Upload
vanessa-wright
View
218
Download
1
Embed Size (px)
Citation preview
Using Information for Health Management; Part I- Health Information Systems Strengthening
2
3
Learning objectives
– the information cycle; tools and processes for turning data into action
– the relationship between data use and data quality
– hierarchy of standards / essential data set– common reasons for compromised data quality,
and various counter measures– different information products for
communicating different meanings
Data, information, knowledge
Data – observations and measurements
Information (processed data)– facts extracted from a set of data (interpreted data),
– data brought together to demonstrate facts– Meaningful and useful
Knowledge– contextualized information
– actionable
42
42 / 60 = 70%Fully Immunized Children < 1y
Where 60 is the target population
42 / 60 = 70%Fully Immunized Children < 1yFor month May 2010Organization Unit: Whatever
With additional contextual insight we may act upon this information
Org.Unit Immunization Coverage May 2012
Whatever 70
Notsogood 40
Verybad 15
Superduper 98
Spatial /organizational context
Month 2012 Immunization Coverage for WhateverJan 83Feb 80Mar 70Apr 52May 64Jun 60Jul 54
Aug 43Sep 37Oct 39
Temporal context
Information cycle; from data to action
10
Data collection
11
Input: Using data sources and tools to collect quality data
Common problems:- Too much to collect- Poor understanding of data collection tools- Timeliness of reporting- Low data quality
Output: relevant data
Processing
12
Input: Relevant data
Processes:-Quality checks-Aggregation to relevant levels-Calculation of indicators- Analysis of data => information
Common problems:- Much irrelevant data- Low data quality- Limited knowledge of data needs and analysis
Output: data converted to information
Presentation
13
Input: Analysed information, to be presented to communicate meaning
Common problems:- Importance of information is lost through poor presentation- Data can be misrepresented or misinterpreted
Output: various information products- Charts- Tables- Maps
Action
14
Input: Information products
Process: Interpretation of information, prioritizing action, making action plan, and execute action through budgets and plans with goals and targets
Common problems:-Incomplete information-Lack of good targets-No plan for evaluation
Output: Decision-making for health management
Data collection – at the source of data creation (point of care)
Service data collected by nurses and doctors in-between attending to patients
Usually several (manual) steps before it is in any database/storage– Tally sheets– Tally sheet totals at end of month– Monthly summary forms which are
reported to the next level
Often too much to collect for already overworked staff
15
What data elements should be collected?
Cannot be obtained elsewhere (e.g. survey) Are easy to collect (cost vs usefulness) Do not require much additional work or time Can be collected relatively accurately Is part of one or more indicators
Comparability of collected data
Stable standardised definitions– To ensure spatial comparability between different facilities,
districts, provinces and nations– To ensure comparability over time
Note:Revising poor indicators /data sets /data elements may not be advisable due to cost and loss of backward comparability
18Balancing varying information needs
Essential data sets (EDS)
Hierarchy of standards
EDS Example: vaccination data
Input (community)– Staff attendence, vaccines, to whom, when, where
Process (district)– Vaccination of children
Output (province)– Coverage of child immunization
Outcome (national)– Immunization rates going up
Impact (international)– Healthier children, less disease. Synergies
Where do we get data from?Routine data collection
– Routine health unit and community data• Activity data about patients seen and programmes run,
routine services and epidemiological surveillance• Semi-permanent data about the population served, the facility
itself and staff that run it– Civil registration (vital events being integrated with health e.g.
India)Non-routine data collection
– Surveys– Population census (headcounts proportion/facility catchment’s
area)– Quantitative or qualitative rapid assessment methods
Data Sources in the HMN data warehouse concept
Client record cards
Tally sheetsEasy way of counting identical events that do not
have to be followed-up (e.g. headcounts, children weighed)
25
RegistersRecord data that need follow-up over long periods such as ANC,
immunisation, Family Planning, Tuberculosis (TB)
Reports
weekly,monthly,quarterly
Processing; assuring data quality and calculate indicators
- Turning data into information- How to assess data quality?- What are indicators, and why do we need
them?
28
Why checking data is vital?
• Use of inaccurate data leads to– Wrong priorities (focus on the wrong data)– Wrong decisions (not applying the right actions)– Garbage in = garbage out
• Producing data is expensive– Waste of resources to collect poor data
Routine data should be..
Reliable: Correct, Complete, Consistent
Timely: fixed deadlines for reporting
Actionable: no action = throw data away
Comparable:same numerator and denominator definitions used by all data processers BUT striving for comparability can compromise local relevance
Complete data?
Spatial: submission by all (most) reporting facilities
Temporal: can you do analysis over time?
Does provided services cover the full population? Many indicators depend on population figures as denominators
Correct data?
Are we collecting the data we need?
The data values seems sensible/plausible?
The same definition applied uniformly?
Are there any preferential end digits used?
JAN FEB MARCH APRIL MAY JUNE JULY
250 230 245 225 230 240 250
Consistent data?
Data in the similar range as this time last year or similar to other organization units
No large gaps or missing data
No multiplicity of data (same data from multiple sources –which one to trust?)
Timely data?
Late reports weaken the potential for comparison, action can be too late,but still useful for documenting trends
What are the causes of poor data quality?
- Too many forms to fill out that are not useful to health workers
- Data collection tools are poorly designed and hard to understand
- Too many steps of manual aggregation and transfer of figures
- Limited feedback on data quality to those who collect it
- Data is not used
35
Accuracy enhancing principles Capacity building through training
(90% of HISP activities)
User-friendly collection/collation tools Feedback on data errors (but not only!) Availability of processed data Local Use of information Essential Data Set Clear definitions; standards Information sharing
CHIEFDOM LEAGUE TABLE 2ND QUARTER APRIL – JUNE 2009
20.893.632.4114.34391.4Total
14143.228.3100.038.273.82954.9Bumpeh
12123.738.677.8101.268.02961.1Upper Banta
12123.760.5100.057.453.72671.8Ribbi
884.364.086.689.492.64049.8Kori
884.336.5100.077.693.24580.4Kargboro
884.386.5100.0140.769.73555.6Kamaje
884.332.193.092.4110.33761.4Bagruwa
664.735.6100.0120.8201.64888.3Lower Banta
664.778.2100.046.796.552118.4Kowa
334.833.091.791.7106.846140.3Timidale
334.871.375.093.4162.75590.3Kaiyamba
334.845.9100.086.390.557134.9Dasse
225.048.1100.086.2154.362124.3Fakunya
115.393.386.696.6170.94598.2Kongbora
RankingRankingAverage Score
% Exclusive Breastfeeding
at Penta3
% MMRC Submitted
% 2nd Dose of IPT
% 3rd ANC Visit
% PHU Delivery2nd
Quarter
% FullImmunized 2nd Quarter
Chiefdoms
What can be done to improve data quality?
Systemic changes
– Essential dataset
– Feedback routines
– Information for action
– Promote information use at all levels
Data validation mechanisms
– Min/Max rules in software
– Data validation rules, check for consistency in logic of data
– Completeness and timeliness reports 38
Minimum and Maximum Values
0
500
1000
1500
2000
2500
3000
Jan Feb March April May June July
Num
ber
Minimum and Maximum Values
Maximum
Minimum
Indicators
Calculated by combining two or more pieces of data, so that– They can measure trends over time– They can provide a yardstick whereby facilities / teams
can compare themselves to others (spatial, organizational)
– monitor progress towards defined targets– Good for measuring change
To do this, indicators need to have a numerator and denominator
39
- measure service COVERAGE and QUALITY
40
Indicator types
– Example: How is maternal mortality rate defined?“the number of maternal deaths per 1,000 women of reproductive age
in the population (generally defined as 15–44 years of age).” – What about maternal mortality ratio?“the number of maternal deaths per 100,000 live births in same time
period.”
Numerator: Number of deaths assigned to pregnancy-related causes during a given time interval
Denominator: Number of live births during the same time interval
Multiplier: 100,000
Millenium Development Goals have a set of proposed indicators [weblink]
denominatorindicator =
numeratorX 100 = %
42
Atop the line – numerators(activities / interventions / events / observations / people)
a count of the event being measured
How many occurrences are there:
morbidity (health problem, disease)
mortality (death)
resources (manpower, funds, materials)
Generally raw data (numbers)
43
Under the line – denominators (population at risk)
size of target population at risk of the event
What group do they belong to:
- general population (total, catchment, target)
- gender population (male / female)
- age group population (<1, >18, 15-44)
- cases / events – per (live births, TB case)
An ideal An ideal indicator indicator RAVESRAVES !!! !!!
44
45
Indicators should be
RELIABLE gives the same result if used by different people
APPROPRIATE fits with context, capacity, culture and the required decisions
VALID truly measures what you want to measure
EASY feasible to collect the data
SENSITIVE immediately reflects changes in events being measured
Essential indicators: determines the essential data set at each level
Hierarchy of standards
Indicator Operationalization
Defining the sources of the data – both numerator & denominator (how is it to be collected?)
Determining the frequency of collection and processing of the indicator (How often should it be collected, reported, analyzed?)
Determining appropriate levels of aggregation(To where should it be reported and analyzed?)
Setting levels of thresholds and target
What will be the nature of the action (decision) once the indicator reaches the threshold?