Domains of Excellence: A CDC Framework for Developing High-Quality, Impact-Driven Public Health Science Publications.

CONTEXT: The Centers for Disease Control and Prevention (CDC) has a long history of using high-quality science to drive public health action that has improved the health, safety, and well-being of people in the United States and globally. To ensure scientific quality, manuscripts authored by CDC staff are required to undergo an internal review and approval process known as clearance. During 2022, CDC launched a scientific clearance transformation initiative to improve the efficiency of the clearance process while ensuring scientific quality. PROGRAM: As part of the scientific clearance transformation initiative, a group of senior scientists across CDC developed a framework called the Domains of Excellence for High-Quality Publications (DOE framework). The framework includes 7 areas ("domains") that authors can consider for developing high-quality and impactful scientific manuscripts: Clarity, Scientific Rigor, Public Health Relevance, Policy Content, Ethical Standards, Collaboration, and Health Equity. Each domain includes multiple quality elements, highlighting specific key considerations within. IMPLEMENTATION: CDC scientists are expected to use the DOE framework when conceptualizing, developing, revising, and reviewing scientific products to support collaboration and to ensure the quality and impact of their scientific manuscripts. DISCUSSION: The DOE framework sets expectations for a consistent standard for scientific manuscripts across CDC and promotes collaboration among authors, partners, and other subject matter experts. Many aspects have broad applicability to the public health field at large and might be relevant for others developing high-quality manuscripts in public health science. The framework can serve as a useful reference document for CDC authors and others in the public health community as they prepare scientific manuscripts for publication and dissemination.

CDC's Guiding Principles to Promote an Equity-Centered Approach to Public Health Communication.

A public health practitioner's mission is to protect and promote the health of all people in all communities. Components of being successful in that mission include understanding who is at risk of negative outcomes, identifying effective actions to promote and protect health, and communicating information accordingly. Information must be scientifically rigorous, provide appropriate contextualizing information, and refer to and visually represent people through words and images in respectful ways. Public health communication objectives include that the audience accepts, understands, and acts on the information to protect and promote health. This article describes the impetus for, development of, and public health applications and implications of principles to guide communication efforts. CDC's Health Equity Guiding Principles for Inclusive Communication is a web-based resource published in August 2021 that offers - but does not mandate - guidance and recommendations for public health practice. The resource can help public health practitioners and their partners consider social inequities and diversity, think more inclusively about the people they serve, and adapt to the cultural, linguistic, environmental, and historical situation of each population or audience of focus. Users are encouraged to have conversations about the Guiding Principles as they plan and develop communication products and strategies in collaboration with communities and partners and build a shared vocabulary consistent with how communities and groups of focus see and understand themselves, because words matter. As the public health field renews its focus on shifting the paradigm toward equity, a language and narrative shift is a vital intervention.

Leveraging Social and Structural Determinants of Health at the Centers for Disease Control and Prevention: A Systems-Level Opportunity to Improve Public Health.

CONTEXT: Social and structural determinants of health (SDOH) have become part of the public health and health care landscape. The need to address SDOH is reinforced by morbidity and mortality trends, including a recent multiyear decrease in life expectancy and persistent health disparities. Leadership on SDOH-related efforts has come from public health, health care, private philanthropy, and nongovernmental entities. STRATEGY: The Centers for Disease Control and Prevention (CDC) has been addressing SDOH through both disease- or condition-specific programs and crosscutting offices. Guidance from public health partners in the field has led the CDC to consider more strategic approaches to incorporating SDOH into public health activities. IMPLEMENTATION: The CDC's crosscutting SDOH Workgroup responded to external recommendations to develop a specific vision and plan that aims to integrate SDOH into the agency's infrastructure. The group also sponsors CDC forums for sharing research and trainings on embedding SDOH in programs. The group created a Web site to centralize CDC SDOH research, data sources, practice tools, programs, and policies. PROGRESS: The CDC has shown strong leadership in prioritizing SDOH in recent years. Individual programs and crosscutting offices have developed various models aimed at ensuring that public health research and practice address SDOH. DISCUSSION: Building sustainable SDOH infrastructures in public health institutions that reach across multiple health topics and non-health organizations could increase chances of meeting public health morbidity and mortality reduction goals, including decreasing health disparities. Although public health priorities and socioeconomic trends will change over time, experience suggests that social and structural factors will continue to influence the public's health. The CDC and state, tribal, local, and territorial public health institutions have played important leadership roles in the system of community and service organizations that interface with communities they mutually serve to address SDOH. Continued capacity-building could help grow and sustain an SDOH infrastructure that advances this work.

Racial and Ethnic Disparities in Incidence of SARS-CoV-2 Infection, 22 US States and DC, January 1-October 1, 2020.

We examined disparities in cumulative incidence of severe acute respiratory syndrome coronavirus 2 by race/ethnicity, age, and sex in the United States during January 1-October 1, 2020. Hispanic/Latino and non-Hispanic Black, American Indian/Alaskan Native, and Native Hawaiian/other Pacific Islander persons had a substantially higher incidence of infection than non-Hispanic White persons.

Racial and Ethnic Disparities in COVID-19 Incidence by Age, Sex, and Period Among Persons Aged <25 Years - 16 U.S. Jurisdictions, January 1-December 31, 2020.

The COVID-19 pandemic has disproportionately affected racial and ethnic minority groups in the United States. Whereas racial and ethnic disparities in severe COVID-19-associated outcomes, including mortality, have been documented (1-3), less is known about population-based disparities in infection with SARS-CoV-2, the virus that causes COVID-19. In addition, although persons aged <30 years account for approximately one third of reported infections,§ there is limited information on racial and ethnic disparities in infection among young persons over time and by sex and age. Based on 689,672 U.S. COVID-19 cases reported to CDC's case-based surveillance system by jurisdictional health departments, racial and ethnic disparities in COVID-19 incidence among persons aged <25 years in 16 U.S. jurisdictions¶ were described by age group and sex and across three periods during January 1-December 31, 2020. During January-April, COVID-19 incidence was substantially higher among most racial and ethnic minority groups compared with that among non-Hispanic White (White) persons (rate ratio [RR] range = 1.09-4.62). During May-August, the RR increased from 2.49 to 4.57 among non-Hispanic Native Hawaiian and Pacific Islander (NH/PI) persons but decreased among other racial and ethnic minority groups (RR range = 0.52-2.82). Decreases in disparities were observed during September-December (RR range = 0.37-1.69); these decreases were largely because of a greater increase in incidence among White persons, rather than a decline in incidence among racial and ethnic minority groups. NH/PI, non-Hispanic American Indian or Alaska Native (AI/AN), and Hispanic or Latino (Hispanic) persons experienced the largest persistent disparities over the entire period. Ensuring equitable and timely access to preventive measures, including testing, safe work and education settings, and vaccination when eligible is important to address racial/ethnic disparities.

Counties with High COVID-19 Incidence and Relatively Large Racial and Ethnic Minority Populations - United States, April 1-December 22, 2020.

Long-standing systemic social, economic, and environmental inequities in the United States have put many communities of color (racial and ethnic minority groups) at increased risk for exposure to and infection with SARS-CoV-2, the virus that causes COVID-19, as well as more severe COVID-19-related outcomes (1-3). Because race and ethnicity are missing for a proportion of reported COVID-19 cases, counties with substantial missing information often are excluded from analyses of disparities (4). Thus, as a complement to these case-based analyses, population-based studies can help direct public health interventions. Using data from the 50 states and the District of Columbia (DC), CDC identified counties where five racial and ethnic minority groups (Hispanic or Latino [Hispanic], non-Hispanic Black or African American [Black], non-Hispanic Asian [Asian], non-Hispanic American Indian or Alaska Native [AI/AN], and non-Hispanic Native Hawaiian or other Pacific Islander [NH/PI]) might have experienced high COVID-19 impact during April 1-December 22, 2020. These counties had high 2-week COVID-19 incidences (>100 new cases per 100,000 persons in the total population) and percentages of persons in five racial and ethnic groups that were larger than the national percentages (denoted as "large"). During April 1-14, a total of 359 (11.4%) of 3,142 U.S. counties reported high COVID-19 incidence, including 28.7% of counties with large percentages of Asian persons and 27.9% of counties with large percentages of Black persons. During August 5-18, high COVID-19 incidence was reported by 2,034 (64.7%) counties, including 92.4% of counties with large percentages of Black persons and 74.5% of counties with large percentages of Hispanic persons. During December 9-22, high COVID-19 incidence was reported by 3,114 (99.1%) counties, including >95% of those with large percentages of persons in each of the five racial and ethnic minority groups. The findings of this population-based analysis complement those of case-based analyses. In jurisdictions with substantial missing race and ethnicity information, this method could be applied to smaller geographic areas, to identify communities of color that might be experiencing high potential COVID-19 impact. As areas with high rates of new infection change over time, public health efforts can be tailored to the needs of communities of color as the pandemic evolves and integrated with longer-term plans to improve health equity.

Disaggregating Data to Measure Racial Disparities in COVID-19 Outcomes and Guide Community Response - Hawaii, March 1, 2020-February 28, 2021.

Native Hawaiian and Pacific Islander populations have been disproportionately affected by COVID-19 (1-3). Native Hawaiian, Pacific Islander, and Asian populations vary in language; cultural practices; and social, economic, and environmental experiences,† which can affect health outcomes (4).§ However, data from these populations are often aggregated in analyses. Although data aggregation is often used as an approach to increase sample size and statistical power when analyzing data from smaller population groups, it can limit the understanding of disparities among diverse Native Hawaiian, Pacific Islander, and Asian subpopulations¶ (4-7). To assess disparities in COVID-19 outcomes among Native Hawaiian, Pacific Islander, and Asian populations, a disaggregated, descriptive analysis, informed by recommendations from these communities,** was performed using race data from 21,005 COVID-19 cases and 449 COVID-19-associated deaths reported to the Hawaii State Department of Health (HDOH) during March 1, 2020-February 28, 2021.†† In Hawaii, COVID-19 incidence and mortality rates per 100,000 population were 1,477 and 32, respectively during this period. In analyses with race categories that were not mutually exclusive, including persons of one race alone or in combination with one or more races, Pacific Islander persons, who account for 5% of Hawaii's population, represented 22% of COVID-19 cases and deaths (COVID-19 incidence of 7,070 and mortality rate of 150). Native Hawaiian persons experienced an incidence of 1,181 and a mortality rate of 15. Among subcategories of Asian populations, the highest incidences were experienced by Filipino persons (1,247) and Vietnamese persons (1,200). Disaggregating Native Hawaiian, Pacific Islander, and Asian race data can aid in identifying racial disparities among specific subpopulations and highlights the importance of partnering with communities to develop culturally responsive outreach teams§§ and tailored public health interventions and vaccination campaigns to more effectively address health disparities.

Addressing Health Equity in Public Health Practice: Frameworks, Promising Strategies, and Measurement Considerations.

This review describes the context of health equity and options for integrating equity into public health practice. We first discuss how the conceptualization of health equity and how equity considerations in US public health practice have been shaped by multidisciplinary engagements. We then discuss specific ways to address equity in core public health functions, provide examples of relevant frameworks and promising strategies, and discuss conceptual and measurement issues relevant to assessing progress in moving toward health equity. Challenges and opportunities and their implications for future directions are identified.

Race/Ethnicity, Underlying Medical Conditions, Homelessness, and Hospitalization Status of Adult Patients with COVID-19 at an Urban Safety-Net Medical Center - Boston, Massachusetts, 2020.

As of July 5, 2020, approximately 2.8 million coronavirus disease 2019 (COVID-19) cases and 130,000 COVID-19-associated deaths had been reported in the United States (1). Populations historically affected by health disparities, including certain racial and ethnic minority populations, have been disproportionally affected by and hospitalized with COVID-19 (2-4). Data also suggest a higher prevalence of infection with SARS-CoV-2, the virus that causes COVID-19, among persons experiencing homelessness (5). Safety-net hospitals,† such as Boston Medical Center (BMC), which provide health care to persons regardless of their insurance status or ability to pay, treat higher proportions of these populations and might experience challenges during the COVID-19 pandemic. This report describes the characteristics and clinical outcomes of adult patients with laboratory-confirmed COVID-19 treated at BMC during March 1-May 18, 2020. During this time, 2,729 patients with SARS-CoV-2 infection were treated at BMC and categorized into one of the following mutually exclusive clinical severity designations: exclusive outpatient management (1,543; 56.5%), non-intensive care unit (ICU) hospitalization (900; 33.0%), ICU hospitalization without invasive mechanical ventilation (69; 2.5%), ICU hospitalization with mechanical ventilation (119; 4.4%), and death (98; 3.6%). The cohort comprised 44.6% non-Hispanic black (black) patients and 30.1% Hispanic or Latino (Hispanic) patients. Persons experiencing homelessness accounted for 16.4% of patients. Most patients who died were aged ≥60 years (81.6%). Clinical severity differed by age, race/ethnicity, underlying medical conditions, and homelessness. A higher proportion of Hispanic patients were hospitalized (46.5%) than were black (39.5%) or non-Hispanic white (white) (34.4%) patients, a finding most pronounced among those aged <60 years. A higher proportion of non-ICU inpatients were experiencing homelessness (24.3%), compared with homeless patients who were admitted to the ICU without mechanical ventilation (15.9%), with mechanical ventilation (15.1%), or who died (15.3%). Patient characteristics associated with illness and clinical severity, such as age, race/ethnicity, homelessness, and underlying medical conditions can inform tailored strategies that might improve outcomes and mitigate strain on the health care system from COVID-19.

COVID-19 Among American Indian and Alaska Native Persons - 23 States, January 31-July 3, 2020.

Although non-Hispanic American Indian and Alaska Native (AI/AN) persons account for 0.7% of the U.S. population,* a recent analysis reported that 1.3% of coronavirus disease 2019 (COVID-19) cases reported to CDC with known race and ethnicity were among AI/AN persons (1). To assess the impact of COVID-19 among the AI/AN population, reports of laboratory-confirmed COVID-19 cases during January 22†-July 3, 2020 were analyzed. The analysis was limited to 23 states§ with >70% complete race/ethnicity information and five or more laboratory-confirmed COVID-19 cases among both AI/AN persons (alone or in combination with other races and ethnicities) and non-Hispanic white (white) persons. Among 424,899 COVID-19 cases reported by these states, 340,059 (80%) had complete race/ethnicity information; among these 340,059 cases, 9,072 (2.7%) occurred among AI/AN persons, and 138,960 (40.9%) among white persons. Among 340,059 cases with complete patient race/ethnicity data, the cumulative incidence among AI/AN persons in these 23 states was 594 per 100,000 AI/AN population (95% confidence interval [CI] = 203-1,740), compared with 169 per 100,000 white population (95% CI = 137-209) (rate ratio [RR] = 3.5; 95% CI = 1.2-10.1). AI/AN persons with COVID-19 were younger (median age = 40 years; interquartile range [IQR] = 26-56 years) than were white persons (median age = 51 years; IQR = 32-67 years). More complete case report data and timely, culturally responsive, and evidence-based public health efforts that leverage the strengths of AI/AN communities are needed to decrease COVID-19 transmission and improve patient outcomes.

Measurement of Health Disparities, Health Inequities, and Social Determinants of Health to Support the Advancement of Health Equity.

Reduction of health disparities and advancement of health equity in the United States require high-quality data indicative of where the nation stands vis-à-vis health equity, as well as proper analytic tools to facilitate accurate interpretation of these data. This article opens with an overview of health equity and social determinants of health. It then proposes a set of recommended practices in measurement of health disparities, health inequities, and social determinants of health at the national level to support the advancement of health equity, highlighting that (1) differences in health and its determinants that are associated with social position are important to assess; (2) social and structural determinants of health should be assessed and multiple levels of measurement should be considered; (3) the rationale for methodological choices made and measures chosen should be made explicit; (4) groups to be compared should be simultaneously classified by multiple social statuses; and (5) stakeholders and their communication needs can often be considered in the selection of analytic methods. Although much is understood about the role of social determinants of health in shaping the health of populations, researchers should continue to advance understanding of the pathways through which they operate on particular health outcomes. There is still much to learn and implement about how to measure health disparities, health inequities, and social determinants of health at the national level, and the challenges of health equity persist. We anticipate that the present discussion will contribute to the laying of a foundation for standard practice in the monitoring of national progress toward achievement of health equity.

Vital signs: leading causes of death, prevalence of diseases and risk factors, and use of health services among Hispanics in the United States - 2009-2013.

BACKGROUND: Hispanics and Latinos (Hispanics) are estimated to represent 17.7% of the U.S. population. Published national health estimates stratified by Hispanic origin and nativity are lacking. METHODS: Four national data sets were analyzed to compare Hispanics overall, non-Hispanic whites (whites), and Hispanic country/region of origin subgroups (Hispanic origin subgroups) for leading causes of death, prevalence of diseases and associated risk factors, and use of health services. Analyses were generally restricted to ages 18-64 years and were further stratified when possible by sex and nativity. RESULTS: Hispanics were on average nearly 15 years younger than whites; they were more likely to live below the poverty line and not to have completed high school. Hispanics showed a 24% lower all-cause death rate and lower death rates for nine of the 15 leading causes of death, but higher death rates from diabetes (51% higher), chronic liver disease and cirrhosis (48%), essential hypertension and hypertensive renal disease (8%), and homicide (96%) and higher prevalence of diabetes (133%) and obesity (23%) compared with whites. In all, 41.5% of Hispanics lacked health insurance (15.1% of whites), and 15.5% of Hispanics reported delay or nonreceipt of needed medical care because of cost concerns (13.6% of whites). Among Hispanics, self-reported smoking prevalences varied by Hispanic origin and by sex. U.S.-born Hispanics had higher prevalences of obesity, hypertension, smoking, heart disease, and cancer than foreign-born Hispanics: 30% higher, 40%, 72%, 89%, and 93%, respectively. CONCLUSION: Hispanics had better health outcomes than whites for most analyzed health factors, despite facing worse socioeconomic barriers, but they had much higher death rates from diabetes, chronic liver disease/cirrhosis, and homicide, and a higher prevalence of obesity. There were substantial differences among Hispanics by origin, nativity, and sex. IMPLICATIONS FOR PUBLIC HEALTH: Differences by origin, nativity, and sex are important considerations when targeting health programs to specific audiences. Increasing the proportions of Hispanics with health insurance and a medical home (patientcentered, team-based, comprehensive, coordinated health care with enhanced access) is critical. A feasible and systematic data collection strategy is needed to reflect health diversity among Hispanic origin subgroups, including by nativity.

Preventing Suicidal Behavior Among American Indian and Alaska Native Adolescents and Young Adults.

From 2009 to 2018, overall suicide rates in the United States increased by 20.3% and increased by 43.5% among non-Hispanic American Indian and Alaska Native (AI/AN) communities. Combining years 2009 through 2018, suicide rates per 100 000 population among non-Hispanic AI/AN adolescents and young adults aged 15 to 34 years were 2 to 4 times higher than those of adolescents and young adults of other races and ethnicities. An estimated 14% to 27% of non-Hispanic AI/AN adolescents attempted suicide during that time. The elevated rates of suicidal behavior among non-Hispanic AI/AN adolescents and young adults reflect inequities in the conditions that create health. In this topical review, we describe school-based educational efforts that are driven by local AI/AN communities, such as the American Indian Life Skills curriculum, that teach stress and coping skills and show promise in reducing suicidal ideation attempts and fatalities among AI/AN adolescents. Using a social-determinants-of-health lens, we review the availability and quality of employment as an important influencer of suicidal behavior, as well as the role of the workplace as an environment for suicide prevention in AI/AN communities. Working with tribal, state, local, and federal colleagues, the public health community can implement programs known to be effective and create additional comprehensive strategies to reduce inequities and ultimately reduce suicide rates.

Postpartum intrauterine device insertion and postpartum tubal sterilization in the United States.

OBJECTIVE: The purpose of this study was to estimate US rates of postpartum intrauterine device (IUD) insertion and postpartum tubal sterilization. STUDY DESIGN: Data from the 2001-2008 Nationwide Inpatient Sample were used to identify delivery hospitalizations with IUD insertion or tubal sterilization procedure codes. RESULTS: Estimated rates of postpartum IUD insertion and postpartum tubal sterilization were 0.27 and 770.67 per 10,000 deliveries, respectively. Although the rate of IUD insertion was similar across age groups, the rate of tubal sterilization increased with age. Nonetheless, 15% of tubal sterilizations occurred among women who were ≤ 24 years old. IUD insertion was more likely among women who delivered at teaching hospitals (odds ratio, 3.02; 95% confidence interval, 1.43-6.37); tubal sterilization was more likely among women without private insurance (odds ratio, 2.04; 95% confidence interval, 1.97-2.11). CONCLUSION: Among US postpartum women, IUD insertion occurs considerably less frequently than tubal sterilization, even among younger women for whom poststerilization regret is a concern.

Complications of common gynecologic surgeries among HIV-infected women in the United States.

OBJECTIVE: To compare frequencies of complications among HIV-infected and-uninfected women undergoing common gynecological surgical procedures in inpatient settings. METHODS: We used 1994-2007 data from the Nationwide Inpatient Sample of the Healthcare Cost and Utilization Project, a nationally representative sample of inpatient hospitalizations. Our analysis included discharge records of women aged ≥15 undergoing hysterectomy, oophorectomy, salpingectomy for ectopic pregnancy, bilateral tubal sterilization, or dilation and curettage. Associations between HIV infection status and surgical complications were evaluated in multivariable logistic regression models, adjusting for key covariates. RESULTS: For each surgery, HIV infection was associated with experiencing ≥1 complication. Adjusted ORs ranged from 2.0 (95% confidence interval (CI): 1.7, 2.2) for hysterectomy with oophorectomy to 3.1 (95% CI: 2.4, 4.0) for bilateral tubal sterilization with no comorbidity present. HIV infection was positively associated with extended length of stay and infectious complications of all of the surgeries examined. For some surgeries, it was positively associated with transfusion and anemia due to acute blood loss. Among HIV-infected women, the odds of infectious and other complications did not decrease between 1994-2000 and 2001-2007. CONCLUSION: HIV infection was associated with elevated frequencies of complications of gynecologic surgeries in the US, even in the era of HAART.

Validation of algorithm to identify American Indian/Alaska Native pregnant women at risk from pandemic H1N1 influenza.

Pregnant women and American Indian and Alaska Native people are at elevated risk of severe disease and mortality from 2009 pandemic influenza A/H1N1. We validated an electronic health record-based algorithm used by Indian Health Service to identify pregnant women in near real-time surveillance of pandemic influenza A/H1N1. We randomly selected a stratified sample of 515 patients at 3 Indian Health Service-funded hospitals with varied characteristics. With comprehensive review of patients' electronic health records as the gold standard, we calculated the positive predictive value and sensitivity of the pregnancy algorithm. The sensitivity of the algorithm at individual hospitals ranged from 94.1-96.0%. Positive predictive value ranged from 94.4-98.3%. Despite differences among hospitals on key characteristics, the pregnancy algorithm performed nearly equivalently with high positive predictive value and sensitivity at all facilities. It may prove helpful for surveillance during future epidemics and for targeting interventions for pregnant women and infants.

Collection of Data on Sex, Sexual Orientation, and Gender Identity by U.S. Public Health Data and Monitoring Systems, 2015-2018.

We aimed to assess Centers for Disease Control and Prevention (CDC) data systems on the extent of data collection on sex, sexual orientation, and gender identity as well as on age and race/ethnicity. Between March and September 2019, we searched 11 federal websites to identify CDC-supported or -led U.S. data systems active between 2015 and 2018. We searched the systems' website, documentation, and publications for evidence of data collection on sex, sexual orientation, gender identity, age, and race/ethnicity. We categorized each system by type (disease notification, periodic prevalence survey, registry/vital record, or multiple sources). We provide descriptive statistics of characteristics of the identified systems. Most (94.1%) systems we assessed collected data on sex. All systems collected data on age, and approximately 80% collected data on race/ethnicity. Only 17.7% collected data on sexual orientation and 5.9% on gender identity. Periodic prevalence surveys were the most common system type for collecting all the variables we assessed. While most U.S. public health data and monitoring systems collect data disaggregated by sex, age, and race/ethnicity, far fewer do so for sexual orientation or gender identity. Standards and examples exist to aid efforts to collect and report these vitally important data. Additionally important is increasing accessibility and appropriately tailored dissemination of reports of these data to public health professionals and other collaborators.

Perceived control over condom use among sex workers in Madagascar: a cohort study.

BACKGROUND: Women's perceived control over condom use has been found to be an important determinant of actual condom use in some studies. However, many existing analyses used cross-sectional data and little quantitative information exists to characterize the relationships between perceived control and actual condom use among sex worker populations. METHODS: We assessed the association between measures of perceived condom use control and self-reported use of male condoms employing data from a longitudinal pilot study among 192 sex workers in Madagascar. RESULTS: In multivariable models, a lack of perceived control over condom use with a main partner and having a main partner ever refuse to use a condom when asked were both associated with an increased number of sex acts unprotected by condoms in the past week with a main partner (RR 1.86; 95% CI 1.21-2.85; RR 1.34; 95% CI 1.03-1.73, respectively). Conversely, no measure of condom use control was significantly associated with condom use with clients. CONCLUSION: Perceived control over condom use was an important determinant of condom use with main partners, but not clients, among sex workers in Madagascar. Programs working with sex workers should reach out to main and commercial partners of sex workers to increase male condom use.