RECAP 2023: HOW TO CONDUCT META – ANALYSIS AND SYSTEMATIC REVIEW

Jaferlou R. Mariano

First Year Resident

Department of Family and Community Medicine

Justice Jose Abad Santos General Hospital

May 06, 2023

Observational Studies

Four types of study designs

1. Series of cases
2. Cross-Sectional
3. Case-control
4. Cohort Studies

CROSS – SECTIONAL STUDIES is the most frequently used observational studies. Cross-sectional studies
determine the frequency (or level) of a particular attribute, such as a specific risk factor, disease or any
other health-related event, in a defined population at a particular point in time. It can be done by using
a survey or analysis of an existing data. Cross-sectional studies are typically used to determine the
current status or “situation”.

Purpose:

 Assessing practices, attitudes, knowledge and beliefs

 Basis for designing appropriate public health measures
• Useful for planning health interventions
• Measure the difference in status “before” and “after”
• Establish association between a risk factor and a disease or outcome of disease
• Validation of diagnostic tests - This can be done by comparing the result of an instrument with a
reference standard both done cross-sectionally in the same population at the same time.

Descriptive - provide estimates of prevalence of disease, traits such as smoking behavior, people’s
attitudes, knowledge or health behavior.

Analytical - assess associations between different variables like association between a risk factor and
disease outcome.

Prevalence - the proportion in a population that at the time of the study has the disease, risk factor etc.

Relevance – examples is the Burden of disease for society, Establish utilization of health

Services, Guide allocation of health resources

Advantages

• Faster and are inexpensive

• Conducted either before planning a cohort study or a baseline in a cohort study

Limitations

 Difficult to derive “causal relationships”

 No temporal relation (reverse causality)
 not sufficient to show disease trends
 impossible to apply if the disease or risk factor or both are rare
Steps in Conducting Cross-sectional Study

1. Establish the Purpose or Objective

a. Start a few important questions
i. What need to know?
ii. Why do you need?
iii. What will happen as a result?
iv. Can you get information from existing instead on conducting a survey?
b. What are you aiming?
i. Presence of disease or risk factor
ii. Knowledge, attitudes, behaviors and practices
iii. Perceptions of knowledge, attitudes, behaviors and practices
iv. Association of risk factors to development of disease
v. Validation of test

2. Develop your Objectives

a. The general objective states the general purpose or goal of the study
b. Develop the specific objectives using SMART criteria
i. Specific
ii. Measurable
iii. Attainable
iv. Relevant
v. Time bound

3. Develop a Hypothetical Results (Dummy Tables)

a. Often it is useful to construct a hypothetical report or dummy table before developing
the data collection. It will focus on what you need to know and what data you need to
collect and help you develop your sampling frame, sample size and method of sampling.

4. Define the Population

a. Selection bias is an issue in cross-sectional studies.

5. Develop the Data Collection Method

a. What or who should be the study population
i. Inclusion criteria
ii. Exclusion criteria – those who fit the inclusion but should be excluded because
the data they might contribute will not be accurate
b. When and where should the study population be recruited
c. How should the study population be selected
i. Accuracy is the very close estimate to the TRUE value in a population. It
depends on Representativeness of the sample population and reliability of the
data collected
ii. Precision is the variation that a result would have if it were calculated from
several samples
 1. Confidence interval (CI) – narrow interval more precise. It depends very
much on the size of the sample
d. To avoid these biases, sample size computation and random selection are very important
e. Can use a Sample Size using Epi Info
f. The best approach is to use random sampling.
g. Chance determines who will be included in the sample
i. Sampling frame (complete list of the study population)
ii. Simple random (pull names from a hat)
iii. Systematic random (every 5th name)
iv. Convenience sampling

6. Collect the data

a. Survey questionnaire
i. Ask the same question in the same way to all respondents
ii. Can be face-to-face interview, telephone, email or web-based
iii. Challenge is to avoid information bias
iv. Pilot test your questionnaire
b. Tips in Developing Questionnaire
i. Title the questionnaire
ii. Include simple instructions
iii. Use plain language and be direct
iv. Be brief, focus on “need to know” questions and minimize “nice to know”
information
v. Start with non-threatening questions and most important questions
vi. Make sure answer choices correspond to the questions
vii. Balance the “negative” or “low” answer choices with “positive” or “high”
choices
c. Face-to-face survey data collection yield the best results
i. Certain they are the actual respondent
ii. Explain the survey, the questions and choices
iii. Electronic or web-based surveys are also becoming popular
iv. Response rate of 60% to 80% should be the target

7. Analyze the Data

a. Descriptive statistics counts, percent, means and standard deviation
b. Analysis of association can be done if response rate is high
c. Divide the subjects according to the risk factor and outcome into four distinct groups.
d. Statistical Treatment for 2x2 Tables
i. Chi-square and Fisher exact test
ii. Odds ratio
iii. Pearson or Spearman correlation
iv. Sensitivity, specificity, predictive values and ROC curve
8. Write the Report
a. STROBE guidelines
i. Title – PIO and study design
ii. Abstract – Background, Objectives, Design, Results, Discussion, Conclusion
iii. Introduction
iv. Define disease
v. Prognostic factors that affect the disease outcome
vi. Differences of the local setting from previous study
vii. Pre-specified hypothesis on the prognostic factor in the local setting
viii. Objectives - state general and specific objectives
ix. Methodology
1. Study design
2. Subjects and setting
3. Instruments and Questionnaire
4. Design and development
5. Describe the questionnaire
6. Pilot testing
7. Sample size
8. Statistical Method
x. Results
1. Narrative data
A. Participants
B. Outcome
C. Main results
xi. Discussion
1. Summarize key results in relation to the study objectives
2. Interpretation of results, implication to family and community practice
3. Discuss similarity or difference of the results to other studies and
generalizability of the results to other population group
4. Discuss the limitation and potential bias of the results
5. Conclusion and recommendations
ATTENDANCE: