Transcript Variable Case Control Effect n=451 n=1353 OR (95% CI)

Identifying Predictors for Pertussis
Disease in Texas Infants Utilizing
Surveillance and Birth Certificate Data
from 1999-2003
Lucille Palenapa
IDEW Presentation
Fri, Jun 27, 2008
Overview
• Review project
– Objective
– Methodology
• Results
• Recommendations
• Q&A
Pertussis Challenges
• Difficult to diagnose & distinguish
• Often ruled out based on vaccine
history
• Lab results often not reliable
• Difficult to obtain sterile sample
**Need to facilitate earlier identification
of disease
– Establish risk factors for pertussis disease
Texas Birth Certificate Data
• Utilized in past research to assess risk
factors for infectious diseases
• Comprehensive source of infant,
maternal and paternal characteristic
data
Objective
• Identify significant risk factors for
pertussis disease in infants <12 months
of age in Texas utilizing surveillance
and birth certificate data from 19992003
Methodology
• Case-infant defined:
– Infant reported to DSHS as a confirmed or
probable pertussis case
– <12 months at onset
– Born in Texas from 1999-2003
• Control-infant defined:
– Randomly selected by same date of birth as
case-infant
– Not reported as pertussis case to DSHS
– Born in Texas from 1999-2003
Exclusion criteria
• Infants who were:
– Ruled out or lost to follow-up
– Not born in Texas
– >12 months at disease onset
– > 1 case of pertussis in 5-year study
(counted only once)
Methodology
• Retrospective case-control study
• 5 year data (1999-2003)
• DSHS pertussis surveillance data
• Control group
– Texas birth certificate data
• DSHS IRB submission
Methodology (Data Collection)
• After IRB approval, collected:
– Surveillance Data:
• Identified infants reported to DSHS as (confirmed
and probable) pertussis cases who were <12
months at onset for each respective year (19992003)
– Birth certificate data:
• Received large birth data files (avg 300-365K
records w/ avg 200 fields /year)
Methodology (cont.)
• SAS Statistical Software Utilized:
– Recode and reformat birth certificate data
into readable format
– Match cases to birth certificate data:
• Matched on name and date of birth
• Quality assurance done by hand to ensure match
of surveillance case to birth data
– 451 cases
– 3 controls randomly selected for each case
based on same date of birth
– 1353 controls
Methodology (cont.)
• 15 variables initially chosen for analysis
– 2 variables (birth length & medicaid
participation) eliminated because data
completeness <50%
• Variables chosen based on indications
of biological feasibility and on previous
epidemiological studies
Table 1. Variables Selected for Analysis
from Texas Birth Certificate
Variable
% missing
Selected
(Yes/No)
51.2
0.0
0.0
0.1
1.9
0.3
0.0
No
Yes
Yes
Yes
Yes
Yes
Yes
(Infant Charac.)
Birth length
Birth sequence
Birth type
Birth weight
Gestational age
No. siblings living
Sex
Table 1. Variables Selected for Analysis
from Texas Birth Certificate (cont.)
Variable
% missing
Selected
(Yes/No)
0.0
0.1
2.2
0.4
0.2
64.9
4.9
0.7
Yes
Yes
Yes
Yes
Yes
No
Yes
Yes
(Maternal Charac.)
Age
Alcohol Use
Education
Hispanic Origin
Marital Status
Medicaid
Prenatal Care
Tobacco Use
Analysis
• Descriptive statistics
• Multivariate logistic regression model:
• Reducing confounders:
– 10% odds ratio analyses, removal of
variables not causing at least 10% change
in OR
Results
Table 2. Mean Values of Continuous Variables
Describing Infant and Maternal Characteristics
by Case Status
Variable
Case
Control
n=451
n=1353
Birth weight (gms)
3182.6
3306.7
Gestational age (wks)
40.1
39.4
No. siblings living
1.5
2.1
Age (yrs)
24.8
26.4
Education (yrs)
13.7
14.1
Infant Charac.
Maternal Charac.
Table 3. Distribution of Categorical Variables
Describing Maternal Characteristics by Case Status
Variable
Case
Control
n=451
n=1353
Does mother use alcohol?
Yes
5 (1.1)
No
445 (98.9)
Education
>12 yrs
188 (42.7)
<12
252 (57.2)
Is mother of Hispanic origin?
Yes
242 (54.0)
No
206 (45.9)
Maternal Status (Mother married?)
Yes
277 (61.4)
No
174 (38.6)
8 (0.6)
1334 (99.4)
414 (31.3)
911 (68.8)
640 (47.4)
709 (52.6)
932 (69.0)
418 (30.9)
Table 3. Distribution of Categorical Variables
Describing Maternal Characteristics by Case Status
(cont.)
Variable
Case
Control
n=451
n=1353
Number of prenatal visits
>1
420 (97.9)
0
9 (2.1)
Mother’s Race
White*
399 (88.5)
Black
41 (9.1)
Native-American
1 (0.2)
Asian
9 (2.0)
Pacific Islander
0 (0.0)
Unknown
1 (0.2)
Does mother smoke cigarettes?
Yes
42 (9.3)
No
408 (90.7)
1264 (98.2)
23 (1.8)
1160 (85.8)
146 (10.8)
1 (0.1)
39 (2.9)
4 (0.3)
3 (0.2)
70 (5.2)
1271 (94.8)
Table 4. Crude Odds Ratios and 95% Confidence
Intervals for Infant Pertussis According to Maternal
Characteristics
Variable
Case
Control
Effect
n=451
n=1353
OR (95% CI)
Maternal age (yrs)
<19
103
20-29
238
30-39
103
>40
7
Maternal education (yrs)
<8
48
>8 and <12
280
>12 and <15
60
>16
52
Is mother of Hispanic origin?
Yes
242
No
206
178
749
401
25
2.3 (1.6-3.1)
1.2 (0.9-1.6)
Referent
1.1 (0.5-2.6)
125
698
219
283
2.1 (1.3-3.3)
2.2 (1.6-3.0)
1.5 (0.9-2.2)
Referent
640
709
1.3 (1.1-1.6)
Referent
Table 4. Crude Odds Ratios and 95% Confidence
Intervals for Infant Pertussis According to Maternal
Characteristics (cont.)
Variable
Case
Control
Effect
n=451
n=1353
OR (95% CI)
Mother married
Yes
No
277
174
932
418
Referent
1.4 (1.1-1.7)
Does mother smoke
cigarettes?
Yes
No
42
408
70
1341
1.9 (1.3-2.8)
Referent
Table 5. Crude Odds Ratios and 95% Confidence
Intervals for Infant Characteristics
Variable
Case
Control
Effect
n=451
n=1353
OR (95% CI)
Infant birth sequence
1st
>2nd
426
25
1320
33
Referent
2.3 (1.4-4.0)
Infant birth type
Single
Multiple
426
25
1320
33
Referent
2.4 (1.4-4.0)
Infant birth weight (gms)
<1499
8
1500-2499
50
>2500
393
22
61
1270
1.2 (0.5-2.7)
2.7 (1.8-3.9)
Referent
Table 5. Crude Odds Ratios and 95% Confidence
Intervals for Infant Characteristics (cont.)
Variable
Case
Control
Effect
n=451
n=1353
OR (95% CI)
Infant gestational age (wks)
<37
110
>38
327
245
1087
1.5 (1.2-1.9)
Referent
Infant sex
Male
Female
218
233
710
643
Referent
1.2 (0.9-1.9)
No. of siblings living
0
1-4
>5
146
290
14
526
787
23
Referent
1.3 (1.1-1.7)
2.2 (1.4-4.4)
Final Data Analyses
• Multivariate logistic regression:
– 6 variables selected as significant
predictors of pertussis disease
• 10% adjusted odds ratio:
– No variable eliminated
– 6 variables still remained
Table 6. Crude Odds and Adjusted Ratios and 95%
Confidence Intervals of Significant Predictors for
Pertussis
Variable
Case
Control
Effect
n=451
n=1353
OR (95% CI)
422
25
1290
33
Referent
2.4 (1.4-4.0)
Referent
1.9 (1.0-3.4)
Infant birth weight
<1499
8
1500-2499
48
>2500
391
20
60
1243
1.2 (0.5-2.7)
2.7 (1.8-3.9)
Referent
1.4 (0.6-3.4)
2.1 (1.4-3.3)
Referent
No. of siblings living
0
144
1-4
289
>5
14
520
780
23
Referent
1.3 (1.1-1.7)
2.2 (1.4-4.4)
Referent
1.8 (1.4-2.3)
3.1 (1.5-6.5)
Infant birth type
Single
Multiple
Adj. Effect
OR (95% CI)
Table 6. Crude Odds and Adjusted Ratios and 95%
Confidence Intervals of Significant Predictors for
Pertussis (cont.)
Variable
Case
Control
Effect
n=451
n=1353
OR (95% CI)
Maternal cigarette use
Yes
42
No
405
Maternal age
<19
20-29
30-39
>40
101
237
102
7
Maternal Hispanic Origin
Yes
242
No
205
Adj. Effect
OR (95% CI)
70
1253
1.9 (1.3-2.8)
Referent
2.1 (1.3-3.5)
Referent
175
734
390
24
2.3 (1.6-3.1)
1.2 (0.9-1.6)
Referent
1.1 (0.5-2.6)
3.0 (2.1-4.4)
1.3 (1.0-1.7)
Referent
1.0 (0.4-2.5)
628
1.3 (1.1-1.6)
1.3 (1.0-1.5)
695
Referent
Referent
Significant Predictor Variables
Variable
Adj. OR
(95% CI)
-Number of siblings >5
-Maternal age <19 yrs
-Infant low birth weight
(1500-2499 gms)
-Maternal cigarette use
(Yes)
3.1 (1.5-6.5)
3.0 (2.1-4.4)
2.1 (1.4-3.3)
2.1 (1.3-3.5)
Discussion & Recommendations
Number of siblings living
• No current literature w/ exact findings
• Similar findings:
– As no. of older siblings increased, delay in
immunization increased for household infants and
younger siblings (Reading, Surridge, & Adamson,
2004)
– Infants from larger household size less likely to be
fully immunized, more likely to have delayed
immunization (Li & Taylor, 1993; Peckham, Bedford,
Senturia, & Ades, 1989)
– Later born siblings more likely to have delayed
immunization than firstborn children (Higgins, 1990;
Kaplan, Macie-Taylor, & Boldsen, 1992; Schaffer &
Szilagyi, 1995)
– Later born children more prone to infectious disease
(Kaplan, 1990)
Significant Maternal Variables
• Maternal Age
– Similar findings for young maternal age
(Izurieta et al., 1996)
– Adolescent aged mothers found to have
significantly lower levels of antibodies for
pertussis than older mothers (Gonik, 2005
& Healy, 2006)
Significant Infant Predictors
• Low birth weight (LBW) infant
– Biologically feasible
– Similar findings, LBW infants more likely to
develop pertussis than normal birth weight
infants (Langkamp & Davis, 1996)
Maternal Cigarette Use
• Similar findings established:
– Maternal smoking increases the likelihood
of respiratory infections in infants (Ahmer
et al., 1999; Ahmer, et al., 1998; Geng,
Savage, Razani-Boroujerdi & Sopori, 1996;
Saadi, et al., 1996; Stocks & Dezateux,
2003)
• Smoking during pregnancy assoc w/
several adverse outcomes:
–
–
–
–
Premature delivery
Spontaneous abortion
Growth restriction
Increased risk of SIDS
Limitations of Study
• Difficult to gauge the effects of many
covariates with the statistical
procedures used
• Problem of multiple comparisons
present, acceptance criterion may have
been satisfied purely by chance
• Use of birth certificate data to predict
health outcomes has had mixed reviews
• Method for reporting disease
Recommendations
• Increase awareness and knowledge of
serious dangers of pertussis
• Specially targeted education campaigns
should include a focus on:
– Infant households with large number of
siblings
– Teen mothers
– Low birth weight infants
– Pregnant mothers and mothers with infants
who smoke
Recommendations (cont.)
• “Cocooning” strategy
– “Immunization of family members and close
contacts of the newborn”
– Post-partum
• Keep infants upto-date on
vaccinations
• Tdap booster
vaccine for
adolescents and
adults
Benefits of Study
• Aid clinicians by establishing risk
factors to facilitate earlier recognition
of disease and earlier consequent
prophylaxis treatment of patient and
close contacts
• Reduce health care costs associated
with pertussis
• Reduce lost productivity
Benefits of Study (cont.)
• Protect those most vulnerable, the
future of Texas…
Study Contributors
•
•
•
•
Rita Espinoza
Marilyn Felkner
Richard Taylor
Eric Miller
THANK YOU!
Lucille Palenapa
(512) 458-7111 x.6611
[email protected]