.

Table 1
Survey questions.
Cancer risk perception and knowledge
 Cancer is most often caused by a person’s behavior or lifestyle
 It seems like everything causes cancer
 I'd rather not know my chance of getting cancer
 There’s not much you can do to lower your chances of getting cancer
 There are so many recommendations about cancer prevention, it’s hard to know which ones to follow
 When I think about cancer, I automatically think about death
 How likely are you to get cancer in your lifetime compared to others your age?
Physician communication behavior – In the past 12 months, how often did your PCP …
 Give you the chance to ask all health-related questions
 Give you the attention you needed to your feelings/emotions
 Involve you in the decisions about your health care as much as you wanted
 Make sure you understood the things you needed to do to take care of your health
 Explain things in a way you could understand
 Spend enough time with you
 Help you deal with feelings of uncertainty about your health or health care

For the MENA sample, the survey was distributed at 12 diverse settings of MENA interest across three southeast Michigan counties. MENA women completed the survey using online forms or paper forms, in Arabic or English, and at home or by interviewer. The remaining sample was obtained via telephone and online methods. A landline or cellphone random dial phone interview conducted by Harris Interactive Inc sampled White and African American/Black adults in the Rogel Cancer Center catchment area, which comprises about two-thirds of the state. An online sample was recruited through Dynata which oversampled African American/Black residents in the catchment area. All respondents received an incentive at completion.

Screening outcome measures

Request a detailed protocol

Cervical cancer screening was considered up to date if the respondent indicated a Pap and/or HPV test within the past 3 years (US Preventive Services Task Force et al., 2018). For CRC screening, up to date included fecal occult blood test or fecal immunochemistry test within the past year or a colonoscopy within the past 10 years in accordance with the United States Preventive Services Task Force (USPSTF) guidelines (US Preventive Services Task Force et al., 2016).

Statistical analysis

Request a detailed protocol

The four screening populations were defined as those women up to date with both CRC and cervical cancer screenings, with cervical cancer screening alone, with CRC screening alone or neither screening test. The demographic data were analyzed by means and frequencies. The answers to the communication behavior questions and the risk-related questions were collapsed (always/usually vs. sometimes/never) and dichotomized (strongly/somewhat disagree vs. somewhat/strongly agree) for Kruskal–Wallis analysis across the four screening populations. Multinomial logistic regression modeling was used to determine the significantly different communication and risk predictors for the four screening populations adjusting for the demographic variables. It is important to this study that comparator screenings were more than the no screening option, which has been the most commonly reported method of analysis. It is clear from the screening rates that women screen for CRC and cervical cancer differently. Allowing the analysis to include each of these screens alone as a comparator may provide insights into associations with the differential screening outcomes. Statistical analysis was completed with Statistica v 13.2 (Dell Inc, 2016).

Results

There were 394 women, average age of 57.7 years (SD 4.5), who completed the CRC and cervical cancer screening questions (Table 2). Women declared their race/ethnicity as 69% White, 19% Black, 9% MENA, and 3% others including Asian/Pacific islander, Hispanic, American Indian, and multi-race. This distribution is very similar to the population distribution of southeastern Michigan, with some over-representation of MENA women (Detroit-Warren-Dearborn MI Census Reporter, 2019). The majority of the study population were Christian, had a college education, were married, had a chronic disease, and had an income between $50,000 and $99,999 also consistent with the median income for southeast Michigan (Detroit-Warren-Dearborn MI Census Reporter, 2019).

Table 2
Demographic descriptors of all screening populations.
Age, years (mean, SD)57.74.5
Race/ethnicityN%
 White27168.8
 African American/Black7519.0
 Other*112.8
 MENA (Middle Eastern North African)379.4
Religion
 Christian24163.6
 Jewish143.7
 Muslim266.9
 No religion/no belief in god9825.9
Education
 High school or less8521.6
 Some college12331.2
 College graduate13835.0
 Post college education4812.2
Marital status
 Married/partnered23659.9
 Single15639.6
Income
 <$10,000287.1
 $10–49,99914837.6
 $50–99,99914035.5
 ≥$100,0007017.8
Any chronic condition§29173.9
  1. *Other includes Asian or other Pacific Islander, Hispanic, American Indian, and multi-race. This group was further excluded from our analyses.

    Christian includes Chaldean, Coptic, Assyrian, Catholic, Jehovah Witness.

  2. Married/partnered includes married and living as married; single includes divorced, widowed, separated, and never been married.

    §Chronic conditions include diabetes, hypertension, heart disease (heart attack, angina, congestive heart failure), lung disease (asthma, emphysema, chronic bronchitis), arthritis, depression, or cancer.

Religious affiliation among White women was 65% Christian (including Chaldean, Catholic, Assyrian, Coptic, and Jehovah Witness), 1% Muslim, 14% Jewish, and 29% no religion/no belief in god. Among Black women only two religions were reported, 77% Christian, and 23% no religion/no belief in god. Among MENA women 34% reported being Christian and 66% reported being Muslim.

Among all women, those declaring Muslim religion participated significantly less often in dual cervical cancer and CRC screening than women of other religions (p<0.05, Supplementary file 1 – 'Screening outcomes by total population religion and by MENA ethnicity'). However, among MENA women, neither Muslim nor Christian religions were associated with preferential cervical cancer or CRC screening, with both religions having the greatest percentages in the cervical cancer-only screening group.

Moreover, women of different religions had significantly different proportions of agreement for the cancer risk perceptions and the physician communication behaviors, where Muslim women among all religions had the lowest agreement that she was involved with her physician in decisions about her health care as much as she wanted (Supplementary file 2 – 'Agreement with cancer risk perceptions and physician communication behaviors by self-declared religion'). Supplementary file 3 – 'Cancer screenings by religion and race/ethnicity' shows that MENA ethnicity, not a religious affiliation, is significantly associated with cervical cancer and/or CRC screening patterns.

Cancer risk perceptions and knowledge among all women

Women in the four screening groups had significantly different proportions of agreement for cancer risk perceptions and knowledge (Table 3).

Table 3
Risk perceptions and cancer knowledge by screening categories.
Both cervical cancer and CRC screeningNeither screeningCervical cancer screening aloneCRC screening alone
N=394N=213N=52N=82N=47
Cancer risk perceptions and knowledgeN%N%N%N%
Cancer is caused by behavior or lifestyle
 Strongly/somewhat disagree12860.1%2958.0%4558.4%3063.8%
 Somewhat/strongly agree8539.9%2142.0%3241.6%1736.2%
Everything causes cancer
 Strongly/somewhat disagree7434.7%1529.4%2835.9%1634.0%
 Somewhat/strongly agree13965.3%3670.6%5064.1%3166.0%
I’d rather not know my chance of getting cancer
 Strongly/somewhat disagree16075.8%3261.5%5368.8%3880.9%
 Somewhat/strongly agree5124.2%1838.5%2431.2%919.2%
There’s not much you can do to lower your chances of getting cancer (1)
 Strongly/somewhat disagree18084.9%3567.3%5977.6%3676.6%
 Somewhat/strongly agree3215.1%1732.7%1722.4%1123.4%
There are so many recommendations about cancer prevention, it’s hard to know which ones to follow
 Strongly/somewhat disagree7434.9%1325.0%1924.7%1327.7%
 Somewhat/strongly agree13865.1%3975.0%5875.3%3472.3%
When I think about cancer, I automatically think about death (2)
 Strongly/somewhat disagree11051.9%2848.1%2633.8%1940.4%
 Somewhat/strongly agree10248.1%2751.9%5166.2%2859.6%
How likely are you to get cancer in your lifetime compared to others your age?
 Much less, less likely or about the same17783.5%4790.4%6583.3%3574.5%
 More or much more likely3516.5%59.6%1316.7%1225.5%
  1. (1) p<0.05 [H(3, N=387)=9.04].

    (2) p<0.05 [H(3, N=388)=8.22].

In particular, those with neither cancer screening had the highest agreement (strongly/somewhat) rate (33%) for the fatalism construct of ‘There’s not much you can do to lower your chances of getting cancer’ (p=0.03 H(3,N=387) = 9.04), whereas those who had completed both CRC and cervical cancer screening had the lowest agreement rate (15%). Likewise, ‘When I think about cancer, I automatically think about death’ had the lowest agreement rate (48%) among those who had completed both CRC and cervical cancer screening (p=0.04 H(3,N=388)=8.22) and the highest among those who had completed only cervical cancer screening (66%).

Communication behavior among all women

Dually screened women had significantly more positive rankings of the PCP communications of any screened population. The difference in the rankings of the PCP communication behaviors is seen in the comparator group where compared to a single screen, especially cervical cancer screening alone, women agreed significantly more often that positive PCP communication behavior was supportive for her in six of the seven questions (Table 4).

Table 4
Primary care physician communication behaviors associated with the screening categories.
Both cervical cancer and CRC screeningNeither screeningCervical cancer screening aloneCRC screening alone
N=394N=213N=52N=82N=47
Communication behaviors
Chance to ask all health-related questions (1)
 Always/usually19592.9%3394.3%6282.7%3786.1%
 Sometimes/never157.1%25.7%1317.3%614.0%
Give you the attention you needed to your feelings/emotions (2)
 Always/usually18890.8%3186.1%5473.0%3073.2%
 Sometimes/never199.2%513.9%2027.0%1126.8%
Involved you in the decisions about your health care as much as you wanted (3)
 Always/usually19192.7%2976.3%6081.1%3583.3%
 Sometimes/never157.3%923.7%1418.9%716.7%
Make sure you understood the things you needed to do to take care of your health (4)
 Always/usually19393.7%3694.7%6483.1%3786.1%
 Sometimes/never136.3%25.3%1316.9%614.0%
Explain things in a way you could understand
 Always/usually19293.7%3386.8%6584.4%3888.4%
 Sometimes/never136.3%513.2%1215.6%511.6%
Spend enough time with you (5)
 Always/usually18891.3%3186.1%5875.3%3479.1%
 Sometimes/never188.7%513.9%1924.7%920.9%
Help you deal with feelings of uncertainty about your health or health care (6)
 Always/usually16188.5%2887.5%5474.0%3075.0%
 Sometimes/never2111.5%412.5%1926.0%1025.0%
  1. (1) p<0.05 [H (3, N=363)=7.888].

    (2) p<0.001 [H (3, N=358)=17.98].

  2. (3) p<0.01 [H (3, N=360)=12.89].

    (4) p<0.05 [H (3, N=364)=9.207].

  3. (5) p<0.01 [H (3, N=362)=13.629].

    (6) p<0.05 [H (3, N=327)=10.49136].

The second comparator group, those completing neither screen, likewise showed that the dual screening group also ranked the seventh question ‘Has your PCP involved you in the decisions about your health care as much as you wanted?’ significantly higher than other screening groups.

Unadjusted multinomial modeling

All cancer risk perception statements were used to model the four cancer screening behaviors. Table 5 shows the significant associations.

Table 5
Predictors of screening by cancer risk perception and knowledge.
Both screens compared to neitherBoth compared to a cervix only
Cancer risk perception and knowledgeaORL95U95aORL95U95
There’s not much you can do to lower your chances of getting cancer
 Strongly/somewhat disagree1.651.162.351.180.851.66
 Somewhat/strongly agree1.01.0
When I think about cancer, I automatically think about death
 Strongly/somewhat disagree1.00.731.361.411.071.86
 Somewhat/strongly agree1.01.0
  1. Bold/red is significant.

    No other screening population comparison was significant.

  2. No other beliefs about cancer were significant.

    Adjusted for all cancer risk perception and knowledge questions.

Those women who had completed both CRC and cervical cancer screening compared to neither screen were significantly more likely to disagree that ‘there is not much you can do to lower your chances of getting cancer’ (aOR 1.65; 95% CI: 1.15, 2.35). On the other hand, the women completing both a CRC and cervical cancer screen compared to completing only the cervical cancer screen were significantly more likely to disagree that ‘when they think about cancer, they automatically think about death’ (aOR 1.41; 95% CI: 1.07, 2.86).

Among the PCP communication behaviors, only one behavior was significantly associated with screening outcomes in two comparator groups: ‘Did your PCP involve you in the decisions about your health care as much as you wanted?’ (Table 6).

Table 6
Communication behaviors predicting screening.
Both screenings compared to neitherBoth screenings compared to cervix only
Communication behavioraORL95U95aORL95U95
Did your PCP involve you in the decisions about your health care as much as you wanted
 Always/usually1.991.263.141.721.162.55
 Sometimes/never1.01.0
  1. PCP means primary care physician.

    Bold/red is significant.

  2. No other communication behavior was significant.

    No other screening population comparison was significant.

  3. Adjusted for all communication behavior questions.

Those women who had completed both CRC and cervical cancer screening compared to neither screen were significantly more likely to agree that ‘her PCP involved her in the decisions about her health care as much as she wanted’ (aOR=1.99; 95% CI: 1.26, 3.14). In addition, women who had both screens compared to only the cervical cancer screen were significantly more likely to also agree about her PCP-dyadic involvement in health care decision-making (aOR=1.72; 95% CI: 1.16, 2.55).

Multivariate multinomial modeling

When the multinomial population models were adjusted for age, race, and having any chronic disease, in addition to the cancer risk perceptions and the communication behaviors, three screening comparisons had significant predictors (Table 7).

Table 7
Communication behavior, age, and race predict screening.
Both screens compared to neitherBoth compared to cervix aloneCervix only compared to CRC alone
aORL95U95aORL95U95aORL95U95
Age1.050.971.141.131.061.210.840.770.93
Race/ethnicity
 White1.01.01.0
 Black1.740.873.461.761.033.000.490.201.24
 MENA0.330.150.700.300.160.564.591.1318.67
Did your PCP involve you in the decisions about your health care as much as you wanted?
 Always/usually1.961.223.151.581.042.421.010.591.73
 Sometimes/never1.01.01.0
  1. PCP means primary care physician.

    Bold/red is significant.

  2. Adjusted for age, race/ethnicity, and PCP-patient dyad decision-making.

None of the cancer risk perceptions were significant, nor were having any chronic diseases. The only significant predictor for women having both screens compared to neither screen was the PCP communication behavior to involve the woman as much as she wanted in her health care (PCP-patient dyad decision-making) (aOR=1.96; 95% CI: 1.22, 3.15). However, even after adjustment for the PCP-patient dyad decision-making communication style, being of MENA ethnicity, itself, was also significantly associated with less likelihood of dual vs. no screening compared to White women (aOR=0.33; 95% CI: 0.15,0.70).

Women who had both screens compared to only the cervical cancer screen were significantly older (aOR=1.13; 95% CI: 1.06, 1.21) and more involved with their PCP in decision-making (aOR=1.58; 95% CI: 1.04, 2.42). Compared to White women, Black women were significantly more likely to have both cancer screens compared to only the cervical cancer screen (aOR=1.76; 95% CI: 1.03, 3.00) and MENA women were significantly less likely (aOR = 0.30; 95% CI: 0.15, 0.56).

Women who had only a single cervical cancer screen compared to those with a single CRC screen were significantly younger (aOR=0.84; 95% CI: 0.77, 0.93). In addition, MENA women compared to White women were significantly more likely to only screen for cervical cancer compared to CRC cancer (aOR=4.59; 95% CI: 1.13, 18.67).

Discussion

There are significant gaps in the screening rates for both cervical cancer and CRC. We focused on understanding the many combinations of screening uptake that occur and found different predictors for dual or single CRC and cervical cancer screening among women in southeast Michigan. In prior work, having a chronic disease increased CRC or cervical cancer screening rates compared to no screen (Harper et al., 2021aCofie et al., 2018; Borrayo and O'Lawrence, 2016) and other studies have shown that cancer fatalism is a major barrier to screening uptake (Clarke et al., 2021; Jun and Oh, 2013). In this work, neither having chronic diseases nor cancer risk perceptions, such as cancer fatalism, had an influence on any cervical cancer or CRC screening behavior pattern. We did however find that positive PCP communication including the woman in her health care as much as she wants (PCP-patient dyad decision-making) was associated with completing both CRC and cervical cancer screening compared to no screen, but even more importantly also compared to only cervical cancer screening – the first new insight from this work.

When one screen alone was completed, women indicated no associated involvement by the PCP in their health care decisions, whereas if there is significant PCP-patient dyad decision-making, there is a higher likelihood that both screens are completed. This creates a hypothesis that comprehensive cancer screening may best be accomplished through a PCP-patient relationship and not through targeted single cancer site screenings. To date, in the United States, specific organ cancer screenings are directed by the respective specialties. For instance, gynecologists dominate the screening method and frequency recommendations for cervical cancer screening just as gastroenterologists dominate for CRC screening. Even the USPSTF guidelines, based in primary care, consider single cancer screenings alone, without the possibility of synergistic multi-organ screenings. In addition, single cancer screenings have been the mainstay of US population programs to date with some success (Wong and Miller, 2019; Pollack et al., 2020). But our results may indicate that a single cancer screen when done outside the PCP-patient relationship is perceived more as a test to be done rather than part of general wellbeing. We hypothesize a solution of home-based or self-screening options to target dual cervical cancer (primary HPV testing) and CRC (fecal immunochemical testing with DNA) screening to be used in the PCP-patient dyad decision-making process (El Khoury et al., 2021; Gorin et al., 2021). The self-screening option could also allow the general internist who does not provide gynecologic exams the ability to be effective in this PCP-patient dyad. Moreover, we hypothesize that the PCP-patient dyad relationship may further enhance follow-up after any abnormal screenings (Tsui et al., 2019; Peterson et al., 2016).

The second new insight from this work is that age continues to be an important predictor of these two screens, but in opposite directions. Within the 50–65 years’ age range, younger women participate more in cervical cancer screening and older women participate more in CRC screening. We hypothesize a two-level system solution: one patient-centric and one physician-centric. The patient-centric phenomenon is based in the woman’s choice of her PCP. Women who choose a gynecologist as the PCP have less opportunity for dual cancer screening options, as the physician relationship is likely to be focused on the reproductive years and end at the time of menopause in her fifties, and certainly by 65 when there are no further routine reasons for gynecologic care. Likewise, we hypothesize that women who choose a general internist as the PCP are more likely to focus on other medical issues as most general internists do not provide gynecologic exams. Furthermore, we hypothesize that women who have a family physician as a PCP are the group most likely to have both screens completed as both cancer preventions are part of the foundational PCP-patient relationship of family medicine care. While it is patient-centric to choose the PCP, it is physician-centric for test ordering. We hypothesize for future work that family medicine physicians order dual screening more often than gynecologists or general internists.

The third new insight is the documentation of the disparities in cancer screening among MENA women which are not well reported due to US Census race/ethnicity categories. Currently race/ethnicity must be indicated by having the respondent write in their MENA heritage rather than having it as a separate preidentified racial/ethnic category. We have shown in this work that MENA women are rarely screened for CRC, be it alone or in addition to cervical cancer screening. Past work also shows MENA women have little CRC screening (Harper et al., 2021a). In addition, we have shown that it is not religion itself, but rather the cultural ethnicity of MENA that is associated with the behaviors. We hypothesize that both physician and patient education in the MENA community about patient involvement in decision-making and the screens themselves will improve both cervical cancer and CRC screening.

Finally, we are the first to show for Black women in southeast Michigan that there are significant strong associations between the positive PCP-patient dyad in decision-making and the completion of both CRC and cervical cancer screening compared to cervical cancer screening alone even after adjustment for age. We hypothesize that the PCP-patient dyad for Black women occurs in patient-centered medical homes as the health care structure in which they seek care leading to dual screening.

In summary, this work shows that race/ethnicity, age, and physician communication behavior are three independent influencers associated with completing both CRC and cervical cancer screening compared to cervical cancer screening alone. This work provides a baseline for several future studies.

Strengths and limitations

Strengths

This study compared differential screenings for dual cervical cancer and CRC whereas all prior literature has only compared a single screening to no screening. In addition, these two specific internal organs (cervix and colorectum) were intentionally paired in a novel way because their barriers are quite similar: Risk factors for CRC screening in racially diverse populations include less than a high school education, low income, no health insurance, and no regular health care provider (Larsen et al., 2017; Lin et al., 2021). Other barriers to CRC screening include fatalism, religious beliefs, lack of self-worth, sexually related concerns, history of sexual abuse, and past negative experiences with screening (Alcalá et al., 2021; Knight et al., 2015). These are the same barriers that others have shown for cervical cancer screening (Cadman et al., 2012; Harper et al., 2020). This study provides the baseline data needed to explore screening uptake when both cervical cancer and CRC screenings are provided as an at-home screening option compared to in-office procedural appointments.

A second major strength of this study is the inclusion of MENA women who have little representation in translational clinical research. We have been able to show new deficits in CRC screening not previously identified for MENA women, establishing a baseline prevalence.

Moreover, our methodologic approach is a strength in that epidemiologic risk factors for more than one cancer screen do not exist; and those studies of a single cancer screening do not consider the association with communication and risk perception understanding. Understanding the differences for completion of one screen but not another is important to establish as prior screening strategies targeting a single screen may have missed the opportunity to present options for other cancer screenings. Finally, the population weighting previously done on this database (Harper et al., 2021a) provides generalizability of results for Southeast Michigan.

Limitations

This is a cross-sectional survey implemented in the community, by Internet and phone providing a onetime sampling of self-reported health outcomes; as such, it is not possible to infer causal relationships between constructs or items in the survey. Likewise, self-reported health outcomes are not equivalent to validated medical record or claims database verification (St Clair et al., 2017). Health outcomes are mistakenly overestimated by about half of those who have not received the screen, whereas they are accurately reported by those who are screened (Anderson et al., 2019; Bonafede et al., 2019). An additional limitation is the self-report nature of the health descriptors. There has not been any literature indicating the veracity of self-report descriptors including health states, education, income, or other health behaviors. Both health descriptors and outcomes have been linked to social acceptability and expected responses, as well as subject to linguistic misunderstandings either from language translation or meaning of the question itself, and inaccurate recall (St Clair et al., 2017). While this is a limitation, none of our covariates were significantly associated with any outcome after adjustments in the model.

Finally, our work applies to the United States and the health care structure it provides. These results may not be applicable to other national health care systems.

Conclusions

Optimizing PCP-patient decision-making in health care may increase opportunities for both cervical cancer and CRC screening either in the office or by self-sampling. Understanding the effects of age and the different interventional strategies needed for MENA women compared to Black women will inform future intervention trials aimed to increase both cancer screenings.

Data availability

The complete raw data-set collected in the study including the participants' data set, syntax file and log files for analysis has been uploaded to https://www.openicpsr.org/openicpsr/project/148081/version/V1/view.

The following data sets were generated
    1. Resnicow K
    (2021) openICPSR
    Population Health Assessment in Cancer Center Catchment Areas.
    https://doi.org/10.3886/E148081V1

References

  1. Book
    1. American Cancer Society
    (2020)
    Colorectal Cancer Facts & Figures 2020-2022
    Atlanta: American Cancer Society.
  2. Book
    1. Anderson J
    2. Bourne D
    3. Peterson K
    4. Mackey K
    (2019)
    Evidence Brief: Accuracy of Self-Report for Cervical and Breast Cancer Screening
    Washington (DC): Department of Veterans Affairs.
    1. Borrayo BD
    2. O'Lawrence H
    (2016)
    A post analysis of a preventive and chronic healthcare tool
    Journal of Health and Human Services Administration 39:15–40.
  3. Software
    1. Dell Inc
    (2016) Dell Statistica (Data Analysis Software System), version 13.2
    Dell Statistica (Data Analysis Software System).
    1. El Khoury C
    2. Haro E
    3. Alves M
    4. O'Dwyer MC
    5. Meixner K
    6. Albiac LC
    7. Capizzano JN
    8. Ramakrishnan M
    9. Salada C
    10. Jimbo M
    11. Gorin SS
    12. Sen A
    13. Harper DM
    (2021)
    Patient-centered home Cancer screening attitudes during SARS-CoV-2 pandemic
    Journal of Patient-Centered Research and Reviews In press.
  4. Report
    1. Harper DM
    2. Plegue M
    3. Sen A
    4. Gorin SS
    5. Jimbo M
    6. Patel MR
    7. Resnicow K
    (2021a)
    Predictors of Screening for Cervical and Colorectal Cancer in Women 50-65 Years Old in a Multi-Ethnic Population
    Preventive medicine reports.
  5. Report
    1. Lin JS
    2. Perdue LA
    3. Henrikson NB
    4. Bean SI
    5. Blasi PR
    (2021)
    Screening for Colorectal Cancer: An Evidence Update for the US. Preventive Services Task Force. Evidence Synthesis No. 202. AHRQ Publication No. 20-05271-EF-1
    Agency for Healthcare Research and Quality.
  6. Conference
    1. Semaan G
    (2015)
    The acculturation modes of arab americans: an empirical study on the effects of gender, religion, nationality and sojourner status
    Intercultural Communication Studies. pp. 174–191.
  7. Conference
    1. Yassine M
    2. Wing D
    3. Wojcik C
    4. Tan-Schriner C
    (2010)
    Special Cancer behavioral risk factor survey 2008 Michigan Department of Community Health
    Michigan Cancer Consortium.

Decision letter

  1. Wadih Arap
    Reviewing Editor; Rutgers Cancer Institute of New Jersey at University Hospital, United States
  2. Mone Zaidi
    Senior Editor; Icahn School of Medicine at Mount Sinai, United States
  3. Farah Usmani
    Reviewer; United Nations Population Fund (UNFPA), United States

Our editorial process produces two outputs: i) public reviews designed to be posted alongside the preprint for the benefit of readers; ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Acceptance summary:

The authors study demographic factors influencing the uptake of cervical and colorectal cancer screening among 50- to 65-years-old women. Towards those goals, they adapted and administered behavioral questions from the Health Information National Trends Survey to a multiethnic population sample in Southeast Michigan. Self-reported cancer screenings for the cancers (as defined by the USPSTF updated guidelines) served as the variable outcome. The study has public health merit in its identification of distinct predictors for cervical cancer and colorectal cancer screenings. The insights from this work on screening behavior differences among women, the perception of cancer risks and impact of positive provider communication, point to the need for exploring new ways for more holistic and integrated cancer prevention with a targeted focus.

Decision letter after peer review:

Thank you for submitting your article "Cancer risk perception and knowledge and physician communication behaviors: specific influences on cervical cancer and colorectal cancer screening in women 50-65 years old" for consideration by eLife. Your article has been reviewed by 2 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Mone Zaidi as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Farah Usmani (Reviewer #1).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

The Referees found the work of potential interest for publication in eLife but have also made specific recommendations for your consideration, which are summarized as follows:

1. In a general journal and for the readers that may not be essentially familiar with the demographics in the State of Michigan, some historical context should be given for the presence of a relatively large MENA population in the region.

2. In reference to the study finding that younger women participate more in the cervical cancer screenings while the older women participate more in the colorectal cancer screenings, the authors should examine hypotheses to deconstruct this empiric age-specific and/or cancer-specific division(s). For instance, a plausible reason for decreased uptake of cervical cancer screening by older women could perhaps be the perception that they have been menopausal and symptomless, and hence would be free from reproductive tract disorders including cancers.

3. If available, the data on the religious beliefs of the participants should be included in Table 1. Also, potential correlative differences in MENA women (e.g., between Muslim versus Christian) should be addressed in the Discussion.

4. As per the stylistic specifications of eLife, titles of eLife manuscripts should not use a colon and two-part titles should also be avoided. Please revise your title with this advice in mind.

Reviewer #1 (Recommendations for the authors):

With reference to the finding that younger women participate more in the cervical cancer screen and older women participate more in the CRC screen, the authors may like to examine hypotheses to deconstruct the age divide in cancer screening.

A possible reason for decreased uptake of cervical cancer screening by older women could be the perception that they have been menopausal for some years now (and symptomless), and hence would be free from reproductive tract disorders including cancers.

Reviewer #3 (Recommendations for the authors):

In a general journal and for the readers that may not be familiar with Michigan demographics, perhaps some historical context should be given for the presence of a relatively large MENA population in the region.

The transition from a multi-ethnic population to African-American/Black and MENA is not very clear in the manuscript. Recommend some text revision in the Introduction and/or Discussion so that the non-epidemiology readers may follow it.

Would it be worth to remind GPs and family medicine providers that pelvic should be an integral part of the physical exam?

Was the role for religious beliefs investigated? For example, what were there differences, if any, between Muslin versus Christian MENA women? Otherwise a central investigational opportunity for understanding and intervention in this setting has been missing.

Could the authors speculate on the reason(s) for their finding of the age- and cancer-related in the screenings?

Ref. 1. There is a typo in the name of Dr. Sheri Seinfeld Gorin.

https://doi.org/10.7554/eLife.70003.sa1

Author response

Essential revisions:

The Referees found the work of potential interest for publication in eLife but have also made specific recommendations for your consideration, which are summarized as follows:

1. In a general journal and for the readers that may not be essentially familiar with the demographics in the State of Michigan, some historical context should be given for the presence of a relatively large MENA population in the region.

We are happy to provide a brief historical context to this important population in Michigan. We have provided two paragraphs in the Introduction that provides a better perspective of the MENA immigration in southeast Michigan

2. In reference to the study finding that younger women participate more in the cervical cancer screenings while the older women participate more in the colorectal cancer screenings, the authors should examine hypotheses to deconstruct this empiric age-specific and/or cancer-specific division(s). For instance, a plausible reason for decreased uptake of cervical cancer screening by older women could perhaps be the perception that they have been menopausal and symptomless, and hence would be free from reproductive tract disorders including cancers.

We appreciate the reviewers' interest in this age dichotomy within our internal age range. This is a first notification of this health gap that we are happy to have established in the literature. Our Discussion section has been rewritten to include the third paragraph which is devoted to the hypothesis generating questions that come from our results.

In the US, our first hypothesis is that this phenomenon is driven from the provider side, not the patient side. Hence, gynecologists, general internists and family medicine physicians are the front line physicians for cancer screenings.

Women in the US have the opportunity to declare any one of these specialties as their Primary Care Physician (PCP).

Our hypothesis is that the woman stays with her gyn if she started early in her reproductive life with a gyn as her PCP, and the gyn is focused on cervical cancer screening through the age of 65 where it is no longer recommended. Likewise, the woman who chooses a general internist for her PCP usually is referred to a gyn for the pelvic exam as this is not in the scope of practice for most general internists — nor is their office equipped for the requirements for chaperoning, and the equipment needs of a table with stirrups, specula and sampling devices. In addition, as women gather new medical diseases, such as hypertension, diabetes, COPD, rheumatoid arthritis, she is more likely to switch to the general internist providing therapeutic care as her current PCP. This is often around the age of 50, where the focus of the health care is on the already established diseases and not on screening for new cancers, especially not those that require a speculum exam.

On the other hand, family medicine physicians practice the full scope of care and are in the position to provide both gyn and GI cancer screenings. Our next project will be to evaluate the provider type who orders the screening tests from the large claims databases such as Medicare or Optum to see where the correlation between physician type and dual or single cancer screens occurs.

3. If available, the data on the religious beliefs of the participants should be included in Table 1. Also, potential correlative differences in MENA women (e.g., between Muslim versus Christian) should be addressed in the Discussion.

We are grateful for this suggestion as did collect this information and were able to reconstruct the Results section with this perspective.

Three supplementary tables have been added to the Results section. Supplementary table I shows that while overall Muslims screen differently than all other religions, the MENA women do not show differences in screening by Muslim vs Christian religion. We describe in the 4th paragraph of the Discussion section that this reflects on the culture of MENA being more important than the religion.

Supplemental table 2 shows the multivariate multinomial modeling of religion vs.

race/ethnicity to predict dual, single or neither screening. This table shows that screening is not associated with religion when the model is adjusted for both religion and race/ethnicity. This supports the 4th paragraph of the Discussion section.

Supplemental table 3 shows the differences in cancer risk perception and physician communication behaviors by religion. This table shows that Christians and Jewish women have a significantly higher agreement that there is not much you can do to lower your chances of getting cancer than do Muslim women. Similarly, when the Jewish women think about cancer, a large portion automatically think about death, es eciall com ared to Muslim women. Finally, Muslim women are the least to be involved with their PCP in a dyad decision making relationship. These results are incorporated into Table 6 where adjustments for religion, race, age and all univariate predictors were made. The results of Table 6 show that the culture of MENA women overrides the religion in significant prediction for cancer screenings.

4. As per the stylistic specifications of eLife, titles of eLife manuscripts should not use a colon and two-part titles should also be avoided. Please revise your title with this advice in mind.

We have changed the title of the manuscript to be in accordance with eLife stylistic policy.

Reviewer #3 (Recommendations for the authors):

In a general journal and for the readers that may not be familiar with Michigan demographics, perhaps some historical context should be given for the presence of a relatively large MENA population in the region.

See response to Essential revisions #1.

The transition from a multiethnic population to African-American/Black and MENA is not very clear in the manuscript.

We have rewritten the manuscript to clarify that our study population is comprised of White, Black and MENA women only.

Recommend some text revision in the Introduction and/or Discussion so that the non-epidemiology readers may follow it.

Would it be worth to remind GPs and family medicine providers that pelvic should be an integral part of the physical exam?

We appreciate this thought, but evidence has evolved to show that the sensitivity and specificity of a pelvic exam does not provide any information to the PCP that would be diagnostic, hence the practice of pelvic exams routinely has been abandoned.

Was the role for religious beliefs investigated? For example, what were there differences, if any, between Muslin versus Christian MENA women? Otherwise a central investigational opportunity for understanding and intervention in this setting has been missing.

Including the variable of religion into the analysis has been very important to be able to state clearly that the results are culturally and ethnically derived, not by religion.

Could the authors speculate on the reason(s) for their finding of the age- and cancer-related in the screenings?

We agree that these two aspects of our results provide us with several hypothesis-generating studies for the future. We have hypothesized that the age dichotomy is generated from the woman's characteristics — if she has chosen a gyn as a PCP she is likely to get a cervical cancer screen in her 50's but the patient will be gone to her practice by the time she is 65; similarly if the woman has chosen a general internist for her PCP, it is unlikely that she will have a cervical cancer screen between 50 and 65 but will have a CRC screen.

Likewise we hypothesized that the dual screening vs single vs no screenings comes from the type of physician ordering the test rather than by a patient's characteristics, as a general internist is unlikely to remember to refer out for the cervical cancer screening and cannot do it in the office. We hypothesize that home based cervical and CRC screenings would alleviate the issue of in office screening.

Ref. 1. There is a typo in the name of Dr. Sheri Seinfeld Gorin.

Dr. Gorin confirmed the spelling of her name in the reference section.

https://doi.org/10.7554/eLife.70003.sa2

Article and author information

Author details

  1. Diane M Harper

    1. Department of Family Medicine, University of Michigan School of Medicine, Ann Arbor, United States
    2. Department of Obstetrics & Gynecology, University of Michigan School of Medicine, Ann Arbor, United States
    3. Department of Women’s Studies, University of Michigan, College of Literature, Science and the Arts, Ann Arbor, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Writing - original draft, Writing - review and editing
    For correspondence
    harperdi@med.umich.edu
    Competing interests
    Senior editor, eLife
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7648-883X
  2. Madiha Tariq

    Arab Community Center for Economic and Social Services (ACCESS), Dearborn, United States
    Contribution
    Project administration, Writing - review and editing
    Competing interests
    No competing interests declared
  3. Asraa Alhawli

    Arab Community Center for Economic and Social Services (ACCESS), Dearborn, United States
    Contribution
    Project administration, Writing - review and editing
    Competing interests
    No competing interests declared
  4. Nadia Syed

    Arab Community Center for Economic and Social Services (ACCESS), Dearborn, United States
    Contribution
    Project administration, Writing - review and editing
    Competing interests
    No competing interests declared
  5. Minal Patel

    1. Department of Health Behavior & Health Education, University of Michigan School of Public, Ann Arbor, United States
    2. Outreach and Health Disparities Research, University of Michigan Rogel Cancer Center, Ann Arbor, United States
    Contribution
    Conceptualization, Funding acquisition, Investigation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  6. Ken Resnicow

    1. Outreach and Health Disparities Research, University of Michigan Rogel Cancer Center, Ann Arbor, United States
    2. Center for Health Communications Research, University of Michigan, School of Public Health, Ann Arbor, United States
    Contribution
    Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared

Funding

National Cancer Institute (P30CA046592-29-S4)

  • Ken Resnicow

National Center for Advancing Translational Sciences (UL1TR002240)

  • Diane M Harper

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This work was supported by NIH through the Michigan Institute for Clinical and Health Research UL1TR002240 and by NCI through the University of Michigan Rogel Cancer Center P30CA046592-29-S4 grants.

Ethics

Human subjects: This study was approved by Protocol #: HUM00159558 approved March 10, 2019 by the University of Michigan.

Senior Editor

  1. Mone Zaidi, Icahn School of Medicine at Mount Sinai, United States

Reviewing Editor

  1. Wadih Arap, Rutgers Cancer Institute of New Jersey at University Hospital, United States

Reviewer

  1. Farah Usmani, United Nations Population Fund (UNFPA), United States

Publication history

  1. Received: May 3, 2021
  2. Preprint posted: May 7, 2021 (view preprint)
  3. Accepted: July 29, 2021
  4. Version of Record published: August 24, 2021 (version 1)

Copyright

© 2021, Harper et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 240
    Page views
  • 21
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Epidemiology and Global Health
    2. Genetics and Genomics
    Jiahui Si et al.
    Research Article

    Background: Identifying environmentally responsive genetic loci where DNA methylation is associated with coronary heart disease (CHD) may reveal novel pathways or therapeutic targets for CHD. We conducted the first prospective epigenome-wide analysis of DNA methylation in relation to incident CHD in the Asian population.

    Methods: We did a nested case-control study comprising incident CHD cases and 1:1 matched controls who were identified from the 10-year follow-up of the China Kadoorie Biobank. Methylation level of baseline blood leukocyte DNA was measured by Infinium Methylation EPIC BeadChip. We performed the single cytosine-phosphate-guanine (CpG) site association analysis and network approach to identify CHD-associated CpG sites and co-methylation gene module.

    Results: After quality control, 982 participants (mean age 50.1 years) were retained. Methylation level at 25 CpG sites across the genome was associated with incident CHD (genome-wide false discovery rate [FDR] < 0.05 or module-specific FDR <0.01). One SD increase in methylation level of identified CpGs was associated with differences in CHD risk, ranging from a 47% decrease to a 118% increase. Mediation analyses revealed 28.5% of the excessed CHD risk associated with smoking was mediated by methylation level at the promoter region of ANKS1A gene (P for mediation effect = 0.036). Methylation level at the promoter region of SNX30 was associated with blood pressure and subsequent risk of CHD, with the mediating proportion to be 7.7% (P = 0.003) via systolic blood pressure and 6.4% (P = 0.006) via diastolic blood pressure. Network analysis revealed a co-methylation module associated with CHD.

    Conclusions: We identified novel blood methylation alterations associated with incident CHD in the Asian population and provided evidence of the possible role of epigenetic regulations in the smoking- and BP-related pathways to CHD risk.

    Funding: This work was supported by National Natural Science Foundation of China (81390544 and 91846303). The CKB baseline survey and the first re-survey were supported by a grant from the Kadoorie Charitable Foundation in Hong Kong. The long-term follow-up is supported by grants from the UK Wellcome Trust (202922/Z/16/Z, 088158/Z/09/Z, 104085/Z/14/Z), grant (2016YFC0900500, 2016YFC0900501, 2016YFC0900504, 2016YFC1303904) from the National Key and Program of China, and Chinese Ministry of Science and Technology (2011BAI09B01).

    1. Epidemiology and Global Health
    2. Microbiology and Infectious Disease
    Samuel Kariuki et al.
    Research Article

    Background: Understanding the dynamics of infection and carriage of typhoid in endemic settings is critical to finding solutions to prevention and control.

    Methods: In a 3 year case-control study, we investigated typhoid among children aged <16 years (4,670 febrile cases and 8,549 age matched controls) living in an informal settlement, Nairobi, Kenya.

    Results: 148 S. Typhi isolates from cases and 95 from controls (stool culture) were identified; a carriage frequency of 1%. Whole-genome sequencing showed 97% of cases and 88% of controls were genotype 4.3.1 (Haplotype 58), with the majority of each (76% and 88%) being multidrug-resistant strains in 3 sublineages of H58 genotype (East Africa 1 (EA1), EA2, and EA3), with sequences from cases and carriers intermingled.

    Conclusions: The high rate of multidrug-resistant H58 S.Typhi, and the close phylogenetic relationships between cases and controls, provides evidence for the role of carriers as a reservoir for the community spread of typhoid in this setting.

    Funding: National Institutes of Health (R01AI099525); Wellcome Trust (106158/Z/14/Z); European Commission (TyphiNET No 845681); National Institute for Health Research (NIHR); Bill and Melinda Gates Foundation (OPP1175797).