Transcript Document

Tran The Trung
Department of Endocrinology
University of Medicine and Pharmacy HCMC
Cross-sectional Study: Introduction
• Cross-sectional = prevalence study
• All measurements are made at once
• Suitable goal:
• Prevalence*
• Describing the variables and their distribution
patterns*
• Associations between variables (then infer
cause & effect based on the author hypotheses,
not on the design study)
Cross-sectional studies
• Snap shot
• Measure exposure and outcome variables at one
point in time.
• Main outcome measure is prevalence
P = Number of people with disease x at time t
Number of people at risk for disease x at time t
Prevalence=k x Incidence x Duration
Cross-sectional study structure
• Population  Sample
• Measurements : Once
• Variables: characteristics, factors,
diseases,…
• One variable:
• Prevalence (nominal, categorical variables)
• Distribution (numeric variables)
Cross-sectional study structure
Population
Sampling
Sample
Prevalence
• Numerator: Number of people who have the
disease at one point in time
• Denominator: Number of people (at risk) at that
point
=> Cross-sectional study: the only suitable design
to estimate Prevalence
Association between variables
No risk factor
No disease
Risk factor
No disease
Sample
No risk factor
Disease
Risk factor = predictor variable
Risk factor
Disease
Ex.
Age, sex
Disease = outcome variable
Income, marital status
Which variables are predictors ?
Habit (smoking)
Which are outcomes ?
Weight
Cross-sectional studies - Strengths
• Useful baseline assessment
• Generalizable results if population based sample
• Study multiple outcomes and exposures
• Immediate outcome assessment and no loss to
follow-up, therefore faster, cheaper, easier
• Can measure prevalence
• Hypothesis generating for causal links
• Serial surveys eg, Census
Cross-sectional studies - Weaknesses
• Provide limited information.
• Cannot establish sequence of events
• Not for causation or prognosis
• Look for biological plausibility in causal links
• Impractical for rare diseases if pop based sample.
Could use in rare disease registry.
• Prone to bias (selection, measurement).
Strengths and weaknesses
• No waiting time
• Fast & inexpensive
• No loss of follow-up
• Prevalence of a disease or
risk factor
• Relations of variables
• May be the first step in a
cohort study
• Difficulty of establishing
causal relationships from
data
• Impractical to study rare
disease (can be done on
rare disease if sample is
collected from high risk or
diseased patients).
• Not measure incidence (no
information for prognosis &
natural history)
Bias in cross-sectional studies
Selection Bias
Is study population representative of target population?
Is there systematic increase or decrease of prevalence?
Measurement Bias
Outcome
• Misclassified (dead, misdiagnosed, undiagnosed)
• Length-biased sampling
• Cases overrepresented if illness has long duration and are underrepresented
if short duration.(Prev = k x I x duration)
Risk Factor
• Recall bias
• Prevalence-incidence bias
• RF affects disease duration not incidence eg, HLA-A2
Cross-sectional studies - Uses
• Prevalence used in planning
• Individual:
Pre-treament probability for Rx and Dx
• Population:
Health care services
• Describe distribution of variables
• Examine associations among variables
• Hypothesis generating for causal links
• Prediction
Analyses used in studies
Crosssectional
Case-control
Cohort
Prevalence
Incidence
Yes
No
No
No
No
Yes
Correlation
Yes
Yes
Yes
Regression
OR
RR
HR
Yes
Yes
No
No
Yes
Yes
No
No
Yes
Yes
Yes
Yes
Survival
No
No
Yes
Calculation
• Prevalence:
• Disease D = (a + c)/(a+b+c+d)
• Factor A = (a + b)/(a+b+c+d)
• Association:
• OR = ad/bc
Disease D
Factor
A
+
+
-
a
b
-
c
d
EXAMPLE: A CROSSSECTIONAL STUDY
Abstract
• The objective of this study was to assess the extent of diabetic control
and its associated factors among Vietnamese patients with diabetes
mellitus (DM).
• The study was conducted among 652 outpatients who were recruited at
a public general hospital (People Hospital 115) and a private clinic
(Medic Center) in Ho Chi Minh City, Vietnam.
• Median age of participants was 57 years from People Hospital 115, and
60 years for participants from Medic Center. 39% of patients at People
Hospital 115 and 33% of patients at Medic Center had Hemoglobin A1c
(HbA1c) greater than 8%. However, 55% and 45% of these patients
from each facility reported they are in good control... Overall,
Vietnamese diabetic patients in this study exhibited poor plasma
glucose control.
• Physician education designed to improve monitoring of glucose levels
and diabetic complications, and patient education aimed at raising
awareness about actual diabetic control are indicated in this population.
Purpose of Study
• The purpose of this study was to investigate:
• (a) characteristics of diabetic outpatients in the medical
practices of an urban area in Vietnam,
• (b) factors associated with diabetic control among these
patients, and
• (c) capacities of these patients to self-manage DM.
Methods
• This cross-sectional study was conducted from December
17, 2007 to January 17, 2008 in Ho Chi Minh City,
Vietnam.
• Study participants were diabetic outpatients who visited
endocrinologists at a private clinic (Medic Center) and a
public hospital (People Hospital 115) during the survey
period.
Methods
• General variables:
• date of birth, sex, anthropometries, family history, and health habits
(tobacco smoking and alcohol consumption).
• DM-related variables:
• previous visits, year of diagnosis, year medication was initiated,
type of DM, glycemic measurements (fasting or casual blood
glucose concentration [mg/dL], and HbA1c [%] measured within 6
months), type of DM treatment (diet alone, sulfonylurea, alphaglucosidase inhibitors, biguanides, thiazolidine derivatives,
phenylalanine derivatives, insulin, or other treatments), and the
presence or absence of diabetic complications (diabetic retinopathy
within the last 12 months, proteinuria, diabetic gangrene, and
atherosclerotic disease).
Methods
• HbA1c: with less than 6.5% being recognized as good
control and 8.0% or higher as poor control.
• Hypertension
• Lipid-related data
• Participants were interviewed about their perception of
well-being, DM-related distress, evaluation of selfmanagement, and perception of diabetic control.
Results
• A total of 658 diabetic patients (257 at People Hospital
115 and 401 at Medic Center) were invited to participate
in the study.
• Of these, 652 patients (253 at People Hospital 115 and
399 at Medic Center) agreed to participate.
• At People Hospital 115, the median age of participants
was 57 years (28-82) and 51.0 % of patients were male.
• At Medic Center, the median age was 60 years (24-92)
and 22.3% of patients were male (Table 1).
Results
• Median fasting plasma glucose concentration was 123.5
mg/dL (70-375) at People Hospital 115 and 131 mg/dL
(49-441) at Medic Center.
• Median HbA1c was 7.5% (5.2-16.2) at People Hospital
115 and 7.3% (5.0-13.4) at Medic Center.
Summary
• Cross-sectional study ~ Prevalence study
• The only design to assess a prevalence.
• Other statistic can also be calculated in cross-sectional
study (depend on the aims of the study), including:
comparison (t-test, Chi-square, ANOVA), correlations
(Pearson-r, Spearman-rho), regression (linear and logistic
OR).
• But not RR (and HR also)!