1. Dept. of Control of Neglected Tropical Diseases
Rapid Monitoring of Treatment Coverage:
The Coverage Supervision Tool (CST)
Training material
2. I. Background
II. 9 Steps of the CST
III. Country Experiences
IV. Discussion of ‘best implementation
practices’
Outline of Training Sessions
3. Coverage Surveys
Coverage Evaluations
District-level, statistically rigorous,
Implemented periodically
Coverage
Monitoring
Coverage Surveys
3
Simple, inexpensive, &
rapid for routine use
Coverage
Supervision Tool
(CST)
Feedback from Program Managers:
• Coverage Surveys are time consuming and
expensive → can’t be done everywhere all the time
• Coverage surveys are too late to improve current
round MDA
• Teams found simplicity of LQAS appealing
• Supervisors complained of having no tools for
supervision
4. Coverage Supervision Tool
4
Quick, simple and inexpensive tool for monitoring and
supervising MDA; implemented by district supervisors
Objective
Primary
Uses
• Classifying coverage as likely above/below the
threshold
• Supervising CDDs and sub-district planners/organizers
• Detecting issues with compliance and the drug
distribution
• Identifying sub-districts in need of mop-up activities
5. Coverage Supervision Tool
5
Conducted by district or sub-district level supervisors
Who?
When?
< 2 weeks of the MDA round (to allow time for immediate action
and mop-up if necessary)
Where? Targeted supervision areas (sub-district or smaller)
How?
LQAS; interviewing 20 people selected randomly from within the
Supervision Area
6. Coverage Supervision Tool (CST)
6
As a monitoring and supervisory tool the CST can be used to ensure:
– Villages/communities are not missed
• And to conduct mop-up activities when necessary
– Identify problems with the supply and drug distribution systems
• And to strengthen these systems to improve performance of the next MDA round
– Individual compliance is high
• And where it isn’t, to identify and address reasons for the non-compliance
– CDDs are accurately recording their work
• And when they aren’t to identify and address the reasons for the discrepancy
8. How is CST unique from other
monitoring tools?
CST Rapid Coverage
Monitoring
In-Process
Monitoring
Sampling Random Purposive Purposive
Team Internal ? External
9. Coverage Supervision Tool (CST)
9
Overview:
Step 1: Identify population to survey
Step 2: Identify supervisory areas (SA)
Step 3: Obtain a list of all households using a) registers or b)
household enumeration
Step 4: Randomly select 20 households
Step 5: Selection of Individuals
Step 6: Interview Individuals
Step 7: Interpretation of Results
Step 8: Develop an Action Plan
Step 9: Implement the Action Plan
10. CST: Planning
10
Checklist:
Questionnaire (at least 1 per SA per team)
Random number table (ideally laminated)
Coin (for random selection)
Chalk
Examples of the medication (and any
additional visual aids – e.g., dose pole)
CST 2-pager quick guide
Clipboard (1 per team)
Household enumeration sheets
Notebook for scratch paper
Pencils
Action Plan handout (at least 1 per SA)
11. CST: Planning
11
Team Composition
• CST is an internal monitoring tool
• Designed for district and sub-district supervisor
implementation
• Team composition: CST Planner and >1
Enumerator
12. CST: Planning
12
Cost: $250-$1,000 per supervisory area*
*But if the supervisor does it as part of his/her supervisory
activities in their supervisory area then the only cost is
training
Days: ½ - 2 days per SA
15. Step 1: Identify the Survey Population
15
Survey Population = The population for which an estimate of
preventive chemotherapy coverage is desired
Disease Survey Population
Lymphatic filariasis Everybody living in the survey area
Onchocerciasis Everybody living in the survey area
Schistosomiasis
May vary, based on national treatment priorities and could include:
- School age children(5-14 years)
- High risk adults
Soil-transmitted
helminthiasis (STH)
May vary, based on national treatment priorities and could include:
- Preschool age children (1-4 years)
- School age children (5-14 years)
- Women of child-bearing age
- Everybody living in the survey area at the time of MDA (for LF)
Trachoma Everybody living in the survey area
16. Step 1: Identify the Survey Population
16
Is it possible to use the CST to monitor >1 drug package?
YES, however this is only possible when the drug packages are
being distributed at the same time through an integrated
MDA.
In such instances it will be necessary to clearly define the
survey populations for each drug package.
- e.g., DEC and ALB - everybody
PZQ - children 5-14 years
17. Step 1: Identify the Survey Population
17
Complete Exercise 1
(in the accompanying Participant’s Guide)
19. Step 2: Identify Supervision Areas
19
Supervision Area (SA) = corresponds to the smallest
administrative or geographic unit for which a first-
level supervisor is responsible. This typically the
catchment area of someone who supervises the
community drug distributors.
20. Step 2: Identify Supervision Areas
20
SA1
SA2 SA3
SA4
SA5
SA6
SA7
SA8
SA9
SA11
SA10
District
21. How to determine which SA(s) to pick for the
Coverage Supervision Tool?
1. You suspect MDA coverage was poor
2. Recent migration or expansion make denominator
estimates uncertain
3. You want to supervise the work of the drug distributors
or their direct supervisors
4. Random selection
Step 2: Identify Supervision Areas
22. How to Use a Random Number Table
22
For the CST a random number table may be needed
for the following Steps:
Step 2. Identify the supervision area
Step 4. Randomly select 20 households
Step 5. Select one person to interview
23. How to Use a Random Number Table
23
Random Number Table Instructions:
1. Make sure each item (e.g. , supervision area/household/segment) in your list is
assigned a number. Determine how many digits are needed in your random
number. The total number of digits will be equal to the maximum number items
from which you are selecting. For example if the district has 127 supervision
areas then the random number will need to have 3 digits (1-2-7).
2. Close your eyes and use a pointed object, such as a pen or pencil, to touch the
random numbers in the table. Your starting point is the number closest to where
you touched the random number table.
24. How to Use a Random Number Table
Random Number Table Instructions Continued:
3. Read the number of the digits required from left to right, starting with the number
that is closest to the tip of your pen. Numbers that are larger than the total
number of items (e.g., supervision areas/households/segments) will be discarded
and the process should be repeated until you get a number that is less than or
equal to the total number.
4. If, in selecting a random number from the table, the end of the row is reached
before the desired number of digits is obtained, the selection of remaining digits
should continue with the beginning of the next row.
24
25. How to Use A Random Number Table
Example:
Suppose there are a total of 72 SAs listed in the district,
therefore you need to pick a random number between 1 –
72. Because the total number has 2 digits (7 – 2) you will
need to read 2 digits from the random number table. Close
your eyes and touch the random number table with the tip
of a pen and read the number that is closest. Suppose your
pen lands on the number “8” and the next number to the
right of it is “1”. This means your selected number is 81.
Because the number 81 is > 72, you must continue reading
the table to the right. The next number is 19. Since 19<72,
this number is valid and means that the 19th SA is selected.
25
26. Step 2: Identify Supervision Areas
Large Supervision Areas
The CST becomes impractical and time-consuming in SAs that are
very large. If the selected SA is large (e.g., >5,000 people) it
should be subset into groups of <5,000 people, based on existing
administrative boundaries (e.g., villages, hamlets, blocks, zones,
barrios, etc.), and one or more of these subsets should be
randomly selected for the CST.
27. Step 2: Identify Supervision Areas
Large Supervision Areas – Example 1
Supervision Area Village Estimated Population
SA #1 Gumu 2105
Assosa 3066
Shula 3480
Randomly pick 1 village to target CST
28. Step 2: Identify Supervision Areas
Large Supervision Areas – Example 2
Supervision Area Village / Block Estimated Population
SA #2 Concepción 10,934
Block A.1
Block A.2
Block B.1
Block B.2
Block B.3
Block C.1
Block C.2
Randomly pick 1-2 blocks in which to
target CST
30. OBTAIN A LIST OF ALL HOUSEHOLDS USING A)
REGISTERS OR B) HOUSEHOLD ENUMERATION
STEP 3:
31. Step 3: Obtain a list of all households using a)
registers or b) household enumeration
Keep in Mind:
• Only 20 people selected per survey population per SA
• Want everyone in the survey population to have a similar chance
of being chosen
Selection needs to be random
32. Does an accurate SA register(s) exist?
Step 3: Obtain a list of all households using a)
registers or b) household enumeration
33. How do I know if the register/census accurate?
1. Is the register routinely updated?
2. Are migrant or foreign-born populations that currently
live in the village included in the register?
3. Was register completed independently of a health
campaign?
If the answer to at least two questions is “Yes” then
the register may be accurate
Does an accurate SA register(s) exist?
Step 3: Obtain a list of all households using a)
registers or b) household enumeration
34. Pros of using a register/census:
Can result in large time saving
Simple to select one HH
Cons of using a register/census:
If register/census is
inaccurate the survey results
could be biased
Ultimately, the decision to use an existing
register/census is up to the supervisor conducting the
CST but should be consistent throughout the SA
Should we use registers?
Does an accurate SA register(s) exist?
Step 3: Obtain a list of all households using a)
registers or b) household enumeration
35. No
Scenario A) Registers
1. Obtain all registers in the SA
2. Assign a sequential number
to each HH in the register
Does an accurate SA register(s) exist?
Step 3: Obtain a list of all households using a)
registers or b) household enumeration
Yes
36. Step 3: Obtain a list of all households using
a) registers
37. Step 3: Obtain a list of all households using
a) registers
1. Upon arriving in an SA, request to see the village
register/census
2. Check if each household in the register is assigned a
sequential number
3. If households are not numbered, the CST team should number
each household in the register with pencil
4. If multiple registers are required to cover everyone in the SA,
the households should be numbered sequentially across the
registers, with no skipped numbers and no repeats
38. Step 3: Obtain a list of all households using
a) registers
4. If multiple registers are required to cover everyone in the SA,
the households should be numbered sequentially across the
registers, with no skipped numbers and no repeats
HHs: 1 - 152 HHs: 153 - 383 HHs: 384 - 471
The SA has a total of
471 households
39. No
Scenario A) Registers
1. Obtain all registers in the SA
2. Assign a number to each HH
in the register
Does an accurate SA register(s) exist?
Step 3: Obtain a list of all households using a)
registers or b) household enumeration
Yes
Scenario B) HH Enumeration
Enumerate all households in
the SA
No
40. 40
A Rapid Household Census Approach
1. Each member of the survey team should pair up with a local
volunteer
2. Divide the village into sections so that each pair is assigned
one section of the village
3. Assign each pair a letter code (e.g. “A”, “B”)
4. Each pair numbers ALL the households in their section using
chalk to write the household code on each door
(e.g. ‘A-1”, “A-2”, “A-3”, …”A61”,”A62”)
5. After each team has numbered all houses in their section, the
lists are combined to determine the TOTAL number of houses
in the village.
Step 3: Obtain a list of all households using a)
registers or b) household enumeration
42. Example Continued…
Assign a code to each household in the
segment and write this code on house with
chalk (if acceptable):
“Team code” + Number
Example B -1
B-2
B-3
B-4
.
.
.
B78
47. Step 4: Randomly Select 20 Households
• Pick 20 unique random numbers between:
1 – total #of households in the SA
• The selected numbers correspond to the nth households in the
register or on the cumulative list from the rapid enumeration
47
A brief pause to practice using a Random Number Table
Note: if the SA spans multiple villages it is necessary to enumerate
all households in the entire SA first and then select the 20 random
numbers once the total number of households is known
48. Step 4: Randomly Select 20 Households
Once the 20 households have been selected, the team
should have a local guide take them to each of the selected
households. If there are multiple teams, the 20 households
can be divided between the team members.
48
A brief pause to practice using a Random Number Table
51. Step 5: Selection of Individuals
Upon arrival at the selected household :
i. Introduce team and explain the purpose of visit
ii. List all the individuals living in the household
who are part of the survey population
(regardless of whether they are present at the
time of the visit)
iii. Randomly pick one of these individuals to
interview by drawing slips of paper from a hat or
using a random number table
52. Step 5: Selection of Individuals
M 33yrs F 30yrs F 16yrs F 11yrs M 8yrs F 7yrs
Suppose your survey population is
children 5 – 14 years old
Survey Population
1. F 11yrs
2. M 8yrs
3. F 7yrs
53. Integrated CST
1. Go to the selected HH, list all members in survey population #1
(e.g., all ages), and pick ONE person randomly to interview
2. Next list all people in survey population #2 (5-14 yrs) and pick
ONE to interview
A-3
1. Martin – 65
2. Josefa – 64
3. Juana – 29
4. Maria-13
5. Louis-12
6. Abdel-8
1. Maria-13
2. Louis-12
3. Abdel-8
Survey Pop. #1(everyone ) Survey Pop. #2 (5-14)
Step 5: Selection of Individuals
54. Step 5: Selection of Individuals
M 33yrs F 30yrs F 16yrs
What if there is nobody in Survey
Population (5-14 years)?
Proceed to next house in the
village/register until you find a
household that has >1 child 5-14 years
55. If the selected individual is not present…
Will they return later
the same day?
Return later in the day to
interview the person
Can they be reached
via cell phone?
Can someone else in the HH
respond on their behalf?
Interview the
person via cell
Allow the HH member to
provide a proxy response
Advance to the next numbered
HH as replacement
Yes
Yes
Yes
No
No
No
58. Step 6: Interview Individual
• Use the data collection form to interview the
selected individual(s) to determine whether
or not they were offered and if so,
swallowed the drug(s)
• It is important to bring samples of the drug(s)
to show the interviewee and help aid with
recall
61. CST Decision Rule Table
Step 7: Interpretation of Results
61
Disease
Survey Population
Threshold for
coverage
Decision Rules: Based on the number covered out
of 20 people sampled
Good
Coverage*
Cannot Conclude
Coverage was
Good
Inadequate
Coverage*
Lymphatic Filariasis Everybody 65% >=16 11-15 <=10
Onchocerciasis Everybody 80% >=19 14-18 <=13
STH / Schistosomiasis SAC (5-14yrs) 75% >=18 13-17 <=12
Trachoma Everybody 80% >=19 14-18 <=13
Survey populations for which
coverage thresholds are set
Coverage thresholds
specified by WHO
*Based on alpha = 0.1
The numbers in the columns correspond to the
number of people (out of the 20 interviewed)
answering “yes” to the coverage question
62. Good
(> the threshold)
Cannot conclude
coverage was good
Inadequate
(< the threshold)
Coverage
Conclusion:
Suggested Next
Steps:
Interpretation: It is very likely that the true
coverage in the SA is at or
above the target threshold.
Not enough information to
conclude with statistical
confidence if coverage was
above or below threshold.
It is very likely that the true
coverage in the SA is below
the target threshold.
•Do these results agree with the
reported coverage (e.g., is the
reported coverage also above
the threshold)? If not, try to
identify reasons for the
discrepancy.
•Share these positive findings
with those involved in the
MDA.
•If persons were identified who
should have swallowed the
drugs but did not, review the
reasons why and take
whatever steps are indicated to
raise coverage even higher
than it already is.
•Try to understand reasons why
coverage may have been poor
by looking at the “no”
responses from the CST.
•Investigate why coverage may
have been poor by looking at
the “no” responses from the
CST (e.g., was it insufficient
supply? poor compliance? drug
distributor performance?
insufficient social
mobilization?).
•Does the reported coverage
show that coverage is above the
target threshold? If so, it is
important to identify reasons for
the discrepancy; consider
conducting a data quality self-
assessment.
•Investigate why coverage may
have been poor by looking at
the “no” responses from the CST
(e.g., was it insufficient supply?
poor compliance? drug
distributor performance?
insufficient social mobilization?).
•Is a mop-up campaign needed?
63. This number is compared with the decision rule table
Note that the target
thresholds are based on
the % who swallowed
MDA; this is the number
that should correspond
with the reported coverage
63
64. Discuss the following scenarios as a group. Determine the
appropriate interpretation, according to the decision rule table,
and follow-up recommended
64
Scenario 1: You are using the CST to classify coverage for
lymphatic filariasis, interpret the following results:
Exercise 8
Number who were offered the drug: 17
Number who swallowed the drug: 17
Reported coverage for the implementation unit: 88%
Step 7: Interpretation of Results
65. 65
Scenario 2: You are using the CST to classify coverage for soil-
transmitted helminthiases. Interpret the following results:
Number who were offered the drug: 19
Number who swallowed the drug: 16
Reported coverage for the implementation unit: 70%
Step 7: Interpretation of Results
66. 66
Scenario 3: You are using the CST to classify coverage for
schistosomiasis. Interpret the following results:
Number who were offered the drug: 11
Number who swallowed the drug: 11
Reported coverage for the implementation unit: 90%
Step 7: Interpretation of Results
67. Step 7: Interpretation of Results
67
Scenario 4: You are using the CST to classify coverage for
trachoma. Interpret the following results:
Number who were offered the drug: 16
Number who swallowed the drug: 16
Reported coverage for the implementation unit: 98%
69. Step 8: Develop an Action Plan
• Helps the district and provincial-level supervisors come
up with an actionable plan to improve MDA performance.
• Action Plan must be developed immediately following the
CST to allow time for mop-up activities if necessary.
• Treatment mop-up is indicated in any SA classified as
having ‘inadequate’ coverage
• District supervisor may choose to limit the mop-up to
only those SA(s) where coverage was classified as
‘inadequate’ or he/she may choose to extend the mop-up
activities to other parts of the district.
70. Step 8: Develop an Action Plan
Once the survey results are tallied it is important to complete the CST Action
Plan document to help interpret the results and identify the next steps that
need to be taken to improve the program.
70
71. Potential Actions
Conduct treatment mop-up
Provide refresher trainings to drug distributors and/or
first-level supervisors
Improve community registers
Increase social mobilization efforts or try new strategy
Adapt information/education materials before next
round to target common reasons for non-compliance
Congratulate and/or publicly acknowledge drug
distributors and first-level supervisors doing well
Step 8: Develop an Action Plan
73. Step 9: Implement the Action Plan
• Some actions, like treatment mop-up, will require
immediate mobilization to implement
Improved coverage effective MDA
• Other actions may take place during period between
MDA rounds or immediately proceeding next MDA
• Up to national program how completed Action Plans
are shared
75. Country CST Experience
Country Disease(s) Method Classification Key Results
Ethiopia LF & Oncho Community Good: 4
Indeterminate: 1
More social
mobilization needed;
Incorporated CST into
National Plan 2017
76. Country CST Experience
Country Disease(s) Method Classification Key Results
Ethiopia LF & Oncho Community Good: 4
Indeterminate: 1
More social
mobilization needed;
Incorporated CST into
National Plan 2017
Nigeria LF & Oncho Community Good: 0
Indeterminate: 4
Inadequate: 3
Potential for CST led to
↑ coverage; training
and redelineation of
CDD areas needed
77. Country CST Experience
Country Disease(s) Method Classification Key Results
Ethiopia LF & Oncho Community Good: 4
Indeterminate: 1
More social
mobilization needed;
Incorporated CST into
National Plan 2017
Nigeria LF & Oncho Community Good: 0
Indeterminate: 4
Inadequate: 3
Potential for CST led to
↑ coverage; training
and redelineation of
CDD areas needed
Philippines LF & STH (school
& community
platform)
Community Good: 2
Indeterminate: 3
Inadequate: 1
Compliance issues
identified; CST during
MDA practical;
registers should be
prerequisite
78. Country CST Experience
Country Disease(s) Method Classification Key Results
Ethiopia LF & Oncho Community Good: 4
Indeterminate: 1
More social
mobilization needed;
Incorporated CST into
National Plan 2017
Nigeria LF & Oncho Community Good: 0
Indeterminate: 4
Inadequate: 3
Potential for CST led to
↑ coverage; training
and redelineation of
CDD areas needed
Philippines LF & STH (school
& community
platform)
Community Good: 2
Indeterminate: 3
Inadequate: 1
Compliance issues
identified; CST during
MDA practical;
registers should be
prerequisite
Nigeria STH & SCH Community
vs. school
(School/Comm.)
Good: 6/0
Indeterminate: 9/11
Inadequate: 0/4
Found unregistered
schools; nonattending
children untreated; CST
much faster in schools;
80. Summary of CST Steps
80
Step 1: Identify population to survey
Step 2: Identify supervisory areas (SA)
Step 3: Obtain a list of all households using a) registers or b)
household enumeration
Step 4: Randomly select 20 households
Step 5: Selection of Individuals
Step 6: Interview Individuals
Step 7: Interpretation of Results
Step 8: Develop an Action Plan
Step 9: Implement the Action Plan
81. Limitations of the CST
81
• The CST cannot be used to generate an estimate of coverage,
but rather can only be used to classify coverage as likely
good/poor
• The CST is not an equal probability sample
• The CST has poor power, meaning it will often classify SAs
with coverage that is truly above or below the threshold as
being “consistent with the threshold.”
82. Benefits of the CST
• Inexpensive and can be easily implemented by sub-district
supervisors
• Can be adapted for any of the PC NTDs, including integrated
• Internal self-assessment, not external audit
• Timing of tool provides opportunity for mop-up
• Can identify gaps in social mobilization or drug distribution
82
83. Key Points for Discussion
• What timing is optimal?
– How does this affect ability to do mop-up?
– How does this affect classification accuracy?
• Is there a place for a school-based CST?
• Should proxy responses be allowed?
85. Review Quiz
• What is the name of the tool we have discussed today?
• How soon after MDA should you implement the tool?
• What is a supervision area and how do you choose one?
• Who is included in survey population #1?
• Who is included in survey population #2?
• How many people are sampled per survey population in one
supervision area?
• How many people are sampled per village?
• When the supervision area is small (<5,000 people) how do we
determine which individuals to sample within a village?
85
86. Review Quiz
• When the supervision area is large(>5,000 people) how do we determine
which individuals to sample within a village?
• If an accurate register exists, how do we determine the maximum range for
our random number?
• If there are 377 households in the register how many digits should our
random number have?
• Should we include people who are ineligible for MDA in the LF survey?
• What happens if the selected household is empty?
• What do you do if the household selected only has grown-ups (>15 years)
living there?
• What if the person selected is not present?
86
87. Review Quiz
• What is meant by “target threshold”?
• What do you do after you have collected data on all 20 individuals?
• Why do we ask if the drug was offered and if it was swallowed as
separate questions?
• If 16 people swallowed the drugs for LF (and the target pop. is the entire
population) how should we classify coverage?
• What should you do if the reported coverage is 95% but using the CST
you find only 13 people who report having swallowed the drug?
• What actions might you take if the number of people who swallowed
the drug is classified as “inadequate”?
• What is the last step of the CST?
87
88. Day IV Summary: Technical Orientation on Tools for Monitoring and Evaluation,
World Health Organization, HQ Geneva, from 17 to 21 October 2016
Coverage Supervision Tool (CST)
National NTD
programmes/
Member States
NTD Implementing
partners/
Member States
2017
Country
Case
studies