CASE REPORT CASE SERIES CROSS-SECTIONAL ECOLOGIC

Definition Single occurrence/ unique cases Larger collection of case series Measures disease at a single point in time Examines rates of disease in relation to a
Lowest level but first line evidence Both exposure and outcome are factor on a POPULATION LEVEL
ascertained simultaneously
AKA Prevalence Study (Point prevalence /
Period prevalence)
Unit of Study Individuals < 3 Individuals >3 Individuals Population / group

Design Process Describe individuals based on
epidemiology variables person time place
Describe individuals based on
epidemiology variables and ANALYZE
similarities and differences
All persons in the population at the time
of ascertainment are included
Uses Descriptive and Inferential statistics
(point and interval estimation)
Depends of study.
Summary measures are calculated to
characterize the entire population

Advantages Easy and fast Easy and fast Generalizability Economic

Not costly Not costly For Hypothesis Generation Wide Range of exposure levels
Describes rare and unique cases Describes rare and unique cases Ability to examine contextual effects on
health
For Hypothesis Generation
Disadvantages Uncontrolled Uncontrolled Cannot infer temporal sequence bet Data problems
Not generalizable Not generalizable exposure and disease Ecologic Fallacy (individual vs group level
Unknown future outcome Unknown future outcome Not possible to determine incidence association)
Associations cannot be evaluated Associations cannot be evaluated Length-biased sampling Ecologic bias (Aggregation bias,
Selection BIAS Selection BIAS Specification bias)

EXPLORATORY
MULTIGROUP DESIGN
Definition Compare the rate of
disease among many
places during the same
period
Data Spatial autoregressive
Analysis modeling
Spatial autocorrelation
Purpose Search for spatial patterns
that might suggest an
environmental etiology or
more specific etiologic
hypotheses
Examples mapping the age-adjusted
mortality rate due to
tuberculosis across
regions in the Philippines
in 2015
OTHER TYPES:
EXPLORATORY MULTIGROUP
DESIGN (MIGRANT STUDIES)
Comparison of rates between
migrants and their offspring and
residents of their countries of
emigration and immigration
Spatial autoregressive modeling
Spatial autocorrelation
Suggest influence of certain types
of risk factors for disease under
study
Obesity among US immigrants
from Japan have almost similar
rates with Americans but lower
rates among native Japanese –
influence of environmental or
behavioral risk factor
EXPLORATORY MIXED DESIGN
combines the basic features of
the exploratory multiple-group
study and the time-trend design
ECOLOGIC STUDY
ETIOLOGIC MULTIGROUP DESIGN
Assess the ecologic association between
the average exposure level or
prevalence and the rate of disease
among many groups
ordinary least square, linear model, etc
Suggest influence of certain types of risk
factors for disease under study
Studying the association between
background gammaradiation and the
incidence of childhood cancers between
1975 and 1985 in the region surrounding
a nuclear plant
ETIOLOGIC MIXED DESIGN
- assess the association between change
in average exposure level or prevalence
and change in disease rate among many
groups
- Two types of comparisons made
simultaneously: change over time
within groups and differences among
group
EXPLORATORY TIME-TREND
DESIGN
Comparison of the disease rates
over time in one geographically
defined population
Autoregressive Integrated Moving
Average (ARIMA)
Evaluating effect of intervention
in a population, forecasting future
rates and trends
demonstrating the temporal
trend of dengue cases in Manila
for 10 years
ETIOLOGIC TIME-TREND DESIGN
Assess the ecologic association
between change in average
exposure level or prevalence and
change in disease rate in one
geographically defined population
Time-series regression
Suggest influence of certain types
of risk factors for disease under
study
Determine the associations
between average annual
absorbed dose of radiation fallout
from weapons testing and the
incidence rate of childhood
leukemia in three European
countries between 1945 and 1984
 Descriptive Cross-sectional Cohort Case-Control Interventional

Properties Descriptive Analytic Analytic Analytic Analytic

Cross sectional Prospective (Longitudinal) Retrospective Prospective
Observational Observational (Longitudinal) Experimental
Observational
Known N/A Both Exposure and Exposure is known Outcome is known Exposure is manipulated
Outcome are UNKNOWN Outcome is followed up Exposure is asked or
Both are simultaneously confirmed
determined

Definition Study magnitude Includes all persons in the EO EO EO
and distribution of population Archetype of all epidemiologic studies Concerned about
disease with Closed vs fixed cohort BURDEN of disease Internal Validity
respect to person, Can measure incidence (Randomization and Blinding)
place and time External Validity
(Generalizability)
Concerned about
BURDEN of Random sampling vs Purposive sampling
disease

Uses Hypothesis Hypothesis testing Hypothesis testing Hypothesis testing Hypothesis testing
generation To estimate prevalence / Measures association thru relative risk Causal factors
Planning, BURDEN Measures potential impact thru attributable Risk vs protective Study population
monitoring, and Monitor changes factor Mode of transmission Field vs Clinical
eval (successive cross-sectional of disease
Trends studies) Intervention
Determine association Preventive vs Therapeutic
between coexisting
variables Unit of analysis
Individual vs Community

Advantages Resource efficient
Does not suffer attrition (no
dropouts or loss to
followup)
Generalizable
Disadvantage Temporal ambiguity
s Survival bias
Non-response bias
Temporal sequence is clearly established Quick and inexpensive Strong evidence of cause and effect because of
Assessing effects of RARE EXPOSURES Suited for diseases randomization and blinding
Allow examination of multiple effects with long latency
Assessing effects of Greater control to the exposure variable
RARE OUTCOMES
Allow examination of
multiple etiologic
factors
Time consuming and expensive Inefficient for rare Ethical issues
Bias to losses to follow-up exposure Takes more time and resources
Cannot generate Result may not be generalizable to a wider
incidence of disease population if inclusion/exclusion criteria are

Misc Types: Measure of disease
Case report occurrence
Case series Prevalence in popn
Cross-sectional Prevalence in exposed (Pi)
(Prevalence Study) Prevalence in unexposed
Ecologic Studies (Po)
Prevalence Ratio (PR)=
Pi/Po
PR = 1 not associated
PR >1 positively associated
PR <1 negatively associated
Those who are <exposure>
are (PR) times more likely
to have <disease> than
those <no exposure>
If PR <1 get inverse or 1/PR
Those who are <not
exposed> are (1/PR) times
more likely to have
<disease> than those
<exposed>
Prevalence Difference (PD)
= Pi – Po
PD = 0 not associated
PD > 0 positively associated
For every (factor) persons
there are (PD) more
persons among those
<exposed> than those
<unexposed>
PD < 0 negatively
associated
For every (factor) persons
there are (PD) less persons
among those <exposed>
than those <unexposed>
Temporal ambiguity specfiic
Information / Recall
bias
Relative Risk – strength of association of E and Measure of association Measure of disease frequency
O Incidence proportion – followup was complete
Odds Ratio Incidence rate – losses to follow up
Attributable fraction – effect of removing the
exposure
Relative Risk (RR – Risk Ratio Rate Ratio) Odds Ratio Relative Risk (RR)
OR = a*d / b*c RR = Itreatment / Icontrol
IP (CI) = (new cases / popn at risk) * F OR = 0
IR (ID) = (new cases / PT) * F No association %Eff = (Icontrol – ITreatment) / Icontrol
Relative risk = Iexp / Iunexp OR > 1 (exposure is %Eff = 1 – RR
Risk Ratio = CIexp / CIunexp harmful)
Rate Ratio = IDexp / IDunexp There was a (%Eff) relative reduction in
The odds of <disease> among the <treatment> compared
<exposure> among to the <control>
<disease> was (OR)
RD = CIexp - CIunexp times more compared
For every <factor>, there are (RD) more deaths to the odds of
from <disease> among exposed than <exposure> among
unexposed <no disease>
AFexp = (Iexp – Iunexp) / Iexp The estimated risk of
Among the exposed, (AF)% of the outcome are <disease> is (OR) times
due to the exposure more among
<exposure> compared
AFpop = (Ipop – Iunexp) / Ipop to <no exposure>
If the exposure was removed from the
population (AF)% of the disease in the popn OR < 1 (exposure is
will be averted protective)
The odds of <no
Attributable Number = (AF) x cases (exposed exposure> among
or population) <disease> is (1/OR)
times more compared
to the <no disease>
The estimated risk of
<disease> is (1/OR)
times more among the
<no exposure>
compared to the
<exposure>