Saving Lives With Better Gestational Age Estimation

Saving Lives With Better Gestational Age Estimation: Improving the Accuracy of Recall and Reporting of the Date of Last Menstrual Period (LMP) in Rural Bangladesh

Background (brief): Burden: Bangladesh has a high maternal (194 per 100,000 live births) and newborn mortality (28 per 1000 live births). In 2010, prematurity represented ~14% of all births and directly and indirectly contributed to 45% of all neonatal deaths. Gestational age (GA) is a key determinant of newborn survival and long-term impairment. Accurate estimation of GA facilitates timely provision of essential interventions to improve maternal and newborn outcomes. Knowledge gap: Last menstrual period (LMP) is a simple, low-cost self reported information, recommended by the World Health Organization for estimating gestational age but has issues of recall mainly among poorer, less educated women and women with irregular menstruation, undiagnosed abortion, and spotting during early pregnancy. Several studies have noted that about 20-50% of women cannot accurately recall the date of LMP. Relevance: The goal of this study is to improve maternal and newborn outcomes by increasing the accuracy of gestational age estimation, using menstrual based dating, that is vital for providing timely and necessary obstetric and newborn care interventions. The study will determine the efficacy of three community based interventions using e-platform targeted to improve the recall and reporting of the date of last menstrual period in a rural resource poor setting. The innovative e-platform based interventions, if successful, can provide substantial evidence to scale-up in a low resource setting where e-Health and m-Health initiatives are proliferating with active support from all sectors in policy and implementation. Hypothesis : Implementation of conventional and e-platform based interventions will lead to a 30% improvement in recall of the date of the LMP in adolescent girls and married women in rural Bangladesh. Intervention triggered improvement in LMP date recall among pregnant women in rural Bangladesh will improve the accuracy in GA estimation. Objectives: The main objectives of the study are to: Determine whether a set of conventional and e-platform based interventions improve recall of the date of the LMP in adolescent girls and married women in rural Bangladesh. Determine whether intervention triggered improvement in LMP date recall in rural Bangladesh improves the accuracy in GA estimation or not Methods: A 4- parallel arm, superiority, community based cluster randomized controlled trial comparing three conventional and e-platform based interventions is proposed to improve recall of GA with a control arm. The trial will be conducted among adolescent girls and recently married women (within past two years) with no children or a single child in Mirzapur sub-district of Bangladesh.The interventions include (i) education and calendar for recording date of LMP (ii) education and SMS based system for recording dates of LMP and reminders using mobile phone (normal) (iii) education and smart-phone based application for recording dates of LMP with an inbuilt reminder system. Outcome measures/variables: The trial has two primary outcomes, (i) improvement in the recall of LMP date among enrolled participants in the three different intervention compared to control arm and (ii) increased accuracy of LMP-based gestational age measurement, as compared to USG, following the intervention.

Background: Bangladesh has a high maternal (194 per 100,000 live births) and newborn mortality (28 per 1000 live births)(1-3). In 2010, prematurity represented ~14% of all births and directly and indirectly contributed to 45% of all neonatal deaths(4, 5). Accurate estimation of gestational age (GA) is a critical for providing optimal clinical care during the prenatal, delivery and postnatal period and for public health interventions(6). Knowing GA to the week is essential for determining the adequacy of fetal growth, correct diagnosis of preterm or post-term labor and also for the correct use of many interventions (e.g., antenatal corticosteroids therapy, tocolytics, continuous positive airway pressure) targeted to improve maternal and newborn outcomes(7). Three methods of GA estimation are primarily used - menstrual based dating (LMP), ultrasonography (USG) and neonatal assessment, all of which have some strengths and weaknesses(8). USG, based on biometric measurements of the fetus, is considered to be the gold standard when done before 20 weeks of gestation i.e., when biologic variations in fetal size and the effects of growth restriction are low(9). However, the accuracy of USG decreases when done in the latter trimesters, particularly in settings with high intrauterine growth restriction, as in Bangladesh(10). However, USG is often unavailable in low-resource countries, especially in rural areas. Even if it is available, the equipment can be of poor quality and operated by personnel with minimal expertise(10). In addition, pregnancies are often not identified until the 2nd or 3rd trimesters and women often seek antenatal care late in their pregnancies, which further limits the use of USG to assess GA(10-12). Alternate methods of assessing GA are therefore preferred in such settings. One approach is to use neonatal assessments that are based on standardized scoring systems of physical and neuromuscular maturity. These are, however, less precise then obstetric estimates and also require skilled personnel which is frequently not practical(13). Dating based on the LMP is another alternative. It is a simple, low-cost method recommended by the World Health Organization(14). However, several studies have shown that that approximately 15-45% of pregnant women are unable to recall their LMP accurately resulting in an unreliable estimate and digit preference(15-17). Recall period is crucial determinant influencing accuracy of LMP. Literature suggests that LMP recall is more precise with shorter recall period.(15) Accuracy in recall of LMP is also limited due to irregularity or individual variations in the length of the menstrual cycle, short birth spacing, preconception amenorrhea and implantation bleeding(18, 19). This is further complicated by late presentation for first antenatal care consultation and providers with minimal skills in probing for an accurate LMP date. Reliance on LMP alone has shown a tendency to overestimate GA at the extremes of gestation which is more common in women with no or low education and those living in poverty(15, 20). Despite the limitations, studies conducted in Bangladesh and Guatemala has suggested LMP to be the preferred method for determining GA in low-resource settings(10, 21). The study therefore aim to determine the efficacy of three interventions targeted to improve the recall and reporting of the LMP among women in rural Bangladesh. The interventions include both conventional LMP dating calendar as well as an innovative e-platform - mobile phones and smart phone applications. Although calendar, or menstrual dairies has already been tried in clinical settings for recording menstrual dates, public health implication of using these tools for recording LMP is yet to be explored(15, 22). Moreover, use of m health for providing health education has already shown impact on improving coverage of maternal health services.(23) Additionally, providing educational and health promotion messages, tele-consultation, prescription and referral through mobile phones are becoming increasingly popular in Bangladesh due to the nationwide coverage of mobile network, availability of cheap hand sets, increasing use of smart phones and cheap call rates. The growing investment of the public and private sector in Information and Communications Techonology (ICT) has seen a substantial access and acceptance of technology among the general public, especially the poor. In 2013, there were 105 million active mobile phone subscribers with 75% of household in rural areas owning a mobile phone(3, 24). The current Government's mandate to build Digital Bangladesh also includes provision of Quality Healthcare to all citizens through innovative application of ICT. A recent WHO report indicated that health education was delivered by the health ministry to 98% of its target population through SMS. All these initiatives support the potential financial viability for eHealth and mHealth in Bangladesh and raises clear opportunities in terms of healthcare provision and its scale-up(24). Hypothesis : Implementation of conventional and e-platform based interventions will lead to a 30% improvement in recall of the date of the LMP in adolescent girls and married women in rural Bangladesh. Intervention triggered improvement in LMP date recall among pregnant women in rural Bangladesh will improve the accuracy in GA estimation by 15%. Specific Objectives: The specific objectives are to: Determine whether a set of conventional and e-platform based interventions improve recall of the date of the LMP in adolescent girls and married women in rural Bangladesh. Determine whether intervention triggered improvement in LMP date recall and reporting in rural Bangladesh improves the accuracy in GA estimation or not. Study Design: In order to achieve the objectives, a 4- parallel arm, superiority, community based cluster randomized controlled trial comparing conventional and e-platform based interventions is proposed to improve recall of GA with a no intervention or control group. The trial will be conducted among two groups of participants- adolescent girls and recently married women (within past two years) with no children or a single child in a rural community in Bangladesh. The four arms are: Arm 1: Control arm. Arm 2: Education on the importance of remembering the date of LMP and a paper based calendar to help remember and record the date of LMP in each month. Arm 3: Education and a cell-phone based SMS alert system where participants will be asked to text their LMP dates every month, free of cost. In situations, where the participants fail to text, she will be provided SMS reminders. Arm 4: Education and smart-phone based application for recording menstrual dates every month. Setting: The study will be carried out in Mirzapur sub-district of Tangail district located at the northwest of Dhaka city, the capital of Bangladesh. Administratively Mirzapur has 13 unions (the lowest administrative body), with 219 villages covering a area of 375 sq. km and a population of 3,37,496. icddr,b has an existing demographic surveillance system (DSS) in Mirzapur that covers 288,395 population from 10 unions, data in which is collected by 17 surveillance workers in a four monthly round. The trial will be implemented in the existing surveillance area for ease of identification and enrollment of study participants. Table 1 also provides estimates on the number of adolescents in 2014 and married women with no or single child at the start of 2014. Participants: Adolescent girls (15-17 years) and recently married women (within past two years) with no children or a single child residing in the selected villages, willing to participate in this study will be recruited. Randomization, invitation, recruitment and consent: The trial administrator for this study will be responsible for randomizing clusters in the study arms. The DSS area consists of 1360 blocks, each with a population of 212 and is assigned to 17 surveillance workers. For the purpose of this trial 136 clusters will be formed each with 10 blocks i.e., each cluster will have a population of 2120. The trial administrator will randomize the clusters to four arms (each with 34 clusters) using a computer generated randomization schedule. To ensure that clusters of each DSS data collector have a chance of receiving any of the interventions, randomization will be stratified by the surveillance workers' area. This design will control for potential confounding factors (observed and unobserved) as well as selection bias especially by random assignment of the interventions and controls at cluster level. We will try to ensure that there will be minimal chance of spill over as participants from the same community or blocks will receive the same intervention and we will try to keep geographical buffers between the clusters. Following randomization of clusters, the required sample of adolescents and married women per cluster will be selected from a list of eligible participants generated from the DSS database. Those who agree to further contact will be recruited in the study after written informed consent. Later LMP date of the recruited participants will be collected fortnightly for 3 months in all the arms followed by baseline assessment. Interventions will be rolled out after the baseline assessment and the person conducting or participating in the survey will not be aware of the intervention group. The sample size calculations have been elaborated in the section on Sample size . Intervention activities: (i) Intervention package and duration Three interventions will be tested in this study. These are described below: Intervention 1: This intervention will entail education and documentation of menstrual bleeding on a paper based calendar. The participants will be counselled on importance of remembering LMP. In addition, a free calendar will be provided to the participant who will be asked to record menstrual bleeding and spotting dates in each month. The women will be asked to record "no bleeding" if she did not bleed for any particular month. Likewise, if anybody forgets to record, she will asked to keep a record of it in the subsequent month. Female counselors from the study team will conduct the group or individual education sessions. Intervention 2: Participants of this intervention will receive education and a cell-phone based recording and SMS alert system. Cell-phones will be provided free of cost to participants who will be asked to text the bleeding dates of their menstruation and spotting every month within 3 days of the start of the bleeding. Study team will collaborate with a mobile phone company and the charge of SMS for reporting LMP dates will be free for the user. In situations, where the participant fails to text, she will be provided with several SMS reminders after the due date. Intervention 3: Participants of this intervention arm will receive education and a free smart-phone with an application for recording menstruation dates and an inbuilt reminder system. Experienced programmer from icddr,b will be involved in developing the application. Again, participants will be asked to record bleeding dates each month and will upload the data in the central server via internet. If participants fail to record the dates and upload the data, an automatic reminders will be sent. (ii) Enrolment in the intervention The study participants (adolescent girls & recently married women) will be selected using DSS database and will be recruited in the LMP surveillance after getting written informed consent. Thereafter, they would be offered either of three interventions: menstrual tracking calendar, SMS or smart phone based LMP recording system (as per intervention design). Following enrolment in the study, group or individual counseling session will be conducted by female counselor to educate the participants on importance of remembering LMP dates and on use of the tools for recording of LMP in each month. The roll out of interventions will be cluster specific and will follow the same sequence in which the baseline survey will be done. (iii) Delivering the interventions Education will be provided by female counselors in individual or group sessions immediately after baseline survey. The topics of education will include importance of remembering and recording menstrual bleeding patterns and dates. Calendar, mobile phones and smart phones will be given to the recruited participants at free of cost and the study will bear the cost of SMS and internet for uploading data. The counselors will also teach the participants how to record dates using the different tools. The intervention will be reinforced intensively every month during the first 3 months and later intensity of reinforcement will be adjusted depending on the compliance of the use of tools. Methods of data collection: 1. Data collection will employ the following techniques: 2. Household surveys in all arms 3. LMP surveillance in all arms 4. Data extraction from the intervention tools. 5. Pregnancy surveillance in intervention arms (i) Household surveys: Household surveys will be conducted with eligible adolescents and recently married women in all arms at two points of time: baseline- before the start of the intervention and at end of project, after 12 months in the intervention arms. Any household with no participants/guardian present at the time of the survey will be visited a second and third time. If these additional visits are unsuccessful then the participants will be excluded. Face-to face interviews will be conducted by trained data collectors using a structured questionnaire (annex X) to elicit the required information. Outcome variables: Recalled and reported dates of LMP for the past 3 months prior to the survey and their level of certainty in reporting the dates Explanatory variables: Demographic characteristics of participants including age, sex, education and occupation Household socio-economic profile including housing status, assets, monthly income, ownership of mobile phone & accessibility to mobile phone etc. Birth history and use of contraceptives of married women Information on menstrual irregularity, gynaecological problems Acceptability, challenges and adherence to the tools (where applicable) (ii) LMP surveillence in all the arms: LMP dates of the recruited participants will be collected actively every two weeks so that the actual date of LMP is recorded in all arms for 3 months before baseline and endline survey. No other information will be elicited from the participants. (iii) Data extraction from the intervention tools: Data will be extracted from calendar records, SMS and smart-phones. The data from mobile SMS will be obtained every month from the mobile company whereas users of the smart-phone will be asked to upload their data via the internet connected to the cell-phone. For participants using the calendar, data collectors will collect the information from the enrolled household every two months. (iv) Pregnancy surveillance in intervention arms: Pregnancy surveillance will be continued throughout the intervention period among recently married women enrolled in the intervention arms. In case of missed period reporting on calendar or failure of SMS or smart phone based LMP reporting, a surveillance worker will visit the enrolled recently married women's house. Surveillance worker will be complete each round every 8 weeks. Any women having missed period within one year of recruitment will have the pregnancies confirmed through testing and their recalled LMP date collected. They will be followed until 20 weeks into their pregnancy with USG assessments being done between 10-13 weeks of gestation. The study will bear the cost of USG and associated travel. Study team: (i) Central The study will be conducted by central team of investigators and field research team. The Central team will be composed of one principle investigator, one MNCS expert as an external co-principle investigator and 2 research investigators (1 full time equivalent). The Central team will be responsible for overall study design, seeking ethical approval, tools development, field implementation and quality assurance. (ii) Intervention implementation team The intervention implementation team will be comprised of female counselors, immediate supervisors and a manager. It was estimated that 15 counselors (5 per intervention arm) will be required to train/educate the participants (2400; see sample size section for details) in the intervention arms in a period 1-2 months. As mentioned earlier, female counselors will conduct education sessions twice during project implementation period; at the beginning and six month after intervention. There will be one immediate supervisor for the counselors in each arm and over all intervention activity will be monitored by a field manager. (ii) Data collection team The study will require approximately 3200 participants to estimate the required effect size (detailed in the section on sample size calculation). Based on these assumptions, it was estimated that the study will require fifteen data collectors to carry out the surveys in two months in the study area. Additionally, 10 data collectors will be deployed to collect regular LMP data from all the arms for 3 months before baseline and enline survey and 5 data collectors for pregnancy surveillance throughout the project period. The project will employ 2 immediate supervisors of data collectors and a manager to monitor the activities of the data collection team. Sample size: Sample size has been calculated to observe expected changes in the primary outcome measure using the formula for comparison of two proportions in the presence of clustering. The investigators expect that any of the intervention arms will have a minimum 30% relative improvement in accuracy of LMP recall in comparison to control arm and our intention is not to compare between interventions due to resource constraints. Assuming that in the rural context of Bangladesh approximately 55% of women can accurately recall LMP, to measure a 30% expected change in recall after the implementation of the interventions, we will require 238 participants from 34 clusters per arm assuming 5% level of significance and 90% power. Accounting for 15% refusal and 20% loss to follow-up we will require approximately 400 adolescents and 400 recently married women (within past 2 years) with no children or one child in each arm. Hence, 1600 adolescents and 1600 recently married women will be required in four arms. . The second outcome aims to assess the accuracy of LMP based GA assessment in the intervention arms compared to USG based GA. As mentioned a subset of the recently married women, who become pregnant within a year of enrollment will recruited for this study. The study conducted at Shishu Hospital in Bangladesh (10), have shown that the intra class correlation coefficient of LMP based GA measurement versus USG based GA measurement was 0.84. Considering a change of 15% from the above estimates due to the intervention, a sample of 120 pregnancies will be required assuming 80% power, 5% level of significance and 20% refusal or loss to follow-up. Women who become pregnant from the pool of 1200 married women from the three intervention arms will be followed up until 16 weeks of gestation and will be asked to undertake USG for GA measurement between 10-13 weeks. Surveillance data from Mirzapur shows that approximately 11% of recently married women will become pregnant within the next year. Hence, The investigators assume to get approximately 120 pregnant women out of 1200 married women in the intervention groups. Data analysis: Data analysis will be done following intention to treat analysis under the guidance of an experienced statistician. Analyses will be conducted at the individual level, and will be adjusted for cluster randomization. For the primary outcome we will first validate the tools by comparing the distribution of recorded LMP dates between the intervention arms at different points in time (baseline and endpoint) with actual LMP dates collected during LMP surveillance. We will examine the proportion of women exactly reporting the accurate LMP dates. Differences in differences method will be used to ascertain the change in proportion of women accurately reporting LMP dates and their level of certainty on reporting LMP dates. Accuracy of LMP dates will be ascertained if survey and surveillance dates matches within +/- 1 day variation for an individual. Level of certainty of reporting LMP dates will be assessed using Likart scale. The analysis will be adjusted for confounders such as education, socio-economic status and length in recall time. Additionally, we will explore the change in mean and standard deviation in reported LMP compared to the actual LMP by recall periods. For the second outcome we will compare GA calculated by LMP and USG (gold standard) using the mean (students t-test) and distribution. Convergent validity will also be assessed between the estimates of GA by LMP and USG using intra-class correlation coefficients, Lin's concordance coefficient and Bland-altman analysis for exact comparison of continuous values. Stata® software version 13 will be used for all analyses. Data safety & Monitoring: We anticipate no risk of interventions. Validity and integrity of the data will be ensured by appropriate research design, use of pretested and validated tools for data collection and by quality assurance. In order to protect the safety of participants, all trial related information will be kept confidential and stored securely at the central office in icddr,b. Coded identification will be used to anonymyse and depersonalize the data. The linking code, electronic data files and paper forms will be stored in a separate locations under password protections or lock and key. Access to the data will be to the small number of individuals including the investigators, statisticians, quality control and audit. The trial results will be communicated and published irrespective of the outcome of the trial. Ethical Approval: Ethical approval for the trial has been obtained from the Institutional review Board of icddr,b. The trial will be conducted following the ethical principles in the Declaration of Helsinki and good practice guidelines on the proper conduct of research. Informed consent process: Participants will be informed about the objective of the study along with associated risk/benefits and will be asked to participate voluntarily. Informed written consent will be taken from the participants. For illiterate participants thumb print will be taken. If the participant agrees to participate, they will be recruited for the study. Questionnaires will be administered in the local language for data collection. Perceived risks and benefits: There are no risks to participation in this study. The topics addressed do not relate to illegal, sensitive or stigmatized behavior. Safeguards to protect any recognized vulnerability of the study: The risk to invasion of privacy will be addressed by the use of study identification numbers and the secure storage of study records in locked cabinet at study headquarters. Confidentiality of the data will be assured at all steps of the study including data collection, data management, access to data and use of the information. The survey data will be retained according to the Orgnization's policy. Only the de-identified version of the data will be accessible for analysis. Reimbursement or compensation to study participants: No direct monetary compensation will be provided to the participants in this study. However, the intervention is expected to increase the knowledge of the participants regarding importance of LMP date recording. Responsiveness of the project to community needs and priorities: This study is expected to provide critical evidence to understand the efficacy of three different interventions for LMP date recording and thereby improving the accuracy of gestational age measurement. As such the project not only aims to benefit the community but also will generate evidence to support further scale-up at national level.

The participants will be counseled on importance of remembering LMP. In addition, a free calendar will be provided to the participant who will be asked to record menstrual bleeding and spotting dates in each month. The women will be asked to record "no bleeding" if she did not bleed for any particular month. Likewise, if anybody forgets to record, she will ask to keep a record of it in the subsequent month. Female counselors from the study team will conduct the group or individual education sessions.

Cell-phones will be provided free of cost to participants who will be asked to text the bleeding dates of their menstruation and spotting every month within 3 days of the start of the bleeding. Study team will collaborate with a mobile phone company and the charge of SMS for reporting LMP dates will be free for the user. In situations, where the participant fails to text, she will be provided with several SMS reminders after the due date.

Smart phones with an application and an inbuilt reminder system will be provided free of cost to participants who will be asked for recording menstruation dates. Experienced programmer from icddr,b will be involved in developing the application. Again, participants will be asked to record bleeding dates each month and will upload the data in the central server via internet. If participants fail to record the dates and upload the data, an automatic reminder will be sent.

The participants will be counseled on importance of remembering LMP. In addition, they will be provided a free calendar to record their LMP dates.

The participants will be counseled on importance of remembering LMP. They will be asked to send the LMP date via SMS. Participants will be reminded via SMS if they have not sent their dates (max -3 reminders)

The participants will be counseled on importance of remembering LMP. Participants will be asked to record the LMP on a calendar app. This will be transferred to central server via internet Participants will be reminded via SMS if they have not sent their dates (max -3 reminders)

Improvement in the recall of LMP date among enrolled participants in the intervention arms compared to control arm

This outcome will be assessed by comparing recall date of LMP from surveys with recorded dates extracted from the intervention tools and actual LMP dates collected from LMP surveillance. In order to do so, first we will compare the distribution of LMP of the enrolled participants in the intervention arms recorded in the intervention tools along with the actual LMP dates collected by LMP surveillance for validation of the tools. Later differences in differences method will be employed to measure the change in coverage of accurately recalled LMP date and level of certainty for recalling LMP dates

The second outcome will be assessed by comparing gestational age measured by LMP with GA from USG assessments. The study will not attempt to make any comparison across interventions.

Distribution of recorded LMP dates from the tools will be compared with actual LMP dates collected from LMP surveillance

Inclusion Criteria: For Adolescent girls: Permanent resident of the study area (Adolescents who are usual residents of the households and living or intend to live in the area for 6 or more months) Aged between 15-17 years at the time of recruitment Parent or legal guardian is present at the time of recruitment and is willing to give assent for the adolescent's participation For Recently married women: Permanent resident of the study area (Recently married women who are usual residents of the households living or intend to live in the area for 6 or more months) Married within the past two years with no children or a single child Exclusion Criteria: For Adolescent girls: Married at the time of recruitment Parent or the legal guardian is mentally/physically unable to give assent For Recently married women: 1. Pregnant or exclusively breastfeeding at the time of recruitment

Salam SS, Ali NB, Rahman AE, Tahsina T, Islam MI, Iqbal A, Hoque DME, Saha SK, El Arifeen S. Study protocol of a 4- parallel arm, superiority, community based cluster randomized controlled trial comparing paper and e-platform based interventions to improve accuracy of recall of last menstrual period (LMP) dates in rural Bangladesh. BMC Public Health. 2018 Dec 10;18(1):1359. doi: 10.1186/s12889-018-6258-z.