Improving reporting of infant deaths, maternal deaths and stillbirths in Haryana, India

Abstract Underreporting hampers the accurate estimation of the numbers of infant and maternal deaths and stillbirths in India. In Haryana state, a surveillance-based model – the Maternal Infant Death Review System – was launched in 2013 to try to resolve this issue. The system is a mixture of routine passive data collection and active surveillance by specially recruited and trained field volunteers. The volunteers gather the relevant data from child day-care centres, community health centres, cremation grounds, hospitals, the municipal corporation’s offices and primary health centres and regularly visit health subcentres. The collected data are triangulated against the standard death registers and discussions with relevant community members. The details of any unregistered death are rapidly uploaded on the system’s web-based platform. In April 2014, we made field observations, reviewed records and conducted in-depth interviews with the key stakeholders to see if the system’s performance matched the state government’s planned objectives. The data collected indicate that implementation of the system has led to quantitative and qualitative improvements in reporting of infant and maternal deaths and stillbirths. Completeness and consistency in the reporting of deaths are essential for focused policy and programmatic interventions and there remains scope for improvement in Haryana via further reform and changes in policy. The model in its current form is potentially sustainable and scalable in similar settings elsewhere.


Introduction
Each year in India, there are approximately 28 million pregnancies, 26 million live births, 67 000 maternal deaths and a million neonatal deaths. 1 There are about 42 infant deaths per 1000 live births, 2 five stillbirths per 1000 deliveries 3 and 178 maternal deaths per 100 000 live births. 4 The National Health Mission, which was launched in 2005, marked a turning point in the history of India's health-care system. Although the mission had many objectives, some of its main goals were reductions in infant and maternal mortality as well as a general improvement in the quality of health-care services via a sustainable system.
The northern state of Haryana is India's second wealthiest state, in terms of its annual per capita income -about 2139 United States dollars in 2012-2013. 5 Despite this relative prosperity, Haryana records rates of infant mortality -42 infant deaths per 1000 live births 2 -and stillbirth -nine stillbirths per 1000 births 3 -and a maternal mortality ratio -146 maternal deaths per 100 000 live births 4 -that are similar or worse than the national mean values. It is, however, unlikely that the official national and state records of infant and maternal mortality and stillbirth are accurate. Although such records may be supplemented by data collected by routine health services or in sample surveys, they are usually largely based on vital statistics. 6 In most low-and middle-income countries, the incomplete registration of births and deaths results in inaccurate vital statistics. 7 In India, the registration of births has recently improved but the registration of deaths remains generally poor -although there have been interventions to improve the reporting of maternal deaths in the states of Kerala, Punjab, Tamil Nadu and West Bengal. 8 Surveillance can help in the control and prevention of diseases of public health importance by facilitating planning, evaluation and the formulation of research hypotheses. 9 In Haryana, a surveillance-based model -the Maternal Infant Death Review System -was launched in 2013 to try to improve state-wide estimates of infant and maternal mortality and stillbirth rates. Below, we describe the review system and the results of an evaluation of its impact that we conducted, according to the relevant guidelines of the United States Centers for Disease Control and Prevention. 9 The review system Genesis Although the Indian health management information system is intended to provide facility-level reports on maternal and infant deaths, is not yet very robust and the information it provides remains incomplete. 10 In 2012, the National Health Mission in Haryana therefore designed a centralized system of passive surveillance to gather better information on maternal deaths, infant deaths and stillbirths from facilities at various levels of the health-care system. In April 2013, the addition of active surveillance to this system was piloted in Karnal district. After a brief evaluation indicated that three deaths were being reported in Karnal for every one being detected by passive surveillance (V Chayal, Post Graduate Institute of Medical Sciences, Rohtak, India, personal communication, 2015), the combination of passive and active surveillancei.e. the Maternal Infant Death Review System -was gradually rolled out across the state. From September 2013, the whole of Haryana was covered by the review system.
Abstract Underreporting hampers the accurate estimation of the numbers of infant and maternal deaths and stillbirths in India. In Haryana state, a surveillance-based model -the Maternal Infant Death Review System -was launched in 2013 to try to resolve this issue. The system is a mixture of routine passive data collection and active surveillance by specially recruited and trained field volunteers. The volunteers gather the relevant data from child day-care centres, community health centres, cremation grounds, hospitals, the municipal corporation's offices and primary health centres and regularly visit health subcentres. The collected data are triangulated against the standard death registers and discussions with relevant community members. The details of any unregistered death are rapidly uploaded on the system's web-based platform. In April 2014, we made field observations, reviewed records and conducted in-depth interviews with the key stakeholders to see if the system's performance matched the state government's planned objectives. The data collected indicate that implementation of the system has led to quantitative and qualitative improvements in reporting of infant and maternal deaths and stillbirths. Completeness and consistency in the reporting of deaths are essential for focused policy and programmatic interventions and there remains scope for improvement in Haryana via further reform and changes in policy. The model in its current form is potentially sustainable and scalable in similar settings elsewhere.

Description
The review system combines passive surveillance on infant and maternal deaths and stillbirths with active surveillance of the same events (Fig. 1). For the passive surveillance, frontline health-care workers and medical officers are asked to report, to their district-level authorities, every abortion, delivery, infant death, pregnancy, maternal death, stillbirth and vaccination and all antenatal and postnatal care and contraception provided in their area and community. The district-level managers are then asked to upload this information daily on the review system's web-based platform. The standard case definitions of the World Health Organization (WHO) are used to categorize the deaths as infant, maternal or stillbirth. [11][12][13] The active surveillance component of the review system is managed by 13 trained surveillance field volunteers, who were specially recruited and trained. These volunteers are asked to visit all the higher level facilities -i.e. the child day-care centres known as anganwadi centres, community health centres, cremation grounds, hospitals, the municipal corporation's offices and primary health centres. They retrieve information from registers and discussion with staff, on any unreported maternal deaths, infant deaths and stillbirths. Having collected any relevant data, the volunteers must then visit the relevant lower level facilities, such as health subcentres, to verify the data with frontline health workers and check antenatal-care and postnatalcare registers. During verification, if the volunteer confirms the discovery of an infant or maternal death or stillbirth that has not been reported so far, that event is recorded on the web-based platform. In discussions with the relevant frontline health workers, the volunteers investigate why each previously unreported death or stillbirth had not been reported and educate the workers on the importance of reporting all deaths and stillbirths.
All women who die within nine months of registering for antenatal care or within two months of registering for postnatal care are classified as maternal deaths by the volunteers.
The volunteers are also encouraged to interact with community members and to visit households in the village or urban area where a death of interest has taken place, to verify the cause of death whenever there is any doubt.
Although some of the state's private hospitals initially refused to share information on in-hospital deaths and stillbirths, all subsequently agreed to share such information with the state government and volunteers provided the hospitals' names were not disclosed. The hospitals' managers were assured that no legal action would be taken against them as a result of the information they shared.
As a single death may be reported to the review system more than once, for instance by different health workers at district and state levels, the review system's platform is designed to highlight and remove duplicate data.

Evaluation
In April 2014, we made field observations, reviewed records and conducted in-depth interviews with the key stakeholders and the volunteers to see if the review system's performance matched the state government's planned objectives. We held several rounds of formal and informal discussions to investigate the process, strengths and challenges of implementing the review system and the review system's effectiveness and robustness. We investigated how each death registered by the review system had come to be reported and compared how the perceived trends in infant and maternal mortality in Haryana between 2012 and 2014 differed according to the source of the primary data.

Effectiveness
Interviews with the relevant officials at state and district levels indicated that the review system had been effective in providing the village-level data needed to make programmatic decisions. The interviewees also discussed how the review system had facilitated supportive supervision.

Implications
It appears that Haryana's surveillancebased review system has at least partially addressed the issue of the underreporting of maternal deaths, infant deaths and stillbirths in the state. In complex environments, it becomes essential to have robust measures to record and report mortality -along with other information -to assist health-care monitoring.
In many settings, health information systems and related data-sharing mechanisms need to be improved. Haryana's surveillance-based review system now rapidly provides detailed and relatively accurate information about the health status of women and infants. Reporting from multiple sources has reduced the likelihood of missed deaths and, presumably, enabled better decisionmaking. Previous systems for recording mortality, such as the civil registration system, have not been very effective.
In some countries, it is not uncommon for more than two thirds of maternal deaths to go unreported in official records. [14][15][16] Low-cost surveillance by key informants can be used as an effective method to monitor trends in maternal mortality, especially in areas with poor vital registration. 17 In 2004, Tamil Nadu became the first Indian state to establish a system for the mandatory registration and reporting of all maternal deaths, within 24 hours of death, by public and private health facilities. 18 Kerala and West Bengal have also launched interventions to improve the reporting of maternal deaths and appear to have seen reductions in maternal mortality over time. 19,20 In terms of the maternal mortality ratio, Kerala and Tamil Nadu met the 2015 national target -of 100 maternal deaths per 100 000 live births -and Maharashtra is close to achieving the same target. 21 Haryana's review system is an attempt to establish a health information system that is similar to those used in Kerala, Maharashtra and Tamil Nadu but tailored to Haryana's health milieu.
Underreporting of deaths remains an issue across the globe. [22][23][24][25][26][27][28][29][30][31] A case study in rural Indonesia demonstrated the usefulness of local maternal and child health registers as sources of information in measuring and reporting perinatal mortality and stillbirths, in combination with local vital registration systems. 32 Even in an urban area of a developed high-income country, such as the United States of America, the use of computer-assisted active surveillance revealed 14 new maternal deaths, resulting in an 88% increase in the ascertainment of maternal deaths. 33 Other innovative strategies employed to improve the reporting of deaths include record linkage, a retrospective survey among doctors, use of an interview census and confidential forms, specific inquiries on maternal deaths, a household census of birth and infant deaths, matching of hospital records with death certificates and a retrospective population-based survey. 15,16,31,[34][35][36][37][38][39][40][41][42][43][44][45] Maternal death review, as implemented in India and several other countries, has led to local policy changes and improvements in the quality of maternal health services, even in challenging settings. Building on this success, WHO introduced maternal death surveillance and response -a continuous action cycle that links routine identification, notification, quantification and determination of causes with actions to prevent future deaths. 46 While notification of maternal deaths has been mandated by national policy in over 50 countries, only 17 countries have a national mandate for perinatal death reviews. 47 Although Haryana's review system has helped improve the quality of the death reports using frontline health workers, there is scope for further improvement in the recording, reporting and notification of deaths. One suggestion is that guidelines similar to those for WHO's maternal death surveillance and response are followed as a state-level or national policy. One benefit of Haryana's review system was the creation of a uniform definition of stillbirth across all health worker groups -and this clear definition may well have contributed to the increase seen in the number of stillbirths reported (Fig. 2).
The potential importance of strengthening death -and health -reporting systems, to improve the availability, completeness and quality of the data cannot be overemphasized. Close interaction of health staff with their local communities provides a good foundation for the improvement of death reporting, 48 especially when passive and active forms of surveillance run in parallel. In Haryana, it remains unclear if the review system will maintain its efficiency if and when the surveillance field volunteers are withdrawn. Refresher training and regular monitoring of the volunteers may be needed to improve their effectiveness. In many places, the frequency of the volunteers' field visits has gradually fallen. Many of the frontline field workers involved in the review system appear to have very heavy workloads and little time for the detailed documentation of deaths. This problem may have been exacerbated by a lack of awareness about the importance of accurate reporting and inconsistencies in community access to local health-care services. 32 Further health reforms and policy reviews may well be needed, at both district and state levels, if good data on infant and maternal deaths are to be collected in the long term. An external evaluation of the cost-effectiveness of the review system is still needed.
In general, routine civil registration systems lack political priority and this often leads to inadequate associated policies, poor management and underfunding. Although more effective systems of death reporting may be more complex and require institutional agreements across many governmental departments, they can be made to work given strong regional momentum and leadership. 47,49 Given the encouraging results already achieved with minimal external support, Haryana's successful integrated review system should probably be extended to other states. ■ Funding: This work was supported financially by the National Health Mission in Haryana.
Competing interests: None declared.