A Verbal / Social Autopsy ( VASA ) Child Mortality Inquiry to Investigate Under-Five Mortality Determinants in Slums of Karachi , Pakistan : A Mix Methods Interventional Study

Background Pakistan stands along top ten countries responsible for two-third of global child mortality burden. To improve Child Mortality (CM) estimates in Pakistan, there is an exigent need to understand the in-depth reasoning behind mortalities. In Pakistan, the data on cause-ofdeath (CoD) for a large number of child mortalities in Pakistan is either not available or not suitable for use. Pakistan uses an alternative technique called Verbal autopsy (VA). Since VA only determines the biological cause of death (BCoD), and does not capture multitude of other modifiable social, cultural and health system determinants, it has a limited importance. Such non-biological determinants are captured by another technique i.e. Social Autopsy (SA). Objective By integrating VA with SA, extended set of mortality related determinants will be explored, in addition to strengthening death notification coverage and assigning CoD in such missed out mortalities. Method Under the working group of Health Advocacy Council for Women and Children (HACWC) in collaboration with Child Registry of Pakistan (CROP), mixed method (QUAN-QUAL) interventional study will be carried out in urban-slums of Karachi city using VASA-integrated under-five mortality investigative technique. Parents of dead children will be interviewed. Second stage of study will be followed by qualitative interviews with different cadres of stakeholders. Expected Outcomes By focusing the specific determinants related to case-management and care-seeking process (identified by The Pathway to Survival Framework-TPtoSF) an in-depth understanding of related determinants can be achieved, which will help in crafting potential interventions required to limit the barriers that increase the chances of CM in developing nations.

from now.Although the world has failed to achieve this goal, however, over the past 25 years, the global under-five mortality rate (U5MR) has improved substantially (i.e 53% reduction; i.e. with a 91 under-five deaths per 1000 live births in 1990 to 43 in 2015) (Turab et al., 2014;UNICEF-Report-2014;UNICEF-Report-2015).Just one-third of 195 countries have achieved Millennium Development Goal-4.This reduction trend has left an underlying disparity of estimates among developing and developed nations, with majority of child deaths concentrated in Sub-Saharan Africa (SSA) and South Asian (SA) countries (UNICEF-Report-2014; UNICEF-Report-2015). Almost half of global under-five mortality burden is clustered in only five developing countries (India, Nigeria, Pakistan, Democratic Republic of the Congo and China) (Nisar, Y. B., & Dibley, M. J. 2016; Report, 2014; UNICEF-Report-2014; UNICEF-Report-2015).In the planning for post-2015-target era, discussions are geared up now on targeting these high burdened under-five mortality countries with more holistic approach in a practically sustainable manner.

Child Mortality in Pakistan
Over the past 25 years, Pakistan has appeared to be one of the dangerous places on earth for a child to survive till age five (M B Siddiqui, 2016).Here, almost 46 percent of under-five deaths occurred during the first month (neonatal period) of their lives (UNICEF-Report-2015). Pakistan has curbed down its U5MR (since 1990) from the value of 139/1000 live births to almost 81/1000 live births (in 2016); however, still its current U5MR is 35/1000 live births above the target (of 46/1000 live births) set for Millennium Development Goal-4 that was due to achieve in 2015 (Report, 2014;UNICEF-Report-2014;UNICEF-Report-2015).There is also a huge rural-urban disparity of child mortality estimates within country (rural geographies have 44% higher U5MR than urban estimates) (PDHS, 2012(PDHS, -2013)).

Weak Mortality Notification in Karachi and Verbal Autopsy
Karachi, being the highly populous metropolitan city of Pakistan, has almost 62% of its population residing in urban slums with severely compromised socioeconomic profiles (D'Souza, R., M. 2003;Shaikh, H. 2016).Likewise, Pakistan, the urban slums of Karachi city has weak maternal and child death notification coverage (Marsh, D., et al. 1993a;Jafarey, N., S., et al. 2009).This is due to several reasons.Since, often a large number of children die at home and their occurrences fail to appear in national vital statistics register (PDHS, 2012(PDHS, -2013)), the data on CoD of such missed out deaths does not exist and not available for policy and planning.Even, the available data on notified deaths are incomplete, unreliable or lacking in major variables, leaving the entire data as compromised (Marsh, D., et al. 1993a;Marsha, Sadruddin, Fikree, Krishnan, & Darmstadt, 2003).To summarize, the data on CoD for a large number of child mortalities in Pakistan is either not available or not reliable for use.As an alternative to identify such deaths, knowing their causes and assigning a certified CoD; verbal autopsy (VA) has been informally piloted in several of the slums of the city (Marsh, D., et al., 1993a;1993b;PDHS, 2012PDHS, -2013;;Jafarey, N., S., et al., 2009).However, majority of such verbal autopsies have been attempted for research purpose only and were not followed in routine practice (Marsh, D., et al., 2003;Nausheen S., et al., 2013).

Social Causes of Child Mortality and Social Autopsy
Although VA has given a huge amount of worth full data and much resources and energy has been devoted in developing and employing VA technique (Aleksandrowicz, L., et al., 2014;Edmond, K. M., et al., 2008;Kalter, H. D., et al., 2015;M B Siddiqui, 2016;Marsha et al., 2003) 1984).TPtoSF provides a more holistic way to capture and organise the social causes of child mortalityspecific factors affecting case management, access, utilization and delivery of healthcare services at community, outreach and health facility level (Waiswa, P., et al. 2012;Waldman, R., et al. 1996).The SA component based on TPtoSF gives a much broader platform for investigator to investigate such non-biological determinants of death (Kalter et al., 2011;Waldman, R., et al. 1996).
The social autopsy is not only the technique for identifying social determinants of death, but as an interventional tool is a much powerful means of empowering community (by data sharing with them) for making them understand and expressing concern behavior in overcoming barriers and social determinants, linked with deaths in their localities (Biswas, A., et al., 2016;Biswas, A., et al., 2016;Kalter et al., 2011).

Verbal Autopsy/Social Autopsy integration
By investigating the mortality events using VA and SA technique (based on TPtoSF), ce one may be able to identify the barriers and role of different determinants in death causation and ultimately be able to connect non-biological determinants with the biological cause of death.However, each death should be investigated using this integration to get a generalized view of the situation in the community.Only then the gathered data may be helpful to comprehend for policy implications.Such integration of VA with SA and practicing it over a sustained period of time as a single procedure on each mortality event in the community, can give us extended set of determinants to focus, which can drive the policy towards improvement in child survival estimates in Pakistan.Literature proves that such integration has given compelling results for improving maternal and child survival estimates (D'Ambruoso, L., et al., 2010;Kalter et al., 2016;Koffi, A. K., 2015;Koffi, A. K., et al., 2016).

Rationale of the study
Several attempts have been undertaken across the globe, where the VASA integration technique have been attempted to capture the broad range of data related to child mortalities (Bensaid et al., 2016;Kalter et al., 2016;Koffi, A. K., et al., 2016), however, the literature lacks the evidence of the use of integrated VASA technique in the urban slums of Karachi (which are the underserved areas of Karachi) using Pathway to Survival framework.Additionally, Pakistan's traditional mortality investigations only focuses on the biological causes of death sparing the social events around the death incidents, and hence misses out a huge amount of information which is of worth importance in terms of policy (M B Siddiqui, 2016).Therefore, considering such deficiencies in our very local context, this study has been designed to conduct the verbal autopsy-social autopsy (VASA) of child mortality cases exploring the the process of preventive measures adopted before the illness and process of casemanagement and healthcare delivery after the illness (in terms of The Pathway to Survival framework) for fatally ill children, which were later died.

Research Hypothesis
This study intends to conduct a pilot testing of integrated VASA under-five mortality investigation (build on Pathway to Survival Framework) in Karachi slums, to see whether this investigation can feasibly be conducted (and culturally acceptable) in Karachi slums; be able to identify the indepth reasoning of under-five mortality in Karachi slums; be able to develop specific interventions for improving child survival of urban slums, and be able to reduce the under-five mortality events Karachi slums using interventions.

Research Objectives
The main aim of the study is to conduct the integrated VASA under-five mortality investigation in urban slums of Karachi.This integration will be used to (1) identify the biological causes of death and social determinants linked with the death incidents (especially the case-management and careseeking process); (2) to identify potential interventions for improving the casemanagement and care-seeking process by community interaction.
(3) to identify the effect of such interventions in reducing the child mortality estimates in slums under investigation.

Conceptual Framework
The Pathway to Survival framework is the most comprehensive, as it provides a more holistic view of the process and the weaknesses along the route of healthcare delivery process.The framework supports the implementation of Integrated Management of Childhood Illness (IMCI) guidelines through focusing on the continuum of health delivery process by defining the case management practices and care-seeking process both inside and outside the home.The framework underscores that a death of a child represents a breakdown in one or more of these steps of this "Pathway."It implies the need for accepted standards of quality of care, both within the community and in health facilities.Although no human endeavor is as straightforward as the Pathway proposes, it was useful for concentrating ideas.One might say that the children may take a "detour to death" along the Pathway.The interventions should correct that detour.

Study site
The VASA Under-five mortality study will be conducted in the urban slums of Karachi.No updated data exist on the trends on child survival indicators in such slums involving the biological and social causes of death.

Study population
The parents of all those children who died either as stillbirths till the age of five years (with the duration of time between death and point of data collection should not be more than one year), died due to any illness (excluding the physical injury caused by accident) will be enrolled (after consent) in the study.

Study design
The study is a mix-method interventional study, involving both the quantitative and qualitative components.The study will be identifying the biological causes and social determinants of under-five deaths.This will be followed by development of interventions.These interventions will be introduced in 06 randomly selected slums compared with 06 controlled slums.These interventions will be given for complete one-year duration for each intervention followed by assessment of effect of these interventions in terms of prevalence and incidence of under-five deaths after one year period of interventions.

Sample size
Based on the total population living in slums of Karachi, the sample size was calculated (using Creative Research System) to be 384 with 95% Confidence level and 5% absolute precision.The sample size has been inflated to 400 cases as a round off figure.From the data of Local town health office, we will be identifying the death statistics and household addresses in these 12 regions.We will also be sending local community health workers to identify the additional deaths (in addition to local town health office records) in these areas using snowball technique.Each identified area/slum have been allocated a quota based on proportionate quota sampling technique (proportionally of the overall sample size required for the study).For each identified area (slums), the quota will be filled through simple random sampling technique.

Pre-Survey Household Visits (PSHVs)
Before the actual data collection, Pre-Survey Household Visits will be paid to all the identified households.During the PSHV, consent will be taken from parents of deceased children.Households will be selected based on the inclusion and exclusion criteria.

Inclusion and exclusion criteria
1.All those households where there has been any child death, between 0 to 59 months, i.e. stillbirths till under-five years of age will be included in the study.2. All the parents will be enrolled for interview (both parents without any psychiatric disorder and only nonpregnant mothers at the time of interview).3.Only those household where the deceased children died because of any illness will be included in the study.Any child death which was not a result of any illness will not be recruited.4. All households where stillbirths and under-five deaths occurred within the last one year from the start of data collection date (to minimize recall bias) will be recruited.Any death outside these dates will not be recruited.5.All parents with the possibility of emotional distress will not be included in the study.Such parents will be referred to counselor for counseling sessions.

Data Collection Tool
The data collection tools have been taken from Child Health Epidemiology Reference Group (CHERG), which includes VA and SA components.The CHERG's VA and SA (based on TPtoSF) components will be integrated as a single tool, which will be initially translated (and back-translated) in urdu (local language) and then validated according to Pakistan population.
(1) Validation of VASA questionnaire The validation processes will involve the use of face validity, content validity, discriminant validity and reliability and internal consistency (using inter-item correlation analysis and use of Cronbach's alpha correlation coefficient). (

2) Pretesting validated questionnaire
The translated and validated version of the questionnaire will be pre-tested in the community (other than study site but having similar baseline characteristics).
(3) CAPI Questionnaire (PDA compatible) Finally, the pre-tested questionnaire will be converted into computer-assisted personal interview (CAPI) software.The software will be loaded in Personal Digital Devices (PDAs) and will be used during data collection exercise.

Data Collection & Interventions
The study will be carried out in two phases.
The phase-1 will include surveying the recruited households and performing an interview based detailed quantitative survey with the parents of deceased children.To provide a mourning period to the family, and to minimize the recall bias, the gap between death and interview should not be more than 06 (six) weeks.Phase-1 will identify the biological CoD and the social determinants (barriers in health care access) linked with death.
The most frequently encountered barriers in health delivery process that led to the deaths of these children will be shared with these groups.In the same meeting, a complete set of possible interventions for overcoming these delays and barriers will be also be shared with these groups.The participants will then be asked to address each of these interventions by performing a SWOT analysis (considering issues specific to their community).The whole discussions will be transcribed and managed manually.The second phase will not only help us in exploring us the possible interventions required for the slum community, but it will help us in developing recommendations on establishing a sustained death notification system and also in creating awareness about the determinants which potentially affect the child mortality and ultimately empowering them to increase their health related responsiveness and accountability.Each of the interventions will be trial and tested for a year duration to see the effect on the child mortality.Interventions will be given to six randomly selected slums and other six will be kept as controls.

Data Analysis Plan
The electronic data will be shared with three reviewers.The biological cause of death will be given by the verbal autopsy reviewer.The CoD will be assigned according to ICD-10 classification.The output of CAPI data will be in excel.Descriptive statistics will be calculated.To check the normality of the quantitative variable, Shapiro-Wilk test will be applied.Mean ± SD or Median (IQR i.e.Interquartile range) will be calculated as appropriate.To check the internal reliability of all variables Cronbach's alpha will be applied.Frequency and percentage will be calculated for social and biological causes.
To check the association among all categorical variable, Chi square test/Fisher's exact test (if frequency is found less than five) will be used to see the association of biological and social factors linked with the death of the child.Discriminant analysis will be applied to identify the major biological and social causes of stillbirth and death after birth.CHAID Analysis-Chi-Squared Automatic Interaction Detection analysis.Finally, p-value will be taken as 0.05 as significance at 95% confidence interval.Confounders will be controlled through stratification of age of mother and child, parity, duration of illness of child, duration of marriage, whether delivery conducted by trained Healthcare professionals or informal and untrained community ladies to see the effect of these on outcome variable i.e. the death of the child.Post stratification chi square test will be applied to see the association.The final data analysis will reveal the biological cause of death and social factors causing delay and failures in case-management care-seeking process.

Outcome Indicators
The outcomes of the study include: 1. Biological causes of under-five deaths in the rural areas of Sindh.

The social contributors (based on
Pathway to survival framework) of child mortalities, that plays role in causing delay and failure in case management and healthcare delivery process.3. Recommendations for counteracting these contributors to improve the child survival in rural areas of Pakistan.4. List of interventions to reduce child mortalities.5. Effect of all these interventions (separately) on child survival estimates.
To get the true representation and generalized results, we have selected points across all the slums in Karachi, where within 05 km of residential addresses have at-least one household belonging to the most commonly encountered ethnic background residents of Pakistan.From the data of Bureau of Population Statistic-Karachi; Ministry of Health; and Metropolitan Corporation, we came across 12 such areas across the city, where there is almost equal representation of all these ethnic backgrounds in the residential population.These areas correspond to different slums.The corresponding respective slums are Chenesar goth, Mujahid colony, Hijrat colony, Hazara Colony, Sultanabad, Bilal Colony, Shireen Jinnah Colony, Yaseenabad, Landhi slums, Khud Ki Basti, Safora Goth, Malir slum.