1

INDEPTH Workshop Report

Report prepared by Ms. Samuelina Arthur (INDEPTH), Ernest Obed Nettey (Kintampo HDSS) and

Martin Bangha (INDEPTH)

Workshop Title Page: MADIMAH 3 AND MORTALITY DATA ANALYSIS WORKSHOP:

Data Preparation for All-Cause and Cause-Specific Mortality Analysis

a) From 30th

May to 6th

June, 2012

b) At Miklin Hotel, East Legon, Accra, Ghana

c) Hosted by INDEPTH Secretariat

d) Under the leadership of Philippe Bocquier (Université Catholique de

Louvain), Peter Byass (Umea University), Samuel Clark (University

of Washington ), Steve Tollman (Wits University)

e) Funded by INDEPTH (with support from IDRC Asia)

Names of INDEPTH

Member HDSSs that

participated

The first part which ran from 30th June to 2 July 2012 brought together participants

who attended the MADIMAH 2 workshop that was held in Maputo, Mozambique

28-30 Oct 2011 as well as participants from the Burkina specific data preparation

workshop held in April 2012. A total of about 32 participants from 15 HDSS

attended this first part. These were joined on 2 July by the participants from the

member HDSSs that had not participated in the previous data preparation workshops

(MADIMAH 1 & 2) that were held in Accra and Maputo. Taken together, well over

80 participants including representative from 31 member HDSSs in 14 countries

attended the two workshops. The INDEPTH secretariat was also represented along

with partners from several other collaborating institutions. The list of participating

HDSSs and collaborating institutions is provided below. For detailed list of

participants see Appendix A

HDSS/Collaborating

Institution

Country Number of

participants

Agincourt South Africa 5

Africa Centre (ACDIS) South Africa 1

Bandim Guinea-Bissau 1

Bandafassi Senegal 1

Ballabgarh India 2

Butajira Ethiopia 2

ChiliLab Vietnam 2

DodaLab Vietnam 1

FilaBavi Vietnam 2

Dikgale South Africa 3

Dodowa Ghana 3

Farafenni Gambia 1

Ifakara Tanzania 1

Iganga/Mayuge Uganda 2

Kaya Burkina Faso 2

Kilifi Kenya 2

Kintampo Ghana 4

2

Kisumu Kenya 2

Manhica Mozambique 2

Magu Tanzania 2

Mlomp/Niakhar Senegal 1

Nanoro Burkina Faso 2

Nairobi Kenya 2

Navrongo Ghana 4

Nouna Burkina Faso 2

Ouagadougou Burkina Faso 2

Rakai Uganda 2

Rufiji Tanzania 1

Taabo Cote D‟Ivoire 2

Sapone Burkina Faso 2

Vadu India 3

INDEPTH Secretariat Ghana 4

31 INDEPTH member

HDSSs

14 Countries 69 Participants

Names of INDEPTH

Partners that participated

Collaborating Institution Country Number of

participants

Ghana Health Services (GHS) Ghana 1

Ghana Statistical Services

(GSS)

Ghana 1

Ministry of Health (MOH) Ghana 1

School of Public Health (UG) Ghana 1

Université Catholique de

Louvain

Belgium 1

Umea University Sweden 1

Swiss TPH Switzerland 1

University of Washington USA 1

The two workshops were facilitated by experts from different institutions namely:

PHILIPPE BOCQUIER, Université Catholique de Louvain, Belgium

PETER BYASS, Umea University, Sweden

SAM CLARK, University of Washington, Seatle, USA

STEVE TOLLMAN, Agincourt/Wits University, South Africa

With the assistance of the following: Afolabi Sulaiman and Cho Kabudula

(Agincourt); Kazienga Adama and Karim Derra (Nanoro); Ovalho Augusto

(Manhica); Ernest Nettey and Charles Zandoh (Kintampo), Frank Odhiambo

and Amek Nyaguara (Kisumu); Patricia Elung\„ata and Marilyn Wamukoya

(Nairobi); Martin Bangha (INDEPTH Secretariat)

Background a) The rationale/justification for the two stage workshop

The first NDEPTH mortality data analysis was held in July 2000 at the Hotel

Splendide, Ouagadougou hosted by Nouna HDSS. At that meeting, 18 HDSSs

presented their data. Two years later, the first INDEPTH book title Population and

3

Health in Developing Countries was published out of this meeting. Since then,

INDEPTH members HDSSs have increased substantially with more data

accumulated over the years. However, attempts to update this first INDEPTH book

alongside other efforts to publish a comprehensive compendium on fertility and/or

migration using data from most member centers have been unsuccessful.

The reasons were to be found in the need to ensured data quality as well as

harmonization of the various datasets for comparability. Indeed, efforts have been

invested in data quality by different working groups (WG) with considerable success

(e.g. Migration and Urbanization WG and the Mortality Analysis WG). All these

efforts have underscored data management issues and differences in formats as

major obstacle to all cross-HDSS analysis. These led to the recognition that event

history analysis format was the best way to go. In 2011, the Migration and

Urbanization Working Group (MUWG) was supported to launch the 2nd

phase of

work under the theme: “Multi-centre Analysis of Dynamics in Migration and Health

(MADIMAH). The official launch of MADIMAH was marked by the holding of its

first data preparation workshop in Accra in April 2011, during which event-history

files were created for the 15 or so participating HDSSs. INDEPTH later supported a

second data preparation workshop for the same group in Maputo during which

training in event history analyses (micro-data longitudinal analysis) for migration,

mortality and fertility studies was undertaken.

Following the MADIMAH 2 workshop, the facilitator approached INDEPTH for a

3rd

workshop on grounds that this was necessary to come up with the full event

history that will enable the computation of migration, mortality and fertility

indicators before they can proceed with analysis. In March 2012, during the

INDEPTHStats data quality meeting, it was unanimously agreed that the

comprehensive approach of ensuring the harmonization and standardization of data

for cross-HDSSs analysis is best by the event history format. This recommendation

was adopted by the Board.

In line with the foregoing a mortality analysis workshop was scheduled to bring all

the member centres to speed on the data preparation and management format. As

such, this workshop target all centres (all leaders were expected to attend) so as to

discussed these and other strategic decision. Since the previous MADIMAH (1&2)

workshop had targeted only a few centres, it was agreed that MADIMAH 3 should

be held as a training of trainers‟ workshop just before the main mortality analysis

workshop. The MADIMAH 3 workshop targeted previous workshop participants.

b) Workshop announcement

The workshop was announced through the regular INDEPTH information channels:

via the INDEPTH Network web site and email to all the HDSS member centre

leaders (see Appendix b for the detailed announcement).

c) Participants’ selection and application process

Participants for the MADIMAH 3 (training of trainers‟) workshop were mainly

members of the Migration and Urbanization WG who had participated at the two

4

preceding workshops (MADIMAH 1 & 2) together with the Burkina participants

who had benefited from a data preparation workshop in April 2012. For the

Mortality analysis workshop, all member HDSSs (centre leaders) were requested to

nominate the appropriate participants who registered online. The centre leaders were

also requested to attend since strategic decisions had to be taken on the mortality

datasets.

Objectives MADIMAH 3 Data Preparation Workshop

The Objectives of the MADIMAH

To finalise consistency checks and complete data preparation as well as

introduce participants to time merging procedure

To include fertility variables (in particular delivery events) and observation dates

in the event history data preparation and to generate fertility outputs in Events

History analysis

To train facilitators (trainers) for the general mortality analysis workshop

MORTALITY Analysis Workshop

As indicated in the foregoing background, following the recommendations of the

INDEPTHStats data quality meeting in Accra March 2012, there were strategic

deliberations at the levels of the Secretariat, Board and SAC. These deliberations

reached a unanimous conclusion that the event history format is the best way to go

for all INDEPTH member HDSSs. Hence the decision to bring the data analysts and

centre leaders from all member HDSSs to Accra for the crucial mortality analysis

workshop. The overall aim of mortality workshop was the harmonization of

datasets by jointly adopting the event history analysis format for INDEPTH

members‟ data on mortality, migration, fertility and cause of death data as well as

agreeing on datasets specifications for online repository (iSHARE), indicators for

INDEPTHStats (display) and more importantly the strategy to increase the scientific

productivity and relevance of the INDEPTH Network ( detailed analysis plan for

cross-centre, cross-national and cross-continental scientific papers).

Specific objectives of the workshop include:

To understand and agree on strategic issues (regarding iSHARE,

INDEPTHStats, INDEPTH Data Specification, INDEPTH data access

and sharing policy, Annual core minimum datasets, confidentiality

issues, etc)

To understand, agree and finalize the events history format for

longitudinal HDSS data, i.e. mortality, fertility and cause of death.

Bring the rest of the member HDSSs to speed with the data preparation

(MADIMAH 1-3 procedures)

To work on cause of death, all-cause and cause-specific mortality

To agree on analysis plan for papers in scientific journals on cross-

centre, cross-national and cross-continental all-cause and cause-specific

mortality, levels, patterns and trends.

5

Activities carried out: MADIMAH 3 Workshop

Saturday 30th

June 2012

In the opening remarks by the Executive Director of INDEPTH, Dr. Osman

Sankoh stressed the workshop objectives: that MADIMAH 3 is the last of the

series of data preparation and also meant as a trainer of tutors. The latter will

have to share the knowledge with the wider INDEPTH scientific family during

the mortality analysis session. He expressed gratitude to Philippe for his

availability and willingness to share the knowledge and to those making

themselves available to be tutors. He also expressed pleasure at the presence of

two female scientists in the room from Nairobi HDSS.

Finalization of mortality analysis and preparation of fertility data

To kick off the workshop, Prof. Philippe Bocquier (Universite Catholique de

Louvain) gave some background information about the MADIMAH project. He

stressed on the need for harmonization of data from each HDSS to come out

with a common data structure so that all member HDSSs can produce

comparable data using the same data structure and analytical technique. He

stressed on the idea of analysis-oriented data management which has become a

standard for INDEPTH members expected to adhere to the core minimum data

requirements. This approach in the long run will benefit all member HDSSs to

be able to produce the minimum data requirement. He also acknowledged the

fact that the participants were not all at the same level since not all participants

were part of MADIMAH 1 and 2. In particular, he made allusion of HDSSs from

Burkina who were not part of the initial group but had benefitted from a country-

specific data preparation workshop in April and are now catching up.

Once we have a standardized data across HDSSs, then comparative analysis can

be done as well as data can be shared easily on the iSHARE platform.

Checking of various consistencies and acceptable level of error: Programs on

preparing the data for the workshop were sent to participants ahead of the

workshop. Phillippe pointed out that it is very important to run the consistencies

matrices before analyzing the data. The recommendation for observations with

errors is to be dropped during analysis if they are not much, after which these

errors can be resolved. However, it is important to try as much as possible to

identify and resolved such inconsistencies. Discussions and recommendations

from previous workshops indicate an error level of 1% may be acceptable.

The participants were taken through the event matrix table with emphasis on

some of the important errors that are generated by the matrix. Participants were

made to understand that after identifying and solving the errors in the data, there

is the need to run the matrix again to see the number of errors that still remain in

the data. The final solution in the event matrix is to drop records that still have

problems. This is not recommended if the numbers are still high, however effort

should be made to resolve them. Deleted records should be coded in a way that

the time they contributed is not discarded even though there may be problem of

how they entered or exited. If this is not done, a bias may be introduced into the

analysis.

The most important problem is where a person enters twice at different dates.

This affects both the numerator and the denominator. (Double events are

problems because they can bias the estimates: indeed reconciliation are the most

6

difficult cases to resolve). After checking of data, various demographic rates are

now computed which should at least follow the gold standard pattern

(plausibility). The matrix of errors is very effective in identifying problems in

the sequencing of events. It also serves as a monitoring system. It provides an

indication of the quality of data and the areas where more efforts are needed to

improve data quality.

In sum day 1 consisted of going through the mortality and migration (in and out)

output as well as the INDEPTH core minimum data specification, identifying

inconsistencies and how to define/resolve them.

Sunday 1st July 2012

Data preparation for fertility analysis

Taking into account fertility: Unlike migration and mortality, fertility is not

related to residential status of the individual, but fertility is very crucial for

knowledge on the growth of the population. Besides, it can be a consequence of

migration and mortality. While not directly linked to residency, fertility like

migration (possibility of several residency episodes) is a renewable event

(possibility of several delivery episodes per individual). The preceding data

preparation workshops (MADIMAH 1 & 2) did not have time to tackle fertility.

Consequently, day 2 of the workshop focused on how to include into the EHA

data events that are not directly linked to residency.

The main event considered here being the delivery (DLV). The group was taken

through steps required to bring data structure and variables to proposed

INDEPTH data structure and variable naming conventions. Again, participants

were taken through the TMERGE routines which are a user generated ado

program for STATA. Here, it is assumed that all inconsistencies have been

checked regarding the ordering of events in the residency file before merging

with the delivery file. Merging the individual or residency file and the DLV file

is not easy since the latter is not based on the individual Ids. Hence a special

kind of merge is required that matches according to time.

Harmonization was not also considered during the previous data preparation

workshops (MADIMAH 1 & 2) since the aim at first was just to ensure that all

the centers have consistent files. However, henceforth, there is need to

harmonize all datasets to ensure that all the members are speaking the same

language and that data files/variables names are same across different datasets.

Using exactly the same datafiles that they had used to produce mortality and in-

/out-migration rates, the participants were requested to harmonize all the

variables according to the INDEPTH data specification standards.

Monday 2nd

July 2012

Work on data preparation for fertility analysis continued. In particular, the focus

was on how to handle the likely case where the delivery events occur at the same

time or date as the change of residency status. The logic here is that, in such

cases where two different events occur exactly the same day for the same

person, there is often a likelihood that one event will disappear from the file in

favour of the first recorded or given priority in the analysis. The strategy adopted

in order to capture all possible events as close to the date/time of occurrence as

possible was to record such events with a fraction of time difference in timing

7

(event time).

In sum, participants were taken through the updated procedure for merging

residency episodes and delivery events by time and generated preliminary

fertility indicators. At the end, most participants were able to produce the first

smoothed curve/graph of ASFR directly in STATA and others could generate

the table of ASFRs from their data.

The group adjourned after lunch break to reconvene the next day for Mortality

analysis. Before the adjournment, facilitators for the mortality workshop were

identified. Participants were actually requested to volunteer and we had in all 11

who volunteered to assist as facilitators during the mortality analysis.

MORTALITY ANALYSIS WORKSHOP

With the arrival of participants from the remaining HDSSs, the mortality workshop

picked up the next day. As noted, it aimed at the creation of event history analysis

format for INDEPTH‟s longitudinal data on mortality, fertility, migration and cause

of death, agreeing on data specifications for iSHARE (online repository for HDSS

data) and INDEPTHStats (displaying HDSS indicators). Another major goal was to

have a detailed analysis plan for cross-centre, cross-national and cross-continental

scientific papers to increase the Network‟s productivity.

Day 1: July 3rd

Tuesday

Prof. Marcel Tanner, Chair of the INDEPTH Board, presided over the opening

session. In a welcome address to the participants the Executive Director of

INDEPTH, Dr. Osman Sankoh, reiterated the main goal of the workshop which, he

said, was to increase the scientific productivity and relevance of INDEPTH. Among

the expected outputs he mentioned include: agreed data specifications; a detailed

analysis plan; a plan for the launch of INDEPTHStats; clarity on iSHARE; and an

understanding of the Secretariat‟s mandate to expedite INDEPTH publications. He

also mentioned that the activities of the workshop would also facilitate the revision

of the first INDEPTH publication entitled: “Population and Health in Developing

Countries” which was published exactly ten years ago.

In his opening address, the INDEPTH Board Chair, Prof. Marcel Tanner urged the

participants to raise the scientific productivity of the Network and provide the global

research society with credible scientific information. “Do not compromise on

science, the Network is about science,” “this is a low level meeting with high

content”, he said. He further encouraged the participants and INDEPTH in particular

to steer away from the usual talk of “high-level meetings with low content”. That

INDEPTH and its members centre should adopt new approaches to make their work

more efficient. In particular he stressed on data quality. In essence, we need to

ensure that what we share is of good quality, which makes impact. What the

INDEPTH needs to do is to ensure good data for the network, and aim at lots of

impact/input for the benefit of the people we are working. There is the need for the

update of cause specific determination of mortality. Prof Marcel Tanner ended with

the assuring participant to focus on their work seriously, no compromise on science,

as good science equal good products and good impact. With this we should achieved

a status for INDEPTH „Too dangerous to be ignored” by other institutions and

organizations.

8

Following the opening session, were a series of presentations or updates on various

activities that are critical to the data harmonization and mortality data analysis.

In one of the presentations, Dr. Martin Bangha, the Capacity Strengthening and

Training Manager of INDEPTH, presented the INDEPTH Data Access &

Sharing Policy (iDASP). In particular, he recounted the process involved in the

development, revision and adoption of the policy. He highlighted the levels of

access and data processes distinguished in the policy and acknowledge the

committee that led the development of the policy that is now published and

available on the INDEPTH website.

Dr. Abraham Kobus provided an update on iShare (the online INDEPTH data

repository); the Wellcome Trust funded initiative aimed to assist INDEPTH

scientific working groups and the network scientific activities to develop

documents, extract, harmonise and maintain quality assurance of the analytical

datasets required from centres operational databases. The data repository is

important because it license access to INDEPTH data. More importantly will be

a place for the working group to share data with each other.

Prof. Philippe Bocquier (a member INDEPTH SAC) presented the updated

INDEPTH data specification and minimum core datasets with highlights on

issues that came up during the MADIMAH 3 workshop.

Mr. Somnath Sambhudas presented the INDEPTHStats (online display of

indicators from HDSSs) and Mr. Eliezer Odai-Lartey presented the electronic

VA data capture being developed to help speed up the diagnosis of completed

VAs by physicians.

Dr. Martin Bangha made an update on current status of the data plausibility

checks on mortality and fertility that have been conducted on the datasets

submitted to the INDEPTH Secretariat by the member HDSSs. In effect,

member centres submitted disaggregated HDSS datasets to the Secretariat and a

team of demographers/scientists had been applying plausibility checks based on

life table values and graphs. Reports were written on the checks and forwarded

to the member HDSSs. The update show that there are currently 27 mortality

dataset confirmed as plausible while there are 26 fertility datasets confirmed as

plausible.

After the lunch break, work continued in plenary with Philippe Bocquier

presenting on analysis-oriented longitudinal data management. He went through

a few myths concerning event history analysis (e.g. used for survey data or right-

censored data, etc) and the various terminologies used (e.g. Survival Analysis,

Panel analysis etc). Issues raised and discussed include:

o How to separate and take into account events that occurred on same

day/time to same individual (e.g. multiple DLV)? Concern was expressed

about the potential of artificially separating events that happened on the

same day by adding a day to one of the dates and the fact that this option

could cause dates to be extended beyond censor dates. In response, this

problem could be solved by adding a percentage of a day. Going forward,

this could be decided on by INDEPTH to develop a standard for all.

o How does EHA account for enumeration polygamous men who spend a lot

of time within surveillance area but move often from compound to

compound? How does EHA account for their short multiple residential stays

9

in compound?

o How are the HDSSs working to identify real time emergencies in

populations? EHA, particularly if data was captured electronically should be

able to move toward data outputs in real time. It means the HDSSs to be

relevant at national level should become more a surveillance and response

platform than just a monitoring and evaluation platform.

o Currently, HDSSs does not update vagrant homeless populations because of

residency definitions.

Day 2: July 4th

Wednesday

The first day plenary updates and lectures on EHA gave way to the second day

organized in break out sessions of 3 working groups with the facilitators from the

MADIMAH 3 helping their other colleagues through the practical exercises of data

preparation for events history format: Converting episode files into Event History

Analysis files and checking the structure of these files.

Each group consisted of the MADIMAH facilitators and non-MADIMAH HDSSs.

The Non-MADIMAH HDSSs were guided through to construct the consistency

matrices and reported on percentage of inconsistencies. This exercise on how to

create the consistency matrix was facilitated by Prof Philippe Bocquier.

Towards the close of day 2, there was a scheduled presentation in plenary by Prof

Peter Byass who Chairs the INDEPTH SAC. He introduced the group to issues

concerning Cause of Death determination as a context to a later presentation on the

use of InterVA (a probabilistic approach to analysing cause of death). He took the

participant through the WHO/HMN/INDEPTH/UQ short standard verbal autopsy

(VA) tool, which was about to be released.

Finally to wrap up for the day, the Board Chair (Prof Marcel Tanner who was taking

off that evening) was given the floor to share his impressions with the participants.

Expressing his happiness at working with the various centres and appreciation of the

ongoing workshop, he said he has learnt something. He again stressed that we must

all aim at growing INDEPTH to the level of “Too dangerous to be ignored”. In his

comments, he suggested that this could be well achieved if we focus on the

following interrelated points for the years to come:

Build on assets: As he sees it, the best is yet to come. A lot of the basic work has

been done at the sites as well as at the secretariat. Bringing VA and iSHARE2 is

very important. We have to adapt to the technical possibility and realities of

trying to study population dynamics in LMICs.

Data quality: To influence policy and make an impact, we must have better data

The VA story in particular: We have generated a lot of wonderful

data/information. However we do not want to continue with the long delays.

InterVA provides the platform for us to be able to compare/make comparative

cross-site analysis. If we want to be listened to, contribute country-specific data,

and to be recognized in the global arena, we need the synthetic comparative

analysis. We do not want to make wide comparisons rather, there is need to

adapt to situations and move on.

In conclusion, he expressed the hope that by the time INDEPTH celebrates the

next 10 years, it should have achieved this comparative synthetic analysis,

whereby member HDSSs should not only report deaths but have moved beyond

10

to explaining why they die of as well as explaining the difference from one area

to another. In this way, INDEPTH would acquire the desirable status of “Too

dangerous to be ignored”. The key remains scientific performance with no

compromise on science. While thanking all the participants for the commitment

and efforts he urge them on with the words: “iSHARE 3 Joys: Joy to discover;

Joy to share and comparing more; Joy to translate to action!”

Day 3: July 5thth

Thursday

The computation for residency status and censoring variables ended in a discussion

led by Kobus Herbst with respect to how the date variables was to be formatted

(handling simultaneous events). The suggestion was that dates should be formatted

by second so that simultaneous events can be separated by a few hours so that dates

of events are not unduly changed. By implication, quite substantial changes had to

be made to previous do-files so as to reformat dates to the lowest denominator.

In a plenary that wrapped up activities for day 3, Peter Byass introduced the group to

the use of InterVA-4 package that will be used to analyse cause of death. He spoke

about the evolution of the software through the years, and demonstrated its utility to

generate causes of death in a short time. InterVA-4 is a probabilistic model that

determines cause of the death on the basis of responses to VA questions and finally

determines cause of death with likelihood estimates. He mentioned an electronic

version of InterVA-4, currently in preparation, which will be field tested in August

for data capture and processing cause of death. Peter made clear that it should still

be possible to use longer VA forms and extract the indicators needed for the new

WHO short standard and InterVA-4. He informed participants that WHO, however,

recommends that any research settings seeking more complex forms should include

all the items in the new WHO standard as a minimum core. The main problems

expressed by participants concerned the lack of narrative towards the determination

of cause of death.

Planning Group Sessions:

Simultaneously to the breakout sessions on 4th and 5th

July, a planning group of

SAC, Board, Centres and Secretariat representatives met under the chairmanship of

Prof. Sam Clark (member of INDEPTH SAC) to deliberate on various issues:

including the launch of INDEPTHStats, and propose an INDEPTH analysis plan,

considering potential themes for cross-centre papers.

A. Agreed indicators for display in INDEPTHStats:

Mortality:

1-CDR

2-Child mortality rate (4q1)

3-IMR (1q0), deaths/live births (do both)

4-U5MR (5q0)

5-ASDR

6-Adult mortality (45q15)

7-Life expectancy

8-Neonatal mortality

9- ASCDR (using INDEPTH, Segi, WHO)

Fertility:

11

1-CBR

2-ASFR

3-TFR

4-Mean age of childbearing (maternity, a point measure)

5-Sex ratio at birth

Population

1. Standard Age/Sex groups Pys

2. Standard Age/Sex groups counts of deaths and births

3. Counts of in-/out- migration (standard age / sex)

4. Proportion Under15

5. Proportion 15-59

6. Proportion 60+

7. Overall sex ratio

Migration:

1. Crude In-Migration rate

2. Crude Out-Migration Rate

3. Crude Net-Migration rate (IM-OM)

4. Crude Gross -Migration rate (IM+OM)

5. Age-Specific In-Migration Rates

6. Age-Specific Out-Migration Rates

Cause of Death

Broad categories

B. Publications plans, for the next six months: papers expected on

• Demographic and Health Transitions

• Vaccination and child survival

C. Beyond the six months, papers expected in the following areas:

• Mortality and Fertility

• Migration Dynamics

• Sahelian pattern of childhood mortality

D. With CAUSE OF DEATH information, papers are expected in the

following areas:

• Child mortality

• Adult mortality

• Maternal mortality

• Seasonal mortality

• Aging and NCDs

• Social Determinants of Health (Poverty, Social Inequality, Urban Poverty)

E. Other agreed potential themes for cross-centre papers:

• Newborns

• Health Systems

• Adolescent cohort studies (SRH and Metabolic diseases)

• Epilepsy/VA validation

• HIV/AIDS - ALPHA collaboration

• TB

• Hypertension

12

• Family and household dynamics

• HDSS Methods (Analytical methods, HDSS management methods, VAs, SES,

Well-being, Data management/capture/OpenHDS, Expanding the HDSS

footprint (HDSS light), Measurement, Data linkage, Morbidity and health,

Triangulation of HDSS, DHS and census data)

• Malaria

• INESS

• Indoor Air Pollution

• Capacity strengthening / academic experience

• Social Autopsy

• H3Africa

Day 4: July 6th

Friday

Being the last day, the main highlights was the stock taking on the achievements of

the workshop. Philippe Bocquier facilitated the final stock taking session where the

each HDSS reported on the extent of work accomplished on the creation of events

history data structures. Except for Farafenni HDSS (the Gambia), most HDSSs had

completed the data consistency matrices. As might be expected, MADIMAH sites

had completed most of the required outputs for mortality, out-migration, in-

migration and fertility. Non-MADIMAH sites also completed the mortality and out-

migration outputs (see summary report on the centres in Appendix D).

Workshop Achievements

At the end of the workshop, the accomplishments include but not limited to the

following:

Understanding and agreement on Events History Format for HDSS data for

analysis of mortality, fertility, migration and cause of death

Training of trainers on event history data preparation and consistency checks

Regarding INDEPTHStats, the main indicators were agreed for its launch

Merging and extraction of files and identifications of common errors that occurs

during this process.

Understanding and agreement on INDEPTH core minimum dataset

specifications for iSHARE as well as the way forward for iSHARE

Analysis plan for papers in scientific journals on cross-centre, cross-national and

cross-continental all-cause and cause-specific mortality

Conclusion /

Recommendations

Some of the main Recommendations include but not limited to:

Data currently on iSHARE are not in line with the new INDEPTH data

specification. Suspend all current access to data on the repository. As a transition

to iSHARE 2, consider only centre data that have gone through the plausibility

checks of INDEPTH and that are in line with data specification

Cause of death coding: Use the new WHO/INDEPTH classification and for level

of details, use the chapter headings with selection of important diseases

(Malaria, TB, HIV, Pneumonia, Neonatal)

Cause of death determination: For comparative INDEPTH or cross-centre

studies, InterVA 4 is the method of choice. Other methods like physician coding

can be used for centre-specific analysis/studies

13

INDEPTHStats: Ensure that everything on display is consistent with the data on

iSHARE 2 and to use only datasets that have gone through checks and certified

as plausible

To customize INDEPTHStats, then subject indicators/uploaded information to

internal review (centre leaders) and external review by international experts (UN

group) before making this public. Meanwhile, the list of indicators will be

reviewed every two years

To enhance the scientific productivity of the Network, focus on cross-centre

publications while simultaneously support centre-specific papers.

The workshop also considered the important issue of whether member centres

should acknowledge INDEPTH‟s contribution to their work and agreed that

centres should acknowledge INDEPTH in all their publications, especially those

that make use of the HDSS platform.

By way of conclusion: INDEPTH “Too Dangerous to be Ignored” or “Too

Shiny to be Missed”

The closing remark was made by Prof. Osman Sankoh. He expressed his gratitude to

the board chair, board members, SAC members and all participants who were

attended for their participations, contributions and their support for the Network.

During the stocktaking exercise that was done regarding the quality of the data

achieved by the various HDSSs present, the participants cheered at the revelation

that 50% of the HDSSs had fewer than 0.8% errors, and 90% had fewer than 3.5%

errors even BEFORE corrections! This meant that the Network‟s efforts to support

the member centres over time had paid off. The Board chair, Marcel Tanner, in his

excitement had coined the slogan: "It's too dangerous to ignore INDEPTH." In

response to this achievement, Evasius Bauni of Kilifi HDSS coined his: "INDEPTH

is too shiny to be missed."

Appendices

(See below)

a) List of participants

b) Workshop announcement

c) Workshop programme

d) Summary table on completed event history outputs by HDSS

e) Presentation and other workshop documents (see attached files)

14

Appendix A: List of participants

Name of Participant HDSS Contact

1 Chodziwadziwa Kabudula Agincourt, South Africa cho@agincourt.co.za

2 Sulaimon Afolabi Agincourt, South Africa afolabi@agincourt.co.za

3 Fikre Gashe Butajira, Ethiopia fikreens@yahoo.com

4 Negussie Kabeta Butajira, Ethiopia negdaysun@yahoo.com

5 Solomon Choma Dikgale, South Africa Solomon.Choma@ul.ac.za

6 Nguyen Linh Filabavi, Vietnam nguyenngoc.linh@yahoo.com

7 Obed Nettey Kintampo, Ghana kojoernest@gmail.com

8 Charles zandoh Kintampo, Ghana charles.zandoh@kintampo-hrc.org

9 Frank Odhiambo Kisumu, Kenya fodhiambo@kemricdc.org

10 Nyaguara Amek Kisumu, Kenya namek@kemricdc.org

11 Orvalho Augusto Manhica, Mozambique orvaquim@gmail.com

12 Charfudin Sacoor Manhica, Mozambique charfudin.sacoor@manhica.net

13 Kazienga Adama Nanoro, Burkina Faso kazienga_adama@yahoo.fr

14 Karim Derra Nanoro, Burkina Faso kderra@crun.bf

15 Patricia elung'ata Nairobi, Kenya pelungata@aphrc.org

16 Marilyn Wamukoya Nairobi, Kenya mwamukoya@aphrc.org

17 Niamba Louis Nouna, Burkina Faso niamba_louis@yahoo.fr

18 Ali Sie Nouna, Burkina Faso sieali@yahoo.fr

19 George Wak Navrongo, Ghana gwak@navrongo.mimcom.org

20 Abraham Oduro Navrongo, Ghana aoduro@navrongo.mimcom.org

21 Paul welaga Navrongo, Ghana pwelaga@yahoo.com

22 Cornelius Debpuur Navrongo, Ghana cdebpuur@navrongo.mimcom.org

23 Bruno Lankoande Ouagadougou, Burkina Faso blankoande@issp.bf

24 Roch Millogo Ouagadougou, Burkina Faso rmillogo@issp.bf

25 Lankoande Malik Sapone, Burkina Faso lankoande.cnrfp@fasonet.bf

26 Sombie Seydou Sapone, Burkina Faso sombie.cnrfp@fasonet.bf

27 Bhushan Girase Vadu, India bhushangirase@gmail.com

28 Somnath Sambhudas Vadu, India sanjay.juvekar@gmail.com

29 Philippe Bocquier SAC/UCL Belgium philippe.bocquier@uclouvain.be

30 Stephen Tollman Agincourt, South Africa stephen.tollman@wits.ac.za

31 Kathleen Kahn Board/Agincourt, South Africa Kathleen.Kahn@wits.ac.za

32 Samuel Clark SAC/U. Washington, USA work@samclark.net

33 Seth Owusu-Agyei Kintampo, Ghana seth.owusu-agyei@kintampo-hrc.org

34 Abraham Herbst Africa Centre, South Africa kherbst@africacentre.ac.za

35 Amabelia Rodrigues Bandim, Guinea-Bissau a.rodrigues@bandim.org

36 Ousmane Ndiaye Bandafassi, Senegal ousmane.ndiaye@ird.fr

37 Sanjay Rai Ballabgarh, India drsanjay.aiims@gmail.com

38 Shashi Kant Ballabgarh, India skant76@gmail.com

39 QuyenEN Bui Chililab, Vietnam buituquyen@gmail.com

40 Bich Tran Chililab, Vietnam thb@hsph.edu.vn

41 Marianne Alberts Dikgale, South Africa marianne.alberts@ul.ac.za

42 Timotheus Darikwa Dikgale, South Africa Timotheus.Darikwa@ul.ac.za

43 Tran Thanh Dodalab, Vietnam trankimthanh20034444@yahoo.com

44 Doris Sarpong Dodowa, Ghana doboakye@gmail.com

45 Elizabeth Awini Dodowa, Ghana awini.elizabeth@gmail.com

46 Margaret Gyapong Board/Dodowa, Ghana mgyapong@dhrc-ghs.org;

47 Momodou Jasseh Farafenni, Gambia mjasseh@mrc.gm

48 Phuc Ho Dang Filabavi, Vietnam hodang54@yahoo.com

49 Amri Zomboko Ifakara, Tanzania samri@ihi.or.tz

50 Dorean Nabukalu Iganga/Mayuge, Uganda d_nabukalu@yahoo.com

15

51 Aristide BADO Kaya, Burkina Faso arbado@irss.bf

52 Simon Tiendrebeogo Kaya, Burkina Faso stiendrebeogo@irss.bf

53 Evasius Bauni Kilifi, Kenya EBauni@kilifi.kemri-wellcome.org

54 Carolyne Ndila Kilifi, Kenya cndila@gmail.com

55 Coleman Kishamawe Magu, Tanzania kishamawe@yahoo.com

56 Laetitia Douillot Mlomp, Senegal laetitia.douillot@ird.fr

57 Tom Lutalo Rakai, Uganda tlutalo@rhsp.org

58 Joseph Ssekasanvu Rakai, Uganda joesseka@rhsp.org

59 Sigilbert Mrema Rufiji, Tanzania smrema@ihi.or.tz

60 Siaka KONE Taabo, Côte d'Ivoire Kone_siakao@yahoo.fr

61 Nahoua Baikoro Taabo, Côte d'Ivoire awa_bai@yahoo.fr

62 Peter Byass SAC/Umea University, Sweden peter.byass@epiph.umu.se

63 Sanjay Juvekar Board/Vadu, India Sanjay.juvekar@gmail.com

64 Tola Heng FilaBavi, Vietnam tola_ray@yahoo.com

65 Marcel Tanner Board/Swiss TPH marcel.tanner@unibas.ch

66 Jacques Emina Indep. Consultant, Belgium jacques.emina@gmail.com

67 Peter waiswa Iganga/Mayuge, Uganda pwaiswa2001@yahoo.com

68 Eliezer Odei-Lamptey Kintampo, Ghana Eliezer.Lartey@kintampo-hrc.org;

69 Cynthia Bannerman Ghana Health Services (GHS)

70 Reginald Odai Ministry of Health (MOH)

71 Richmond Ayeteey School of Public Health (SPH)

72 Representative?? Ghana Statistical Services

73 Osman Sankoh INDEPTH Secretariat Osman.sankoh@indepth-network.org

74 Martin Bangha INDEPTH Secretariat Martin.bangha@indepth-network.org

75 Titus Tei INDEPTH Secretariat Titus.tei@indepth-network.org

76 Samuelina Arthur INDEPTH Secretariat Samuelina.arthur@indepth-

network.org

16

Appendix B: Workshop announcement

Related Workshops

Titles: 1) MADIMAH 3

2) Analysis of All-Cause and Cause-Specific Mortality at

INDEPTH member HDSSs Dates: MADIMAH 3

a) Arrival: b) Workshop: c) Departure:

Mortality Analysis d) Arrival: e) Workshop: f) Departure:

29 June 2012 30 June – 2 July 2012 6 July 2011 2 July 2012 3 July – 6 July 2012 6 July 2011

Venue: a) Hotel: b) City/Country:

Same venue for both workshops Miklin Hotel, East Legon, Accra, Ghana Accommodation at: Miklin Hotel and MJ Grand Hotel, East Legon

Facilitators Overall responsibility:

Prof. Peter Byass (INDEPTH SAC Chair)

Prof. Steve Tollman (INDEPTH Principal Scientist)

Prof. Sam Clark (INDEPTH SAC) MADIMAH

Prof. Philippe Bocquier (INDEPTH SAC)

Dr. Mark Collinson (INDEPTH Migration and Urbanisation WG)

A group of new tutors to be trained

Objectives: A. MADIMAH 3 1. Include FERTILITY in the events history preparation 2. Finalise checks and complete data preparation 3. Train tutors for the general mortality analysis group

B. Mortality Analysis 1. Deal with strategic issues (iSHARE, INDEPTHStats, INDEPTH Data

Specification, INDEPTH data access and sharing policy, Annual minimum data, confidentiality issues, etc)

2. Bring the rest of HDSSs to speed with MADIMAH 1-3 3. Finalise Events History Format. Agree final data sets 4. Work on Cause of Death 5. Analysis all-cause and cause-specific mortality 6. Agree cross-site and centre-specific papers for international journals

on mortality levels, patterns and trends

Expected Outcomes: A. MADIMAH 3 1. Mortality datasets prepared for analysis based on events history

format. 2. A group of INDEPTH scientists trained to train others.

B. Mortality Analysis 1. Final datasets for mortality analysis – INDEPTH stamped 2. Agreement on Cause of Death analysis 3. Preliminary analysis results obtained

17

4. Further analysis plan for multisite and centre-specific publications agreed

5. Responsibilities for further analysis and writing agreed 6. Agreement on the launch of INDEPTHStats

Proposed topics for discussion

a) Events History Analysis b) Cause of Death c) Life Tables d) Mortality Levels, Patterns and Trends e) Maternal Mortality f) Infant and Child Mortality g) Adult Mortality h) Old-age Mortality i) Seasonal Trends in Mortality j) Model Life Tables k) INDEPTH Population Standard l) Cross-site and centre-specific publications

Expected Participants: 1. ALL INDEPTH Member HDSSs (Data Analyst plus HDSS leader are expected to attend.

2. Centres with multiple HDSSs may send one analyst for each HDSS)

Deadline for Nomination of Participants to the Secretariat

Reasonable tickets to workshops can be obtained if booking for participants is done early enough. Hence, centre leaders are kindly urged to nominate participants on or before the deadline stated below. (Friday, 25 May 2012)

Hosts:

INDEPTH Secretariat

Requirements: a) Nominations must be made by centre leaders only to ensure that the right people attend the workshop

b) Centre leaders are kindly requested to attend this workshop as strategic decisions on mortality datasets will be taken.

c) All participants are required to bring along laptops to the workshop.

Travel advisory: a) Visas:

b) Overnight stays

a) We will issue out invitation letters to confirmed participants to apply

for visas, where applicable. Please inform the Secretariat if you require any assistance.

b) For overnight stays, please see travel support below.

Travel Support: a) The Secretariat will cover the cost of a return economy class ticket from home country to the city where the workshop will be held.

b) Reasonably priced non-refundable tickets will be purchased. When you receive a booking you must check to ensure that everything is fine. Once confirmed by you, any changes that will lead to extra cost must be covered by the member centre.

c) Travel insurance cannot be covered by INDEPTH d) In-country travel from centres to the airport will be covered. Please

inform the Secretariat of the cost. Evidence of payment for the travel must be produced.

e) Should overnight stay both in-country and while in transit be necessary, please inform the Secretariat before starting the journey so that adequate preparation can be done for reimbursements. Original receipts will be requested by the Secretariat.

f) The Secretariat will usually cover accommodation for the workshop

18

days plus half boarding. A per diem will be paid for dinner and incidentals. All other services at the hotel (mini bar, telephones, etc) must be paid for directly by the participants.

Reimbursements a) Please bring along to the workshop originals of all expenses incurred for your travel to the workshop.

b) To avoid delays, you may scan the receipts and email before starting your journey to the Secretariat.

c) We prefer to transfer funds to institutions’ bank accounts. However, if you have pre-financed the trip privately, we will refund to your personal bank account.

INDEPTH Staff responsible for these workshops

Dr. Martin Bangha (Capacity Strengthening and Training Manager) Email: martin.bangha@indepth-network.org Supported by:

1. Caroline Tekyi-Mensah: caroline.tekyi-mensah@indepth-network.org 2. Peter Asiedu: peter.asiedu@indepth-network.org

Other issues: Participants may be asked kindly by the Secretariat to bring along information materials from their centres. These may include strategic plans, annual reports, publications and brochures and financial reports on sub-grants.

19

Appendix C: Workshop programme

MADIMAH 3 and Mortality Analysis Workshop Miklin Hotel, East Legon: 30th June – 6th July 2012

Date/Time Activities

Fri June 29 Arrivals in Accra

Sat 30 June Finalization of mortality analysis and preparation of fertility data

09:00-10:30 Completing data checks for migration/mortality: conformity to

INDEPTH “Core minimum micro-data set specifications”

10:30-11:00 Health Break

11:00-13:00 Producing life-table indicators for mortality workshop using Excel

13:00-14:00 Lunch

14:00-16:00 Sites updates on their fertility files

16:00-16:00 Health Break

16:30-17:30 Preparing data: adding fertility events to residency file

Sun 01 July Finalization of fertility data and first fertility analysis

09:00-13:00 Organizing data for fertility analysis

11:00-11:30 Health Break

11:30-13:00 Organizing data for fertility analysis

13:00-14:00 Lunch

14:00-16:00 Producing fertility indicators

16:00-16:30 Health Break

16:30-17:30 Producing fertility indicators

Mon 02 July Way forward

09:00-11:00 Reporting on fertility analysis and MADIMAH next steps

11:00-11:30 Health Break

11:30-13:00 Deliberations: identifying trainers and leaders for mortality workshop

13:00-14:00 Lunch

14:00-16:00 Meeting of leaders and trainers for mortality workshop

20

Appendix D: Summary Table on completion of expected outputs by various members HDSSs

HDSS Site

Percent Complete (30 sites)

MADIMAH Group (15)

Mortality Group

(15)

Consistency matrices (order of events)

Matrix before corrections (with % errors) 96.7% 15 14

Initial mean percentage errors 1.8%

Initial median percentage errors 0.8%

Initial 9th decile percentage errors 3.5%

Matrix after desk corrections (with % errors) 93.3% 15 13

Mortality indicators

Mortality nMx table 86.7% 14 12

Mortality lx table 86.7% 14 12

Table of nax 86.7% 14 12

Graph of age-specific mortality rates (log-scale)

90.0% 15 12

Excel Life-Table using Stata output* 30.0% 6 3

Out-Migration indicators

Out-Migration Rates table (5-year age groups) 53.3% 13 3

Graph of out-migration age-specific rates 56.7% 14 3

In-Migration indicators

In-Migration Rates table (5-year age groups) 40.0% 12 0

Graph of in-migration age-specific rates 46.7% 14 0

Fertility indicators

Age-specific fertility rates (ASFR) table 33.3% 10 0

Graph of age-specific fertility rates 33.3% 10 0

Mean age at childbearing 30.0% 9 0

21

Appendix E: List of presentations