Transcript 28 Data Interpretation.ppt

Data Interpretation
Objectives
• To describe interpretation of epidemiological
data
• To classify the sub-group analysis based on
hypothesized
-exposure/risk/determinant with the
-outcome/factor per study objectives
• To apply the type of measure of disease
occurrence and association
Measurements in epidemiological study
•
•
•
•
•
Main ideas and concepts
What is being assessed?
What are we answering?
What variables are included?
Calculations / Understanding ?
Data analysis
• Place your objectives in front of you
• Characterize the items you have mentioned in
objective (to determine….) by the variables
that are determining
• Prevalence (mainly the 3rd yr studies are to determine
prevalence of KAP; or any other factor
• determinants; as hypothesized that any two factors will
be associated
• Outcome; as hypothesized
• List the variables / determinant /sub-groups that will be
compared
Methods: Review your methods and data
•
•
•
•
•
•
•
•
•
Study design – does it justify to the research question?
Study setting – internal and external validity concerns?
Sampling / inclusion / exclusion criteria; selection biases?
Subjects : demographic, socioeconomic characteristics; selection bias?
Variables: clarity of defining exposure, outcome, other variables?
Data management: information bias; how data were managed?
Data collection; questions vague, missing information?
Measurement error: instruments calibrated; data collectors trained? ;
Statistical methods: summary statistics given; appropriate statistical tests
used?
5
Epidemiological Study design
• Cross sectional study design:
• Be cautious about associations between factors
and risks and outcomes
• As exposure and outcome are collected at the
same time; terms like being overweight is a risk
for an outcome ‘arthritis’ when data for weight
and arthritis were collected at the same time.
• Use ‘risk’ in case control, cohort & experimental
studies
Selection of Study Participants : Examples
(review selection criteria to assess representativeness)
• “Participants in the Women’s Health Study were a random
sample of all women ages 55 to 69 years derived from the state
of Iowa automobile driver’s license list in 1985, which
represented approximately 94% of Iowa women in that age
group....
• “We aimed to select 5 controls for every case from among
individuals in the study population who had no diagnosis of
autism or other pervasive developmental disorders (PDD)
recorded in their general practice record and who were alive
and registered with a participating practice on the date of the
PDD diagnosis in the case.
7
Variables: Example
• “Only major congenital malformations were included in the
analyses. Minor anomalies were excluded according to the
exclusion list of European Registration of Congenital Anomalies
(EUROCAT).
• If a child had more than 1 major congenital malformation of 1
organ system, those malformations were treated as 1 outcome
in the analyses by organ system ...
• In the statistical analyses, factors considered potential
confounders were maternal age at delivery and number of
previous parities. Factors considered potential effect modifiers
were maternal age at reimbursement for antiepileptic
medication and maternal age at delivery”
8
Data Sources/Measurement: Example
• “Total caffeine intake was calculated primarily using U.S.
Department of Agriculture food composition sources. In these
calculations, it was assumed that the content of caffeine was 137
mg per cup of coffee, 47 mg per cup of tea, 46 mg per can or
bottle of cola beverage, and 7 mg per serving of chocolate candy.
This method of measuring (caffeine) intake was shown to be valid
in both the NHS I cohort and a similar cohort study of male health
professionals... Self-reported diagnosis of hypertension was found
to be reliable in the NHS I cohort”
• “Samples pertaining to cases and controls were always analyzed
together in the same batch and laboratory personnel were unable
to distinguish among cases and controls”
9
Measures of disease occurrence : Prevalence
Comparison of smoking consumption pattern of KSU students 2009
Study variables
Cigarette No. Shisha No. Shamma No. Others No.
(%)
(%)
(%)
(%)
Total No. (%)
Gender
Male
Female
Total
475 (51.1)
344 (37)
40 (4.3)
61 (6.6)
929 (81.8)
72 (34.8)
69 (33.3)
10 (4.8)
56 (27.1)
207 (18.2)
547 (48.1)
413 (36.3)
50 (4.4)
117 (10.2)
1136 (100)
Source: Mandil A. Bin Saeed AA, Ahmed S, Al-Dabbagh R, AlSaadi M, Khan M. Smoking among
university students: A gender analysis Journal of Infection and Public Health (2010) 3, 179—187
Explain the descriptive characteristics in data
Descriptive data:
Demographic Characteristics of Study Participants by
Sunlight Exposure Group
Variables
Sunlight exposure groups
Total
(n = 54)
Low
(n = 19)
Moderate
(n = 18)
High
(n = 17)
33 (9)
31 (7)
41 (11)
35 (10)
6 (31.6)
7 (38.9)
16 (94.1)
29 (53.7)
13 (68.4)
11 (61.1)
1 (5.8)
25 (46.3)
Illiterate
0
0
9 (52.9)
9 (16.7)
Literate
19 (100)
18 (100)
8 (47.1)
45 (83.3)
Age, yrs
[Mean(SD*)]
Sex [n (%)]
Male
Female
Literacy rate
[n (%)]
SD=Standard Deviation
Source: Humayun Q, Iqbal R, Azam I, Siddiqui AR, Khan A, Baig-ansari N. Development and validation of sunlight exposure measurement
(SEM-Q) in adult population residing in Pakistan. BMC Public Health 2012, 12:421 doi:10.1186/1471-2458-12-421
Sun Exposure tested by interview /questionnaire
and comparison with Serum Vitamin D levels and
Variables
Sunlight exposure groups
Low (n = 19)
Moderate
(n = 18)
High (n = 17)
Serum vitamin D
(ng/ml) [Mean(SD)]
9.8
(4.7)
11.1
(4.6)
17.0
(6.5)
Time (minutes)
outdoors in summer
[Mean(SD)]
54.5
(30.0)
81(62.7)
331.2
(63.8)
Mean (SD) time
outdoors in winters
59.7
(32.5)
89.4
(65.0)
310
(85.0)
Source: Humayun Q, Iqbal R, Azam I, Siddiqui AR, Khan A, Baig-ansari N. Development and validation of sunlight exposure measurement
(SEM-Q) in adult population residing in Pakistan. BMC Public Health 2012, 12:421 doi:10.1186/1471-2458-12-421
Comparison of age and pretreatment ALT and Alpha-fetoprotein tests in
HCV patients with high and low grades of inflammation on liver biopsy.
ALT test
Characteristics
ALT U/L Mean (SD)
In males
In females
Age
< 40 years
≥ 40 years
Alfafetoprotein ng/mL
(median & IQR)
Liver Biopsy with
High grades of
inflammation
(moderate and severe)
Test response
n
Liver Biopsy with
Low grades of
inflammation
(minimal and mild)
Test response
n
38
49
133.92(75.4) 241
107.85 (76.4) 267
98.26(68.2)
74.26 (63.2)
53
50
51.4%
48.5%
3.3
(0.2 – 11.7)
357
212
62.7%
37.2%
2.2
(0.2 – 16.5)
4
46
P- Value
0.003b
0.001b
0.031a
0.65c
Statistical Tests: a:Chi Square; b: Student’s t test, c: Mann Whitney Test
ALT=Alanine transaminase
Source: Mirza S, Siddiqui AR, Hamid S, Umar M, Bashir S. Extent of liver inflammation in predicting response to Interferon alpha & Ribavirin in
chronic hepatitis C patients: a cohort study Journal: BMC Gastroenterology 2012 Jun 14;12:71. doi: 10.1186/1471-230X-12-71.
Mean pretreatment ALT at various liver
inflammation grades
ALT in U/L for:
Minimal
inflammation
(mean ± SD)
Mild
inflammation
(mean ± SD)
Moderate
inflammation
(mean ± SD)
Severe
Inflammation
(mean ± SD)
Males
Females
85.1 ± 59
59.1 ± 36.7
112.4 ± 74.5
85.5 ± 75.3
129.6 ± 74.5
102.9 ± 69.5
166 ± 81.7
137.1 ± 111.85
Source: Mirza S, Siddiqui AR, Hamid S, Umar M, Bashir S. Extent of liver inflammation in predicting response to Interferon alpha & Ribavirin in
chronic hepatitis C patients: a cohort study Journal: BMC Gastroenterology 2012 Jun 14;12:71. doi: 10.1186/1471-230X-12-71.
Mean levels of pretreatment ALT by inflammation grades in males & females:
Source: Mirza S, Siddiqui AR, Hamid S, Umar M, Bashir S. Extent of liver inflammation in predicting response to Interferon alpha & Ribavirin in
chronic hepatitis C patients: a cohort study Journal: BMC Gastroenterology 2012 Jun 14;12:71. doi: 10.1186/1471-230X-12-71.
Correlation: Increase in mean ALT with increase
in liver inflammation grades
Measure of association
Risk Factors for Diarrhea in Children less than
5 years in Low-income Settlements in Karachi
A case control study
– Cases: children <5 years with diarrhea/dysentery
– Controls: healthy children matched to cases on
age and gender from the same community
Inclusion Criteria
CASE
• Diarrhea1, or Dysentery2 of <7day
• No antibiotic use within the last 7
days of enrolment
• Moderate-to-severe diarrhea,
defined as at least one of the
following:
– a. Sunken eyes, more than
normal
– b. Loss of skin turgor
– c. Intravenous rehydration
administered or prescribed
CONTROL
• No diarrhea within 7 days of
enrollment
• Should not have taken antibiotics
in the previous one week
• Age, gender and neighborhood
matched to index case
– Concomitant: within 14 days of
presentation of the index case
1Defined
2
as 3 or more abnormally loose stools during the previous 24 hours.
Presence of blood in stools
18
Household characteristics of diarrhea of cases and controls
Household characteristics
Cases (n = 154)
n (%)
Controls (n = 268)
n (%)
OR
(95% CI)
Socio-economic status *
First tertile (upper)
47 (30.5)
99 (37)
1
Second tertile (middle)
58 (37)
99 (38)
1.3 (0.8-2.2)
Third tertile (lower)
49 (32)
70 (26)
1.5 (0.9-2.5)
65 (42)
142 (53)
1
27 (17.5)
63 (24)
17 (11)
19 (7)
5.6 (±4.6)
5.3 (±3.5)
1.0 (0.9-1.1)
2.25 (± 1.2)
2.42 (±1.9)
0.9 (0.8-1)
Caretaker’s educational status
No school education
Primary education
Secondary education
Crowding index **
Mean no. of children < 5
years in HH (± SD)
* Wealth index: index based on proportionate weighted sum of household assets
** Number of people in HH / Number of rooms in HH
0.9 (0.5-1.5)
Water and sanitation practices in the household of children with diarrhea and
asymptomatic matched controls.
Water and sanitation
practices
Cases (n = 154)
n (%)
Controls (n = 268)
n (%)
Unadjusted OR
(95% CI)
Water source in last 2 weeks
Piped water
87 (56.5)
156 (58)
1
Bought / tanks
51 (33)
76 (28.4)
2.6 (0.9-8.0)
Public tap/ rain water/
borehole
11 (7)
9 (3.3)
1.3 (0.7-2.6)
Other sources
5 (3.2)
27 (10)
0.3 (0.1-0.9)*
No
48 (31)
50 (18.6)
1.81 (0.94-3.51)*
Yes
27 (17.5)
51 (19)
1
79 (51)
166 (62)
0.90 (0.51-1.60)
Fetch drinking water everyday
sometimes
Untreated drinking water given to child in last 2 weeks
No
110 (71)
199 (74.5)
1
Yes
44 (29)
68 (25.5)
1.1 (0.7-1.7)
*:p=0.055 (not significant; interpretation?
cont’d: Water and sanitation practices in the households of study participants
Cases (n = 154)
n (%)
Controls (n = 268)
n (%)
Unadjusted MOR
(95% CI)
Yes
82 (53)
110 (41)
1
No
72 (46.7)
157 (59)
1.9 (1.2-3)
6 (4)
10 (4)
2.4 (0.6-10)
Boil
46 (30)
74 (27)
1
Leave in the sun/alum
2 (1.3)
8 (2.6)
0.5 (0.1-2.6)
Filtration (cloth/ other filters)
38 (25)
30 (11.2)
2.7 (1.3-5.7)
No treatment
68 (44)
156 (58)
0.6 (0.4-1)
112 (73)
164 (62)
1
Bury / scatter in yard
13 (8)
14 (5)
1.4 (0.6-3.3)
Bush /open sewer /field
29 (19)
88 (33)
0.4 (0.2-0.7)
Treatment of drinking water
Sometimes
Method used to treat drinking water
Method of stool disposal
Toilet
21
Finalizing data analysis
• Writing an abstract
Predicting tobacco use among high school students by using the global
youth tobacco survey in Riyadh, Saudi Arabia.
Eligible persons
Place/setting
Time
Good response-low selection bias
OBJECTIVE: To identify the predictors that lead to cigarette smoking among high school
students by utilizing the global youth tobacco survey in Riyadh, Kingdom of Saudi Arabia
(KSA). METHODS: A cross-sectional study was conducted among high school students
(grades 10-12) in Riyadh, KSA, between April 24, 2010, and June 16, 2010. RESULTS: The
response rate of the students was 92.17%. The percentage of high school students who had
previously smoked cigarettes, even just 1-2 puffs, was 43.3% overall. This behavior was more
common among male students (56.4%) than females (31.3%). The prevalence of students
who reported that they are currently smoking at least one cigarette in the past 30 days was
19.5% (31.3% and 8.9% for males and females, respectively). "Ever smoked" status was
associated with male gender (OR = 2.88, confidence interval [CI]: 2.28-3.63), parent
smoking (OR = 1.70, CI: 1.25-2.30) or other member of the household smoking (OR = 2.11, CI:
1.59-2.81) who smoked, closest friends who smoked (OR = 8.17, CI: 5.56-12.00), and lack of
refusal to sell cigarettes (OR = 5.68, CI: 2.09-15.48). CONCLUSION: Several predictors of
cigarette smoking among high school students were identified.
Outcomes
Defined; clear
Low information bias
Vague predictor
Who is selling?
Predictors shown by data; OR, 95% CI
23
EXERCISE – SELF WORK
Massive outbreak of viral gastroenteritis associated with consumption of
municipal drinking water in a European capital city.
On 24 August 2008, an outbreak alert regarding cases of acute gastroenteritis in
Podgorica triggered investigations to guide control measures. From 23 August
to 7 September, 1699 cases were reported in Podgorica (population 136000);
an estimated size of the outbreak is about 10000-15000 corresponding to an
attack rate of ~ 10%. We conducted an age-and neighborhood-matched casecontrol study, microbiologically analyzed fecal and municipal water samples and
assessed the water distribution system. All cases (83/83) and 90% (89/90) of
controls drank un-boiled chlorinated municipal water [matched odds ratio
(mOR) 11·2, 95% confidence interval (CI), l·6-∞]. Consumption of bottled water
was inversely associated with illness (mOR 0·3, 95% CI 0.1-0.8). Analyses of
fecal samples identified six norovirus génotypes (21/38 samples) and
occasionally other viruses. Multiple defects in thè water distribution System
were noted. These results suggest that the outbreak was caused by fecally
contaminated municipal water. It is unusual to have such a large outbreak in a
European city especially when the municipal water supply is chlorinated.
Therefore, it is important to establish effective multiple-barrier watertreatment Systems whenever possible, but even with an established
chlorinated supply, sustained vigilance is central to public health.
Massive outbreak of viral gastroenteritis associated with consumption of
municipal drinking water in a European capital city.
Time
Affected population
Setting/place
Design
On 24 August 2008, an outbreak alert regarding cases of acute gastroenteritis in
Podgorica triggered investigations to guide control measures. From 23 August to 7
September, 1699 cases were reported in Podgorica (population 136000) and we
estimated the total size of the outbreak to be 10000-15000 corresponding to an attack
rate of ~ 10%. We conducted an age-and neighbourhood-matched case-control study,
microbiologically analysed faecal and municipal water samples and assessed the water
distribution System. All cases (83/83) and 90% (89/90) of controls drank unboiled
chlorinated municipal water [matched odds ratio (mOR) 11·2, 95% confidence interval
(CI), l·6-∞]. Consumption of bottled water was inversely associated with illness (mOR
Attack0·3, 95% CI 0.1-0.8). Analyses of faecal samples identified six norovirus génotypes
rate (21/38 samples) and occasionally other viruses. Multiple defects in thè water
distribution System were noted. These results suggest that the outbreak was caused by
faecally contaminated municipal water. It is unusual to have such a large outbreak in a
H0 European city especially when the municipal water supply is chlorinated. Therefore, it is
important to establish effective multiple-barrier water-treatment Systems whenever
possible, but even with an established chlorinated supply, sustained vigilance
is central
Not clear ?
to public health.
Lab results
Predictors shown by data; OR, 95% CI
26