3. Background and classic LQAS
MC-LQAS
Application to malaria control within the research
activity con
4.
5. LQAS today is a statistical quality control method
Developed in the 1920’s attributable to Dodge and
Romig’s work. Mainly to control quality of
industrially produced goods on the principle that:
◦ Supervisor inspects a lot of goods from a production unit
or assembly line
◦ If number of defective goods exceeds a pre-determined
allowable number, then the lot is rejected; otherwise
classified as acceptable quality
◦ Number of allowable defective goods is based on a
production standard and statistically determined sample
size
6. Transitioned into health systems to assess health
care services, health behaviors and disease
burden.
Production standard is a predetermined population
coverage target set by managers
Lot consists of a supervision area e.g. a
community or health facility catchment
LQAS data collected at multiple time points can be
used to measure the spatial variation or behavior
change
8. Implemented as part of stratified random sampling
design
Uses small samples often 19 per strata or lot
Sample determines whether coverage by a health
intervention reaches a specific target by using a
statistically determined decision rule(DR)
DR is the minimum number of individuals in the
sample that should have received an intervention
9. Classic LQAS uses one decision rule, sample size 19
and 2 threshold values, that define lower and upper
regions.
Each lot is then classified as ‘Acceptable’ or
‘Unacceptable’ against the target.
Since LQAS is based on rigorous random sampling,
results from the catchment area can be aggregated for
provincial or national level coverage.
Statistical underpinning is the operating characteristic
(OC)curve
10. Operating Characteristic Curve for
Probability to Accept
Sample of 19 and Decion Rule of 13
1
0.8
0.6
0.4
0.2
0
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3
Supervision Area Coverage
11.
12. Popular tool
◦ Ease of use
◦ Straight forward implementation
◦ Rapidity of results
◦ Sound statistical underpinning
13. A
Reached
B
Target
C
D
E
Below the Target
F G Or Below Average
Valadez 2011
14. Maintain the program at the
current level
Identify Supervisors and Health Reached
Workers that can help other Health Target
Workers improve their performance
Identify the reasons for
program problems
Below the Target
Develop targeted Or Below Average
solutions
Valadez 2011
15. Control of misclassification (α-alpha error & β-beta
error)
Requirement for finer classification in disease
control and treatment recommendations e.g. WHO
treatment guidelines for schistosomiasis are linked
to three way classification of prevalence of
infection
Inevitable extension to LQAS
16. Focuses into three classification of ‘low’, ‘middle’,
‘high’
Defines two decision rules e.g. ( d and d ) to yield
1 2
least misclassification error for a given sample size
(n)
Probability of correct classification remains high at
upper and lower thresholds
On analysis, classify ‘low’ if the successes x from
total n observations is less than or equal to d1
;classify ‘high’ if x is greater than d2; otherwise,
classify ‘middle’
17.
18. Uses sample size of n=28, decision rules d1=2 and
d2=10
With d2=10, elicits grey region around the upper
threshold of 40% favouring classification of
category 3( high) over category 2 (Middle).
Thus, grey regions ranging from 0.06 to 0.15 and
0.30 to 0.45 respectively. That is a better trade-
off , on divides of producer and consumer risks
19. Sample of 28, if 2 or fewer of these observation are
malaria RDT+, then the area is classified as
category 1, termed ‘low’.
For 10 or more counts malaria RDT+, area is
classified as ‘category 3 termed ’high’
Counts between 3 and 9 classify area as category 2
termed ‘Middle’.
Design gives 80% chance of correctly classifying a
given locale at each of the listed thresholds. A
double sample size of 56 increases the power but
often obtain similar results
20. Malaria prevalence threshold values are set at
PfPR of 10% and 40%.
Locale with below 10% is of low prevalence, 10%-
40% is moderate prevalence, above 40% is high
prevalence
MC-LQAS methodology classifies areas into these
three categories using RDTs for PfPR.
MC-LQAS measures malaria intervention indicators
and classify locale.
MC-LQAS data maps locale malaria prevalence
21. Classifications of ‘low’, ‘middle’ and ‘high’ for link
interventions to the prevalence detected
Category 3(high) is targeted for complete set of
malaria interventions(IPT, ITNs, case management
and IRS)
Category 2 (middle) receive ITNs, IPT and case
management
Category 3 (low) maintain strategies towards
elimination agenda.
The reverse is true for performance indicators
measured in terms of achieving set targets.
22. • Reliable malaria density data is lacking in most
programs at levels where management decisions
are made.
• Research contributes to M&E of the malaria control
program’s impact on the prevalence at sub-district
or lower levels (parish), classifying these areas to
target cost-effective control interventions.
• Test MC-LQAS for malaria control (1st Time Use)
23. Aim : To assess malaria prevalence for priority cost-
effective and targeted interventions
Objectives
1. To classify and map malaria prevalence at the parish
level within the district.
2. To validate the utility of Multiple Classification LQAS
(MC-LQAS) during the survey.
3. To measure malaria control performance indicators
and coverage within the sub- counties and parishes.
4. To disseminate findings as evidence for decisions to
prioritize malaria intervention strategies.
24.
25. Ethical application completed and community
assent sought
Trained research assistants
Data collection through questionnaires and blood
samples for malaria test and Hb estimation
Sampling conducted to identify eligible child of
ages 6months to 9 years.
26.
27.
28.
29.
30.
31.
32. Analyzed 448 cases, 6 months to 9 years
◦ Malaria prevalence
◦ Malaria outcome indicators
Demonstrated high prevalence of malaria &
anemia, low coverage of interventions and their
performance, all with marked variations
Such variations is often masked from aggregate
measures reported in large country surveys
33.
34.
35.
36.
37.
38.
39.
40. MC-LQAS is effective to monitor malaria
endemicity and control interventions providing
reliable data and classification that can aid target
interventions.
It can be replicated
41. Use data generated as baseline and re-define
targets to monitor progress
Draw attention of the malaria situation to the
malaria control program
Replicate the studies
42. CCM management and staff
CCM Executive Director, Filippo Spagnuolo & Head of
Programs, Valeria Pecchioni
Professor Joseph Valadez, LSTM
Dr Olives, University of Washington
Professor Feiko ter Kuile, LSTM
ChildFund International, Uganda
MSH Uganda
Uganda Christian University, Mukono
Family support