The 2010 Demographic and Health Survey revealed that in 5 years, Tanzania reduced maternal mortality from 578 to 454 maternal deaths per 100,000 live births [35, 36]. While emergency obstetric care (EmOC) is available at regional hospitals, a recent study of 22,243 live births in rural districts of Tanzania found that only 29% of deliveries occurred in a hospital [3].
This project was a component of the Empower II Project conducted by the Ifakara Health Institute (IHI), which implements Maternal, Newborn and Child Health (MNCH) interventions in real world settings to produce learning for improvements and scale up. Empower I lasted from July 2007 to June 2012 and Empower II launched in March 2013. Collectively, Empower works in four rural regions of Tanzania and targets a total population of 3,919,342 of which 813,092 are women of reproductive age. The NASG/CUG component described below was conducted from November 2014 to June 2016. Members of the University of California, San Francisco (UCSF) Safe Motherhood Program served as consultants for the NASG portion of the project.
This project component addressed challenges around prompt management of OH and around OH referral between facilities through the introduction of the NASG (LifeWrap-NASG, Hong Kong) along with a Closed User Group (CUG) mobile phone network for public dispensaries, health centers, and three public referral hospitals, as well as three private referral hospitals in the eight districts of Geita, Nyang’hwale, Kalambo, Sumbawanga, Singida, Ikungi, Shinyanga, and Kahama (Fig. 1 for a map of the regions). A total of 280 facilities were enrolled: 17 Comprehensive Emergency Obstetric Care (CEmOC) facilities and 263 Basic Emergency Obstetric Care (BEmOC) facilities (17 health centers and 246 dispensaries). To qualify as a CEmOC referral facility, a site needed to have an operating theater and the capacity to perform a blood transfusion. BEmOC level health centers served a population of 50,000 people and provide inpatient care. Dispensaries provided the lowest level of care, serving a population of 10,000 people, and offering delivery beds and outpatient services. While dispensaries were intended to provide BEmOC care, not every dispensary was completely able to do so (the majority were not able to perform these signal functions: parenteral anticonvulsants, removal of retained product, and assisted instrumental delivery).
In addition to the NASG, the project introduced mobile phones on a CUG network. Five hundred eight cell phones were supplied, three per CEmOC facility, 2 per BEmOC facility, and four for each of the 8 Council Health Management Teams (CHMT) with phone service from the Airtel telecommunications company. Phones on a CUG network may call other phones within the network without airtime fees, so that health workers could communicate about patients, call for an ambulance, and contact IHI project staff for information and support.
The study trained participants to use the NASG for pregnant and postpartum women with hypovolemic shock secondary to obstetric hemorrhage of any etiology: ectopic pregnancy, molar pregnancy, complications of abortion, retained placenta, placenta previa, uterine atony, uterine rupture, and genital trauma. Women were classified as having hypovolemic shock if they had any amount of bleeding combined with either a systolic blood pressure of ≤80 mmHg or altered consciousness. Women were included whether the hemorrhage began at home, at the facility, or at another site.
Training
Three phases of training were conducted: a Training of Trainers (TOT) in October 2014, a series of Phase 1 trainings for CEmOC facilities was conducted in November 2014, and a series of Phase 2 training for BEmOC facilities was conducted in February 2015. Sessions specific to the NASG were 1.5–2 days in length and taught a curriculum developed by the UCSF Safe Motherhood Program [curriculum and training videos in English and a video of training in Swahili are available at: http://www.safemotherhood.ucsf.edu/resources/videos/, the NASG/CUG content was incorporated into scheduled Empower Emergency Obstetric Care (EmOC) trainings.
At the TOT 19 trainers were certified to support ongoing cascade trainings as well as project supervision. Only trainers who were able to pass a skills test and had a written exam score of > 70% were certified to train for the project. Phase 1 consisted of four trainings, one per region at CEmOC facilities. Each facility sent four to five CEmOC providers, who were trained in clinical management of women with hypovolemic shock with the NASG and how to train other facility staff in NASG use. A phased approach was used, beginning project implementation only at district hospitals and other CEmOC health facilities for the first 3 months of the program (December 2014 – February 2015), before moving to Phase 2 training and introduction of NASGs to lower level facilities that referred patients to these CEmOC facilities. Phase 2 consisted of 16 trainings in Swahili for BEmOC health center and dispensaries, four trainings per region. The trainings were conducted in March–April 2015, BEmOC facilities sent two staff members to represent each dispensary and 3–4 staff for each health center. These trainings were integrated with another scheduled Empower training on maternal health service provision. A total of 863 staff were trained in Phase 1 and Phase 2 by IHI, UCSF, and TOT trainers, including surgeons, obstetricians, medical officers, midwives, clinical assistants, registered nurses, clinical officers, medical attendants, anesthetists, anesthesiologists, and enrolled nurses.
At the start of Phase 2 meetings were held with IHI project staff, district medical officers, and hospital administrators to review the logistics for implementation, as well as to hear their experiences and challenges during the initial 3 months.
In December 2015, 20 ambulances were equipped with NASGs (one per ambulance) and 21 drivers received training on applying and transporting women with NASGs. As an additional training tool for all levels of facilities, four-minute low resolution (7 MB) and high resolution (11 MB) videos that instructed on NASG application for average, short, large, and unconscious patients were uploaded to the CUG phones during supervision to be used to refresh trained staff or train new staff.
All phases of the training included multiple simulations for a variety of shock and hemorrhage situations (how to transport, placing on conscious and unconscious women, removing when a woman is stable and remains stable, replacing if a woman goes back into shock, etc.). Simulations have been documented to improve retention and team work [37, 38]. We also held unexpected “emergency simulations” during lunch and break times, when participants did not expect to be performing a simulation. An actor playing the husband would carry in an “unconscious and dying,” severely shocked patient/actress and participants had to quickly form a team and demonstrate how they would apply the NASG, manage the hemorrhage and shock of the unconscious and dying actress, and make a referral/transport if necessary.
At the trainings facility staff were informed that they would be receiving the CUG network phones, which became property of the Regional Medical Office. Those who accepted the CUG phones on behalf of the facility signed a contract outlining obligations for maintenance and use, as well as conditions in case of loss or damage. During the Phase 2 training, an introduction to the electronic data collection forms was presented. CEmOC facilities were given onsite training for the electronic forms when their CUG phones were distributed.
Phones distributed for use with the CUG network were Java-enabled Nokia phones. The closed user group was a subscription service through the Airtel company, which provided SIM cards for mobile phones to make calls at no cost to other phones on the network. The CUG phones were also programmed with a data entry system created on OpenXdata. Data was transmitted from the phones via General Packet Radio Service (GPRS) using an http protocol to send data to a local server running a MySQL database for storage and management. The data could be accessed remotely using a secure web-based platform located at: http://maternal.esurveillance.or.tz/.
Entry Criteria/Enrollment: Women were included in the study if they presented with or developed hypovolemic shock secondary to OH at any participating facility. Staff were instructed to apply the NASG for women presenting with signs of obstetric hemorrhage and hypovolemic shock, whether the patient would be referred out or managed on site.
Hemorrhage Management: All facilities had received training on EmOC through the Empower project. Evidence-based guidelines for PPH management included prophylaxis with uterotonics, treatment for PPH with uterotonics, Intravenous fluids, and referral/transport of women who continued to bleed despite uterotonic treatment.
Outcome Indicators: As this was an implementation project, the main outcomes were increased and appropriate use of NASG and CUG network. We set project targets for improvement at 75% increase from the baseline of 0, since both technologies had never been used before. Process indicators included: percentage of facilities with an NASG available and locatable in < 10 min, percentage of health facilities in compliance with established NASG disinfection protocols, percentage of facilities with a CUG network phone available, percentage of facilities able to show a charged CUG network, percentage of facilities that conducted regular training drills, and percentage of phones lost. We also compared baseline case fatality rates (CFR) from OH with endline CFR and predicted a decrease of 25%.
Data collection
From July –August 2014, IHI staff traveled to each of the 280 sites to collect baseline data. This consisted of reviewing all available hard copy registries to determine the CFR, looking at cases of reported OH and number of deaths attributed to OH for the 12 months from July 1, 2013 – June 30, 2014, the facility-based numbers were compared to the official Regional Reports. Baseline data were also gathered on availability of electricity, presence of cell phones for referrals, and transportation availability for all sites. The number of annual deliveries was gathered from district reporting.
Paper-based data collection on NASG use at the CEmOC facilities occurred from Dec 2014-March 2015, however as not all sites had begun NASG implementation, this data was used for training purposes only and not included in the analysis. CUG network phones were rolled out to all facilities during March–April 2015 during the Phase 2 BEmOC trainings. Both paper forms and electronic data entry forms were written in Swahili.
Facility staff were instructed to complete a case report for each woman with signs of obstetric hemorrhage. This data was recorded on a hard copy register for each site, and later entered into the electronic OpenXdata data entry form on the CUG phone and submitted to the central server by designated health care providers who had been trained for each facility. When there were discrepancies in the electronic records, data was verified against the paper-based register, and if additional verification was required, against other maternity and gynecology registries. In addition, sites sent month-end reports summarizing the number of deliveries, the number of obstetric hemorrhages, and the number of maternal deaths. This data was used to ensure 1) that all facilities were sending data on a regular basis 2) that the total number of OH reported was equal to the number of detailed OH case reports received. In cases of discrepancies or when the month-end report had not been submitted, the project team called facility staff using the CUG network to request the reports.
In addition to the patient data, the data form contained a free text comment field which allowed the provider to supply any additional information. Other fields allowed providers to give the reason they did not use uterotonics, the NASG, or the CUG phone for each patient if applicable.
Besides the quantitative electronic data on patient outcomes, we used a supervision tool to record the process indicators described above to ensure that the NASGs and CUG phones were at the facility and in working condition. These forms were completed by project staff during three rounds of supervision, with all 280 sites visited during each of the three rounds. The final forms were completed during the April–May 2016 supervision as an endline measure.
On-site supervision was conducted three times throughout the 15 months: from July – August 2015, December 2015 – January 2016, April – May 2016. Each site received three visits by project staff, at which time they were asked to demonstrate their skills, electronic reports were verified against paper records, and troubleshooting was performed for the phones. The physical condition of the NASGs and phones were verified and recorded on the supervision forms along with the number of staff trained, the number of staff not yet trained, and the frequency of drills.
Qualitative data collection
Feedback was sought to further enhance training and use of CUG and NASG. In addition to the patient data, the data forms contained the free text comment field noted above, which allowed the provider to supply any additional information. Qualitative informal interviews were conducted with facility staff during all three rounds of routine supervisory visits for sites, focusing on sites with apparent overuse (use on women not meeting criteria) or with lower numbers of uses than predicted.
For 2 weeks in July 2015, formal interviews were conducted with 19 facility staff, 15 of them working at the dispensary level, about their experiences with NASG. Thirteen of the staff had been trained, but not yet had a patient that required application of the NASG. Questions focused on ease of use, reasons for overuse or non-use, suggestions for improving training, and NASG acceptability. Due to the timing during the earlier stage of the project, most of the interviews focused on providers’ confidence and theoretical concerns following their training, as few providers had prior experience.
Data analysis Quantitative data analysis was conducted using R (R Core Team (2017). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/) and Stata 12.1 (StataCorp. 2011. Stata Statistical Software: Release 12. College Station, TX: StataCorp LP). Cohort study Risk Ratio (RR) methods within Stata were used to compute relative risk and Chi-Square statistics for the association of the use of the NASG with case fatality rates. Qualitative data analysis was conducted by seeking themes within the various responses to open-ended fields and to the structured one to one interviews.