Addressing data quality The Nigerian Experience

1
Addressing data qualityThe Nigerian Experience

• Louisa Williamson
• DHIS Conference, East London, October 2006

2
The context . . .

• Effective health management relies on information
  that is relevant, timely of good quality.
• Health service providers managers have long
  stated that they cannot access reliable
  information
• The problem is essentially that of poor data
  quality data is generally incomplete and not
  timeously submitted to the state ministry

3
Data Handling characterised by

• vertical programmes . . . NPI, TBL, HAST, DSN,
  FP
• use independent data handling systems to manage
  their information needs . . . usually funded by
  donors
• own defined data sets
• own defined data collection tools
• often, own defined staff logistical
  infrastructure

4
Data Flow current practices across states
NPI
FMOH Level
DSN
TB
HIV/AIDS
SMOH Level
SHMB
Health Programmes - LGA SMOH level
LGA level
NPI
TB
HIV/AIDS
DSN
Hospitals
PHC Facilities
Tons of data little or no sharing limited use
of information
5
Impact . . .

• Poor data quality
• differences in data values submitted across
  forms
• falsification of data
• Duplication of recording effort
• non-submission
• data incomplete
• Fragmented management
• non-sharing of data
• Poor use of human resource
• Inequity in resource allocation, logistics
  management support

6
The approach . . .

• Two key strategies
• strengthening of data handling processes
• building the capacity of technical cadres to
  manage the programme
• Three initiatives were implemented
• Development of systems structures to handle
  data
• Building capacity of staff to handle data use
  information for local management
• Establishment of monitoring systems structures
  to maintain the information system

7
Building streamlined data handling systems
structures
Std. reports
HIS Methodology 5 Rs
Recent feedback
Reliable data
HIS models practices
correct
Reproducible systems
3 Cs of good quality data
complete
Tools
Retrievable Information
consistent

• Relevant data

Indicator based Essential Dataset (EDS)
8
Good data quality helpful hints

• small, essential dataset clear, standardised
  definitions
• good tools facilitate easy collection collation
  of data
• local analysis of data using relevant indicators
• regular feedback on both data information
• discussion of information at facility team
  meetings

9
EDS aligned with core services - streamlined
tools
10
Set target trend lines
11
Forums for regular data review
12
6 Develop staff capacity to handle data
Improved capacity to use software tools
13
Tools to Improve data quality
14
Mentoring, supervision support
15

• Local level data analysis
• data validation
• interrogation of data

16
Data Olympics tool
17
Data handling monitoring tool
18
DHISS Tools to Monitor Data Quality (3 Cs)

• Compulsory pairs - COMPLETENESS
• identifies relationship between 2 data elements
• Compulsory fields COMPLETENESS
• identifies core DEs linked to core services
• Min / Max Ranges CONSISTENCY
• identifies outliers
• Validation rules CORRECTNESS
• describes nature of relationship between 2/gt
  DEs
• Regression Analysis CONSISTENCY
• identifies corrects outliers

19
Tools to monitor Data Quality
10. Regression Analysis
9. Identify data capturer
8. Validation
7. Data report 13 mnths
4. Min / Max Ranges
5. Check box
1. Compulsory pairs
6. Comments box
2. Compulsory fields
3. Data Trend graph
20
Validation rule violation - absolute
21
Immunisation Data what does it tell us about
the NPI service?
22
Maternal health data can it be used?
23
Problem . . . experiential

• Data handling systems structures are in place,
  but there is ongoing poor delivery
• WHY?
• - functionality influenced by people role,
  function, capacity
• WHEN LOOKING AT PEOPLE
• - Identified disconnect between the theory and
  practice of HMIS

24
Lack of clarity . . . HMIS role players

• Producers of information
• facility level clinicians
• clerical / admin staff
• Custodians of data
• data capturers
• information officers
• Users of information
• policy makers - political will
• health managers
• facility level clinicians
• Joe Soap (public)

25
Identification of disconnect between theory
practice
Collection Collation
lack HIS domain knowledge cannot inform
discussion
lack capacity to validate analyse data

• create indicators set targets
• inform std. data element definitions
• streamline data sources tools

Processing
Use of Information
Information

• Regular review of data
• Relate to operational plans
• Monitor service coverage quality

• Data quality checks
• Data validation
• Data analysis

Presentation
No formal mechanism for regular feedback
not part of management team cannot inform
decision making

• Format of tables, graphs reports
• Flow of information
• Feedback mechanisms

26
Research questions

• Background Data handling systems structures in
  place, but ongoing poor delivery why? -
  functionality influenced by people role,
  function, capacity
• How can the disconnect between the theory and
  practice of HMIS be eliminated?
• Aim to explore the context within HMIS which
  role players work
• Develop a framework for identification,
  development support of HMIS role players
• Who are the role players in HMIS?
• What are the roles functions of each role
  player?
• What is the domain knowledge required for each
  role players?
• when, where how should training take place?
• What career pathways exist (can be developed)?