Pregnancy Surveillance Methods within Health and Demographic Surveillance Systems

Introduction

Reduction in neonatal mortality is a key target of the United Nations’ Sustainable Development Goals¹. While the global under-five mortality rate has declined by almost two thirds over the past 30 years², neonatal mortality has declined more slowly. Efforts to accelerate the pace of reduction are hindered by a lack of accurate and timely data on child deaths in the regions where they are most prevalent. Such information is crucial to guiding resource allocation and evaluating the effectiveness of interventions^3,4. In countries with high infant mortality in South Asia and Sub-Saharan Africa, estimates of neonatal mortality based on health care utilization data or clinic reports exclude women and children who do not have access, do not seek care, or die at home⁵. The United Nations Inter-agency Group for Child Mortality Estimation (UN IGME) stillbirth and child mortality estimates are derived from vital registration systems, censuses, and various demographic health survey data⁶. Poor data quality in many developing nations makes accurate estimation more challenging⁷. Furthermore, the use of models to correct such estimates is complicated by lingering questions regarding their applicability to African settings⁸.

Pregnancy surveillance is in theory an important tool in the identification of stillbirths and neonatal deaths and could circumvent some of the methodological flaws of retrospective reports in cross-sectional surveys³. The term pregnancy detection commonly refers to activities that only identify pregnancy, without the follow-up tracking of pregnancies through to their end. Pregnancy surveillance entails activities to identify pregnancies and monitor their outcomes. The latter is often integrated into health and demographic surveillance systems (HDSSs), which collect population-based data about pregnancies and maternal and child health³. HDSSs conduct continuous registration of demographic and vital events in a geographically defined surveillance area (DSA) in low- and middle-income countries⁹. HDSS fieldworkers conduct regular visits to collect demographic and health data from all households under surveillance. Fieldworkers generally collect pregnancy data for all eligible women as part of these regular visits.

The goal of this report is to describe the methods used to conduct population-based pregnancy surveillance low- and middle-income countries. We investigate variation in pregnancy surveillance methods across HDSSs and identify the components that may be linked with data completeness and quality in capturing stillbirths and neonatal deaths. This approach expands the limited literature analyzing pregnancy surveillance methodologies^3,10 and draws on protocols and information from within and outside of the published literature.

Methods

We reviewed published literature, HDSS field manuals and data collection instruments for methods used in pregnancy identification and follow-up (Table 1). Articles published between 2002 and 2019 in English were searched in PubMed and Google Scholar using the search terms: pregnan*, identif*, discover*, population, surveillance, observation, registration, detection, demograph*, maternal, surveys and questionnaires, INDEPTH Network, epidemiology, DHS, DSS, and HDSS.

Following a review of titles and abstracts, we reviewed the main text of 147 articles mentioning pregnancy-related data collection in HDSSs. In total, 37 articles reporting any information about data collection methods or criteria for pregnancy surveillance were retained. An additional 22 articles were located through a manual search of the references of included studies. We also reviewed the websites of two networks: the Child Health and Mortality Prevention Study (CHAMPS) Network is a network of HDSSs in six countries that are focused on identifying the main causes of under-5 mortality (Cunningham, 2019); the International Network for the Demographic Evaluation of Populations and their Health (INDEPTH) is a network of 49 independent HDSSs and 7 associate HDSSs in Africa, Asia, and Oceania¹¹, for online data reports, data collection instruments and field protocols of HDSSs within CHAMPS and INDEPTH. Collectively, we gathered information from 67 sources (Figure 1).

Figure 1. Literature review flowchart.

The review included 52 HDSSs located in 20 countries, which represent the majority of operating sites. For each, we reviewed the criteria used to define women who are eligible for HDSS pregnancy surveillance, requirements for respondents from whom pregnancy information can be collected, visit frequencies for collecting pregnancy information, questions asked for pregnancy identification, supplemental efforts for enhancing pregnancy surveillance beyond regular HDSS household visits, and rules regarding characteristics of interviewers administering pregnancy-related questionnaires.

Results

Discussion

HDSSs provide reference data for health and demographic estimates in countries without civil registration and vital statistics systems; they complement periodic cross-sectional censuses and surveys. Pregnancy surveillance is an important component of health and demographic surveillance activities. The prospective follow-up of pregnancies can produce better estimates of stillbirths and early childhood mortality. We document that the pregnancy surveillance methods used in HDSS are heterogeneous. Methods differ in the eligibility criteria for women to be included in the pregnancy surveillance, the frequency of visit intervals, the use of proxy respondents, and the characteristics required of interviewers. The methods employed can affect the quality and completeness of information on pregnancies and their outcomes. Surveillance methods are not always described in detail, making comparability even more difficult.

HDSS residency requirements determine which women are part of the population under surveillance and are particularly important in areas where migration occurs frequently³². The definition of “residents” can affect the size of the population under surveillance, as well as the capturing of demographic events. Populations located in urban areas that experience high mobility can be more difficult to track through temporary migration patterns. In some communities, it is common for women to temporarily relocate to their parents’ home for the months before and after the birth of a child^49,50. A recent in-migrant may not be immediately recorded as a resident in the HDSS, in which case her pregnancy would not be recorded in that visit. If she is still present in the household in the next round of household visits and meets the residency criterion, she will be recorded and so will her child; however, if the child did not survive to the next HDSS round, the child may not be listed onto the household roster and will thus not be counted. A solution for HDSS with high rates of migration has been to create a temporary residency status, which allows women to be included in surveillance even if it is unclear whether they will become permanent residents.

Many HDSS have restrictions on the marital status and ages of women for whom pregnancy information is collected. These are generally useful, but that can sometimes exclude some women who have a chance to be pregnant from being recorded. It is inappropriate in some communities to ask unmarried women about pregnancy status, and unmarried women are less likely to report their pregnancies overall^13,51. Also women who are very young or at older ages may not disclose pregnancies⁵². This is due to the sensitive nature of pregnancy information. Unmarried women may be more vulnerable to stigma or not ready to disclose an out-of-wedlock pregnancy⁵³. There may be shame around having an unplanned pregnancy; for women who are at risk of pregnancy loss due to their age, they may want to avoid the shame that can accompany pregnancy loss, or being suspected of having terminated the pregnancy⁵³. Unmarried women are likely to be young and at higher risk of adverse pregnancy outcomes⁵⁴. Never-married women under age 25 account for 3.5–60% of births in sub-Saharan Africa⁵⁵. Thus, pregnancies and especially pregnancies with high risk of adverse outcomes can be missed.

Pregnancy surveillance inclusion criteria may need to be designed with the higher risk of negative birth outcomes for older, younger, and unmarried women in mind. Broader inclusion for pregnancy surveillance may improve the tracking and detection of possible adverse birth outcomes for women who might not be counted otherwise. An important consideration for electronic data collection is the preprogrammed rules that set data validation cutoffs in the system based on age or marital status. Overly restrictive rules may impede data collection. For example, if the data collection device does not allow interviewers to record pregnancies for women under the age of 15, or pregnancies to unmarried women, such information will not be registered even if offered. Allowing interviewers to record out of range values makes it possible to record such information.

If the respondent is not home when an interviewer visits, many HDSSs allow for information to be collected from another person in the household, to reduce the need for re-visits. However, especially for sensitive information related to pregnancy, proxy responses may not be accurate. Use of proxy respondents is an important contributor to the undercounting of pregnancies, especially in the first two trimesters when women tend not to discuss a pregnancy with others¹⁰. Pregnancy, stillbirth and early neonatal death are often hidden because they increase women’s vulnerability to social and physical harm through gossip, blame, acts of violence, or beliefs about sorcery and spiritual possession⁵². Therefore, proxy respondents may not be aware of another individual’s pregnancy or feel comfortable sharing such information, leading to underreporting. Even though it is more time-consuming, it is best to only ask women themselves about their own pregnancies. If women are absent at the time of visit, interviewers should return at least twice to speak with her directly before using proxy respondents³². Additionally, many HDSS are now using telephone and text to reach respondents, and these modes may also be appropriate for contacting women about pregnancies.

Because pregnancies only last up to nine months, infrequent data collection increases the risk of missing pregnancies altogether^15,38. The more frequent the HDSS visits, the more likely that pregnancies will be recorded. When pregnancies are missed, the outcomes of pregnancies are often also not recorded. This is especially true for pregnancies that ended in stillbirth or neonatal death. Data collection less than 5 months apart are recommended³². Additionally, more frequent visits can build community engagement and rapport when inquiring about sensitive subjects such as pregnancy. The CHAMPS network recommends visiting each catchment area every few months to improve tracking of both pregnancy and migration-- the latter of which serves to improve tracking of temporarily absent women⁹.

Supplemental pregnancy surveillance activities can provide useful avenues for improving pregnancy detection. Linking health facility data to HDSS increases data when pregnancies are missed by the regular data collection and they can be recorded retrospectively after birth from hospital or health center data³⁵, though only for women who access healthcare. Additionally, networks of informants can supplement HDSS capacity by providing alerts about vital events as they occur, including pregnancy.

Interviewers need to be well trained to collect information in sensitive situations. In some settings, interviewers from outside the community may be better positioned to collect sensitive information. For information regarding pregnancies, female interviewers are highly recommended. Women are more likely to disclose pregnancies to other women^36,56. One study found that interviewers who were female, younger, and conducted more interviews elicited more responses on a survey about social networks, on average⁵⁶. Abortion and stillbirth are also less likely to be reported to interviewers who are men¹⁰. While the impact of interviewer gender on other components of HDSS data collection is inconclusive, the use of female interviewers is preferred for collecting pregnancy data because female respondents disclose more sexual behavior related information to same-gender interviewers^57,58. Cultural norms in Nahuche demand the use of predominantly female interviewers and have resulted in higher quality data and lower refusal rates⁵⁹.

The pregnancy surveillance methods in HDSS are diverse and often lack detailed documentation. There are also questions surrounding the quality of HDSS estimates of neonatal mortality, which are subject to huge variability and are on average lower than corresponding estimates from demographic health surveys⁷. Unreliable data on the vital events of newborns has been described as one of the most challenging issues facing HDSS⁶⁰. Children who survive are likely to be recorded by the HDSS when they are observed in subsequent interview rounds. However, those that die are more easily missed, especially in the absence of a pregnancy notification. Pregnancy surveillance facilitates the follow-up of adverse pregnancy outcomes and early mortality. The inclusion criteria for pregnancy surveillance, respondent and interviewer characteristics, and the frequency of household visits should be considered further for their potential impact on the identification of pregnancies and their follow-up. Improvements to HDSS pregnancy surveillance protocols and completeness have the potential to address well-documented issues of downward bias in early mortality estimates. The standardization of pregnancy reporting protocols and exhaustive capture of pregnancy status information should thus be among the highest priorities for HDSSs.

Data availability

All data underlying the results are available as part of the article and no additional source data are required.