Fluent Essays

What must be done to intervene and ensure that history does not repeat itself for future populations? This week, you examine the impact of the historical roots of social disparities on health of populations in low-income countries. As you go through this week’s Learning Resources, think about what we can learn from history. This week, you consider developing a policy in a country you selected and think about various issues in practicing population health. For your Final Project, share some of your ideas on how you can use the knowledge and insights gained in this course to promote positive social change in a community/country and the world. It is advisable to select a community/country other than the one where you live. To prepare for the Final Project, review all the week’s Learning Resources and consider possible issues you might encounter when implementing a policy. Final Project (7–10 pages), not including the cover and the references: In developing a policy in the country you selected, consider the following: Explain the rationale for selecting the country. Describe the social determinants of health in the country that you would need to address. Explain why you need to address these determinants. Explain the possible public issues you might encounter in health literacy and cultural awareness in this country. Describe the relationship between health inequality/inequities and life expectancy for the population in your selected country. Describe two current efforts in this country (you selected) to reduce health inequities. Explain how you might develop a health policy so that it gets the support of the country you selected. Note:Take into account the culture of the country. Use APA formatting for your Final Project and to cite your resources. Expand on your insights utilizing the Learning Resources. Comment www.thelancet.com/lancetgh Vol 5 June 2017 e557 Smoking status and HIV in low-income and middle-income countries In high-income settings, the prevalence of tobacco use has been shown to be significantly higher in people living with HIV than among HIV-negative individuals of the same age and sex distribution. This at-risk pattern is one of the biggest threats to the number of years of life saved with antiretroviral therapy (ART).1,2 Extrapolation of these findings to low-income and middle-income countries (LMICs) is risky because social, cultural, and behavioural factors influencing tobacco use differ widely across different regions. The epidemiology of tobacco use in HIV-positive individuals in LMICs has been sparsely reported, with limited representativeness and no or biased control populations.3–5 In The Lancet Global Health, Noreen Mdege and colleagues6 report an unprecedented estimation of tobacco use in people living with HIV, using nationally representative samples extracted from the Demographic and Health Surveys (DHS) from 28 countries on three continents. In addition to depicting the burden and diversity of tobacco use, the authors show significantly higher figures of tobacco use in people living with HIV compared with their HIV-negative counterparts, regardless of gender. These results confirm what has already been reported in high-income settings, and emphasise the need for adapted preventive measures and tobacco cessation programmes in LMICs. Countries highly affected by the HIV epidemic usually have underfunded health-care systems and are overburdened with other major epidemics such as malaria and tuberculosis, and are therefore less inclined to invest in preventive measures against non- communicable diseases and their determinants. In this context, smoking-targeted preventive and cessation programmes are often limited or nonexistent. HIV care programmes represent by far the largest chronic care programmes rolled out in LMICs, potentially paving the way for an integrated panel of services targeting non- communicable diseases. Measures directed towards smoking avoidance and cessation can then be introduced and piloted before their extension and adaption to a larger set of health facilities. Although Mdege and colleagues’ analysis6 of publicly available data provides a comprehensive presentation of prevalence estimates of tobacco use in HIV-positive individuals in LMICs, the number of people living with HIV in the study represents less than 0·001\% of the estimated 34 million people living with HIV in 2014 in these parts of the world; this limited size might lead to imprecision and potential bias in the prevalence estimates of tobacco use, especially outside of Africa.7 Although the data were fairly representative of the African region, data for southeast Asia were only available for India, leaving important uncertainties concerning the association between tobacco use and HIV infection in countries particularly affected by tobacco smoking—especially China. This report6 comes at a time when LMICs represent a major target for the tobacco industry.8 Southeast Asia is the widest market for the tobacco industry, and the Chinese tobacco market represents more cigarettes than all other LMICs combined.9 Additional data sources on tobacco use are needed for people living with HIV in LMICs. Achievements made by the international community to enable universal access to ART were accompanied by initiatives providing worldwide data on the follow-up of patients initiating ART. The International Epidemiology Databases to Evaluate AIDS (IeDEA), funded by the US National Institutes of Health, is a unique platform that has so far gathered data on more than 1 700 000 people living with HIV on ART, most of whom live in LMICs. This platform has successfully collected core information on ART exposure, and harmonisation is underway to standardise the collection of basic behavioural risk factors such as tobacco use. Data from observational cohorts participating in IeDEA have already provided regional estimates on tobacco use from west Africa,4 and in the future could contribute to a more robust and complementary estimation of tobacco use in people living with HIV, especially in the context of universal ART.10 Nevertheless, the DHS offer a good opportunity to access a somewhat representative control group of HIV- uninfected people and can be repeated over time using the same methodological approach. This use of DHS data is therefore a unique framework to conduct sound For more on IeDEA see http://www.iedea.org See Articles page e578 http://www.iedea.org http://www.iedea.org Comment e558 www.thelancet.com/lancetgh Vol 5 June 2017 analyses for identification of trends in tobacco use and to measure the effect of smoking prevention and cessation programmes according to HIV infection status. To expand their analysis, Mdege and colleagues could also consider prevalence estimates of tobacco use in younger age groups because these groups are the most susceptible to smoking initiation. Additionally, the low prevalence of tobacco smoking reported in women compared with men in LMICs makes women—along with young people—a particular target for the tobacco industry, whether they live with HIV or not.8 Antoine Jaquet, *François Dabis Institut de Santé Publique, d’Epidémiologie et de Développement, University of Bordeaux, and Inserm, Bordeaux Population Health Research Center, UMR 1219, F-33000 Bordeaux, France [email protected] We are investigators of the West Africa IeDEA collaboration, and declare no competing interests. Copyright © The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license. 1 Reddy KP, Parker RA, Losina E, et al. Impact of cigarette smoking and smoking cessation on life expectancy among people with HIV: A US-based modeling study. J Infect Dis 2016; 214: 1672–81. 2 Mdodo R, Frazier EL, Dube SR, et al. Cigarette smoking prevalence among adults with HIV compared with the general adult population in the United States: cross-sectional surveys. Ann Intern Med 2015; 162: 335–44. 3 Iliyasu Z, Gajida AU, Abubakar IS, Shittu O, Babashani M, Aliyu MH. Patterns and predictors of cigarette smoking among HIV-infected patients in northern Nigeria. Int J STD AIDS 2012; 23: 849–52. 4 Jaquet A, Ekouevi DK, Aboubakrine M, et al. Tobacco use and its determinants in HIV-infected patients on antiretroviral therapy in West African countries. Int J Tuberc Lung Dis 2009; 13: 1433–39. 5 Mwiru RS, Nagu TJ, Kaduri P, Mugusi F, Fawzi W. Prevalence and patterns of cigarette smoking among patients co-infected with human immunodeficiency virus and tuberculosis in Tanzania. Drug Alcohol Depend 2017; 170: 128–32. 6 Mdege ND, Shah S, Ayo-Yusuf OA, Hakim J, Siddiqi K. Tobacco use among people living with HIV: analysis of data from Demographic and Health Surveys from 28 low-income and middle-income countries. Lancet Glob Health 2017; 5: e578–92. 7 UNAIDS. Global AIDS Update 2016. Geneva: UNAIDS, 2016. 8 Gilmore AB, Fooks G, Drope J, Bialous SA, Jackson RR. Exposing and addressing tobacco industry conduct in low-income and middle-income countries. Lancet 2015; 385: 1029–43. 9 Eriksen M, Mackay J, Schluger N, Islami F, Drope J. The Tobacco Atlas, 5th edn. Atlanta, GA: American Cancer Society, 2015. 10 WHO. Consolidated guidelines on the use of antiretroviral drugs for treating and preventing HIV infection: recommendations for a public health approach, 2nd edn. Geneva: World Health Organization, 2016. Smoking status and HIV in low-income and middle-income countries References RESEARCH ARTICLE Open Access Prevalence of arthritis according to age, sex and socioeconomic status in six low and middle income countries: analysis of data from the World Health Organization study on global AGEing and adult health (SAGE) Wave 1 Sharon L. Brennan-Olsen1,2,3,4* , S. Cook1, M. T. Leech5, S. J. Bowe1, P. Kowal6,7, N. Naidoo6, I. N. Ackerman8, R. S. Page1,9, S. M. Hosking1, J. A. Pasco1,3 and M. Mohebbi1 Abstract Background: In higher income countries, social disadvantage is associated with higher arthritis prevalence; however, less is known about arthritis prevalence or determinants in low to middle income countries (LMICs). We assessed arthritis prevalence by age and sex, and marital status and occupation, as two key parameters of socioeconomic position (SEP), using data from the World Health Organization Study on global AGEing and adult health (SAGE). Methods: SAGE Wave 1 (2007–10) includes nationally-representative samples of older adults (≥50 yrs), plus smaller samples of adults aged 18-49 yrs., from China, Ghana, India, Mexico, Russia and South Africa (n = 44,747). Arthritis was defined by self-reported healthcare professional diagnosis, and a symptom-based algorithm. Marital status and education were self-reported. Arthritis prevalence data were extracted for each country by 10-year age strata, sex and SEP. Country-specific survey weightings were applied and weighted prevalences calculated. Results: Self-reported (lifetime) diagnosed arthritis was reported by 5003 women and 2664 men (19.9\% and 14.1\%, respectively), whilst 1220 women and 594 men had current symptom-based arthritis (4.8\% and 3.1\%, respectively). For men, standardised arthritis rates were approximately two- to three-fold greater than for women. The highest rates were observed in Russia: 38\% (95\% CI 36\%–39\%) for men, and 17\% (95\% CI 14\%–20\%) for women. For both sexes and in all LMICs, arthritis was more prevalent among those with least education, and in separated/divorced/widowed women. Conclusions: High arthritis prevalence in LMICs is concerning and may worsen poverty by impacting the ability to work and fulfil community roles. These findings have implications for national efforts to prioritise arthritis prevention and management, and improve healthcare access in LMICs. Keywords: Arthritis, Epidemiology, Prevalence, Socio-demographic characteristics, Low and middle income countries * Correspondence: [email protected] 1Deakin University, Geelong, Australia 2Australian Institute for Musculoskeletal Science (AIMSS), The University of Melbourne-Western Precinct, Level 3, Western Centre for Health Research and Education (WCHRE) Building, C/- Sunshine Hospital, Furlong Road, St Albans, Melbourne, VIC 3021, Australia Full list of author information is available at the end of the article © The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Brennan-Olsen et al. BMC Musculoskeletal Disorders (2017) 18:271 DOI 10.1186/s12891-017-1624-z http://crossmark.crossref.org/dialog/?doi=10.1186/s12891-017-1624-z&domain=pdf http://orcid.org/0000-0003-3269-5401 mailto:[email protected] http://creativecommons.org/licenses/by/4.0/ http://creativecommons.org/publicdomain/zero/1.0/ Background Worldwide, musculoskeletal disorders represent a global threat to healthy ageing [1], and are ranked as the sec- ond most common cause of disability, measured by years lived with disability (YLDs) [2]. Lower and middle in- come countries (LMICs) are not immune to the burden of musculoskeletal diseases, indeed the prevalence of this non-communicable disease (NCD) group is dramatically increasing in LMICs [3]. The 2010 Global Burden of Disease (GBD) study reported that musculoskeletal dis- eases accounted for 19.2\% of all YLDs in LMICs [3]. Despite this, the majority of the global NCD initiatives do not include musculoskeletal diseases [3]. Significantly contributing to the global disability burden associated with the musculoskeletal system are arthritis diseases. Arthritis is an umbrella term that encompasses in ex- cess of 100 different arthritic conditions which are a chronic, painful, and debilitating group of diseases. Arthritis, specifically osteoarthritis, is a significant contributor to global disability burden, and the YLDs attributable to osteoarthritis have increased by 75\% from 1990 to 2013 [2], indicating this disease as a growing problem internationally. In combination with an increasing trajectory of arthritis prevalence [2, 4], growth in YLDs attributable to arthritis is due pri- marily to increased life expectancy worldwide, and prolonged exposure to arthritis risk factors [5]. Compared to higher income countries, many LMICs [6], where two-thirds of the world’s population resides, have a much lower capacity to pay for adequate health- care. Indeed, LMICs have 90\% of the global burden of disease but only 12\% of global health spending [7]. In higher income countries, arthritis is associated with re- duced workplace productivity [8, 9]; however, for resi- dents of LMICs, arthritis imposes a potential additional burden by creating a vicious cycle that subsequently worsens poverty [10]. For example, compared to higher income countries, and in context of scarce medical and social support systems, residents of LMICs with arthritis also experience reduced ability to access, afford or utilize treatments including analgesic and anti-inflammatory pharmacotherapies [11, 12], or arthroplasty for advanced disease [13, 14]. They also have, in context of workforce capacity limitations, less flexibility regarding working conditions or hours [15], and few if any options for early retirement, or social security ‘safety nets’ pertaining to minimum income, including financial and/or material goods. Whilst the majority of research regarding arthritis prevalence has been undertaken in higher income coun- tries, recent data from the 2010 GBD Study provides some evidence that LMICs may have greater arthritis prevalence than higher income countries [16]. Yet, while valuable population level estimates, extrapolation from these GBD estimates is difficult given that they are based on published prevalence and incidence data from a small number of heterogeneous studies spanning different time periods in a limited number of LMIC [17]. Further- more, data from multi-country studies of LMICs that have examined prevalence of arthritis across sociodemo- graphic factors are typically not readily available [18, 19], with the exception of a recent publication, which showed that more years of schooling and greater levels of wealth decreased the odds of having an undiagnosed NCD, including arthritis [20]. Understanding the preva- lence of arthritis across different parameters of socioeco- nomic position (SEP) data would augment our global understanding of global arthritis prevalence, social deter- minants and burden. To date, country-specific arthritis prevalence across parameters of SEP has not been systematically evalu- ated in large, nationally representative samples of populations from LMICs. This information is crucial for planning future healthcare delivery for high bur- den chronic conditions and to ensure sufficient health workforce capacity – both significant concerns in an ageing world [21]. Comprehensive data have been collected in the World Health Organization (WHO) Study on global AGEing and adult health (SAGE) [20, 22, 23], thus providing an important re- source to investigate disease prevalence in large population samples from six LMICs. Using SAGE Wave 1, these analyses were undertaken to determine the prevalence of arthritis in LMICs according to age, sex, and socioeconomic position (SEP). Methods Study population and design SAGE Wave 1 (2007–10) is a longitudinal study with na- tionally representative samples of persons aged 50+ years and a smaller sample of adults aged 18–49 years that in- cludes 44,747 adults aged ≥18 years from China, Ghana, India, Mexico, Russian Federation and South Africa [23]. Multistage cluster sampling strategies were used with households as sampling units. Households were classi- fied into one of two mutually exclusive categories: i) all persons aged 50 years and older were selected from “older” households, and ii) one person aged 18–49 years was selected from each “younger” household. An older or younger household was defined by the age of the re- spondent targeted for individual interview. Household- level and person-level analysis weights were calculated for each country. This research was performed in ac- cordance with the Declaration of Helsinki. The WHO and the respective implementing agency in each country provided ethics approvals. Written, informed consent was obtained from all participants. Brennan-Olsen et al. BMC Musculoskeletal Disorders (2017) 18:271 Page 2 of 12 Data collection in WHO SAGE Using a standardized survey instrument to ensure consistency, and based on standardized methods, inter- viewer training and translation protocols, face-to-face in- terviews were conducted in China (2008–10; response 93\%), Ghana (2008–09; response 81\%), India (2007–08; response 68\%), Mexico (2009–10; response rate 53\%), the Russian Federation (2007–10; response 83\%) and South Africa (2007–08; response 75\%), as previously published [23]. Full details regarding the probability sampling design, cluster sampling strategies and country-specific areas included in SAGE have been pub- lished elsewhere [23]. Briefly, the SAGE questionnaire consisted of household, individual and proxy question- naires, a verbal autopsy, and appendices: the domains of which are summarised in Table 1 [23]. Arthritis status: self-reported and symptom-based For the current analyses, self-reported diagnosis of arth- ritis (lifetime) was based on participant responses to the question; “Have you ever been diagnosed with/told by a health care professional you have arthritis (a disease of the joints; or by other names rheumatism or osteoarthritis)?” As a secondary endpoint, a symptom-based determination of arthritis (yes/no for current within the previous 12 months) was also employed, by applying an algorithm developed by the WHO SAGE study team [23]; questions and the algorithm are presented in Table 2. Socioeconomic position SEP was measured using two key parameters of marital status and educational attainment: the latter used due to the inextricable link between education and skilled vs. un- skilled labour, and thus financial remuneration for work. Self-reported marital status was categorised for analyses into three groups of: (i) never married, (ii) currently mar- ried or cohabitating, and (iii) separated/divorced or widowed. Participants were asked if they had ever been to school; for those that indicated ‘yes’, they were also asked to identify the highest level of education completed. Educa- tion was categorised as (i) ‘no formal schooling’, (ii) less than primary school, or primary school completed, (iii) sec- ondary school completed, or high school (or equivalent) completed, or (iv) college, pre-university or university com- pleted, or post-graduate degree completed. Education levels were mapped to an international standard [24]. Statistical analyses Arthritis (self-reported and symptom-based) prevalence and 95\% confidence intervals (95\%CI) were calculated by implementing household level analysis weights separ- ately for each of the six countries across 10-year age strata (the 20–29 year age group was expanded to also include those aged 18–19 years), sex, marital status and education. Country-specific survey weightings were applied, and weighted prevalence calculated for each country. Adjustment of prevalence estimates for differ- ences in the age structure across countries was accom- plished by age-standardisation, using the direct method of standardisation [25] and the WHO World Standard Population distribution (\%) as standard population [26]. Ten-year intervals were used for age categorisation. Results Country-specific numbers and proportions of the total 44,747 participants (total 57.1\% women), were; China n = 15,050 (33.6\%), Ghana n = 5573 (12.5\%), India n = 12,198 (27.3\%), Mexico n = 2752 (6.1\%), the Russian Federation n = 4947 (11.1\%), and South Africa n = 4227 (9.5\%). Across the entire study population, 5003 women and 2664 men had (lifetime) self-reported arthritis (19.9\% and 14.1\%, respectively), whilst 1220 women and 594 men Table 1 Questionnaire sections included in the SAGE Wave 1 standardized survey instrument [23] Questionnaire section Household roster Questions regarding the dwelling, income, transfers [of family members] in and out of the household, assets and expenditures Individual questionnaire Questions regarding health and its determinants, disability, work history, risk factors, chronic conditions, caregiving, subjective well-being, health care utilization and health systems responsiveness Proxy questionnaire Questions regarding health, functioning, chronic conditions, and health care utilization Verbal autopsy Performed to ascertain the probable cause of death for deaths in the household in the 24 months prior to interview or between interview waves Appendices Includes show-cards to assist with the interviews Table 2 Symptom-based questions and the related algorithm to ascertain prevalent arthritis, developed as part of the World Health Organization SAGE Wave 1 [23] Question number Question text and algorithm 1 During the last 12 months, have you experienced pain, aching, stiffness or swelling in or around the joints (like arms, hands, legs or feet) which were not related to an injury and lasted for more than a month? 2 During the last 12 months, have your experienced stiffness in the joint in the morning after getting up from bed, or after a long rest of the joint without movement? 3 Did this stiffness last for more than 30 min? 4 Did this stiffness go away after exercise or movement in the joint? Algorithm If a participant responded with ‘yes’ to questions 1 and/or 2, and responded with ‘yes’ to question 3 and ‘no’ to question 4, then the participant was categorised as having arthritis Brennan-Olsen et al. BMC Musculoskeletal Disorders (2017) 18:271 Page 3 of 12 were identified as having (within previous 12 months) symptom-based arthritis (4.8\% and 3.1\%, respectively). Table 3 presents the country-specific proportional responses (non-weighted) to the four symptom-based questions (see Table 2), that were included in the algorithm to determine symptom-based arthritis. For women, proportions that reported ‘any pain during the last 12 months’ or ‘any stiffness during the last 12 months’ were lowest for Mexico (28.4\% [95\% CI 26.3\%– 30.9\%] and 23.3\% [95\% CI 20.9\%–26.0\%], respectively) and highest for the Russian Federation (48.4\% [95\% CI 46.4\%–50.4\%] and 50.5\% [95\% CI 48.8\%–52.1\%], respect- ively). For men, the proportions that reported ‘any pain during the last 12 months’ or ‘any stiffness during the last 12 months’ were lowest for Mexico (20.1\% [95\% CI 17.5\%–23.0\%] and 16.1\% [95\% CI\%CI 14.1\%–18.3\%], re- spectively) and highest for the Russian Federation (32.9\% [95\% CI 30.5\%–35.5\%] and 34.6\% [95\% CI 32.4\%– 36.9\%], respectively). Table 4 presents the country-specific and sex-stratified prevalence of self-reported arthritis (weighted), across age strata, educational attainment and marital status. For both sexes in each country, arthritis prevalence increased proportionally with advancing age; with the exception of women from China and men and women from South Africa who had the greatest prevalence in the age group of 60–69 years, all other groups showed a peak in arthritis prevalence in the oldest age group ≥70 years. For women, the prevalence by country ranged from 22.9\% (95\% CI 11.2\%–41.1\%) in Mexico to 45.7\% (95\% CI 39.1\%–52.3\%) in the Russian Federation. For men, prevalence ranged from 9.7\% (95\% CI 6.3\%–14.5\%) in Mexico to 37.8\% (95\% CI 30.3\%–46.0\%) in the Russian Federation. In each country, women who had never been formally schooled or had completed less than primary school had the highest prevalence of arthritis compared to those with a greater level of educational at- tainment. Higher arthritis prevalence was consistently observed for women that were separated, divorced or widowed (range: Russian Federation 36.4\% [95\% CI 29.1\%–44.4\%] to Ghana 11.7\% [95\% CI 8.9\%–15.1\%]) compared to those that were never married or currently married (range: China 0.9\% [95\% CI 0.3\%–3.0\%] to South Africa 12.1\% [95\% CI 5.5\%–24.7\%]). Similar to women, men that had never been formally schooled had the highest arthritis prevalence, with the exception of men from the Russian Federation, for whom the greatest prevalence was observed in those that had completed all or some primary school level education (39.6\% [95\% CI 21.3\%–61.4\%]), however these numbers were small. Compared to other categories, men that were never married had the lowest arthritis prevalence (range: Mexico 0.1\% [95\% CI 0.0\%–0.5\%] to India 3.9\% [95\% CI 1.5\%–9.5\%]). In China and India, men that were currently married had the highest prevalence (11.9\% [95\% CI 9.4\%–14.8\%], and 8.8\% [95\% CI 7.2\%–10.7\%], respectively), whilst for all other countries, men that were separated, divorced or widowed were observed to have the highest arthritis prevalence (highest: Russian Federation 33.5\% [95\% CI 13.3\%–62.3\%]). Table 5 presents the country-specific and sex-stratified prevalence of symptom-based arthritis prevalence (weighted), across age strata, educational attainment and marital status, for each LMIC. Patterns of symptom- based arthritis prevalence were similar to self-reported arthritis for both sexes; however, prevalence was lower than observed for self-reported arthritis. Figure 1 presents a box plot of the age-standardised rates of self-reported arthritis, stratified by sex, across each country (crude and age-standardised rates are presented in Additional file 1: Online Table S1). For five of the six LMICs, the standardised rates of arthritis for men were approximately twice that observed for women; the excep- tion was Ghana, where men had rates three times greater than those observed for women (12\% [95\% CI 11\%–13\%] vs. 4\% [95\% CI 3\%–5\%]). The highest rates of arthritis were observed in the Russian Federation: for men the rate was 38\% (95\% CI 36\%–39\%) and for women it was 17\% (95\% CI 14\%–20\%). Discussion We present the prevalence of arthritis across age, sex and different parameters of SEP in a large population- based study spanning six LMICs. Across the countries and for both sexes, higher arthritis prevalence was con- sistently associated with older age and lower educational attainment, whilst higher prevalence was also observed in women, but not men, that were separated, divorced, or widowed. The pattern between advancing age and increasing arthritis prevalence in LMICs appears similar to the pat- tern observed in higher income countries [27]. However, after age-standardisation, we observed in our current study that the rates of arthritis in LMICs were greater than those reported in higher income countries, specific- ally for men from China, India, the Russian Federation and South Africa. Compared to higher income countries, higher age-standardised rates of arthritis were also ob- served for women from the Russian Federation; however, for the remaining five LMICs, rates appeared to be simi- lar to those observed from higher income countries. Our results indicate the importance of age-standardisation when reporting prevalence data, in order that fair com- parisons can be applied when discussing whether any disparities in diseases exist between countries. In addition to the peak of arthritis prevalence observed in older age groups, we observed a sizeable proportion of arthritis in younger age groups; prevalence that would Brennan-Olsen et al. BMC Musculoskeletal Disorders (2017) 18:271 Page 4 of 12 T a b le 3 Re sp o n se s to th e fo u r q u es ti o n sa in cl u d ed in th e al g o rit h m fo r sy m p to m -b as ed ar th rit is ,s tr at ifi ed b y co u n tr y an d se xb (n o n -w ei g h te d ) W o m en (n = 25 ,1 80 ) C h in a (n = 80 16 ) G h an a (n = 27 49 ) In d ia (n = 74 89 ) M ex ic o b (n = 16 92 ) Ru ss ia n Fe d er at io n (n = 28 06 ) So u th A fr ic a (n = 24 28 ) c A n y p ai n d u rin g la st 12 m o n th s? (Y es ) 29 .1 \% (2 8. 0\% – 30 .2 \% ) 38 .2 \% (3 6. 4\% – 40 .0 \% ) 29 .2 \% (2 8. 0\% – 30 .4 \% ) 28 .4 \% (2 6. 3\% – 30 .9 \% ) 48 .4 \% (4 6. 4\% – 50 .4 \% ) 36 .5 \% (3 4. 6\% – 38 .4 \% ) c A n y st iff n es s d u rin g la st 12 m o n th s? (Y es ) 24 .2 \% (2 3. 2\% – 25 .2 \% ) 43 .5 \% (4 1. 5\% – 45 .6 \% ) 29 .7 \% (2 8. 5\% – 30 .8 \% ) 23 .3 \% (2 0. 9\% – 26 .0 \% ) 50 .5 \% (4 8. 8\% – 52 .1 \% ) 33 .2 \% (3 1. 2\% – 35 .3 \% ) d D id st iff n es s la st fo r > 30 m in ? (Y es ) 24 .7 \% (2 2. 4\% – 27 .1 \% ) 38 .1 \% (3 5. 6\% – 40 .7 \% ) 33 .3 \% (3 0. 9\% – 35 .2 \% ) 26 .1 \% (2 1. 8\% – 31 .0 \% ) 45 .3 \% (4 2. 8\% – 47 .9 \% ) 36 .3 \% (3 3. 3\% – 39 .4 \% ) d D id st iff n es s g o aw ay af te r m o ve m en t? (N o ) 19 .2 \% (1 7. 4\% – 21 .0 \% ) 31 .5 \% (2 8. 9\% – 34 .2 \% ) 25 .4 \% (2 3. 7\% – 27 .3 \% ) 15 .3 \% (1 2. 3\% – 18 .9 \% ) 33 .1 \% (3 0. 5\% – 35 .9 \% ) 19 .8 \% (1 7. 2\% – 22 .7 \% ) M en (n = 18 ,9 14 ) C h in a (n = 69 93 ) G h an a (n = 28 16 ) In d ia (n = 47 09 ) M ex ic o (n = 10 50 ) Ru ss ia n Fe d er at io n b (n = 15 49 ) So u th A fr ic a (n = 17 97 ) c A n y p ai n d u rin g la st 12 m o n th s? (Y es ) 20 .4 \% (1 9. 6\% – 21 .3 \% ) 25 .2 \% (2 3. 5\% – 26 .9 \% ) 23 .4 \% (2 2. 0\% – 24 .7 \% ) 20 .1 \% (1 7. 5\% – 23 .0 \% ) 32 .9 \% (3 0. 5\% – 35 .5 \% ) 25 .3 \% (2 3. 3\% – 27 .5 \% ) c A n y st iff n es s d u rin g la st 12 m o n th s? (Y es ) 17 .2 \% (1 6. 4\% – 17 .9 \% ) 29 .8 \% (2 8. 2\% – 31 .5 \% ) 25 .4 \% (2 4. 1\% – 26 .7 \% ) 16 .1 \% (1 4. 1\% – 18 .3 \% ) 34 .6 \% (3 2. 4\% – 36 .9 \% ) 23 .7 \% (2 1. 9\% – 25 .5 \% ) d D id st iff n es s la st fo r > 30 m in ? (Y es ) 26 .5 \% (2 4. 4\% – 28 .8 \% ) 29 .2 \% (2 5. 6\% – 33 .1 \% ) 29 .0 \% (2 6. 5\% – 31 .6 \% ) 25 .9 \% (1 9. 7\% – 33 .3 \% ) 40 .0 \% (3 5. 0\% – 45 .1 \% ) 30 .1 \% (2 5. 6\% – 35 .0 \% ) d D id st iff n es s g o aw ay af te r m o ve m en t? (N o ) 20 .4 \% (1 8. 1\% – 22 .9 \% ) 25 .2 \% (2 2. 3\% – 28 .3 \% ) 22 .5 \% (1 9. 8\% – 25 .4 \% ) 17 .9 \% (1 2. 6\% – 24 .8 \% ) 29 .4 \% (2 5. 5\% – 33 .7 \% ) 16 .4 \% (1 3. 0\% – 20 .6 \% ) D at a p re se n te d as p ro p o rt io n s w it h 9 5 \% co n fi d en ce in te rv al s (9 5 \% C I) a C o m p le te w o rd in g o f th e sy m p to m -b as ed q u es ti o n s ar e p re se n te d in Ta b le 2 b A p p ro xi m at el y 1 2 \% o f th e sa m p le fr o m th e R u ss ia n Fe d er at io n h ad n o in fo rm at io n re g ar d in g se x o f re sp o n d en ts c P ro p o rt io n s (9 5 \% co n fi d en ce in te rv al s) ar e b as ed o n th e to ta l st u d y p o p u la ti o n fr o m ea ch LM IC d P ro p o rt io n s (9 5 \% co n fi d en ce in te rv al s) ar e b as ed o n th o se th at re sp o n d ed ‘y es ’ to ei th er o n e o r b o th o f th e fi rs t tw o sy m p to m -b as ed q u es ti o n s Brennan-Olsen et al. BMC Musculoskeletal Disorders (2017) 18:271 Page 5 of 12 T a b le 4 C o u n tr y- sp ec ifi c se lf- re p o rt ed ar th rit is p re va le n ce (w ei g h te d ), ac ro ss ag e st ra ta ,e d u ca ti o n al at ta in m en t an d m ar it al st at u s, st ra ti fie d b y se x W o m en w it h se lf- re p o rt ed ar th rit is (n = 50 03 ) C h in a n = 18 51 G h an a n = 35 0 In d ia n = 94 6 M ex ic o n = 20 6 Ru ss ia n Fe d er at io n n = 10 49 So u th A fr ic a n = 60 1 A g e (y ea rs ) 18 – 29 3. 7\% (0 .9 \% – 14 .5 \% ) 4. 4\% (1 .3 \% – 13 .8 \% ) 2. 9\% (1 .9 \% – 4. 2\% ) 0. 4\% (0 .1 \% – 2. 8\% ) 4. 0\% (0 .6 \% – 22 .1 \% ) 8. 9\% (1 .8 \% – 34 .2 \% ) 30 – 39 6. 0\% (3 .8 \% – 9. 5\% ) 3. 0\% (0 .9 \% – 9. 2\% ) 8. 5\% (6 .7 \% – 10 .7 \% ) 1. 8\% (0 .5 \% – 6. 0\% ) 14 .7 \% (7 .0 \% – 28 .3 \% ) 0. 2\% (0 .0 \% – 1. 6\% ) 40 – 49 15 .1 \% (1 1. 2\% – 20 .0 \% ) 3. 6\% (1 .6 \% – 8. 1\% ) 12 .2 \% (9 .6 \% – 15 .3 \% ) 7. 9\% (2 .2 \% – 24 .5 \% ) 21 .4 \% (1 0. 5\% – 38 .6 \% ) 11 .3 \% (5 .6 \% – 21 .4 \% ) 50 – 59 22 .1 \% (2 0. 0\% – 24 .4 \% ) 11 .5 \% (9 .1 \% – 14 .5 \% ) 19 .8 \% (1 6. 7\% – 23 .2 \% ) 6. 6\% (2 .3 \% – 17 .5 \% ) 21 .1 \% (1 5. 6\% – 27 .9 \% ) 29 .2 \% (2 4. 6\% – 34 .2 \% ) 60 – 69 29 .7 \% (2 7. 1\% – 32 .6 \% ) 15 .4 \% (1 2. 1\% – 19 .5 \% ) 21 .4 \% (1 6. 7\% – 26 .9 \% ) 13 .0 \% (8 .8 \% – 18 .7 \% ) 36 .4 \% (2 9. 6\% – 43 .8 \% ) 31 .5 \% (2 5. 7\% – 38 .0 \% ) 70 + 29 .2 \% (2 6. 7\% – 31 .9 \% ) 22 .8 \% (1 8. 6\% – 27 .6 \% ) 23 .5 \% (1 8. 8\% – 29 .0 \% ) 22 .9 \% (1 1. 2\% – 41 .1 \% ) 45 .7 \% (3 9. 1\% – 52 .3 \% ) 26 .5 \% (2 0. 7\% – 33 .2 \% ) Fo rm al ed u ca ti o n a N ev er sc h o o le d 24 .1 \% (1 9. 9\% – 28 .8 \% ) 9. 5\% (7 .0 \% – 12 .7 \% ) 12 .6 \% (1 0. 9\% – 14 .6 \% ) 11 .0 \% (4 .7 \% – 23 .5 \% ) 51 .8 \% (3 1. 0\% – 72 .1 \% ) 17 .5 \% (1 2. 8\% – 23 .5 \% ) ≤ Pr im ar y sc h o o l 18 .1 \% (1 3. 7\% – 23 .6 \% ) 5. 2\% (2 .9 \% – 9. 3\% ) 12 .7 \% (1 0. 5\% – 15 .3 \% ) 7. 4\% (3 .7 \% – 14 .4 \% ) 42 .4 \% (3 3. 0\% – 52 .4 \% ) 31 .1 \% (2 1. 0\% – 43 .5 \% ) Se co n d ar y sc h o o l 13 .0 \% (1 0. 1\% – 16 .5 \% ) 4. 6\% (2 .4 \% – 8. 9\% ) 5. 5\% (4 .0 \% – 7. 5\% ) 3. 1\% (1 .3 \% – 7. 4\% ) 25 .0 \% (2 0. 0\% – 30 .8 \% ) 8. 4\% (4 .8 \% – 14 .3 \% ) C o lle g e 4. 7\% (1 .6 \% – 13 .1 \% ) 1. 6\% (0 .7 \% – 4. 0\% ) 6. 7\% (2 .7 \% – 15 .6 \% ) 1. 6\% (0 .7 \% – 3. 6\% ) 15 .1 \% (1 0. 0\% – 22 .2 \% ) 1. 5\% (0 .6 \% – 3. 6\% ) M ar it al st at u sb N ev er m ar rie d 0. 9\% (0 .3 \% – 3. 0\% ) 7. 8\% (2 .3 \% – 23 .2 \% ) 1. 1\% (0 .4 \% – 3. 0\% ) 1. 3\% (0 .7 \% – 2. 4\% ) 7. 8\% (4 .4 \% – 13 .4 \% ) 12 .1 \% (5 .5 \% – 24 .7 \% ) M ar rie d 14 .7 \% (1 2. 6\% – 17 .2 \% ) 3. 5\% (2 .1 \% – 6. 0\% ) 10 .3 \% (9 .1 \% – 11 .7 \% ) 4. 3\% (2 .5 \% – 7. 3\% ) 17 .4 \% (1 2. 4\% – 24 .0 \% ) 9. 2\% (5 .5 \% – 14 .9 \% ) D iv o rc ed /w id o w ed 25 .2 \% (1 9. 9\% – 31 .5 \% ) 11 .7 \% (8 .9 \% – 15 .1 \% ) 19 .1 \% (1 5. 9\% – 22 .7 \% ) 19 .0 \% (8 .1 \% – 38 .4 \% ) 36 .4 \% (2 9. 1\% – 44 .4 \% ) 19 .3 \% (1 2. 8\% – 28 .1 \% ) M en w it h se lf- re p o rt ed ar th rit is (n = 26 64 ) C h in a n = 11 45 G h an a n = 23 0 In d ia n = 57 8 M ex ic o n = 77 Ru ss ia n Fe d er at io n n = 36 3 So u th A fr ic a n = 27 1 A g e st ra ta (y ea rs ) 18 – 29 1. 3\% (0 .2 \% – 8. 8\% ) − 2. 1\% (1 .0 \% – 4. 7\% ) − − 0. 7\% (0 .1 \% – 3. 4\% ) 30 – 39 5. 5\% (2 .4 \% – 12 .1 \% ) 0. 2\% (0 .0 \% – 1. 4\% ) 6. 1\% (3 .8 \% – 9. 8\% ) − 14 .6 \% (5 .4 \% – 34 .1 \% ) 1. 3\% (0 .3 \% – 5. 8\% ) 40 – 49 12 .0 \% (7 .9 \% – 18 .0 \% ) 3. 7\% (1 .5 \% – 8. 7\% ) 7. 9\% (5 .1 \% – 12 .1 \% ) 2. 9\% (0 .6 \% – 13 .2 \% ) 4. 7\% (1 .3 \% – 15 .9 \% ) 0. 9\% (0 .3 \% – 3. 0\% ) 50 – 59 13 .7 \% (1 1. 8\% – 15 .8 \% ) 7. 4\% (5 .4 \% – 10 .1 \% ) 13 .7 \% (1 1. 3\% – 16 .5 \% ) 0. 9\% (0 .3 \% – 2. 6\% ) 21 .6 \% (9 .5 \% – 42 .2 \% ) 12 .6 \% (9 .3 \% – 16 .8 \% ) 60 – 69 20 .0 \% (1 7. 7\% – 22 .5 \% ) 11 .6 \% (8 .6 \% – 15 .4 \% ) 16 .9 \% (1 3. 8\% – 20 .6 \% ) 8. 0\% (4 .7 \% – 13 .3 \% ) 21 .3 \% (1 5. 2\% – 29 .0 \% ) 28 .2 \% (2 2. 1\% – 35 .2 \% ) 70 + 22 .9 \% (2 0. 7\% – 25 .2 \% ) 16 .7 \% (1 2. 6\% – 21 .7 \% ) 17 .8 \% (1 4. 5\% – 21 .7 \% ) 9. 7\% (6 .3 \% – 14 .5 \% ) 37 .8 5 (3 0. 3\% – 46 .0 \% ) 20 .9 \% (1 3. 5\% – 30 .9 \% ) Fo rm al ed u ca ti o n a N ev er … SYNTHESIS Saving Mothers, Giving Life: It Takes a System to Save a Mother Claudia Morrissey Conlon,a Florina Serbanescu,b Lawrence Marum,c Jessica Healey,d Jonathan LaBrecque,a Reeti Hobson,e Marta Levitt,f Adeodata Kekitiinwa,g Brenda Picho,h Fatma Soud,i Lauren Spigel,j Mona Steffen,e Jorge Velasco,k Robert Cohen,a William Weiss,a on behalf of the Saving Mothers, Giving Life Working Group A multi-partner effort in Uganda and Zambia employed a districtwide health systems strengthening approach, with supply- and demand-side interventions, to address timely use of appropriate, quality maternity care. Between 2012 and 2016, maternal mortality declined by approximately 40\% in both partnership-supported facilities and districts in each country. This experience has useful lessons for other low-resource settings. ABSTRACT Background: Ending preventable maternal and newborn deaths remains a global health imperative under United Nations Sustainable Development Goal targets 3.1 and 3.2. Saving Mothers, Giving Life (SMGL) was designed in 2011 within the Global Health Initiative as a public–private partnership between the U.S. government, Merck for Mothers, Every Mother Counts, the American College of Obstetricians and Gynecologists, the government of Norway, and Project C.U.R.E. SMGL’s initial aim was to dramatically reduce mater- nal mortality in low-resource, high-burden sub-Saharan African countries. SMGL used a district health systems strengthening approach combining both supply- and demand-side interventions to address the 3 key delays to accessing effective maternity care in a timely manner: delays in seeking, reaching, and receiving quality obstetric services. Implementation: The SMGL approach was piloted from June 2012 to December 2013 in 8 rural districts (4 each) in Uganda and Zambia with high levels of maternal deaths. Over the next 4 years, SMGL expanded to a total of 13 districts in Uganda and 18 in Zambia. SMGL built on existing host government and private maternal and child health platforms, and was aligned with and guided by Ugandan and Zambian maternal and newborn health policies and programs. A 35\% reduction in the maternal mortality ratio (MMR) was achieved in SMGL-designated facilities in both countries during the first 12 months of implementation. Results: Maternal health outcomes achieved after 5 years of implementation in the SMGL-designated pilot districts were substantial: a 44\% reduc- tion in both facility and districtwide MMR in Uganda, and a 38\% decrease in facility and a 41\% decline in districtwide MMR in Zambia. Facility deliveries increased by 47\% (from 46\% to 67\%) in Uganda and by 44\% (from 62\% to 90\%) in Zambia. Cesarean delivery rates also increased: by 71\% in Uganda (from 5.3\% to 9.0\%) and by 79\% in Zambia (from 2.7\% to 4.8\%). The average annual rate of reduction for maternal deaths in the SMGL-supported districts exceeded that found countrywide: 11.5\% versus 3.5\% in Uganda and 10.5\% versus 2.8\% in Zambia. The changes in stillbirth rates were significant (�13\% in Uganda and �36\% in Zambia) but those for pre-discharge neonatal mortality rates were not significant in either Uganda or Zambia. Conclusion: A district health systems strengthening approach to addressing the 3 delays to accessing timely, appropriate, high- quality care for pregnant women can save women’s lives from preventable causes and reduce stillbirths. The approach appears not to significantly impact pre-discharge neonatal mortality. INTRODUCTION Despite a 45\% drop in global maternal deathsbetween 1990 and 2015,1 maternal mortality remains an intractable public health problem in many low-resource settings. Only 1 sub-Saharan African country, Rwanda, achieved the target for Millennium Development Goal 5 (reduce by three-quarters, between 1990 and 2015, the maternal mortality ratio).1,2 Attempts have been made to bring high-level a Bureau for Global Health, U.S. Agency for International Development, Washington, DC, USA. b Division of Reproductive Health, U.S. Centers for Disease Control and Prevention, Atlanta, GA, USA. cCenters for Disease Control and Prevention, Lusaka, Zambia. Now retired. d U.S. Agency for International Development, Lusaka, Zambia. Now based in Monrovia, Liberia. e Bureau for Global Health, U.S. Agency for International Development, Washington, DC. Now with ICF, Rockville, MD, USA. f Bureau for Global Health, U.S. Agency for International Development and RTI, Washington, DC, USA. Now with Palladium, Abuja, Nigeria. g Baylor College of Medicine Children’s Foundation-Uganda, Kampala, Uganda. h Infectious Diseases Institute, College of Health Sciences, Makerere University, Kampala, Uganda. i Centers for Disease Control and Prevention, Lusaka, Zambia. Now an inde- pendent consultant, Gainesville, FL, USA. j ICF, Fairfax, VA, USA. Now with Ariadne Labs, Boston, MA, USA. k U.S. Agency for International Development, Papua, New Guinea. Correspondence to Claudia Morrissey Conlon ([email protected]). Global Health: Science and Practice 2019 | Volume 7 | Supplement 1 S6 mailto:[email protected] visibility to the cause, but many countries have not directed sustained political attention or sufficient resources to eliminate preventable maternal mortality3—despite solid evidence of the profound effects a mother’s death has on her family, her community, and on development in general.4,5 The situation is particularly dire in sub-Saharan African countries where 60\% of global maternal deaths occur.1,5,6 In these coun- tries, obstetrical risk is compounded by high fertil- ity rates, raising the lifetime risk of death due to childbirth to 1 in 36, compared with 1 in 8,400 in the European Union.7–9 Newborns fare no better. Globally, the reduc- tion in newborn deaths has not kept pace with the reduction of deaths in children under age 5, with newborn deaths now contributing to nearly half of child mortality.1 The average neonatal mortality rate is 27 deaths per 1,000 live births in low- income countries compared with 3 deaths per 1,000 live births in high-income countries. Eight of the 10 most dangerous places to be born are in sub-Saharan Africa.10 In 2011 the Office of the Global Health Initiative (GHI) within the U.S. Department of State was tasked with designing an endeavor that would bring public and private investment to- gether with committed Ministry of Health (MOH), national, and district leaders to address maternal mortality in sub-Saharan Africa.11,12 It was felt that a highly visible, well-financed, bold initiative similar to the U.S. President’s Emergency Plan for AIDS Relief (PEPFAR), the President’s Malaria Initiative, and Feed the Future was needed to inspire and recruit new public and private actors to the cause, while energizing and mobilizing the global health and development communities. The resulting initiative was Saving Mothers, Giving Life (SMGL), a public–private partnership. SMGL was composed of 6 U.S. agencies: GHI; the United States Agency for International Development (USAID) (which took over oversight of the partner- ship from GHI in July 2012 and responsibility as Secretariat from Merck for Mothers in 2014); the U.S. Centers for Disease Control and Prevention (CDC); the Office of the Global AIDS Coordinator (OGAC); Peace Corps; and the Department of Defense. It also included the Governments of Norway (became inactive in 2014), Uganda, Zambia, and Nigeria (joining in 2015 as the third SMGL country and slated to end in October 2019); Merck for Mothers; Every Mother Counts; the American College of Obstetricians and Gynecologists; and Project C.U.R.E (joined the partnership in 2013). SMGL’s initial goal was to decrease maternal mortality by 50\% in 1 year in SMGL-designated districts in Uganda and Zambia, building on existing national public health plat- forms and systems, and aligning with country maternal health strategies and aspirations.13,14 At the end of the first phase of the partnership, the time frame for the goal was extended to the close of the initiative in 2017. An additional goal of reducing the neonatal mortality rate by 30\% was added in 2013. The Saving Mothers, Giving Life journal sup- plement consists of 11 articles on the SMGL initia- tive. The articles describe the formation and function of the partnership, the SMGL theory of change, programming approach and costs, and the results achieved in Uganda and Zambia where implementation ended in October 2017 (Table 1). It aims to answer key questions about the initia- tive and identify outstanding implementation issues. Results from Nigeria will be reported in 2019 after implementation in that country has ended. THEORY OF CHANGE The SMGL theory of change model was built on a district health systems strengthening approach. It was designed to surmount the critical demand- and supply-side delays that prevent women and newborns from receiving lifesaving care in a timely manner, while strengthening the capacity and resilience of the health care system (Figure 1).15 The governments of Uganda and Zambia, their public health systems, the PEPFAR- and USAID- supported maternal and child health platforms, and private for-profit and nonprofit providers were critical inputs and served as the foundation for SMGL’s contributions to the district maternity care system. Evidence-based interventions were designed to address all key delays, be context- specific, and strengthen the capacity of the district health system. Four outcomes were anticipated: (1) increased use of services and improved self- care, (2) timelier access to appropriate care, (3) im- proved quality and experience of care, and (4) a more robust and resilient district health system. It was hypothesized that if these 4 outcomes were achieved together, SMGL-designated populations would see a substantial decrease in maternal and perinatal mortality. Implementation of the SMGL theory of change followed 7 organizing principles: SMGL’s initial goal was to decrease maternal mortality by 50\% in 1 year in selected districts in Uganda and Zambia. The SMGL theory of change was built on a district health systems strengthening approach. It Takes a System to Save a Mother www.ghspjournal.org Global Health: Science and Practice 2019 | Volume 7 | Supplement 1 S7 http://www.ghspjournal.org 1. Reap system-level synergies by addressing all 3 delays to obtaining lifesaving maternal and newborn care concurrently: delays in seeking appropriate care, delays in reaching services in a timely manner, and delays in receiving quality care at a health facility with the capacity to perform 9 signal emer- gency obstetric and newborn care (EmONC) functions.16–22 2. Recognize the district health system, which extends from community health workers to district hospitals (and to higher levels of care through referrals), as the primary unit for strengthening capacity.23–25 Potential inter- ventions should be assessed in terms of their contributions to improving the functioning of the entire district-level system. 3. Apply a “whole market approach,” which requires identifying and including both public and private inputs (e.g., providers, delivery systems, stakeholders) in planning, execu- tion, and evaluation in a designated district. Together they form the district maternity safety net. 4. Focus on improving services during the most vulnerable period for mothers and newborns— labor, delivery, and early postpartum. Inter- ventions at this time have the possibility of saving the lives of mothers and newborns and preventing fresh stillbirths. The level of fresh stillbirths is often seen as an indicator of the quality of care during labor and delivery. 5. Strengthen the capacity of the health care system to provide comprehensive emergency obstetricand newborn care (CEmONC) within 2 hours of travel time from home or a deliv- ery site for all pregnant women, approxi- mately 15\% of whom will experience a life-threatening complication, many with- out clear predictors.26,27 6. Integrate maternal and newborn health (MNH) services with other reproductive health services, including (1) HIV counseling and testing services to maximize identification and treatment of seropositive pregnant women and prevent mother-to-child trans- mission, and (2) postpartum family planning for women wishing to delay their next pregnancy. 7. Count, analyze, and report all maternal and perinatal deaths along with the cause of TABLE 1. Saving Mothers, Giving Life Supplement Articles Article No. Article Title 1 Saving Mothers, Giving Life: it takes a system to save a mother 2 Impact of the Saving Mothers, Giving Life approach on decreasing maternal and perinatal deaths in Uganda and Zambia 3 Addressing the first delay in Saving Mothers, Giving Life districts in Uganda and Zambia: approaches and results for increasing demand for facility delivery services 4 Addressing the second delay in Saving Mothers, Giving Life districts in Uganda and Zambia: reaching appropriate maternal care in a timely manner 5 Addressing the third delay in Saving Mothers, Giving Life districts in Uganda and Zambia: ensuring adequate and appropriate facility-based maternal and perinatal health care 6 The costs and cost-effectiveness of a district-strengthening strategy to mitigate the 3 delays to quality maternal health care: results from Uganda and Zambia 7 Saving lives together: a qualitative evaluation of the Saving Mothers, Giving Life public-private partnership 8 Community perceptions of a 3-delays model intervention: a qualitative evaluation of Saving Mothers, Giving Life in Zambia 9 Did the Saving Mothers, Giving Life initiative expand timely access to lifesaving care in Uganda? A spatial district-level analysis of travel time to emergency obstetric and newborn care 10 Saving Mothers, Giving Life approach for strengthening health systems to reduce maternal and newborn deaths in 7 scale-up districts in northern Uganda 11 Sustainability and scale of the Saving Mothers, Giving Life approach in Uganda and Zambia It Takes a System to Save a Mother www.ghspjournal.org Global Health: Science and Practice 2019 | Volume 7 | Supplement 1 S8 http://www.ghspjournal.org death; improve completion of facility records and registries; institutionalize maternal and perinatal death surveillance and response (MPDSR) in each district and foster high- level awareness of these reviews among tradi- tional, religious, and political leadership to learn from each preventable death and pro- mote necessary health system and cultural changes. COUNTRY CONTEXT In 2011, Uganda and Zambia were chosen as the first SMGL-supported countries based on (1) their interest to the Global Health Initiative; (2) high levels of maternal mortality—MMR of 420 in Uganda and 262 in Zambia in 20101; (3) solid MOH commitment to decreasing mater- nal and newborn mortality, as evidenced by their Roadmap to Accelerate Reduction of Maternal and Neonatal Mortality and Morbidity and Campaign to Accelerate the Reduction of Maternal, Newborn, and Child Mortality in Africa plans; and (4) the existence of robust PEPFAR- and USAID-supported maternal and child health platforms.28–30 Direct causes of maternal deaths were similar in both countries, with postpartum hemorrhage being the leading cause followed by preeclampsia/eclampsia, sepsis, obstructed labor/ruptured uterus, and complica- tions of unsafe abortions.1 The most deadly indi- rect causes were malaria and HIV.29,31 Inadequate skilled human resources for health were a major constraint to providing effec- tive coverage in both countries.29,31 When SMGL began, the human resources vacancy rate at health facilities in SMGL-supported districts was 40\% in both Uganda and Zambia.11,12,32–34 Uganda and Zambia also shared high HIV rates (7\% and 12\% among adults ages 15 to 49, respec- tively) and their total fertility rates were among the highest in the world (6.2 for both countries) FIGURE 1. Saving Mothers, Giving Life Theory of Change Model Abbreviations: EmONC, emergency obstetric and newborn care; MCH, maternal and child health; MPDSR, maternal and perinatal death surveillance and response; MMR, maternal mortality ratio; NMR, neonatal mortality rate; PEPFAR, U.S. Presidents Emergency Plan for AIDS Relief; SMGL, Saving Mothers, Giving Life; USG, U.S. Government. Source: Adapted from Saving Mothers, Giving Life.57 It Takes a System to Save a Mother www.ghspjournal.org Global Health: Science and Practice 2019 | Volume 7 | Supplement 1 S9 http://www.ghspjournal.org (Table 2). Less than half of births in Zambia, and 57\% in Uganda, were attended by skilled birth attendants and the cesarean delivery rates were low at 5\% in Uganda and 3\% in Zambia. Neonatal mortality rates were 27 and 34 per 1,000 live births in Uganda and Zambia, respec- tively (Table 2). PROJECT DESIGN, IMPLEMENTATION, AND ASSESSMENT SMGL Learning Districts Four districts each in Uganda and Zambia were selected for SMGL support by their MOH based on the large numbers of deliveries and maternal deaths, the availability of existing implementing partners working in the district, and national pri- orities. The 8 districts in total, designated as the SMGL learning districts, were mostly rural and poor.8,11,12,30,31 Figure 2 shows the learning dis- tricts and the scale-up districts. Over the life of the initiative, the 4 learning districts in each coun- try were administratively split further to total 6 learning districts in each country. In Zambia, the 4 initial learning districts were spread across the country with 2 in Eastern Province (Nyimba and Lundazi), 1 in Southern Province (Kalomo), and 1 in Luapula Province (Mansa). The 4-district population was 880,000 with 46,157 deliveries in 2011. Throughout the initiative, 110 health facilities were engaged, 94\% public and 6\% private, including 16 health posts, 88 health centers, and 6 hospitals.11,35 Uganda’s SMGL-supported districts (Kyenjojo, Kamwenge, Kabarole, and Kibaale, aka “the 4Ks”) were contiguous and located in Western Uganda. The population in the 4Ks was 1.75 mil- lion with 78,400 deliveries in 2011. Throughout the initiative, 105 delivering facilities, 61\% public and 39\% private (18 health centers II, 70 health centers III, 11 health centers IV, and 6 hospitals), were supported by SMGL.12,36 SMGL Phases The SMGL initiative was divided into 3 phases: Phase 0—design and startup (June 2011 to May 2012), Phase 1—proof of concept (June 2012 to December 2013), and Phase 2—scale-up and scale-out (January 2014 to October 2017). Phase 0: Design and Startup Initiative design. Design of the SMGL district health systems strengthening approach began in mid-2011 under the aegis of the Global Health TABLE 2. Uganda and Zambia National-Level Indicators at the Start of the SMGL Initiative Indicator Uganda Zambia Maternal mortality ratio (per 100,000 live births) 420a 262a Deliveries in facilities 57\%b 48\%c Births by cesarean delivery 5\%b 3\%c Birth attended by skilled birth attendant 57\%b 47\%c Antenatal care coverage: at least 4 visits 48\%b 60\%c HIV prevalence among adults 15–49 7\%d 12\%d Pregnant women with HIV receiving antiretroviral therapy 61\%d 93\%d Total fertility rate 6.2b 6.2c Modern contraceptive prevalence rate among all women 15–49 21\%b 25\%c Neonatal mortality rate (per 1,000 live births) 27b 34c Abbreviation: SMGL, Saving Mothers, Giving Life. a 2010 data from Trends in Maternal Mortality: 1990 to 2015. Estimates by WHO, UNICEF, UNFPA, World Bank Group and the United Nations Population Division (https://www.who.int/reproductivehealth/publications/monitoring/maternal-mortality-2015/ en/). b 2011 data from Uganda Demographic and Health Survey 2011 (https://dhsprogram.com/pubs/pdf/FR264/FR264.pdf). c 2007 data from Zambia Demographic and Health Survey 2007 (https://www.dhsprogram.com/pubs/pdf/FR211/FR211 [revised-05-12-2009].pdf). d 2011 data from UNAIDS AIDSinfo (http://aidsinfo.unaids.org/). It Takes a System to Save a Mother www.ghspjournal.org Global Health: Science and Practice 2019 | Volume 7 | Supplement 1 S10 https://www.who.int/reproductivehealth/publications/monitoring/maternal-mortality-2015/en/ https://www.who.int/reproductivehealth/publications/monitoring/maternal-mortality-2015/en/ https://dhsprogram.com/pubs/pdf/FR264/FR264.pdf https://www.dhsprogram.com/pubs/pdf/FR211/FR211[revised-05-12-2009].pdf https://www.dhsprogram.com/pubs/pdf/FR211/FR211[revised-05-12-2009].pdf http://aidsinfo.unaids.org/ http://www.ghspjournal.org Initiative. The Global Health Initiative convened a design team of MNH and HIV technical experts in project development, implementation, costing, policy formulation, and monitoring and evalua- tion. The aim was to create a highly visible, bold initiative that would galvanize global action and financial support. A draft SMGL model was devel- oped, guided by GHI principles and informed by extensive examination of the evidence base and modeling from the Lives Saved Tool (LiST). (Supplement 1) A goal was established to reduce maternal mortality in SMGL-supported facilities in Uganda and Zambia by 50\% in 1 year and an implementation plan was formulated. A notable feature of the plan was that partner funding for SMGL implementation was only guaranteed for an initial 12-month period; if performance was deemed subpar, funding for SMGL could end. After country and district selection, the U.S. ambassadors for Uganda and Zambia assigned coordination roles to U.S. agency heads (USAID mission director, CDC director, PEPFAR coordina- tor, Peace Corps lead, and Department of Defense liaison), and interagency working groups were formed. The working groups collaborated with national, provincial, and district MOH-designated SMGL leads (usually district health officers) and implementing partners, forming SMGL country teams. The country teams initially met weekly and then monthly to develop plans and lever- age existing partner programs and capabilities. Country teams then created intensive 1-year workplans for the pilot districts in Uganda and Zambia based on addressing the 3 delays and strengthening the system. The rapid design and execution of the initial SMGL 1-year plan required the participation of existing implementing partners working in SMGL-selected districts. Between Uganda and Zambia, 39 implementing partners were identi- fied, most with set workplans and deliverables (Supplement 2). Under the leadership and super- vision of MOH district health management teams and district health and medical officers, extant implementing partner workplans were adapted to support SMGL country and district plans. Evaluation design. The ability to assess and report health outcomes resulting from SMGL efforts required robust evaluation. The headquar- ters monitoring and evaluation (M&E) commit- tee, composed of specialists from CDC and USAID, developed an ambitious evaluation plan for Phase 1 that was endorsed by the ministries of health and implementing partner representatives in both countries.37 The plan included ongoing FIGURE 2. Saving Mothers, Giving Life-Designated Learning and Scale-Up Districts in Uganda and Zambia Source: Adapted from Saving Mothers, Giving Life.57 It Takes a System to Save a Mother www.ghspjournal.org Global Health: Science and Practice 2019 | Volume 7 | Supplement 1 S11 http://ghspjournal.org/lookup/suppl/doi:10.9745/GHSP-D-18-00427/-/DCSupplemental http://ghspjournal.org/lookup/suppl/doi:10.9745/GHSP-D-18-00427/-/DCSupplemental http://www.ghspjournal.org enumeration of all maternal deaths with verbal autopsies to ascertain cause of death. (See the arti- cle by Serbanescu and colleagues from the SMGL supplement.38) Thirty-one indicators were selected for moni- toring care at all delivering facilities through quar- terly record and registry reviews in SMGL- supported districts in Uganda and Zambia (Supplement 3). In Uganda, these data were col- lected through Pregnancy Outcomes Monitoring Studies; data were also gathered and displayed monthly at selected SMGL facilities in Uganda using a simple matrix referred to as “BABIES” (Birthweight by Age-at-Death Boxes for Inter- vention and Evaluation System), which provided short-loop feedback to improve newborn care. Formative special studies37 included a qualitative study of women’s and communities’ perceptions of childbirth in Zambia and a 2-hour travel-time mapping study in Uganda.39 (See the article by Schmitz and colleagues from the SMGL supplement.40) Baseline assessment. During Phase 0, base- line studies were undertaken in the 8 learning dis- tricts. MMRs were measured through a census with verbal autopsies of deaths among women of reproductive age in Zambia and a Reproductive Age Mortality Survey (RAMOS) in Uganda. (RAMOS uses a variety of sources to identify all deaths of women of reproductive age and decide which of these are maternal- or pregnancy- related.) Health facility assessments (HFAs) of capacity and readiness of the system to provide 9 lifesaving signal functions were undertaken in all public and private delivering facilities in the SMGL-supported districts (Table 3). This enabled planners and implementers to take stock of the existing availability of basic and comprehensive emergency obstetric and newborn care. HFAs were carried out at 3 time points during SMGL: (1) at baseline, to inform SMGL planning and design and to identify needed investments; (2) at the end of the pilot year in 2013 to gauge progress and inform funding and operational decisions during subsequent years; and (3) at endline in 2017 to assess outcomes. Common gaps identified from the baseline HFA included the following: � Delay 1: Demand. ThenumberofGovernment- established community health workers, village health teams (VHTs) in Uganda and Safe Motherhood Action Groups (SMAGs) in Zambia, was inadequate. Women booked late for antenatal care visits and attendance of 4 or more antenatal care visits was low (46\% in Uganda).41 � Delay 2: Access. Women had limited access to comprehensive CEmONC facilities within 2 hours (only 51\% to 55\% of women were able to reach CEmONC within 2 hours using motorized vehicles) due to few operating thea- ters and blood banks, and lack of transport vehicles and referral protocols. Maternity wait- ing homes were often dilapidated and deserted. � Delay 3: Quality. Many maternity blocks in hospitals and health centers were run-down and overcrowded, and they lacked water, elec- tricity, and functioning toilets. Equipment was missing, inoperative, or insufficient for the client load. Facilities lacked 24-hour staffing of skilled birth attendants, anesthetists, and surgeons. � Health Systems Strengthening. In the face of limited quality improvement activities, facili- ties experienced frequent drug and supply stock-outs and weak capture, analysis, and reporting of health outcome data. TABLE 3. Emergency Obstetric and Newborn Care 9 Signal Functions Basic Services Comprehensive Services 1. Administer parenteral antibiotics Perform signal functions 1 through 7 plus: 2. Administer uterotonic drugs (i.e., parenteral oxytocin, misoprostol) 8. Surgery (cesarean delivery) 3. Administer parenteral anticonvulsants for preeclampsia (i.e., magnesium sulfate) 9. Blood transfusion 4. Manually remove the placenta 5. Remove retained products of conception (e.g., manual vacuum extraction, misoprostol, dilation and curettage) 6. Perform assisted vaginal delivery (e.g., vacuum extraction, forceps delivery) 7. Perform basic neonatal resuscitation (e.g., bag and mask) Source: WHO, UNFPA, UNICEF, and Mailman School of Public Health.27 SMGL developed a robust evaluation plan that included ongoing enumeration of all maternal deaths with verbal autopsies to ascertain cause of death. It Takes a System to Save a Mother www.ghspjournal.org Global Health: Science and Practice 2019 | Volume 7 | Supplement 1 S12 http://ghspjournal.org/lookup/suppl/doi:10.9745/GHSP-D-18-00427/-/DCSupplemental http://www.ghspjournal.org These gaps and other district-specific challenges were addressed in SMGL district workplans. Startup. Startup activities began early in 2012. At the national level in Uganda and Zambia, routine meetings were held with the interagency working groups, MOH representa- tives, and implementing partners. Preparations for work with private providers through the Programme for Accessible Health Communication and Education (PACE) project and Marie Stopes International were initiated in Uganda. In Zambia, where the SMGL learning districts were spread out across the country, SMGL district coordinators— often retired midwives—were hired to harmonize all SMGL activities in their district with district health officers and district health management teams, and to serve as a link with implementing partners. During this phase, training commenced for providers and existing government-sponsored community health workers—SMAGs and VHTs. These health workers were recruited from the local community. Groups were a mix of men and women and often included former traditional birth attendants. SMGL provided these volunteers with resources such as gumboots, flashlights, T-shirts, and bicycles. In Zambia, Peace Corps volunteers were recruited and trained as community mobiliz- ers to work with SMAGs to increase demand and organize community transport systems. By the end of the initiative, SMGL-dedicated Peace Corps volunteers were in all 18 SMGL-supported districts. Phase 1: Proof of Concept Results for Phase 1 are based on data for the 12-month period from June 2012 through May 2013. …