The global maternal sepsis study and awareness campaign (GLOSS): Study protocol

Background: Maternal sepsis is the underlying cause of 11% of all maternal deaths and a significant contributor to many deaths attributed to other underlying conditions. The effective prevention, early identification and adequate management of maternal and neonatal infections and sepsis can contribute to reducing the burden of infection as an underlying and contributing cause of morbidity and mortality. The objectives of the Global Maternal Sepsis Study (GLOSS) include: the development and validation of identification criteria for possible severe maternal infection and maternal sepsis; assessment of the frequency of use of a core set of practices recommended for prevention, early identification and management of maternal sepsis; further understanding of mother-to-child transmission of bacterial infection; assessment of the level of awareness about maternal and neonatal sepsis among health care providers; and establishment of a network of health care facilities to implement quality improvement strategies for better identification and management of maternal and early neonatal sepsis. Methods: This is a facility-based, prospective, one-week inception cohort study. This study will be implemented in health care facilities located in pre-specified geographical areas of participating countries across the WHO regions of Africa, Americas, Eastern Mediterranean, Europe, South East Asia, and Western Pacific. During a seven-day period, all women admitted to or already hospitalised in participating facilities with suspected or confirmed infection during any stage of pregnancy through the 42nd day after abortion or childbirth will be included in the study. Included women will be followed during their stay in the facilities until hospital discharge, death or transfer to another health facility. The maximum intra-hospital follow-up period will be 42 days. Discussion: GLOSS will provide a set of actionable criteria for identification of women with possible severe maternal infection and maternal sepsis. This study will provide data on the frequency of maternal sepsis and uptake of effective diagnostic and therapeutic interventions in obstetrics in different hospitals and countries. We will also be able to explore links between interventions and maternal and perinatal outcomes and identify priority areas for action. This is a facility-based, prospective, one-week inception cohort study. During a seven-day period, between 00:00 h, Tuesday 28 November 2017 to 23:59 h Monday 04 December 2017, all women who spend at least 12 h in a participating health care facility (admitted to or already hospitalised) with suspected or confirmed infection during any stage of pregnancy through the 42nd day after abortion or childbirth will be included in the study. Eligibility criteria, and the sources and methods of selection of the participants of this study are provided at three levels: countries and geographical areas within countries, health care facilities and individual participants. This study will be implemented in pre-specified geographical areas of participating countries across the WHO regions of Africa, Americas, Eastern Mediterranean, Europe, South East Asia, and Western Pacific. The invited countries are in Fig. 1. This selection was prepared considering the burden of maternal sepsis and infection-related mortality (based on the latest available WHO estimates, 2015), estimated birth rate and number of births per year (UN data 2015, http://data.un.org/), geographical diversity, and feasibility assessment based on country participation in previous WHO multi-country research, capacity to identify potential country coordinators and current country situation (e.g. not a conflict zone). Researchers and staff from Ministries of Health based in these countries were contacted and invited to implement the study. Countries that were invited to participate in the Global Maternal Sepsis Study and Awareness Campaign. Disclaimer: The boundaries and names shown and the designations used on this map do not imply official endorsement or acceptance by the World Health Organization or the Global Maternal Sepsis Study researchers In addition, six high-income countries were identified through a multinational collaboration of organisations conducting prospective population-based studies of serious illnesses in pregnancy and childbirth, the International Network of Obstetric Survey Systems (INOSS). These countries will apply a slightly modified protocol adapted to their existing surveillance systems. The eligibility criteria for geographical areas, facilities and women will be as described in this protocol. In each participating country, purposive sampling of at least one geographical area was carried out considering the presence of all following criteria: All health care facilities in the geographical area, independently of their administrative organization (public, private, charity, faith-based, social security), presenting at least one of the characteristics below were eligible to participate in this study: All eligible facilities located in the selected geographical areas were invited to participate in this study. For selection of maternity hospitals, a convenient minimal number of births per facility/per year (e.g. minimum of 1000 births/year), or minimal level of care (e.g. tertiary and secondary level, national and district hospitals) was fixed at the country level to ensure a minimal coverage of about 80% of all facility-based births in the geographical area. Inclusion criteria: Women in the participating facilities presenting any of the conditions below during pregnancy, birth, postpartum period or post-abortion (either spontaneous or induced) will be eligible to participate in this study: Reference list of infections associated with systemic repercussions during pregnancy, childbirth, post abortion and postpartum period (modified from ICD-MM, the WHO Application of ICD-10 to deaths during pregnancy, childbirth, and the puerperium) Exclusion criteria: Women presenting the following conditions will not be eligible, unless they present with systemic repercussion due to infection. For example: All women enrolled during the identification week will be followed-up until discharge from the facility, transfer outside the geographical area or death, whichever occurs first. The maximum follow-up period will be 6 weeks for pregnant women if still hospitalised in participating facilities, regardless of the pregnancy outcome at the end of the follow-up period. Infants born to women enrolled in the study will be included and followed-up until hospital discharge, transfer outside the participating area, infant death, or 7 days after birth (if still in the hospital). Appendix 1 lists potential bias that this study may incur and the efforts that will be implemented to address anticipated potential sources of bias, based on the Critical Appraisal Skills Programme – CASP [32]. Data will be collected at the geographical area, facility and individual level using paper forms specially designed for this study. These forms were based on validated tools used in previous multi-country surveys and facility assessment tools, and were customized for this study. The forms were piloted in at least one hospital in the majority of the participating geographical areas. Forms were translated into French, Portuguese, Russian and Spanish and additional official country languages by professional translators as needed. A one-off geographical area questionnaire will be completed by country coordinators to collect information on the main characteristics of the area, including: estimated population size, number of births (or deliveries) and maternal and neonatal deaths, health services organization (e.g. total number of health care facilities), human development index and epidemiology of infectious diseases in the area (endemic diseases and outbreaks). Data will be gathered from civil registries and epidemiological surveillance systems. In each facility, a one-off facility questionnaire will be completed to collect information on structural characteristics of each of the participating facilities: level of specialisation, volume and activity (number of births, maternal and perinatal deaths, frequency of selected obstetric interventions (caesarean section, instrumental vaginal births)), infrastructure (laboratory and other diagnosis services, special or intensive care units, emergency obstetric and neonatal care), resources (monitoring equipment, oxygen, fluid resuscitation, antibiotics, disposables, staff), availability of written protocols for prevention or management of infections, access to water, sanitation and hygiene (WASH) services. Data will be gathered from the heads of department or other authorised staff in the health facility during the data collection period. The individual data form will collect information on eligible women and their neonates, including: socio-demographic characteristics, reproductive history, diagnoses and treatments, fetal and neonatal outcomes, complications and management. Candidates predictors of possible severe maternal infections and sepsis will be also collected (Table 2). Summary of candidate predictors (Adapted from Barton and Sibai [35], Edwards 2015 [25], Albright et al. [36]) Detailed pre-specified clinical and laboratory variables will be collected throughout a 72-h time window before and after suspicion/diagnosis of infection, based on WHO near-miss criteria and obstetric early warning trigger systems (MEOWS) and scoring systems of inflammation (SIRS) and organ dysfunction (SOFA, SOS, APACHE II, MODS, LODS, IGS). Data will be collected from electronic and/or paper maternal and neonatal medical records. In case of doubt or missing information, the health provider caring for the participant could be approached for clarifications or completion of missing information. The medical records will be accessed for up to 3 months after completion of the data collection at the individual level in each facility. Only information on routine clinical monitoring, laboratory and other investigations related to the usual management of suspected and confirmed infections and reported in the medical records will be collected in this study. The study will not require additional collection of any laboratory, diagnostic or other investigations if not performed as part of standard care of included women. There will be no direct interaction of members of the study team with eligible women for other reasons than those of their usual clinical practice, and to inform women about the study, respond to their questions and seek consent when required. A composite of maternal deaths and maternal near-miss cases with reported infection as an underlying or contributing cause. The main analysis that requires a minimum sample size in this study is the development of identification criteria for possible severe maternal infection and maternal sepsis. In this analysis, the diagnostic accuracy of each candidate predictor will be tested against the main outcome of interest (i.e. maternal deaths and maternal near-miss cases with infection as an underlying or contributing cause). Infections are estimated to be an underlying or contributing cause in 25% of all maternal deaths or maternal near-miss cases [2], which corresponds to approximately 25 cases per 10,000 births. Considering the low prevalence of the primary outcome, the resulting sample asymmetry (i.e. number of women with the primary outcome compared to those without the primary outcome), the uncertainty around the prevalence of infections, a convenient and conservative sample of 100 cases with the primary outcomes was selected. This sample corresponds to approximately the upper interquartile range of samples sizes of diagnostic accuracy for the median number of participants with the target condition necessary to determine the test sensitivity (49 events (interquartile range 28–91)) [33]. A convenience sample size was estimated based on the total expected number of births that would have to be monitored to ensure 100 cases with the primary outcome. Based on an average global birth rate of 19.6 live births per 1000 population in a year (UN Data, http://data.un.org) approximately 50 geographical areas with 2,000,000 inhabitants have to be included in the study to cover about 40,000 births in 1 week. Assuming a 7% frequency of infections requiring hospital admission, we expect to have a total sample size of 2800 eligible women included in this study. Additional details are provided in the Fig. 2. Estimated sample size. In grey boxes women to be included in the study.150 geographical areas with 2,000,000 inhabitants, with global birth rate of 19.6 live births per 1000; 2Two million live births per year × mean gestation period (40 weeks/52 weeks of year), not adjusted to account for abortions, miscarriages or stillbirths;3 Includes pregnancy related infection and infections complicating pregnancy, childbirth and the postpartum period (ICD-MM). Regardless of cause of admission (e.g. childbirth) and whether primary or secondary infection (e.g. postoperative, aspiration pneumonia); 4 Based on WHO Multi-country Study 2010–2011 [37] Table 3 shows estimates of the number of women expected to be included per health facility during the one-week inception cohort study, according to activity of health facilities (number of live births). Estimates of number of women expected to be included during 1 week according to volume of health facilities (number of live births/year) In order to achieve Objective 6, an awareness campaign will be launched early November in the facilities participating in the study before data collection. Its aim is to sensitize health care providers on maternal and neonatal sepsis. The specific objectives of the campaign are to improve providers’ awareness of maternal and neonatal sepsis and identification of those cases during the study period in participating facilities, and to foster increased awareness of this condition pre- and post-study period. Public and policy makers will be specifically targeted in a subsequent stage of the campaign. The awareness campaign will include a dedicated website (http://srhr.org/sepsis/), a media/communications toolkit, infographics and social media communications. All these materials will be developed by a communications company with extensive experience in global health campaigns and made available for free to participating sites in 6 languages. Four specific activities are planned around the campaign with the following objectives: These activities are: Surveys will be available in the eight languages, as indicated by preference from country coordinators to ensure a maximum response rate (Arabic, English, French, Italian, Portuguese, Russian, Spanish, and Vietnamese). Snowball sampling will be used to reach health care providers in participating facilities. Country coordinators will be asked to send the link to the survey, or paper questionnaires, to facility coordinators, and at their turn facility coordinators will be asked to recruit other participants in their facilities. Weekly reminders will be sent to all participants during survey collection period. All women will be informed about the implementation of the study in the health facility using posters. Care will be taken to place the information in areas visible to the women and translated into local languages. There will be a statement confirming confidentiality and that all records will be de-identified. The study team will inform all eligible women about the study and the need to review their medical records for this purpose, as well as those of their neonates as soon as they meet any of the inclusion criteria. Women and their families will be informed that they can contact their provider if they have any question about the study or can inform their provider at any time if they want to recuse themselves from the study. All women will be free to refuse participation confidentially and without prejudice. After that, if the woman does not express any objection data will be extracted, including information on her neonate, once she is discharged from hospital. In the case of those women who are unstable upon presentation, the above-stated information will be provided as soon as they are stable and able to understand the materials and/or communicated with their next of kin. For illiterate women the information will be shared with her partner, other family members or any other witness of her choice, and read to them by a study staff member. This study will be performed in accordance with all stipulations of the protocol and in compliance with the International Ethical Guidelines for Health-related Research Involving Humans, 2016, regarding use of routine clinical care data. It is anticipated that written individual consent for inclusion in the study and data collection will not be required in most of the participating countries and/or facilities. Where possible, a modified informed consent process and a waiver of documentation of consent will be requested (opt-out). This waiver of documentation of consent will apply both for documentation of the women’s own consent and documentation of parental or guardian consent for participation of her baby/ies. This will mitigate the risk of selection bias (differences between participants and non-participants) that may be introduced by documented informed consent, although the direction and magnitude of the effect may be difficult to predict [34]. The documentation of informed consent could also affect the total expected number of participants and jeopardise the overall validity of the study if the final sample size is not sufficient to produce the planned analysis. This is an observational study requiring no deviation from routine medical practice, and therefore participants will experience no more than minimal risks and no direct and/or immediate benefits from study participation. Principal risks are those associated with a breach of confidentiality concerning the woman’s participation in the study, but existing routine data will be abstracted anonymously and retrospectively from medical records. The study does not involve interviews, direct observations or any medical or other interventions in patient care. We do not use any biological samples or record genetic information. All pregnant or recently pregnant women are at risk of developing infections so we do not anticipated additional risk (e.g. stigmatization) regarding the examination of medical records of a subgroup of women who actually developed the infection. In addition, no information will be collected that could jeopardise psychological integrity (e.g. psychiatric information). All ethical approvals from national and/or local ethics committees will be obtained before implementation of the study protocol, as required by national legislation. The protocol will be adjusted in case local or national regulations require informed consent from participants or any other changes in the study protocol, including participation of minor participants and whether there is a duty to report errors observed either prospectively or as part of medical chart review to patients and/or authorities. Investigators will always adhere to the most rigorous requirements regarding informed consent and protection of participants and retention of study documentation. Participants will have the right to access, rectify, cancel or oppose on demand the information obtained during the study. The decision not to take part of the study, or withdrawal of participation, will be documented in writing and signed by the woman or her representative. A mechanism will also be put in place to allow women to ask retrospectively to be pulled out of the survey after data extraction. All women and their families will be informed of these mechanisms and that they can withdraw their data at any time and without any charges or losses. Completion of the facility form will be subject to the agreement of the head of department. Agreement from the hospital administration will be obtained, if required. Authorisation from hospital ethical committees will also be sought if necessary. All identifiers for semi-structured interviews and online surveys will be kept confidential. Interviews will be audio recorded, and transcripts will remove all direct identifiers before publication of any results. Informed consent will be obtained from participants before each interview. The audio recordings and transcripts will be deleted after they have been analysed and published. The online surveys will be voluntary and will require participants accept to participate before completing it. General information about the position and geographical location of the survey respondents will be collected. Names and email addresses of respondents to the online survey will be kept in a confidential database to be able to contact them during the post-intervention period. Only the research team will have access to identifiers. Participants will be given the option to recuse themselves from the activities at any point during the interviews, and/or survey. No identifying information will be noted during participant observation, and any conversations resulting from this activity will not be recorded. Results of these activities will be published without attributing responses to any specific person or institution. The Department of Reproductive Health and Research (HRP/RHR), including UNDP/UNFPA/UNICEF /WHO/ World Bank Special Programme of Research, Development and Research Training in Human Reproduction (HRP), at WHO is the sponsor of the study and performs overall coordination. A Technical Advisory Group, comprising HRP/RHR staff, regional coordinators and content experts, was constituted to develop the study protocol, oversee and make decisions related to the implementation and progress of the study, and provide technical guidance. A Data Management Committee comprising HRP/RHR and Centro Rosarino de Estudios Perinatales (CREP) staff developed the data management and analysis plan, and will conduct primary analysis of the data. A regional coordinating committee was set up in each study region based on geographical and language variability, as follows: African English- and Portuguese-speaking countries, African French-speaking countries, the Americas, Asia, Europe and the Middle East. These committees ensure selection of sites and implementation of the study at the regional and national levels. There is a country coordinator responsible in overseeing implementation of the study at the country level. A facility coordinator will be identified in each facility ward to ensure daily identification of eligible subjects, data collection and implementation of the awareness campaign. Modalities of data management including data ownership, collection, storage, protection, analysis, sharing and retention between all the parties involved in the study will be stipulated in a Standard of Operating Procedures for Data Management. The overall data management will be at CREP in Rosario, Argentina. Trained research assistants will visit all wards where eligible women could stay in the participating health care facilities, including but not limited to: gynaecologic, female ward, obstetrics/postpartum/ post-abortion wards, labour wards, adult general medical ward, intensive care unit, high-dependency unit, emergency room, operating theatre, post-operative room, pharmacy (to check if there is any recipient of antibiotics), laboratory (to check for cultures, swabs, antibiotic sensitivity test), infection prevention and control unit (to check for any reported infection in the eligible population) and the mortuary. In each of these units, the medical records of all pregnant or recently pregnant women will be screened daily during the identification week against the eligibility criteria by the responsible nurse or health professional in each hospital or the trained research assistant, and health providers assisting women admitted in those settings will be asked about the presence of women with any of the eligibility criteria. The responsible nurse or health care professional (facility coordinator) in each facility will be asked to flag all potential eligible cases and approached daily for the identification of any women who could be considered as potentially eligible to participate in the study. Potentially eligible women will also be identified using electronic data sources or hospital registries where these are available. Data entry will be centralised at the country level to maximise use of resources, standardize data collection and avoid inclusion of duplicates. Data entry of facility and geographical area forms will be centralised at CREP. Overall monitoring of the study will be performed by regional coordinators, and country coordinators in each participating country. A Manual of Operations will be developed to ensure standardized and accurate data collection among facilities and study investigators. Investigator’s meetings will be organized at the country and/or facility level before initiation of the study to ensure correct implementation of the study protocol and data collection. The total number of women eligible at each facility data will be monitored daily during the identification week and these numbers will be compared to those determined by the data collection. Visual inspection of the completed data collection forms will be performed at the facilities and national level to ensure completeness, reliability and consistency of the data before data entry. A customized online open-source data entry and monitoring system will be developed for the study. The data entry system will minimize data entry errors, delays in data queries and completion of incomplete forms. Data collection and entry procedures will be compliant with the HRP/RHR Standard Operating Procedures and Good Clinical Practice (GCP) guidelines. Random monitoring visits will be organised during and after data collection period to evaluate adherence to the protocol and to perform data quality verification, according to country capacities. Additional visits may be carried out depending upon facility or country activity and performance. A random sample of facilities will be selected and forms will be checked against medical records to ensure accuracy and reliability of data collected. In addition, an independent person will check inclusions and information extracted against hospital/medical records for all cases of maternal admissions to intensive or high dependency care units/beds and all maternal deaths. In addition, at the end of the study period maternal and neonatal admissions will be cross-checked against hospital registries (e.g. retrospective checks against ICD-10-CM and ICD-MM coding strategies, labour ward book, discharge registries) to ensure all eligible cases were included. Upon completion of the study and verification of data, data will be screened for accuracy and completeness, after which the database will be locked from any additional changes. These procedures have been successfully used in previous large multi-country studies coordinated by HRP/RHR. Subject confidentiality and anonymity will be maintained at all times by the sponsor, regional and country coordinators, and staff in participating facilities. This will be ensured by removal of all identifiers from any data collected for this study at the individual and study site levels. No names or other directly identifying information (addresses, dates) will be entered in the regional or global databases including medical data. A unique pre-defined identification number will be attributed to each included participant and study site. Participants’ numbers will be linked to investigator records stored separately and securely making it possible to identify the case to correct missing or erroneous data. Identifying information will be maintained by the responsible person in each hospital in accordance with regulatory agencies requirements and will not be transmitted to the country coordination, centralised data manager (CREP) or WHO. Study sites numbers will be provided to the country coordinators by the central data manager (CREP). Identifying information linking names of the site and study sites numbers will be maintained by CREP and will not be transmitted to the regional coordination or WHO. Strict rules will be established for storage of the data bases and the lists of subjects included in study (on electronic media in locked cupboards with access restricted to principal investigators and study investigators) as well as protection of data files on computers (firewalls, password encryption, etc.). All data will be published as aggregates at the study site, area/country or regional level. It won’t be possible to identify study sites from the published data. For the main analysis, and whenever possible in secondary analysis, geographical areas or countries will not be identify in published data. This information may be however useful depending on the secondary analysis (e.g. antimicrobial resistance patterns across countries). The research consortium will establish rules governing the period of time that identifying information on participants will be maintained and this information will be included and agreed in ethics submissions. A minimal period of 3 years will be applied. Data items that are partially identifying (e.g. dates) will be removed from the regional and global databases before they are pooled into a common database. Specific measures (e.g. timing between diagnosis and receipt of a given intervention) will be calculated before data transfer. HRP/RHR and CREP will conduct data analysis. These units have the technical and personnel capacities to perform all statistical analysis. An analysis plan will be developed before initiation of the study. Primary analysis will be performed on aggregate data and secondary analysis will be performed by region or country as appropriate. Descriptive analysis will be carried out to show the frequencies of maternal and fetal complications, use of interventions, and the relationship between use of interventions and outcomes which will be reported as rates or means and 95% confidence intervals, in each participating country, region and in the pooled sample. Analysis will be stratified by partum status. The following analysis will be undertaken: At the facility level, characteristics of the facilities and the regions will be described using data collected from the facility and country surveys. When applicable, point estimates will be reported at the national, regional or global level. Individual and facility characteristics associated with the use of specific interventions for prevention or management of infections and sepsis will be identified, and differences between regions/countries and health facilities will be investigated. When applicable, analysis will consider the multilevel structure of the data. The association between the interventions and outcomes at the regional, country, unit and individual levels will be investigated by the introduction of variables representing the characteristics at appropriate level of the statistical model. We will apply standard diagnostic accuracy assessments of candidate predictors against the primary outcome of interest including the following approaches: Given the low frequency of the primary outcome, for some candidate predictors secondary surrogate predictors will be considered. Two sets of criteria will be developed, one for identification of possible severe maternal infections at the moment of suspicion or diagnosis of infection, and another one to define maternal sepsis at hospital discharge or death. The interviews will be audio recorded and transcribed word-for-word so the results can be analysed by members of the research team. Interviews will be analysed by looking for themes and categories that emerge from reading the transcripts in addition to hand-written notes taken during the interview process. Through an iterative process of analysis, saturation of categories will occur, as well as development of subcategories or new categories. This analysis will result in a few broad central themes that can be linked to a general analytic framework in constructing a theory on existing barriers and facilitators to provider awareness. Participant observation will help complement the interviews by offering supplemental information on provider behaviour and campaign execution. Descriptive analysis of knowledge, attitudes and practices of respondents to the survey will be compared at baseline and post-campaign. The following means of dissemination will ensure the widest possible distribution: