Transcript michael_kramer_eboh_50th_conference_presentation.pptx

Perinatal/Pediatric Epidemiology at EBOH
• Brief history
• Current “catalog” of faculty & research areas
• Selected methodological contributions
• Impact
History
• Barry Pless arrived in 1975
– Chronic disease in children
– Child injury
•
•
•
•
•
Joined by Larson in 1976 and Kramer in 1978
Moffatt, Dougherty, Ducharme, Duffy (MCH) in 1980s
Ciampi (1985), then Platt (1996) recruited in biostats
Many pediatrician-epidemiologists at MCH since 2000
Kaufman, Basso, Naimi, and Yang in last few years
Current Faculty in Perinatal Epi
• Robert Platt
• Jay Kaufman
• Olga Basso
• Ashley Naimi
• Michael Kramer
Current Faculty in Pediatric Epi
• Beth Foster
• Meranda Nakhla
• Mike Zappitelli
• Maryam Oskoui
• Caroline Quach
• Patricia Fontela
• Jesse Papenburg
• Moshe Ben-Shoshan
• Evelyn Constantin
• Michael Kramer
• Patricia Li
• Robert Platt
Paradox: Intersecting Perinatal Mortality Curves
• First described by Yerushalmy in smokers vs
nonsmokers (AJOG 1964)
• Low birth weight (LBW) ↑ in smokers
• Neonatal mortality ↓ in LBW births to smokers
• Reverse true for births >2500 g
• Cited by tobacco companies for decades
• Observed for all risk factors for LBW or preterm
Crossover for Perinatal Mortality
U.S. Blacks vs Whites, 1997
Perinatal deaths / 1,000 total births
1000
Whites
100
Blacks
10
1
28
29
30
31
32
33
34
35
36
37
38
Gestational age (weeks)
39
40
41 42+
What’s the Appropriate Denominator?
• For total stillbirths, can use total births
• But for GA-specific stillbirth risk, total births
at that GA is inappropriate
– Conditions on birth at that GA
– Reflects proportion of births born dead at that
GA, not the risk of stillbirth at that GA
– All fetuses at that GA are at risk for stillbirth
– Argument made in 1987 (Yudkin et al, Lancet)
GA-Specific Stillbirth Rate
Gestational age (weeks)
10
20
30
42
Livebirth1
Livebirth2
Livebirth3
Livebirth4
Livebirth5
Livebirth6
Livebirth7
Livebirth8
Stillbirth1
Livebirth9
100 per 1,000 fetuses at risk
500 per 1,000 total births
Appropriate Denominator: No Stillbirth Crossover
Stillbirth rate per 1,000 fetuses at risk
2.5
2
Whites
Blacks
1.5
1
0.5
0
28
29
30
31
32
33
34
35
36
37
Gestational age (weeks)
38
39
40
41
42+
Fetuses at Risk and Neonatal Mortality
• Fetuses at a given GA are at risk of live
birth within the next week
• All live births at risk of neonatal death
• All fetuses are at risk of neonatal death
within the next week (Joseph et al 2003)
Neonatal deaths / 1,000 fetuses at risk
Fetuses at Risk: No Neonatal Mortality Crossover
1.5
Whites
1
Blacks
0.5
0
28
29
30
31
32
33
34
35
36
37
Gestational age (weeks)
38
39
40
41
42+
The Preterm Birth Epidemic
Canada, 1981-2010
8.5
8
7.5
7
6.5
Births <37 wk (%)
09
20
07
20
05
20
03
20
01
20
99
19
97
19
95
19
93
19
91
19
89
19
87
19
85
19
83
19
19
81
6
U.S. Trends in Preterm Birth
Non-Hispanic Whites and Blacks, 1981-2012
20
18
16
14
12
10
8
Whites
Blacks
20
11
20
09
20
07
20
05
20
03
20
01
19
99
19
97
19
95
19
93
19
91
19
89
19
87
19
85
19
83
19
81
6
A Socially Contagious Disease
Singleton Preterm Birth, U.S., 2009
Changes in PTB vs Induction
U.S. States, 2002-04 vs 1992-94
4
Change in preterm (%)
3
2
1
0
r=+0.50 (+0.26, +0.68)
-1
-2
-10
-5
0
5
10
Change in induction (%)
15
20
25
Studying Child Health Benefits of Breastfeeding
• Potential for bias due to confounding and reverse causality:
doubt about neurocognitive and growth/obesity benefits
• Best way to minimize bias: RCT
• But randomization to breast- vs artificial feeding is infeasible
and may be unethical
• Initial feeding choice made before birth; prenatal
interventions are difficult and expensive
• Solution: RCT of intervention to promote BF exclusivity and
duration, with analysis by intention to treat
• Overlap of BF behaviours requires very large sample size
PROBIT
PROmotion of Breastfeeding Intervention Trial
A Cluster-Randomized Trial in the Republic of Belarus
Design
• Intervention based on WHO/UNICEF Baby-Friendly
Hospital Initiative
• RCT using cluster randomization
• Clusters randomized: 31 maternity hospitals and one
affiliated polyclinic per hospital
• 17,046 healthy BF newborns >37 weeks and >2500 g
enrolled during postpartum stay
• Sample size based on primary outcome: 10% reduction
in risk of GI infection during infancy
• Births occurred June 1996 to December 1997
Baby-Friendly Hospital Initiative
•
•
•
•
•
•
•
•
•
•
Have a written BF policy
Train staff to implement policy
Inform mothers about BF benefits
Help mothers begin BF within 30 min of birth
Show mothers how to BF and maintain BF
Give healthy newborns breast milk only
Practice rooming-in 24 hours per day
Encourage BF on demand
Give no pacifiers to BF infants
Foster and refer mothers to BF support groups
Duration of Breastfeeding
1
Proportion Still Breastfeeding
0.8
0.6
0.4
0.2
0
0
30
60
90
120 150 180 210 240 270 300 330 360
Age in days
Control
Experimental
Degree of Breastfeeding (%)
Predominant at 3 mo
Predominant at 6 mo
Exclusive at 3 mo
Exclusive at 6 mo
0
10
20
Control
30
40
Experimental
50
60
PROBIT Follow-Up
• PROBIT resulted in 2 cohorts that differed
substantially in exclusivity/duration of BF
– These differences were created by randomization, not
choice of mother or doctor
– This has enabled strong causal inferences with respect
to BF effects on long-term outcomes
• PROBIT II: age 6.5 years, data 2002-2005
• PROBIT III: age 11.5 years, data 2008-2010
• PROBIT IV: age 16 years, data 2012-2015
Impact
• CHIRPP (1990)
• CPSS (1995)
• WHA: exclusive breastfeeding 6 mo (2001)
• Reduction in preterm birth since mid-2000s