1. Ear Health Research Program
BIGDATA
Jemima Beissbarth,
Child Health Division,
Menzies, Darwin.
Ear Health Research Program
Data Collection
Good morning, I'm Jemima Beissbarth, I work for Child Health Division within the the Menzies School of Health in Darwin,

2. Acknowledgements Funding for the building of BIGDATA
RAOM 2017 attendance support
I acknowledge the Traditional Owners of the land we are meeting on and pay my respects to their Elders past and present, and extend this respect to Aboriginal people here today.

3. What is the BIGDATA project?
Project designed to bring all of the Ear Health Research Program (EHRP) data into one dataset to improve statistical power to investigate microbiological and clinical outcomes in Aboriginal children with Otitis Media.
So what is BIGDATA?
BIGDATA is a project designed to bring all of the Ear Health Research Program (EHRP) study data into one dataset to improve statistical power to investigate microbiological and clinical outcomes in children with Otitis Media.
This plan was conceived by Professor Amanda Leach.
Last year, as mentioned funding to further this process was received from CRE_ICHEAR.

4. Ear Research @ Menzies 1992 – 2013 EHRP
12 RCTs (3 additional currently underway)
4 cohort studies
8 years - Spn surveillance in Remote Aboriginal Communities (RACs)
5 years - Spn surveillance in Child Care Centres (CCCs)
Otitis media research began at Menzies in 1992 (is that right?) and since then there has been 12 RCTs, 4 cohort studies, 8 years of pneumococcal surveillance in remote indigenous communities and 5 years of surveillance in urban child care centers.
These completed studies are all contributing their data to BIGDATA.
A further 3 RCTs are currently underway, and will be included as they are completed.
What sorts of data are they contributing?

5. Hygiene intervention, mainly non-indigenous children
12 RCTs
Years
Jul
2001
2003
2011
Study Name NH
COMIT1
BR NOSE
CSOM
CHIPS
CSURE
AATAAC
SSSOM
MOPUP
AAAOM
No. children (randomised) 20
125 34 97
456 32
320 92 53
149
No. Nose / Np swabs
182
1031
201
322
4682
661
169
451
Number ear exams
1183
339
1163
4729
127
390
172
100
471
Number of Ear swabs
218 26
493 13 61
145
192
107 15
Sera
Study included:
Ear exam ü
Nose swab
Ear swabs
Other samples collected
Clinic note review
Hospital note review
Antibiotic use review
Pn imm history
Audiology data
Clinical physical exam
(non-ear)
Lifestyle questionnaire
Using the RCTs as an example, This is a summary of the completed RCTs, with the study acronyms across the top, The ticks indicate per study which activities from the list on the left were conducted during the studies, for example we can see that all the RCTs have included ear examinations, and all but one have included Np or nose swabs. We normally collect data on risk factors for OM, vaccine data and recent antibiotic treatment data.
Hygiene intervention, mainly non-indigenous children

6. Hygiene intervention, mainly non-indigenous children
12 RCTs
Years
Jul
2001
2003
2011
Study Name NH
COMIT1
BR NOSE
CSOM
CHIPS
CSURE
AATAAC
SSSOM
MOPUP
AAAOM
No. children (randomised) 20
125 34 97
456 32
320 92 53
149
No. Nose / Np swabs
182
1031
201
322
4682
661
169
451
Number ear exams
1183
339
1163
4729
127
390
172
100
471
Number of Ear swabs
218 26
493 13 61
145
192
107 15
Sera
Study included:
Ear exam ü
Nose swab
Ear swabs
Other samples collected
Clinic note review
Hospital note review
Antibiotic use review
Pn imm history
Audiology data
Clinical physical exam
(non-ear)
Lifestyle questionnaire
Using the RCTs as an example, This is a summary of the completed RCTs, with the study acronyms across the top, The ticks indicate per study which activities from the list on the left were conducted during the studies, for example we can see that all the RCTs have included ear examinations, and all but one have included Np or nose swabs. We normally collect data on risk factors for OM, vaccine data and recent antibiotic treatment data.
Hygiene intervention, mainly non-indigenous children

7. Surveillance
Years
2003
2004
2005
2010
2012
Totals
2001 pre-PCV7
2003 PCV7
2008
2009
2011
2013
Study Name
MARSi CCC
MARSiii CCC
MARSiv CCC
PROMPT
MARSi/ PROMPT
MARSi
MARSii
MARSiii
MARSiv
MARSv
No. of RACs / CCCs 18 26 23 24 33 27 2 10 21 19 12 14
No. children enrolled
303
473
394
462
607
2239
709
644
819 54
174
253
568
276
371
3868
No. of Nose swabs
299
460
387
426
564
2136
664
818 52
156
232
563
262
367
3114
No. ear exams
429
562
991
698
641
169
249
552
269
363
2993
No. ear swabs 9 11 20
199
163 17 44 73 29 42
600
Ear exam ü
Nose swab
Ear swabs
Other samples collected
Clinic note review
Hospital note review
Antibiotic use review
Pn imm history
Audiology data
Clinical physical exam (non-ear)
Lifestyle questionnaire
This is a summary of the surveillance data, The yellow rectangle is the pneumococcal surveillance in child care centres, and the orange rectangle is the remote community surveillance. Here as the previous slide, the ticks represent the activities performed in each study.

8. Summary of totals 8412 No. of child enrolments 14340 No. of ear exams
No. of nose / np swabs
15935
No. of ear swabs
2102
Sera
609
From the RCTs and the surveillance studies we have 8412 child enrolments. Over ear exams where a diagnosis has been made were performed by research staff.
There Is a collection of almost nose or np swabs which have been microbiologically processed , as well as 2000 plus ear swabs, and six hundred and nine paediatric or cord blood sera samples.
This specimen collection is being added to by the ongoing trials.

9. Data matching Ear exams Swab collection / processing
Data collection forms largely unchanged
Descriptions and coding largely unchanged
Swab collection / processing
Swab storage medium unchanged
Core microbiological methods unchanged
When we are thinking in the context of BIGDATA, we need to consider how robust data collected over a long period of time will be, in terms of continuity of collection questions, skills in staff and methodological changes over time.
In areas like the ear examinations and the swab collection and processing, where methods and data recording has largely not changed we find joining the data together feasible and relatively easy.
Also these are the areas where we have the most data, and this may be where we have statistical power to answer questions we may not have been able to answer with individual study results.

10. Further data Immunisation data Risk factor data
Large amount of data collected, easily adaptable.
Risk factor data
Most challenging data to join across studies.
Antibiotic treatment data
Study medicines
Clinical note review data, with co-morbidities etc
Pneumococcal immunisation data has been collected for a variety of studies, and this also easily integrated across studies.
Risk factor data, like smoking, maternal education and household crowding, is one of the areas which is more challenging. The specific questions asked, and the way the questions are phrased has changed slowly over time, as has the design and descriptions of data storage fields and so this data is more problematic to bring together. Also the risk factor data has the most missing data. Obviously staff do not press parents to answer questions they are not comfortable responding to, and some parents opt not to participate in questionnaires at all.
Antibiotic treatment data has been collected both in the context of medicines given to children as part of a study and also as data collected during clinic record reviews and sometimes also hospital record reviews. We have also collected data on co-morbidities and ear health histories in some studies.
This data will be complex to join across some of the older studies. Recently, as a result of the BIGDATA project, some standardisation of data collection has occurred, so integration of new studies should be easier.

11. How many kids? When cross-sectional and RCT data becomes longitudinal
8412 enrolments
Identify individual children
~5600 enrolments with identifiers, 3500 individuals
Identifiers are being sought where we have none
Some of the remote communities participating in the Ear Health Research program have been a part of many trials. Some children within the BIGDATA cohort have participated in multiple studies.
It is important to make sure we know how many individual children we have and that we have a way of identifying when the same child is in more than one study. This is obviously done with a unique identifier if we have one per child, and then using names and birth dates.

12. BIOBANK 15935 No. of nose / np swabs 2102 No. of ear swabs
From these studies we also have a large biobank, as mentioned we have almost16000 nose swabs and over 2000 ear discharge samples. Most of these samples have been cultured as part of the trial.

13. BIOBANK Spn No. of nose / np swabs 15935 No. of ear swabs 2102
No. of nose / np swabs isolates
8733
No. of ear isolates
265
Np and ear discharge sample processing has produced thousands of isolates. We have nearly 9000 Streptococcus pneumoniae isolates, unique per swab and 265 from ear samples. For these Spn isolates we have generally produced disc diffusion data, some etest data and pneumococcal serotypes.

14. BIOBANK Spn NTHi 15935 No. of nose / np swabs 2102 No. of ear swabs
No. of nose / np swabs isolates
8733
No. of ear isolates
265
NTHi
We also have a lot of nontypeable haemophilus influenzae, and other relevant pathogens not shown here. Around these isolates and others we are building our molecular and genomics program and extending our antimicrobial testing.
No. of nose / np swabs isolates
4017
No. of ear isolates
286

15. Beginning with…. 12 RCTs Spn remote surveillance Spn CCC surveillance
Contributing
ear exams
microbiology
risk factor data
Back to the BIGDATA. We begin with the data collected from each of the studies, contributing the information as mentioned, including ear data, micro data, risk factor data. The primary outcome will be to create an annotated, consolidated dataset containing all data collected in ear health studies.  A guide will be written to describe study designs, explain the potential uses and limitations of the data and which data may suit specific research questions, such as clinically relevant questions about the link between pathogen carriage and resistance and ear health or treatment failure.

16. Building to …. 12 RCTs Spn remote surveillance Spn CCC surveillance
Planned analyses
Antibiotic resistance over time
Vaccine selectiveness (Spn, NTHi)
Mathematic modelling and meta analyses
Spn serotypes
OM trends
12 RCTs
Spn remote surveillance
Spn CCC surveillance
Contributing
ear exams
microbiology
risk factor data
From the combined dataset we will report the antibiotic resistance pattern over time, investigate the selection pressure of pneumococcal vaccines,
A consolidated dataset will also facilitate collaborative efforts in related areas, such as data visualisation, and innovative modelling techniques such as trajectory modelling.
Professor Allen Cheng will be overseeing the much of the analysis and performing the mathematical modelling.
We will looks at Spn serotypes including (age, distribution, geographic trends, time trends)
We will analyse trends in each clinical diagnosis of OM, including unilateral and bilateral disease trends,
This will also facilitate future work linking ear health outcomes with other outcome data (eg learning outcomes) to determine the broader impact of poor ear health.
Additional outcomes such as trends in general health will also be available (respiratory health, antibiotic prescribing patterns, skin conditions).

17. Future directions 12 RCTs Spn remote surveillance Spn CCC surveillance
Planned analyses
Antibiotic resistance over time
Vaccine selectiveness (Spn, NTHi)
Mathematic modelling and meta analyses
Spn serotypes
OM trends
12 RCTs
Spn remote surveillance
Spn CCC surveillance
Contributing
ear exams
microbiology
risk factor data
As a part of the planned analyses, we will design a sentinel surveillance model for otitis media and nasopharyngeal carriage. We would like to know the best number and combination of remote communities to best represent ear disease in the Territory, so we can apply for support to do regular surveillance.
As our data includes all clinical presentations and treatments. We will compare health services management of ear disease with that recommended in national guidelines, to inform policy and practice and assist services to meet best practice in management of ear disease and hearing loss.
Future directions
Designing sentinel surveillance
Informing health literacy
Evaluation of programs

18. Building new research ideas
Future directions 2
Planned analyses
Antibiotic resistance over time
Vaccine selectiveness (Spn, NTHi)
Mathematic modelling and meta analyses
Spn serotypes
OM trends
12 RCTs
Spn remote surveillance
Spn CCC surveillance
Contributing
ear exams
microbiology
risk factor data
Building new research ideas
Collect swabs
examine children,
investigate new intervention strategies
qualitative research
These analyses and future works will build new research ideas and the resulting studies will contribute additional data to the collection.
Future directions
Designing sentinel surveillance
Informing health literacy
Evaluation of programs

19. Building new research ideas
External data
Planned analyses
Antibiotic resistance over time
Vaccine selectiveness (Spn, NTHi)
Mathematic modelling and meta analyses
Spn serotypes
OM trends
12 RCTs
Spn remote surveillance
Spn CCC surveillance
Contributing
ear exams
microbiology
risk factor data
Building new research ideas
Collect swabs
examine children,
investigate new intervention strategies
qualitative research
External data to bring in
Long term heath data (surgery, hearing aids)
Educational outcomes
-Standardised testing results
-Behavioural assessments
-Truancy
-Employment potential
Other Gov. admin data
One of our current projects is to source data from the Northern Territory health and education departments, including Naplan results and long term ear health, including audiology and ear surgery data, to investigate more closely the relationship between ear disease and educational and other long term outcomes.
Future directions
Designing sentinel surveillance
Informing health literacy
Evaluation of programs

20. Thank you
Big thank you to the participants, their families and communities, the field teams who collected all the swabs and clinical data, and the lab teams for the processing.

21. MORE data
A randomised controlled trial of pneumococcal conjugate vaccines Synflorix and Prevenar13 in sequence or alone in high-risk Indigenous infants (PREV-IX_COMBO): immunogenicity, carriage and otitis media outcomes.
This will be incorporated into BIGDATA in early 2018.
Ear assessments and swab analyses from over 400 infants, at 5 time points.
We are continuing to collect data. Next year we will be including the Previx combo RCT data in the collection.

22. Ethics
Each individual project will be submitted for Ethics approval prior to commencement.
So far 2 projects have approval, with a third before the committee currently.
Taking a step back, as part of planning all these analyses, they are going before the appropriate ethics bodies, we have two specific projects approved and another before the committee currently.