This document provides guidance on designing a study to measure disease prevalence. It outlines five key steps: 1) defining study objectives, 2) designating a sampling strategy, 3) preparing data collection tools, 4) data management, and 5) data analysis and reporting. Important considerations for sampling strategy include determining eligibility criteria, constructing a sampling frame, and selecting an appropriate sampling method such as simple random sampling. Sample size is determined based on desired precision, expected prevalence, and confidence level. Statistical analyses are used to evaluate results by testing hypotheses and determining significance based on p-values.

1. Kagira JM
Prevalence & Disease outbreaks
studies

2. DESIGNING A STUDY TO MEASURE PREVALENCE
How can you design a study to measure disease frequency?
FIVE KEY STEPS
Step 1: Define the study objectives
 What are you going to measure? Prevalence? Incidence?
 Unit of interest (or sampling unit) - basic element of the population
that is sampled
 eg., individual animal or a group of animals (such as a pen or herd)
 Case definition – clinical signs or diagnostic test indicating the animal
is disease positive
 Eg: Pig with any nematode egg = case of helminthosis
 Eg: CMT +ve goat = case of mastitis
 Target population – population to which study results will be directly
applicable or extrapolated: eg, Prevalence of mastitis in Juja sub-
county

3. Step 2: Designate the sampling strategy
 What animals will be included in the study?
 Representative sample – a sample of animals that have a
distribution similar to the distribution in the general
population
 Adequate sample size – Exact number required depends
on:
1. Desired precision of the estimate
2. Expected prevalence of infected animals based on prior
knowledge [look for prevalences in similar studies]
 Sampling strategy –detailed description of the selection
process used to obtain a sample from a population

4. Step 3: Prepare the data collection tools and protocol for data
collection
 What information do you need & what tools will you use to collect
it?
 Influenced by objectives, unit of interest & case definition
 Cow with mastitis: a single questionnaire used
 Trypanosomosis case: questionnaire & diagnostic test used
Step 4: Data management
 What system will you use to check and store the data? Egs., Ms
Excel
 Database for data entry, checking and management
Step 5 Data analysis and reporting
 What calculations and analyses are required to obtain the results of
interest and how will these be reported?
 Statistical software (eg., Ms Excel, Statview, SAS) for analyses
 Format for presentation of key results for your target audience e.g.
Dissertation, publication etc

5. Sampling strategy – details/expounded
 Proceeds through series of steps:
i. Define the general population
ii. Select a sampling method
iii. Estimate the required sample size
iv. Nominate the eligibility criteria
v. Construct a sampling frame
 Eligibility criteria - set of criteria that each member of the study population
must meet
 Egs: production type, herd size, management system, age etc
 Sampling frame - list of all units of interest in a general population: needed
for random sampling
 Contains every unit of interest in the general population
 Each unit of interest has unique identification [ eg., all farms with
lactating cows in Juja]

6. SAMPLING METHODS
 Probability (or random)
 Ensure that a representative sample of the general
population is included in the study
 Transparent & unbiased procedure selects units of interest
from the sampling frame
 Each unit of interest in the frame has a known chance of
being selected
 Non-probability (or non-random)
 Researcher determines composition of the study population
 Probability of a unit of interest being selected is not known
 Some groups will be over-represented and others under-
represented eg., piglets vs baconers

7. Probability sampling methods Non‐probability sampling methods
• Simple random • Convenience
• Systematic • Purposive
• Stratified • Haphazard sampling
• Cluster
Probability vs Non-probability
Two principal techniques for random sampling:
• Physical randomisation - process where sampling units are
selected using physical systems that contain random elements
1. Numbered marbles from a bag
2. Use of a die
3. Toss of a coin
• Random numbers - sequence of numbers comprising
individual digits with an equal chance that any number from 0
to 9 will be present
1. Tables of random numbers
2. Computer programs generating random numbers

9.  Simple random sampling: each subject has an equal chance of being
chosen
 Systematic random sampling: selection of sampling units occurs at a
predefined equal interval (sampling interval)
 Used when total number of sampling units is unknown eg., patients attending
emergency unit at hospital, animals slaughtered in Dagoretti SH
 Eg: studying inpatient [dogs] records, requiring 300 records in 12 months
 10 new records daily = 10 × 365 = 3650 records in 1 yr
 Sampling interval (k) = 3650/300 =12
 1 record for every 12 records
 Identify each records with a consecutive number
 Random number from 1 to 12 is taken as starting point, every other 12th record is
sampled eg., 4, 16, 28, 40, 52, and so on…
 Or: sample every 5th animal in a crush
If a sample of five cows was required, five
random numbers between 1 and 10 would be
generated and cows selected on the basis of
the generated random numbers.
Farm level sampling

10. Stratified random sampling
 Sampling frame is divided into groups (strata) & random
selection within each stratum are selected
 Ensures adequate representation of all groups
 Proportional stratified random sampling: number sampled
within each stratum is proportional to the total number within
the stratum
 EG: Prevalence of mastitis in the goats in Thika
 Three production systems: Intensive (70%), extensive(20%), 10% in
town (kiandutu)
 Create a sampling frame for each system (strata). Randomly select
samples from each strata according to the respective %
 Non-random sampling can be done for systems lacking records eg.,
Kiandutu

11.  Determination of average total lactation milk volume (total litres)
produced by dairy cows in Juja Division, Kiambu County
 2 breeds: Aryshire and Friesian
 Stratified random sampling:

12. Cluster sampling
 Sampling frame is divided into logical aggregations (clusters)
and a random selection of clusters is performed
 Individual sampling units (primary sampling units) within the
selected clusters are then examined
 Clustering may occur in space or time
1. Litter of piglets is a cluster formed within a sow,
2. Herd of dairy cows is a cluster within a farm
3. Fleet of matatus is a cluster formed within space (that is,
stage eg., Nairobi Bus station)
 Advantage: Economical
 Disadvantage: standard errors of estimates are often high due
to homogeneity within a cluster

13. Non-probability sampling methods
 Probability of selection of an individual within a population is
not known
 Some groups within the population are over or under-selected
 Include:
1. Convenience sampling: where the most accessible or amenable
sampling units are selected; eg., next to the road
2. Purposive sampling: where the most desired sampling units are
selected; eg., only adults
3. Haphazard sampling: no particular scheme or method of
sampling is used
 Disadvantages:
1. Biased population estimates
2. Extent of that bias cannot be quantified

14. SAMPLE SIZE DETERMINATION
 Number of animals that needs to be included in the study
 Considers:
 Required number appropriate for the study objective
 Practical issues: time, cost, expected level of non-participation or loss
of participants during a longitudinal study

15. Formula for sample size calculation:
n = Z2
α/2 (PQ)
L2
 n = minimum sample size
 Zα = level of confidence required. If 95% confidence required then
Z0.05 = 1.96
 P = Known or expected prevalence based on prior knowledge
 Q = (1-P)
 L = Required level of precision of the prevalence estimate
 Example: Mastitis Project in Thika
 Prevalence (P) of 50% will be considered to maximize the sample size
 Minimum confidence interval of 95%
 Level of precision = 5%
 n = 1.962 (0.5)(1-0.5)
0.052
 n = = 385
 Minimum of 385 lactating goats will be sampled

17. Increase sample size:
After obtaining the sample size; you can
increase sample size
 By 5% if you expect, due to field
conditions:
1. Samples will not be obtained from
some animals
2. Some samples will not be suitable for
processing by the time they reach the
laboratory
 By 40-50% if response rate to your
questionnaire will only be 50-60%

18. EVALUATION OF STATISTICAL RESULTS
 Conclusions of your study are usually based on the results of statistical
analyses
 Inappropriate statistical tests = wrong and misleading conclusions
 Statistical tests used in determining:
 Effective treatments in clinical trials
 Risk factors for disease
Null hypothesis and P value
 Statistical tests are used to investigate the null hypothesis:
 Assumption that no difference exists between groups [eg., prevalence of
nematodes is not different in males vs female goats]
 Statistical tests are used to disapprove this hypothesis
 If a statistical test [eg., t-test, chi-square etc] identifies a difference
between groups disapproving the null hypothesis,
 Difference is reported with a corresponding P value eg., p=0.03
 P<0.05 = significant difference; P>0.05 = no significant difference
 Eg., P=0.97, probability is very high that no difference exists;
 P=0.00002, probability is very high that difference exists between male
and female % +ve of nematodes
READ ONYOUR OWN

19. General principles of hypothesis testing are:
 Formulate a null hypothesis that the effect to be tested
does not exist [Eg., drug A does not have any efficacy
against nematodes]
 Collect data [Expt, Group A [not treated] , B [treated]]
 Calculate the probability (P) of these data occurring if the
null hypothesis were true [Drug not efficacious]
 If P is large, (0.055 …..0.99) the data are consistent with the
null hypothesis
 Conclude that there is no strong evidence that the effect
being tested exists
 If P is small (<0.05), we reject the null hypothesis
 Conclude that there is a statistically significant effect
 There is evidence that the effect being tested exists eg.,
anthelmintic treatment causes significant reduction in egg
counts
READ ONYOUR OWN

20.  Significance level α (alpha): Dividing line between ‘large’ and
‘small’ P values
 Usually α = 0.05
 Significant result = ‘P < 0.05’ ,
 P > 0.05 = not statistically significant
 Eg: Comparison of natural and AI method in conception
 AI: 53 services, 23 resulted in conception
 Natural: 124 services, 71 resulted in conception
 Null hypothesis = conception rates for AI are equal to conception rates
for natural method
 Chi-squared test = compare the two proportions (23/53 and 71/124 ) =
2.86 , df = 1, p value = 0.09
 p>0.05, accept null hypothesis (i.e. no significant difference between the
two methods)
READ ONYOUR OWN
Read notes on Statistics: Chi-square, t-test, ANOVA etc

21.  Statistical test used to evaluate a difference between groups:
READYOUR STATISTICS NOTES

22. DISEASE OUTBREAKS / HERD PROBLEMS
Disease outbreaks
 Outbreak = series of disease events clustered in time
 Questions asked:
1. What is the problem?
2. Can something be done to control it?
3. Can future occurrences be prevented?
 Assumption: disease is not distributed randomly in a
population
 Systematic recording & analysis: pattern, identify the
causes & sources of an epidemic
 Decision: effective control measures

23. PROCEDURE FOR OUTBREAK INVESTIGATIONS
1. Establish or verify a diagnosis
2. Define a case
 Specifies characteristics shared by all affected animals
 Specifies what distinguishes them from non-infected animals
 Ensures that the disease is consistently defined among different
investigation centres and over time
3. Confirm that an outbreak is occurring
 Enhanced surveillance to identify additional cases
i. Heightening awareness to increase passive case reports
ii. Implementing targeted surveillance
 Techniques: contact field practitioners by phone, email
discussion groups (whatApp)
 Large outbreaks: use media releases (print, television, radio)

24. 4. Characterise the outbreak in terms of:
 Time (epidemic curve)
 Identifying the first case (index case) and then graphing subsequent
numbers of cases per day or per week
 Common source epidemic :
 Extremely rapid increase in the number of cases from the index
case eg., food poisoning after attending a wedding
 All the diseased animals were exposed to the source at about the
same time eg., anthrax outbreak in Meru
 Propagated epidemic :
 Number of disease animals is increasing over time
 Typical of contagious disease or prolonged exposure to the agent
via vectors or toxins

25. Point source
Continuous source
Propagated epidemic

26.  Location (spatial distribution):
 Draw a sketch map of the area or the layout of the pens and
the number of cases within pens
 Examine animal movements and recent additions to the
herd or flock
 Look for interrelationships among:
 Cases
 Between location of cases & physical features

27.  Calculate attack rates for animals grouped according to sex,
age, breed
 Number of animals acquiring the disease
Number of animals present at the start of the outbreak
 Collect historical, clinical and productivity data on both cases
and non-cases
5. Analysis of data
 Draw attack rate tables by dividing cohort into exposed and
non-exposed groups
Milk is suspect

28. • Most likely vehicle (HAM): greatest difference in attack rate for
exposed and unexposed individuals.
Eg: salad (91 - 60 = 31%), ham (88 - 15 = 73%)
Eg: Food poisoning data

29. 6. Formulate a working hypothesis :
 Type of epidemic
 Source of epidemic
 Mode of transmission
7. Undertake intensive follow-up investigation:
• Epidemiological analyses (eg., case- control)
• Clinical assessment
• Pathology
• Microbiology
• Toxicology
8. Implement control and prevention measures
9. Report findings with recommendations – to DVS?