Outcomes of preterm birth Stefan Johansson Department of Neonatology, Karolinska university hospital Department of Medical Epidemiology and Biostatistics, Karolinska institutet Stockholm, Sweden
Jan 16, 2016
Outcomes of preterm birth
Stefan Johansson
Department of Neonatology, Karolinska university hospital
Department of Medical Epidemiology and Biostatistics, Karolinska institutet
Stockholm, Sweden
25 October 2006Stefan Johansson 2
25 October 2006Stefan Johansson 3
25 October 2006Stefan Johansson 4
Studies of outcomes of preterm birth- subjected to errors!
systematic errors
random errors
What are systematic and random errors?
25 October 2006Stefan Johansson 5
Studies of outcomes - random errors
Measurement errors may be random few meaurements - the average value could be wrong
Outcome differences may be a random finding uneven sampling of study subjects may result in spurious results
Differences in outcomes are not detected study sample to small
25 October 2006Stefan Johansson 6
BIG IS BEAUTIFUL
25 October 2006Stefan Johansson 7
Studies of outcomes - systematic errors
Selection bias
Information bias
Confounding
25 October 2006Stefan Johansson 8
Studies of outcomes - selection bias
The optimal study would be to include the world’s entire population, but every study have to select their subjects.
What happens if the selected study subjects are not similar to the general population?
RISK OF SELECTION BIAS!!
25 October 2006Stefan Johansson 9
Studies of outcomes - selection bias
”Cardiovascular risk and running - new insight” marathon runners.
”Low mortality among preterm infants” infants in a specialized center
25 October 2006Stefan Johansson 10
Studies of outcomes - information bias
The collection of information is not properly done; misclassification: a preterm infant has several infections but only one is recorded.
Misclassification can be… non-differential: the error is the same for all study subjects differential: the error is not the same for different study groups
Recall bias is a common type of differential information bias: Cancer patients report more stress than healthy control, but both
groups are similarly stressed according to objective stress tests.
25 October 2006Stefan Johansson 11
Studies of outcomes - confounding
Confounding means…something (measured or unmeasured) is important for the associations between you measurements.
Neonatal nurse Pregnancy
25 October 2006Stefan Johansson 12
Studies of outcomes - confounding
The association is confounded by age of neonatal nurses.
Neonatal nurse Pregnancy
Young female
25 October 2006Stefan Johansson 13
Good design pays off
25 October 2006Stefan Johansson 14
Parental characteristics Hospital setting Gestational age Apgar scores Blood testing X-rays Lung diseases Blood pressure Nutrition Infections Drugs Neurological symtoms Noise
Death Motor skills Vision Hearing Blood pressure Blood glucose Allergies Cognitive functions Academic performace Life span
Exposures and outcomes of preterm infants
25 October 2006Stefan Johansson 15
In addition… cohort effects…
Neonatal intensive care is a ”new” speciality. 1970’s mechanical ventilation 1980’s new treatment of premature lung disease
maternal steriod treatment for threatening labour 1990’s high frequency ventilation
gentle nursing strategiestreatment of painnutrition
Preterms born in the 60’s, 70’s, 80’s and 90’s do not represent the same group of people.
25 October 2006Stefan Johansson 16
What do we know from the literature!?
25 October 2006Stefan Johansson 18
Preterm birth and mortality - world-wide
One million infants born preterm die during the first four weeks (26% of neonatal mortality).
Asphyxia
Preterm birth Sepsis
Congenital
Tetanus
DiarrheaOther
Lawn et al, Lancet 2005;365:891-900
25 October 2006Stefan Johansson 19
Contribution of preterm birth to infant mortality
Relative risk [95% CI]* Etiologic fraction %
< 28 weeks 126,7 [124,0-129,5] 35,7
28-31 weeks 16,2 [15,4-17,0] 7,3
32-33 weeks 6,6 [6,1-7,0] 3,2
34-36 weeks 2,9 [2,8-3,0] 6,3
Infant mortality in live born infants < 37 weeks, Canada 1992-1994
*adjusted for age, parity, race, and education Reference group: infants born at term
Kramer et al, JAMA 2000;284:843
25 October 2006Stefan Johansson 20
Contribution of preterm birth to infant mortality
34% of infants deaths attributed to preterm birth (USA 2002)
Of deaths attributed to preterm birth 95% of occured in infants > 32 weeks and <1500 grams two thirds occured during the first 24 hours
Callaghan et al, Pediatrics 2006;118:1566
25 October 2006Stefan Johansson 21
Gestational age and mortality - Sweden
0
10
20
30
40
50
60
-27 wks 28-30 31-33 34-36 37-39 40-42 43-45
Perinatal mortality (%) in Sweden 2003,by gestational age.
25 October 2006Stefan Johansson 22
Trend in mortality of infants < 1500 grams
Horbar et al, Pediatrics 2002;110:143
25 October 2006Stefan Johansson 23
Infant mortality related to preeclampsia
Basso et al, JAMA 2006;296:1357
25 October 2006Stefan Johansson 24
The impact of level-of-care on mortality
Liveborn infants
Mortality Odds ratio 95% CI
Unadjusted- university hospital- county hospital
9241320
14.2 %10.3 %
1.000.70 0.54 - 0.90
Adjusted- university hospital- county hospital
9241320
14.2 %10.3 %
1.001.33 0.98 - 1.81
Johansson et al, Pediatrics 2004;113:1230
25 October 2006Stefan Johansson 25
The impact of level-of-care on mortality
Liveborn infants
Mortality Odds ratio 95% CI
24 - 26 weeks- university hospital- county hospital
262125
29.0 %43.2 %
1.001.84 1.11 - 3.04
27 - 31 weeks- university hospital- county hospital
6621195
8.3 %6.9 %
1.001.09 0.74 - 1.61
Johansson et al, Pediatrics 2004;113:1230
25 October 2006Stefan Johansson 26
What kind of picture emerges…
Many preterm infants die. Mortality risk is inversely associated with gestation/birth weight. Preterm birth contribute greatly to infant mortality rates.
Mortality among the most immature infants has decreased. Preeclampsia related mortality has decreased. Centralizing care of the most immature infants may improve
survival rates.
25 October 2006Stefan Johansson 27
What about outcome in surviving preterm infants?
25 October 2006Stefan Johansson 28
Outcome in adults born preterm
166 adults BW <1000 g vs 145 adults with normal BW,born 1977-1982 in Canada.
Mean gestational age 27 weeks.
Neurosensory impairment/-s identified in 40 adults (27%) Cerebral palsy n=20 Autism n=2 Blindness n=11 Cognitive impairment n=14
Saigal et al, JAMA 2006;295:667
25 October 2006Stefan Johansson 29
Outcome in adults born preterm
Educational attainments was similar in both groups(”highest achievement” excluded those with neurosensory impairment)
<1000 g Normal BW p-value
Total years ofcompleted eduaction
13.9 14.5 .02
Highest achievement .06
< high school 17% 12%
high school 54% 56%
college 24% 18%
university 5% 14%
25 October 2006Stefan Johansson 30
Outcome in adults born preterm
Current employment(”Job classification” excluded those with neurosensory impairment)
<1000 g Normal BW p-value
Full time work 83% 84% .85
Job classification .25
un-/semi-skilled 52% 40%
skilled, technical 35% 41%
management professional
13% 20%
25 October 2006Stefan Johansson 31
Outcome in adults born preterm
Independent living, marital status and parenthood
<1000 g Normal BW p-value
Independent living 42% 53% .19
Marital status .33
single 77% 75%
married 10% 7%
cohabitating 13% 18%
Parenthood 11% 14% .36
25 October 2006Stefan Johansson 32
Outcome in 6 year old children <26 weeks
Infants <26 weeks, born in the UK in 1995.
Severe disability defined as Cerebral palsy IQ less than -3 SD Profound hearing loss Blindness
Mild disability defined as Neurologic signs, minimal
functional impairment IQ between -1 and -2 SD Mild hearing impairment Squint or refractive error
Marlow et al, NEJM 2005;352:9
25 October 2006Stefan Johansson 33
What kind of picture emerges…
Studies of adults born preterm – good outcomes? Studies of children born preterm – poor outcomes?
Why contradicting results… different populations with different different health care systems? a reversed ”healthy worker” effect -
children born < 26w represent a new group of survivors?
25 October 2006Stefan Johansson 34
Conclusions
Outcome of preterm birth… consider methods! High mortality, although decreasing rates/risks.
Conflicting results on long term outcome.
More knowledgeneeded, to predictand promote goodoutcomes.