Improvement of District-Level Maternal Mortality Reporting with - - PDF document

Improvement of District-Level Maternal Mortality Reporting with Reproductive Age Mortality Survey (RAMOS) Joseph Adamako, Anthony Ofosu, Gloria Asare, Tiffany Anthony, Judy Idrovo, Bradley Iott, Andrea Momoh, Rachael Ward, Elisa Warner and Frank Anderson Abstract Objective:. To develop a process for community based maternal mortality surveillance in low resource settings. Methods: District health workers adapted the reproductive age mortality survey (RAMOS) to conduct maternal mortality surveillance among women of reproductive age (WRA) who died between June 2008 and June 2013. The survey was comprised of four simple yes or no questions with two follow up questions. A trained community health nurse conducted verbal autopsy to those women who answered positively to one of four questions. Maternal mortality review committees established the cause of death and contributing factors. Findings: 359 WRA died in the district during the specified time period. A “yes” answer to any

• f the 4 RAMOS questions identified 132 women who experienced either a maternal death, late

maternal death or who died while pregnant or within one year of termination of pregnancy. In the resultant 108 available verbal autopsies, 59 died a maternal or late maternal deaths and 49 died a non-maternal death. The most common causes of maternal death, were hemorrhage (24%) and abortion (17%). Conclusions : The 4+2 RAMOS is a practical method for improving maternal mortality surveillance at the community level, increasing the maternal mortality ratio 158% (from 128 to 359). This study demonstrates that community based surveillance for maternal mortality amongst deaths to WRA is feasible in low resource settings. Increased ascertainment can uncover critical areas for intervention. Scale up to national community health training programs will provide accurate information for national efforts to reduce preventable maternal mortality.

Introduction: Accurate ascertainment of cases of maternal mortality is dependent on a health information system that accurately identifies and records cause of death. Because these systems are not available in some low income countries, estimates of maternal mortality based on survey methods, hospital records and fertility rates provide the data for decision making() The reproductive age mortality survey (RAMOS) has been used in a number of settings to increase ascertainment of maternal mortality (2,3). The RAMOS survey contains 39 questions and spans the many possible causes of mortality to women of reproductive age, some of which pertain to maternal death (4 who ramos instrument). The Ghana Maternal Mortality Survey, which utilized a RAMOS methodology in Accra at health facilities and morgues, found 44% under-reporting of maternal mortality (5). Community based surveillance for maternal mortality using a modified RAMOS survey was conducted in the Sene district of Ghana in 2010 (6). Ten questions were chosen from the RAMOS survey that were most likely related to maternal mortality and could be asked by community health workers to a family member where a woman of reproductive age (WRA) had

• died. A maternal mortality review committee investigated all cases with a positive answer to one
• f the 10 questions, and using the committee's determination of maternal death as the gold

standard, 4 questions emerged as the most predictive: was she pregnant when she died?; was she recently pregnant?; did she have a child less than one year old when she died?; and did she die from miscarriage or abortion? (6) These 4 simple questions were found to have a high predictive value for detecting a maternal

• mortality. Additionally, in the year that community based surveillance for maternal mortality
• ccurred, the number of maternal deaths almost doubled. A similar study in Eritrea showed

similar results (3 ). Methods: This observational study was approved by the Ghana Health Service Ethical Review board and was reviewed by the University of Michigan Institutional Review Board and found to be exempt. Community-based health volunteers and sub-district nursing staff were trained by the Kuntanase Government Health Department in the RAMOS protocol and survey administration. The review

• f data took place between May 2013-August 2014.

In May of 2013, trainers and community health workers discussed the most effective ways to ask the four central questions: if the deceased was pregnant at the time of death, if she was recently pregnant when she died, if she had a child less than one year of age, and if she had a miscarriage

• r abortion. District staff also added 2 additional questions: did she die in a healthcare facility or

at home, and do you know why she died? (RAMOS 4+2). Between May and August 2013, the community based surveillance volunteers and health nurses conducted the interviews in the local language, Twi, and elaborated on any questions that needed further explanations to get the final answer. The community health workers, also called Community Based Surveillance (CBS) workers, assisted in surveillance by identifying all WRA who died in the last five years. Small incentives were given for identifying the women’s families and debriefing them ahead of time as to the details of the RAMOS 4+2 methodology. In addition, the CBS workers attended a training meeting at the Kuntenase District Hospital for a full debriefing of the survey, project training, and a chance to pose questions regarding the material. Permission was asked of each family member whether or not they would participate, and no incentive was given to any interviewee. RAMOS results were available for 357 WRA who died during the period of review. 225 WRA were confirmed as non-maternal deaths as determined by

• RAMOS. 132 WRA were confirmed as possible maternal deaths by answering one of the first

four questions positively. These deaths include maternal deaths, late maternal deaths, pregnancy- associated deaths up to one year. (Table 1) Thirteen families of WRA could not be located for interview or declined interview. Verbal Autopsy Following the administration of RAMOS surveys, community health nurses and midwives conducted verbal autopsies at households where positive answers to one of four indicative questions of the RAMOS survey was found. The verbal autopsy reviews the deceased woman’s regional demographics, obstetric history, hospital attendance, pregnancy preparedness, socio- cultural factors and a written conclusion (7). Upon completion of verbal autopsies, two maternal mortality review committees (MMRC) took place to evaluate and discuss the factors contributing to each death and to determine if the death was the result of a pregnancy complication, or issues that pertained to the pregnancy (maternal deaths or late maternal deaths). MMRCs are group of health professionals that come together and examine medical and technical aspects of care for women who have died. They then identified patterns of adverse outcomes related to non-medical and system related factors. MMRCs consisted of 10-11 participants; an OB/GYN, a pharmacist, midwives, and senior and entry level community health nurses. During the meeting, the RAMOS survey results and verbal autopsy interviews for each woman were analyzed. Definitions of maternal death, pregnancy related death, and late maternal death provided by the World Health Organization and the International Statistical Classification of Diseases and Related Health Problems, 10th revision (ICD-10) were utilized in determining whether the WRA’s death was of maternal causes (Table

1). Additional follow-up was sometimes required for those cases in which a conclusion could not be drawn. At the facility level, hospitals and clinics where the death occurred were visited to

• btain more information and cross check data. At the community level, family members of the

deceased were asked for further details pertaining to the cause of death. The maternal mortality ratio was calculated using the formula in Table 2. Twenty three hospital ascertained deaths were recorded for the period between 2008 and 2013. Using the average number of live births for this period (17,913), the maternal mortality ratio was calculated as 128.4. These numbers were obtained from public record and will be used as a point of reference for data collected moving forward. The positive predictive values (PPV’s) for RAMOS questions one through four were calculated by dividing the number of positive responses among the maternal deaths by the number of positive response answers for all true positive RAMOS (Maternal deaths/ [Maternal deaths +Non-maternal deaths]). In addition, the sensitivity of question six (What do you think was the cause of her death?) was calculated by dividing the number of responses which indicated possible cause of death as a likely maternal death by all determined maternal deaths. In addition, the PPV, NPV, sensitivity, and specificity were calculated for a maternal death- related response to question six (What do you think was the cause of her death?). PPV was calculated by dividing the number of confirmed maternal death cases that gave a maternal death related response to question six by all cases in which gave a maternal death related response to question six. NPV was calculated by dividing the number of confirmed non-maternal cases that gave non-maternal related response to question six by all cases in which gave a non-maternal related response to question six. Sensitivity was calculated by dividing the number of confirmed maternal death cases that gave a maternal death related response to question six by all confirmed maternal death cases. The specificity was calculated by dividing the number of confirmed non- maternal cases that gave non-maternal related response to question six by all confirmed non- maternal cases. Results The average age of the women at the time of death was 30.9 years (range 18-42 years). 18% of the women died within the community, while 78% died in a healthcare facility. The average gravidity among these 59 deaths was 3.4, and the average parity was 2.6. RAMOS survey respondents were most often the mother (24/59), sister (11/59) or grandmother (8/59). The remaining survey respondents were husbands, brothers, or other close family members. Community health workers identified 369 women of reproductive age in the Bosomtwe district who had died during the review period (Chart 1). 13 families were either unable to be located or declined interview. The 4+2 question RAMOS was administered to the 357 remaining families. In 132 cases a family members answered “Yes” to one or more survey questions, while 225 had family members that answered “No” to all four. Of the 132 positive surveys, verbal autopsies were conducted with 118 families to gain

information and identify cause of death. 14 of the cases were lost to follow up. 10 verbal autopsies were found to have been performed erroneously (ie a positive RAMOS that was not accurate). The remaining 108 cases were reviewed by the maternal mortality review committee, which found 59 maternal deaths (maternal deaths and late maternal deaths) and 41 non-maternal

• deaths. 8 cases were undetermined. The most common causes of maternal death, were

hemorrhage (24%), abortion (17%), infection (0.05%), hypertensive diseases such eclampsia and preeclampsia (0.03%). Of these 59 cases deemed maternal deaths, 32% of known maternal deaths had an unknown cause of death. The causes of non-maternal deaths varied widely; ranging from malaria to sickle cell. The corrected maternal mortality ratio for the entire district increased to 329/100.000 live births. The positive predictive values for questions 1-4 are listed in Table 3. The questions that yielded the highest PPV were Q1: “Was she pregnant when she died?” (86.36% PPV), and Q3: “Was she pregnant within the year before she died?” (67.12% PPV). The overall PPV of questions 1-4 was 59%. Question 6, when taken alone, had a PPV of 85%, a negative predictive value (NPV) of 55%, and sensitivity and specificity at 50% and 87%, respectively. Discussion The current study suggests that community based surveillance among deaths to women of reproductive age is feasible and useful in the identification of cases of maternal mortality. The large discrepancy in hospital-based maternal mortality ratios (128) and those in the entire area (329) suggest many maternal deaths remain unreported. The RAMOS 4+2 survey instrument serves as an effective and efficient means to improve surveillance within the Bosomtwe region and has implications for other rural areas. District staff noted that the survey was easy to use and distribute amongst the community health workers. They were able to understand the questions and obtain clear answers from the interviewees. Incorporation into existing community health workers training on a national level would be the most ideal implementation strategy. In addition, the brevity of the survey is important, as interviewing families regarding the loss of a loved one can be difficult. When coupled with a verbal autopsy, this process provides communities with information on the main factors contributing to maternal death. Identifying these causes can serve as a helpful tool in ascertaining improvements that need to be made in medical care, medical outreach, and possible public health interventions. An accurate surveillance system allows public health professionals and government agencies to be aware of common illnesses and causes of death to better serve each population affected. In the absence of a comprehensive health information system that includes birth and death certificates, incorporation of the RAMOS 4+2 methodology into routine surveillance . Positive answers should be followed up with a verbal autopsy to identify cause of death and associated factors. No single RAMOS question identified all maternal deaths. Positive answers to any of the four

dichotomous questions provided the base group for verbal autopsies. We sought to determine the value of the open ended question about cause of death. If families accurately could identify material causes, perhaps the other 4 questions would not be necessary. In this group, however, family members identified only 50% of the maternal deaths accurately. However, we feel that an

• pen-ended question has value as it allows for expanded discussion beyond the 4 yes/no
• questions. It is also useful in eliminating the need for verbal autopsies if the death occurred

during pregnancy but was clearly not related to the pregnancy (such as accident or homicide). During data analysis, it was determined that Question 4, “Did she have a miscarriage or an abortion?” did not specify whether the miscarriage/abortion happened during the most recent pregnancy or during a previous pregnancy. Although this did not affect data analysis, this question should indicate whether an abortion or miscarriage took place during the current pregnancy. Although 6 simple questions were utilized, translation of the survey instrument into the local language required input from community health nurses and CHWs. Tailoring of the questions was needed to adapt of question to the local language idioms that would most accurately reflect the questions’ intent. Such adaptation will be necessary for implementation of RAMOS 4+2 in

• ther communities.

By incorporating RAMOS 4+2 as an active maternal death surveillance program, potential maternal mortality cases that are otherwise unknown to public health officials will be identified and allow for interventions to be targeted to reduce maternal mortality and to improve maternal

• health. In areas where community workers already collect surveillance data, the RAMOS 4+2

could be easily incorporated into monthly activities in rural communities. The WHO consensus statement on preventable maternal mortality calls for significant reductions in maternal mortality ratios by 2030 (ref). The use of community-based surveillance systems will be critical to knowing if we have completely achieved this goal. References:

• 1. World Health Organization. (2014). Trends in maternal mortality: 1990 to 2013. Estimates by

WHO, UNICEF, UNFPA, The World Bank and the United Nations Population Division. Retrieved from http://www.who.int/reproductivehealth/publications/monitoring/maternal- mortality-2013/en/.

• 2. Mungra A, et al. Reproductive age mortality survey to study under-reporting of maternal

mortality in Surinam. European Journal of Obstetrics & Gynecology and Reproductive Biology

• Vol. 77 (p37–39) 1998.
• 3. Determining the Level of Maternal Mortality in Eritrea using RAMOS (Reproductive Age

Mortality Study).Mismay Ghebrehiwet, MD, MPH, PhD1 and Richard H. Morrow MD2

• 4. RAMOS Instrument: http://whqlibdoc.who.int/hq/1987/WHO_FHE_87.7.pdf Last accessed:

February 24, 2015

• 5. Ghana Statistical Service (GSS), Ghana Health Service (GHS), and Macro International.
• 2009. Ghana Maternal Health Survey 2007. Calverton, Maryland, USA: GSS, GHS, and Macro

International.

• 6. Geynisman, Julia, Andrew Latimer, et al. (2011) Improving maternal mortality reporting at the

community level with a 4-question modified reproductive age mortality survey (RAMOS). International Journal of Gynecology and Obstetrics.

• 7. World Health Organization. (November, 2013). Verbal Autopsy Standards: 2012 WHO

Verbal Autopsy Instrument. Retrieved from http://www.who.int/healthinfo/statistics/WHO_VA_2012_RC1_Instrument.pdf.

• 8. http://apps.who.int/iris/bitstream/10665/130776/1/WHO_RHR_14.21_eng.pdf?ua=1&ua=1

Box 1 RAMOS 4+2 Questions Q1: Was she pregnant when she died? Q2: Did she have a child less than 1 year old when she died? Q3: Was she pregnant within the year before she died? Q4: Did she have a miscarriage/abortion? Q5: Did she die at home or in a health care facility? Q5a: If she died in a health care facility, which facility was it? Q5b: If she died at her own home? If not her own home, where? Was it her own home or someone else’s? Q6: What do you think was the cause of her death?

Tables and Figures

Table 1

Definitions of Maternal Death per ICD-10 Maternal death The death of a woman while pregnant or within 42 days of termination of pregnancy, irrespective of the duration and site of the pregnancy, from any cause related to or aggravated by the pregnancy or its management but not from accidental or incidental causes. Pregnancy-related death The death of a woman while pregnant or within 42 days of termination of pregnancy, irrespective of the cause of death. Late maternal death The death of a woman from direct or indirect obstetric causes, more than 42 days, but less than one year after termination of pregnancy.