Is Data from Community Mortality Data Mechanisms and Civil Registration and Vital Statistics Systems Compatible and Co-Usable? Evidence from a pilot Study in Nigeria

doi:10.21203/rs.3.rs-3046594/v1

Download PDF

Research Article

Is Data from Community Mortality Data Mechanisms and Civil Registration and Vital Statistics Systems Compatible and Co-Usable? Evidence from a pilot Study in Nigeria

https://doi.org/10.21203/rs.3.rs-3046594/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Background: Community mortality data mechanisms (CMDMs) -including verbal autopsy programmes and mortality surveillance systems-are taken as pragmatic solutions to the mortality data incapacity of CRVS systems in Low and Middle Income Countries (LMICs). This paper addresses issues related to the compatibility and co-usability of CMDM and CRVS systems data instruments and data. It demonstrates a methodology for the development of a CRVS system compatible community mortality checklist (CMC) instrument applicable to a routine community mortality surveillance system (RCMS). It compares mortality data from the Nigerian CRVS system and a pilot implementation of RCMS using the CMC.

Methods: Development of the CMC was demonstrated with the Nigerian death registration data instrument. RCMS and CRVS generated data on 180 deaths were compared in nine mortality data fields or elements of registered mortality events (ERMEs): age and sex of deceased, timeliness of registration (TOR), place -facility- of death (POD), locality of death, place of registration, death certification, and causes of death (COD).

Results: Differences between RCMS and CRVS mortality data were insignificant in some ERMEs like age, sex and POD of deceased but significant in others like TOR, COD and locality of death. CRVS data were slightly less male biased and more concentrated at older age groups while RCMS data were more evenly spread across age groups.

Conclusion: Data from the two sources are largely compatible. CMC based RCMS can significantly expand coverage of CRVS mortality data in LMICs and help adjust its sex and age bias.

CRVS

CMDM

RCMS

CMC

ERME

Nigeria

LMICs

Effective polices and interventions to prevent, control and manage endemic and emerging health conditions is a continuing imperative in low and middle income countries LMICs. This imperative underpins the necessity for viable, representative, and continuous data on health conditions and mortalities [1, 2]. Civil Registration and Vital Statistics (CRVS) systems are national statistical and administrative infrastructures for continuous and comprehensive collection, documentation, certification, production and dissemination of statistics on vital events related to the health of populations [3, 4]. Unfortunately CRVS systems in LMICs continue to exhibit capacity latency and ineffectiveness in the production of mortality data, i.e. data on deaths and causes of death [5]. Strategies aimed at improving the mortality data capacity and productivity of these systems are usually based on community mortality data mechanisms (CMDMs). CMDMs are designed to collect data on deaths and causes of deaths in communities and often involve verbal autopsies (VA) to determine causes of death (COD). The VAs are implemented within periodic or occasional community surveys or within other data producing structures like health and demographic surveillance systems and community surveillance system (RCMS). It is also possible to have RCMS CMDMs systems focusing on tracing and listing of mortality in communities with simplified approach to determination of CODs [25].

Already formal Policies and interventions promoting the adoption and use of CMDMs to enhance CRVS systems mortality data capacity in LMICs is gaining traction at various levels of governance [6, 22]. At the National level Countries like India, Mozambique, Zambia, Malawi have already adopted some form of CMDMs (especially VA) programmes into their CRVS systems. At the multi-lateral level the African Programme for Accelerated Improvement of Civil Registration and Vital Statistics (APAI-CRVS), and Second Ministerial Conference on Civil Registration and Vital Statistics in Asia and the Pacific have recommended the integration of mortality surveillance into CRVS systems in the two regions [7, 8]. VA based community mortality data acquisition has also been implemented in some other LMICs. But, these have been in intermittent surveys rather than within CMDMs and have been implemented outside of CRVS systems.

1.1 Challenges to the Use of CMDMs to Improve CRVS Systems Mortality Data Capacity

However, strategies aimed at improving the mortality data capacity and productivity of CRVS systems in LMICs using VA based CMDMs and others have to overcome a number of challenges. One challenge pertains to functional, operational and instrumental differences between CMDMs and regular CRVS systems. This challenge with compatibility, in addition to others imposed by resources, technology and socioeconomic and cultural contexts of CRVS systems in LMICs, are major limitations to the use of CMDMs to improve mortality data productivity of CRVS systems in LMICs [5]. Issues related to the functional, operational and instrumental compatibility of standardized verbal autopsy (SVA) based CMDMs and CRVS systems are yet to receive adequate attention in the literature. The contextual, systems, technological and operational issues involved are also just beginning to receive some research attention [9].

Another challenge relates to the reliability and fidelity of data collected through CMDMs in at least two aspects of data quality. The first aspect relates to how representative data from CMDMs are of mortality situations in communities and if they can be used to derive or corroborate established indices of mortality [10]. This is important because of the well-known failure of CRVS systems in LMICs to collect data on most deaths in communities and reliance in these countries on hospital based data for mortality estimates. Thus CMDMs must be shown to be able to bridge the gap in the collection of data on mortalities in communities. The second aspect relates to the compatibility, congruence and correlation of data collected through CMDMs with that from regular CRVS systems operations. Establishing the latter will provide a basis not just for cross validation of data from the two sources but for assuring that data from them can be reliably co-used or integrated to produce indicators and estimates of mortality conditions like death rates, mortality fractions and proportion of deaths ascribable to various causes.

1.3. Paper Objectives and Justification

This paper is motivated by the need to give analytical attention and develop innovative solutions to the multidimensional challenges involved in the use of CMDMs to enhance the mortality data capacity of CRVS systems especially in LMICs. It takes off from two previous studies on the same theme. The first study analysed key issues impinging on the use of SVAs based CMDMs to augment mortality data production of CRVS systems in LMICs [11]. The second envisioned and characterized a CRVS system adapted routine community mortality surveillance (RCMS) as a pragmatic alternative to solution based on standard verbal autopsies [12]. The proposed RCMS is a CMDM based on community mortality checklist (CMC) developed from CRVS systems data instrument. The proposed RCMS mechanism was characterized as a simplified, context (country) specific, CRVS system based CMDM. It is designed to facilitate comprehensive coverage of mortalities, especially deaths in communities which are largely missed by CRVS systems, in LMICs. Its anticipated data production procedure would involve regular community visits through canvassing or notification activities and administration of the community mortality checklist (CMC) data instruments developed from CRVS system data instrument. This paper furthers two objectives related to the use of CMDMs in general and the proposed and CMC RCMS within CRVS system in particular. One objective is to demonstrate a methodology for developing the CMC instrument from a CRVS data instrument. The second objective is to implement a statistical comparism of mortality data collected from a rapid pilot implementation of the CMC based RCMS CMDM and regular CRVS activity. Both objectives have been implemented with respect to the Nigerian CRVS system. Detailed characterization of the Nigerian CRVS system including its functions, institutional and operational structures, socio-economic contexts and the status of mortality data activities are available in other publications [3, 5, 13, 31].

Justifications for the CMC RCMS proposition have been presented in [12]. However it is important to reiterate that the RCMS proposition as well as the objectives and methodology of this paper are hinged on three assumptions on the imperatives for effective CMDM interventions in CRVS systems. One is a need to develop CRVS systems compatible CMDMs. Two is the need to develop CRVS systems compatible data instruments. Three, is the need to establish the compatibility of CRVS and CMDM generated mortality. The first section of this paper concludes with brief discussions on these imperative. The second section commences with a presentation on the methodologies applied to attain the paper’s objectives.

1.4. The Need for CRVS Systems Compatible CMDMs

The literature indicates an absence of comparative systematic organizational, work content, work process and contextual analyses in the selection or development of appropriate CMDMs for CRVS systems in low and middle income countries. Rather it shows a preponderance of studies and projects adopting verbal autopsy based CMDMs. However CRVS mortality data (death registration) field experiences in Nigeria and some other LMICs indicates that CRVS compatible CMDMs needs to meet some basic requirements related to CRVS systems functions, operations, data types and data instruments [13]. Using input-process-output model it can be seen that CRVS systems, as key components of National administrative and statistical systems, fulfil a fourfold mortality (data) function: data collection, documentation, certification and vital statistics production (Fig. 1). Within the nomenclature of CRVS system operations, mortality data collection is not just the tallying of occurrence of deaths and their characteristics as with other mortality data collection processes like surveys and censuses. It is a registration process involving the collection, permanent documentation, statutory certification and production of vital statistics from personalized data fields on each death event. Within the nomenclature each data instance is an event referring to specific experiences in an individual’s life course. These personalized data fields (e.g. name, age, sex, occupation, cause of death-COD-, place of death-POD-,etc.) have been described elsewhere as elements of registered mortality events- ERME [12]. The registration process has a statutorily or administratively prescribed standard format and should enable continuous production and flow of data on mortality patterns and trends from the civil registration activity [14]. The implication is that CRVS compatible CMDMs should facilitate the implementation of the fourfold CRVS functions.

A CRVS system compatible CMDM must also comply with the standard operating procedures of CRVS systems in each country. This includes compliance with national CRVS protocols and operational schedules and routines (timing) [4, 15]. Operational compatibility of CMDM interventions will facilitate and strengthen the implementation of CRVS mortality data and civic administration functions, processes and output. Data type compatibility is also necessary for CMDM effectiveness within CRVS systems. This implies that any CMDM to be used within the CRVS system would need to enable the collection of not just basic mortality data fields (for example age and sex of deceased and cause of death) but requisite data types (like name of deceased, place of death, date of death) to perform CRVS functions of events documentation, certification and vital statistics production. The inability of many VA based CMDMs to facilitate these function is a major limitation.

1.5. The Need for CRVS Compatible CMDM Data Instruments

Developing CRVS systems compatible data instruments is imperative to the successful use of CMDMs within CRVS systems. This has not been given attention in the CRVS literature mainly because most interventions have focused on VA data instruments and have not considered the need to combine VA and CRVS generated data. Developing CRVS compatible CMDM data instruments is indispensable to effective CMDM interventions in CRVS systems in LMICs for at least three reasons. One, it is critical if any CMDM is to enable the effective performance of CRVS systems functions. Two, it is critical also if CMDMs data will be combined with CRVS data for production of vital statistics and other indices of mortality. It is again crucial for inclusive national coverage of mortality events by CRVS systems in LMICs. The need for instrumental compatibility is heightened by differences in CMDM and CRVS instruments. This is especially true in many LMICs where CRVS data instruments are often simplified to match the technical capability of CRVS staff. A congruent issue relates to the need for appropriately sized instruments for easier and cost efficient administration in communities [16]. Thus one approach to the development of CRVS compatible CMDM data instruments is to harmonize CMDM and CRVS instruments. This implies the need for a harmonization methodology taking into account the functions, operational procedures and organization structure of National CRVS as well as the content and lay out of National CRVS death registration instruments. Another approach is to out rightly and ab initio develop CMDM instruments from national CRVS systems data instruments. This approach is easily applicable to new or proposed CMDMs as demonstrated later in this paper.

1.6. The Need to Test and Establish Compatibility of CRVS and CMDM generated mortality Data

At a general level comparing multi source data relates to issues with data harmonization, data integration, data use, data quality standards and data consensus. This is very important in methodologies for compilation of vital statistics and mortality indicators from diverse demographic sources like civil registration, censuses, surveys, other official data repositories and CMDMs [17]. In spite of its necessity studies comparing mortality data from community mortality data mechanisms (CMDMs) including VAs and RCMS with that from regular CRVS systems are not common. Rather comparative analysis of mortality data in the context of CMDMs especially VAs have focused mostly on variations and differences in accuracy of COD predictions between applications or between applications and physician determined CODs [18, 19, 23, 24]). Studies comparing CMDM and CRVS data may be lacking partly because mortality data from CRVS systems in LMICs are usually regarded as inadequate and unviable. Thus CMDM data have often been regarded as alternatives rather than complements to CRVS data. Studies comparing CMDM and CRVS mortality data may also be lacking because programmes to integrate CMDMs (especially VAs) with CRVS systems are still at infancy. As such there is no evidence of comprehensive national implementation of programmes to integrate VAs and other CMDMs into CRVS systems in LMICs where there is a need for them. Existing programmes involve sample civil registration with VAs and do not aim at comprehensive national use of VAs within CRVS systems. However, there is evidence of some improvements in the volume and quality of mortality data from CRVS systems in LMICs in the past few years [20, 21]. This provides strong justification for analysis of CRVS mortality data especially as they compare and contrast with mortality data from other sources.

This section describes the approach to the development of a simplified RCMS CMC data instrument from the Nigerian CRVS data instrument. It also describes the methodology for comparing the RCMS and CRVS system generated data on elements of registered mortality events (ERMEs). Both are based on a rapid pilot community study exploring strategies for enhancing the coverage and completeness of registration of mortality events in Nigeria by integrating simplified, community mortality checklist instruments based routine community mortality surveillance (RCMS) to complement mortality data collection from regular civil registration operations. The study involves iterative development of viable, simplified, context adapted, CMC instruments and pre-testing of the instrument to provide appropriate data for a preliminary statistical assessment of similarities and differences in ERMEs and CODs collected through RCMS and CRVS. The study would also evaluate the RCMS data instruments in terms of content, efficiency of administration, data returns and areas of improvement. Another objective of the study is to gauge social/community knowledge, attitudes and responses to mortality related questions and schemes to enhance mortality data through community surveys. The latter is the subject of upcoming analyses of CRVS knowledge attitude and practices in the area.

2.1. Development of CMC from CRVS Data Instruments

CRVS mortality instruments are generally case sheets. A case sheet is a form for collecting data on an instance or case of an event, for example death. Case sheet forms may be paper (as currently in the Nigerian CRVS) or digital (as in many automated VA or other CAPI applications). Instead of a case sheet, data forms may take a case-list format recording each case in a line in a table. The development of case list instruments benefits from relational model of databases in which cases become elements/line/components or tuples in a table. A table is a collection of related cases in relational database [26, 27]. The content of each table is defined by shared attributes of cases of the phenomenon comprising the table. For the CMC checklist, the shared attributes include characteristics or elements of registered mortality events like name, place, time, sex, place of registration (POR), etc of a death event.

The process of development of case lists from sheets can also be seen in terms of transposition of elements in a table. In this instance, the horizontal axis of a table is transposed into the vertical axis or vice versa. For instance in Fig. 2 the attributes (A,B,C) of each case (1,2,3), that is (A1, B1, C1….A4, B4, C4) listed vertically in each case sheet are transposed horizontally in the case list. These principles were reflected in the development of the CMC data instruments from the Nigerian CRVS system mortality data instrument. However, the structure and content of the instrument was determined by the need to: i) cover key mortality data fields, ii) facilitate the performance of key CRVS systems functions of events registration, documentation, certification and vital statistics production. Generally, the design of any CMC instrument should take into account a number of local factors like the structure and content of national CRVS mortality data instruments, CRVS operational format, data acquisition technology as well disease endemicity and common causes of death.

The development of the CMC instrument was based on a sample of the Nigerian CRVS death registration instrument accessed online from https://data.unicef.org/wp-content/uploads/2017/12/deathRegFormNigeria.pdf. The instrument comprises two major sections: i) event registration/documentation section and ii) main mortality characteristics section. The registration identification/documentation section includes fields for place, date, serial number of registration. The main mortality characteristics section comprises 16 core data fields (ERMEs) divided into two parts: i) particulars of deceased and ii) particulars of informant. Key data fields contained in the deceased section include the name, age, date of death and place of death (POD). Two POD ERMEs are included namely: facility and locality of death. Other key ERMEs include causes of death, evidence of medical certification of death, ethnicity and occupation of deceased. Informant section has three data fields on relationship between decedent and informant, name and location of informant. The CMC case sheet instrument was designed in a table format. Relevant data fields in the CRVS instrument were transposed into appropriate sections in the CMC. However, since the CMC is a compressed instrument, key words, rather than full questions were used to represent a data field. For example instead of the question: “what is the age of the deceased” or “age at death” the word ‘age’ is inserted into the appropriate column in the CMC.

2.2. ERMEs Based Comparism of RCMS and CRVS Generated Mortality Data

Data Collection: The pilot study was designed to collect and compare routine CRVS and CMC RCMS generated mortality and COD data within a geographic area approximating the catchment of a civil registration unit in Nigeria. A study frame based on registration unit catchments was used because it provides the most appropriate spatial and organizational framework for: i) comparing data from the two sources; ii) prospective integration of CMC RCMS and CRVS operations; iii) integration of data from RCMS into regular CRVS system data procedures. The CRVS catchment purposely selected for the study is located in Osun State, SW Nigeria, and has uniform socio-economic, cultural and natural conditions. It includes parts of a major urban center- Ile Ife with estimated population of 250,000- and six adjoining rural settlements ranging in size from 10-15000persons. RCMS data was collected through a rapid/flash community survey. Respondents i.e. informants, were relatives or other persons associated with a mortality event. The informants were selected randomly during canvassing of selected communities. One hundred of such informants were interviewed during the survey by 5 pairs of interviewers in a 5 hour period including travel time between settlements. Ninety-five interviews were satisfactorily completed and used in the analysis. Complementary data on CRVS registered mortality events were sourced from the CRVS registration catchment unit within the study area. The number of CRVS registered mortality events used in the study was determined by three criteria. One is the need to ensure minimal difference in the number of CRVS and CMC RCMS registered events. Two is the need to minimize disparity in date of CRVS and CMC RCMS registered events. Three is to avoid duplication of registration of mortality events. Based on these criteria the 93 CRVS registered mortality events in the one year period preceding the survey were used. Thus a total of 188 registered death events were analysed in the study.

Data Analysis: RCMS and CRVS generated data were compared in 9 elements of registered mortality events (ERMEs) including causes of death (CODs) (Table 1). Two of the 9 ERMEs are numeric variables – age of deceased (in years) and time (timeliness) of registration (in number of days after death) while seven are categorical variables: sex of deceased, place -facility- of death (POD), locality of death, place of registration (POR), certification of death, doctors verification of death and causes of death. Values of the quantitative variables are numeric and not pre-specified. Appropriate values were pre –specified for each qualitative variable. Some of these variables were assigned standard values, for example sex, POD and POR. Initial values for the locality of death variable were actual localities of occurrence of deaths. This was recoded into four categories, namely: i. Ile-Ife, ii. same village, iii. within State and, iv. outside State. The objective was to facilitate analysis and interpretation of results. Data analyses for the numeric variables were implemented with descriptive statistics and F test parameters and with frequency distributions and chi sq. statistics for non-numeric variables.

Table 1

ERME Variables and Values
Variable Types	Variables	Values
Quantitative	Age of deceased	Numeric values in years
Quantitative	Time of registration	Numeric time values measured in number of days from date of death to date of registration
Qualitative	Sex of deceased	i) Male ii) Female
	Place of death	i) Home ii) Hospital iii) Road iv) Others
	Locality of death	i) Ile Ife ii) Same Village iii) Within State iv) Outside State
	Place of registration	i) Urban ii) Rural
	Certification of death	i) Yes ii) No
	Death verified by a doctor	i) Yes ii) No
	Causes of deaths	Various cases as indicated by informants

3.1. Features of RCMS CMC instruments Developed from CRVS Data Instrument: the Nigerian Example

While the Nigerian CRVS death instrument is designed for administration within registration centers, the RCMS instrument, developed from it, is based on the strategy of rapidly collecting essential mortality and COD data fields from households within communities. Put differently, the CRVS instrument is office based while the CMC is a field instrument. The CMC is designed with the objective to be minimally socially intrusive, reduce barriers to interviews and encourage and elicit quick response. Table 2 shows that the objective of compressing and including key mortality data fields (ERMEs) in the Nigerian CRVS case sheet instrument within the CMC case list instruments was achieved. All of the sections (identification/documentation, deceased and informants) in the case sheet instrument are represented in the CMC.

Table 2

Comparism of CRVS and RCMS Data Instruments
Data Sections	ERMEs	RCMS (CMC)	CRVS	Remarks
Identification	Registration Centre, locality	Identification Data Fields	Identification Section	Key identification data fields in CRVS instrument inserted into RCMS CMC
Documentation	DOR, Record Volume, Entry no	Documentation Data Fields	Documentation Section	Key documentation data fields in CRVS instrument inserted into RCMS CMC
Deceased	Name, Sex, Age, POD, DOD, COD, Etc.	Deceased Data Fields	Deceased Section	Key data fields in deceased section of CRVS instrument inserted into RCMS CMC
Informant	Relationship, Name, Residence	Informant Data field	Informant Section	Key data fields in informant section of CRVS instrument inserted into RCMS CMC

The CMC instrument designed from the Nigerian CRVS instrument has four sections Fig. 3. The first section covers key attributes of a deceased: name, sex, age, date of death and place of death (POD). The second is a checklist of possible symptoms shown by the deceased. Most of the symptoms are those commonly reported with endemic diseases in the area. The third section is a checklist of possible causes of death. This include accidents, violence and endemic diseases like malaria, typhoid, diarrhoea, cholera and common and emerging non communicable disease like cancer, diabetes, hypertension, etc. The fourth section includes fields to verify if a recorded death had medically verified, if it was medically certified and the relationship of informant with the deceased. Each CMC sheet is designed to be used within a single community (locality) and for a single day of registration. Unlike other CMDMs, the inclusion of registration/documentation data field in CMC, make the proposed RCMS CRVS compliant. With this death events covered by RCMS can be certified and documented.

3.2. Summary of Pilot Survey Returns

Fifty-one of the 95 interviews conducted through community canvassing during the pilot RCMS survey returns took place in Ile-Ife, the main urban settlement in the study area. The remaining 44 were conducted in various rural settlements canvassed. All of the 93 CRVS registered deaths used were accessed from the National Population Commission registration center located in Ile-Ife, within the study area. Thus 76% and 24% of the sample was sourced from urban and rural locations respectively.

Table 3

Summary of Pilot Survey Returns
	Localities of Registration	Estimated Population	Survey Returns (N)			% of total
	Localities of Registration	Estimated Population	RCMS	CRVS	Both	% of total
Urban	Ile-Ife	250,000	51	93	144	76
Rural	Lami, Store, Baasa, Agaran, Elekiri, Ifewara,	15,000	44	-	44	24
Total			95	93	188	100

How Does RCMS Data Compare with CRVS Data in ERMEs

Mortality data collected (deaths registered) through both sources occurred at a mean age of 65years with a standard deviation of 21, median of 69years and a maximum and minimum of 110years and 0years respectively. Mean age at death registered through CRVS (65years) was marginally higher than that from RCMS (64years). CRVS registered deaths also had a higher minimum and median age at registration. As such differences in the mean age registered by RCMS and CRVS were not significant (F:.800; Sig: .372). The mean number of days between date of death and registration for both sources was 963days (2.5 years- 32 months) with a median of 323 days and standard deviation of 1363. Deaths registered through CRVS (mean: 347 days; SD: 915; median = 70days) were significantly (F: 46.347; Sig:.000) more timely than that registered through RCMS (mean:1567 days; SD: 1470; median = 1130days ). This may be because CRVS are routinely available while the field survey was unscheduled and spontaneous.

Overall males constituted > 68% of the deaths registered while females were about 31%. Deaths registered through RCMS and CRVS had similar sex distribution with male registration dominating each. Sex distribution of deaths registered by both sources has a weak association (Cc = .065) with source of data. Thus the differences in the sex distribution of deaths registered from both sources are not significant. Most deaths registered by both sources occurred at home (68.1%). Approximately 24% occurred in hospitals, while 7% occurred along transport routes and differences in place of death registered by the two sources are not statistically significant. However, a higher proportion of the deaths registered by RCMS (about 75%) occurred at home than those registered through CRVS (61%). Overall, differences in three ERMEs (age at death, sex and place of death of deceased are weakly associated with data source. Conversely, time of registration, locality of death, and certification of death are strongly associated with data source. Thus difference in age at death, sex of deceased and place of death between the two sources is not significant while differences in timeliness and location of registration are significant. Also facility of death is not strongly associated with data source while locality and certification of death are strongly associated with data source. In effect data collected through RCMS and CRVS are similar in terms of some ERMEs:

Table 4

Comparism of RCMS and CRVS Data in Elements of Registered Mortality Events
Mortality Data Elements	Measures/Values	Data Sources			Test Statistics
Mortality Data Elements	Measures/Values	RCMS	CRVS	Both	F/Chi Sq.	Sig	Association
Age at Death (in Years)	Max	110	110	110	.800^a	.372	.065^c
	Min	0	20	0
	Mean	63.6	66.4	65.0
	SD	25.9	18.1	21.3
	Median	65	72	69
Time of Registration (Days After Death)	Max	7760	7521	7761	46.347^a	.000	.447^c
	Min	1	2	1
	Mean	1567	347	963
	SD	1470	915	1363
	Median	1130	70	323
Sex of Deceased (%)	Males	69.5	67.7	68.6	.065^b	.798	.019^d
Sex of Deceased (%)	Females	30.5	32.3	31.4	.065^b	.798	.019^d
Place (facility) of death (%)
	Home	74.7	61.3	68.1	4.742^b	.192	.157^d
	Hospital	20.0	28.0	23.9
	Road	5.3	9.4	7.4
	Others	0.0	1.1	0.5
Locality of Death (%)	Ile-Ife	70.5	96.8	83.5	23.918	.000	.333
	Same Village	8.4	0.0	4.3
	Within State	13.7	2.2	8.0
	Outside State	7.4	1.1	4.3
Location of Registration (%)	Urban	52.7	100.0	76.6	56.235^b	.000	.480^d
Location of Registration (%)	Rural	47.3	0.0	23.4	56.235^b	.000	.480^d
Certification of Death	Yes	29.5%	94.6%	61.7%	95.256^b	.000	.580
Death Verified by Doctor	Yes	45.3	28.0	36.7	8.409^b	.015	.207^d
a = F Statistic, b = Chi Sq. Statistic, c = Eta (nominal by interval association), d = contingency coefficient (nominal by nominal association)

3.3. Age Distribution of Deaths by Sources

Further insight into the characteristics of mortality data collected from the two sources and how they compare can be derived from detailed analysis of the grouped age distribution of the deaths (by both sources). This indicates that deaths occurring between 71–80 years are the most frequently registered. Some 26.1% of all deaths registered occurred within this age group and > 60% of all deaths registered occurred from 61-90years (Fig. 4). Deaths of infants and children (0-10years) and young adults (11–20, 21–30) are the least registered. Age distribution of death data collected through each source generally follows this pattern. However the dominance of age 71–80 in the deaths registered is more pronounced for data collected through CRVS than for RCMS. Notably over > 70% of all deaths registered through CRVS is for deaths occurring at 61years and above. Corresponding figures for registration through RCMS is < 60%. Thus RCMS showed more potential in collecting deaths occurring at younger age than CRVS. The preponderance of death registration for age 71–80 suggests at least two things. One is that most deaths occur at that age. It may also point to a tendency to report deaths at older ages, a situation that is credible given that inheritance/entitlements issues are prominent at this age and the socio cultural practice of ‘celebrating’ deaths of older people.

3.4. Age Distribution of Registered Deaths by Sex of Deceased

The pattern of sex distribution of registered deaths by age groups is also very instructive. Peaking of age of registered deaths at 71–80 years occurs for both males and females (Fig. 5). However peak registration at this age is much more pronounced for males (39%) than for females (16.9%). In addition age distribution of registered female deaths not only exhibits a second peak at 41–50 years but unlike males, > 50% of registered females deaths are age 21–60 years. Conversely > 65% of registered male deaths occurred at 61years and above. Notably > 40% of female deaths are within the reproductive age bracket hinting at the possible causes of death.

3.5. Comparism of Causes and Symptoms Associated with Deaths Registered by RCMS and CRVS

A major challenge with VA/CRVS collected community mortality data is attributing definite causes to particular mortality events. However at a sub population level it is possible to infer the prevalence of various CODs or symptoms associated with registered death events. The COD/symptoms associated with the deaths registered from the two sources- CRVS and RCMS- are shown in Fig. 5. The most common are fevers/headache (30%), senility (22%), road traffic accidents/injuries (11%), high blood pressure/heart failure (7%) and cancers (4%) and coughs (4%). Some important but less reported COD/symptoms include pregnancy/child birth (2%), diarrhoea/vomiting (2%) and wasting. Others like violence, HIV/AIDS and breathing difficulty were associated with 0.5% of the deaths. About 16% of the registrations did not report or specify any COD or symptoms. Source by source, fevers/headaches, senility and accident/injuries were the most commonly reported CODs/symptoms associated with registered deaths. However reports of deaths from the first two were higher with CRVS registered deaths while reports of deaths associated with accidents/injuries were higher with RCMS registered deaths. In addition, deaths associated with pregnancy/child births, cancer, HIV/AIDs, violence and breathing difficulties was registered by only CRVS. On the other hand only RCMS registered coughing and wasting as CODs.

Distribution of symptoms/COD varies with sex and data source (Figs. 6). For both sources combined (Fig. 6b), proportion of deaths associated with some causes/symptoms: fevers/headaches, senility, and high blood pressure are higher for males. Males (17%) also have a higher proportion of deaths with unspecified causes than females (13.6%). Conversely proportion of deaths associated with senility, accidents, diarrhoea/vomiting and wasting are higher for females than males. Some causes of the registered deaths are associated with only one sex: HIV/AIDs, violence and breathing difficulty with males; pregnancy with females. Generally, association of causes of death with data source is more significant than its association with sex (Table 5). Thus variation in causes of the registered death is more significant with data source than with sex of deceased. As such the two sources of data may register significantly different causes of death.

Table 5

Analysis of Variations in Causes of Registered Deaths by Data Source and Sex
Factor	Test Statistics
Factor	Chi Sq.	Sig	Association ^d
Data Source	59.928	.000	.465
Sex	22.250	.175	.325
d = contingency coefficient (nominal by nominal association)

Provision of accurate data on mortalities is a major requirement for effective health systems planning and management. It is also a key element of the global development agenda expressed in the Sustainable Development Goals. However, CRVS systems in LMICs still lack the capacity to produce viable mortality data, thus the need for supportive mechanisms to collect data especially on deaths in communities. This paper addressed two issues critical to the use of RCMS CMDM to enhance the mortality data output of CRVS systems in these countries. One is the need to develop appropriate data instruments not only adapted to the socio-cultural context of each LMIC but also to CRVS systems fourfold function of death registration (death data collection), documentation, certification and vital statistics production. The second is the need to ensure that data from such CMDMs are not only representative of mortality characteristics in communities but also demonstrates congruency, and can be co-used, with data from regular CRVS procedures. In the first issue the paper discussed the development of a CMC instrument from the Nigerian CRVS data instrument for the implementation of a RCMS mechanism. It showed how key mortality data fields (ERMEs) in the Nigerian CRVS data instrument were adapted into the CMC instrument. It also demonstrated how key personal identification and events documentation fields were incorporated into the CMC instrument. This is to enable the parent CRVS system to document and certify deaths registered through the RCMS mechanism.

For the second issue, data from a pilot community survey in SW Nigeria was used to evaluate characteristics of registered mortality events in the area as well as likely differences and similarities in RCMS and CRVS collected data. The survey involved community canvassing to register deaths using the CMC instrument. Results show that average age at death for all (RCMS and CRVS) registered mortality events was 65years; average time after deaths before registration was 963 days; 69% of the registered deaths were males and females 31%. Sixty-eight per cent of the registered deaths occurred at home, 28% in hospitals, and 9% along transport routes. Some 76% were registered in urban areas and 24% in rural areas. About 37% of the deaths were verified by medical personnel (MCCD) and 62% were certified (statutory non-medical certification) by the National Population Commission (NPC). Four main symptoms/causes were associated with the registered mortality events fevers/headaches (30%), senility (22%), accidents/injury (11%), and high blood pressure/heart failure (7%).

Results indicated that differences between the pilot RCMS and CRVS registered events were insignificant in some ERMEs like age and sex of deceased and facility of death but significant in others like time of registration, causes and place of death. Both were ‘weak’ in collecting data on infant/child and female/maternal mortality and in the attribution of definite causes to particular death events. CRVS data were timelier but slightly less male biased and more concentrated at older age groups while RCMS data were more evenly spread across age groups. Results suggest that RCMS can significantly expand coverage of CRVS mortality data and help adjust its sex and age bias. This will inform subsequent phases of the study especially improvements to the RCMS and strategies for integrating it into CRVS.

There is no evidence of previous studies comparing mortality data, especially those collected through community surveillance and CRVS systems, in terms of ERMEs. However, some of the results can be put in the context of other studies of characteristics of registered mortality in some LMICs. For example the Ibadan (SW Nigeria) Study of mortalities amongst a sample of elderly population > 65years recorded a total of 265 deaths over a period of 39 months. These deaths were distributed almost equally between males (49.3%) and females (50.7%). The main causes/symptoms reported in this study were malaria, diarrhoea, unspecified fevers, hypertension and asthma [28]. Another study in SW Nigeria reported deaths for 209 out of 302 sampled households over a ten year period with only 11% of them registered [29]. A survey of 2538 households in Bangladesh recorded 571 deaths that occurred over a 3 year period. Some 17% of the deaths were registered with the country’s CRVS systems with significant disparity between male (26%) and female (5%) deaths registered [30].

These studies collaborate some of the results on characteristics of registered mortality found in this study. This confirms, to some extent, that mortality data collected through community canvassing and surveillance are representative of mortality patterns in communities. It also confirms that data from both sources are compatible. Thus RSCM data can be used to complement that from formal CRVS systems procedures. However given the relatively small number of mortalities, extrapolations from the study must be implemented with caution. This is especially in terms of interpretation of symptoms and causes associated with the deaths. There is a need for larger scale studies to help confirm the results of this study and develop models for integrating CMDM and CRVS operations and data.

Ethical Approval: Not applicable.

Consent to participate: Informed consent was sought and received from all persons interviewed during the study. The objectives and use of the study was fully explained in the local language.

Competing interests: The authors have no competing interests as defined by BMC, or other interests that might be perceived to influence the results and/or discussion reported in this paper.

Authors' contributions: Nnamdi Ifeanyi Maduekwe conceived the project, developed the methodologies, wrote the main manuscript and coordinated the review. Grace Vincent, Mary Oluwadamilola Oladunjoye , Olalekan Luqman Adebayo ,Inyangudo Gideaon Ntieno ,Folorunso Oluwabunmi, Simileoluwa Ageloye, Stephen Oladipo Aloko and Hilda Nwanneka Maduekwe participated in development of objectives, field data collection and analysis, and review of manuscript.

Availability of data and materials: Data used in the study can be accessed through the corresponding author.

Silva, R (2022): Population perspectives and demographic methods to strengthen CRVS systems: Introduction, Genus (2022) 78:8 https://doi.org/10.1186/s41118-022-00156-8
Bauserman, M et al (2020): Maternal mortality in six low and lower-middle income countries from 2010 to 2018: risk factors and trends, Reproductive Health, 17(Suppl 3):173 https://doi.org/10.1186/s12978-020-00990-z
Maduekwe N.I., Banjo O.O. and Sangodapo MO: The Nigerian civil registration and vital statistics system: contexts, institutions, operation. Soc. Indic Res. 2017, 134:651–674
UNDESA (2014): Principles and recommendations for a vital statistics system revision 3. Department of Economic and Social Affairs Statistics Division Statistical Papers, Series M No. 19/Rev. 3.
Maduekwe NI, Banjo OO, Sangodapo MO (2018): Data for the Sustainable Development Goals: Metrics for Evaluating Civil Registration and Vital Statistics Systems Data Relevance and Production Capacity, Illustrations with Nigeria. Soc. Indic Res, 140:101–24. doi:10.1007/s11205-017-1760-8
Bratschi M W (2021): Implementing Verbal Autopsy in the CRVS System, Strengthening Civil Registration and Vital Statistics (CRVS) Systems, Guide 4, Draft for Consultation Produced by Strengthening Civil Registration and Vital Statistics (CRVS) Systems Programme; Downloaded from
APAI CRVS (2019): Integrating mortality surveillance to strengthen civil registration and vital statistics systems, Issues Paper at the Fifth Conference of African Ministers Responsible for Civil Registration Lusaka, 14-18 October 2019
UNESCAP (2021):
de Savigny D (2017): Integrating community-based verbal autopsy into civil registration and vital statistics (CRVS): system-level considerations, Global Health Action, 2017 Vol. 10, 1272882 http://dx.doi.org/10.1080/16549716.2017.1272882
Qi, Jinlei et al (2021): Estimating causes of out-of-hospital deaths in China: application of SmartVA methods, Population Health Metrics 19:25 https://doi.org/10.1186/s12963-021-00256-1
Maduekwe NI, Banjo OO, Sangodapo MO and Abdulazeez A (in preparation): Use of Standard Verbal Autopsies to Improve Mortality Data Capacity of Civil Registration and Vital Statistics Systems in Low and Middle Income Countries: Analysis of Key Issues
Maduekwe N I (in preparation): Improvement of Mortality Data Capacity of CRVS Systems in LMICs: Is Routine Community Mortality Surveillance a Viable Option?
Maduekwe N I, Banjo O. O. and Anyabolu H (2018): Understanding Civil Registration and Vital Statistics as a Component of the Sustainable Development Goals Data Ecosystem in Nigeria, Ife Social Sciences Review, Special Issue on Institutional framework for the Practise of The SDGs, pp. 450-460
Vital Strategies and World Health Organization (2020): Revealing the Toll of COVID-19: A Technical Package for Rapid Mortality Surveillance and Epidemic Response. New York: Vital Strategies
Ghana Statistical Service (Undated): Standard Operating Procedures for Civil Registration and Vital Statistics System of Ghana, downloaded from https://www2.statsghana.gov.gh/docfiles/publications/
Serina, P et al (2015): A shortened verbal autopsy instrument for use in routine mortality surveillance Systems, BMC Medicine, 13:302, DOI 10.1186/s12916-015-0528-8
Global Burden of Diseases 2017 Collaborators (2018): Global, regional, and national age-sex-specific mortality and life expectancy, 1950–2017: a systematic analysis for the Global Burden of Disease Study 2017, Lancet 2018; 392: 1684–735
Murray CJL et al (2014): Using verbal autopsy to measure causes of death: the comparative performance of existing methods, BMC Medicine 2014, 12:5 http://www.biomedcentral.com/1741-7015/12/5
Weldearegawi B, Melaku Y A, Dinant G J and Spigt M (2015): How much do the physician review and InterVA model agree in determining causes of death? A comparative analysis of deaths in rural Ethiopia, BMC Public Health 15:669 DOI 10.1186/s12889-015-2032-7
Wakibi S and Ngure E (2021): An Assessment of Knowledge, Attitude, and Practices of Birth and Death Registration in Kilifi County in the Coastal Region in Kenya, BioMed Research International Volume 2021, Article ID 9963703, 8 pages https://doi.org/10.1155/2021/9963703
Chinganya, Oliver (2018): Civil Registration and Vital Statistics: Emerging Models and Work in Progress, A presentation to the 49th Session of the United Nations Statistical Commission, Side event organized by UNSD, New York: March 8, 2018.
Boyce Ross M. and Reyes, Raquel (2020): Implementing and scaling verbal autopsies: into the unknown, BMC Medicine 18:53 https://doi.org/10.1186/s12916-020-01527-8
Hazard Riley H. , Alam N , Hafizur R C, Tim Adair, Saidul Alam, Peter Kim Streatfield, Ian Douglas Riley and Alan D. Lopez (2018): Comparing tariff and medical assistant assigned causes of death from verbal autopsy interviews in Matlab, Bangladesh: implications for a health and demographic surveillance system, Population Health Metrics, 16:10 https://doi.org/10.1186/s12963-018-0169-1
Jha P., Kumar D., Dikshit R. et al (2019): Automated versus physician assignment of cause of death for verbal autopsies: randomized trial of 9374 deaths in 117 villages in India, BMC Medicine 17, 116
Vital Strategies, United Nations Economic Commission for Africa, United Nations Economic and Social Commission for Asia and the Pacific, and Statistics Norway (2020). Production of a Vital Statistics Report: Guide. NY, Vital Strategies
Maduekwe, N. I. (2011): The Application of Geographic Information Systems to Health and Demographic Surveillance Systems: a Case Study of Oriade HDSS, Unpublished PGD Project Submitted to the regional Center for Training in Aerospace Survey(RECTAS), Ile-Ife, Nigeria
Campbell E.J. and Shin M. (2012): Geographic Information System Basics. V.1.0, downloaded from https://2012books.lardbucket.org › pdfs › geogra... June 6, 2023
Lasisi AO, Esan O, Abiona TA, Gureje O (2018) Causes and Risk of Death Using Verbal Autopsy in The Ibadan Study of Ageing. J Trop Med Health: JTMH-118. DOI: 10.29011/JTMH-118.000118
Akande TM and Sekoni OO (2005): A Survey on Birth And Death Registration in A Semi-Urban Settlement in Middle-Belt Nigeria, European Journal of Scientific Research, Vol 8, No 2, 2005
Haider MM, Alam N, Ibn Bashar M and Helleringer S (2021): Adult death registration in Matlab, rural Bangladesh: completeness, correlates, and obstacles; Genus, 77:13 https://doi.org/10.1186/s41118-021-00125-
Makinde OA, Odimegwu CO, Udoh MO, Adedini SA, Akinyemi JO et al (2020): Death registration in Nigeria: A systematic literature review of its performance and challenges. Global Health Action, 13 (1), doi.org/10.1080/16549716.2020.1811476

No competing interests reported.

Download PDF

Reviews received at journal
03 Nov, 2024
Reviewers agreed at journal
07 Oct, 2024
Reviewers agreed at journal
07 Oct, 2024
Reviewers agreed at journal
06 Oct, 2024
Reviews received at journal
06 Oct, 2023
Reviewers agreed at journal
19 Sep, 2023
Reviewers invited by journal
18 Aug, 2023
Editor assigned by journal
12 Jun, 2023
Submission checks completed at journal
12 Jun, 2023
First submitted to journal
10 Jun, 2023

You are reading this latest preprint version

Is Data from Community Mortality Data Mechanisms and Civil Registration and Vital Statistics Systems Compatible and Co-Usable? Evidence from a pilot Study in Nigeria

Status:

Version 1

Abstract

Figures

1. Introduction

1.1 Challenges to the Use of CMDMs to Improve CRVS Systems Mortality Data Capacity

1.3. Paper Objectives and Justification

1.4. The Need for CRVS Systems Compatible CMDMs

1.5. The Need for CRVS Compatible CMDM Data Instruments

1.6. The Need to Test and Establish Compatibility of CRVS and CMDM generated mortality Data

2. Paper Methodology

2.1. Development of CMC from CRVS Data Instruments

2.2. ERMEs Based Comparism of RCMS and CRVS Generated Mortality Data

3. Results

3.1. Features of RCMS CMC instruments Developed from CRVS Data Instrument: the Nigerian Example

3.2. Summary of Pilot Survey Returns

3.3. Age Distribution of Deaths by Sources

3.4. Age Distribution of Registered Deaths by Sex of Deceased

3.5. Comparism of Causes and Symptoms Associated with Deaths Registered by RCMS and CRVS

4. Discussion/Conclusions

Declarations

References

Additional Declarations

Status:

Version 1