DrPH Team Lead Maternal and Infant Health Branch Division of Reproductive Health pdietzcdcgov 7704885128 Evaluating the 2003 Birth Certificate Data Preliminary Findings from the PRAMS Data Quality Improvement Project ID: 370866
Download Presentation The PPT/PDF document "Patricia M. Dietz," is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Patricia M. Dietz,
DrPHTeam LeadMaternal and Infant Health BranchDivision of Reproductive Healthpdietz@cdc.gov; 770.488.5128
Evaluating the 2003 Birth Certificate Data:
Preliminary Findings from the PRAMS Data Quality Improvement Project Slide2
Pregnancy Risk Assessment Monitoring System
PRAMS PRAMS State-based surveillance system funded by CDCBirth certificates are sampling frameLinks BC to questionnaire dataPRAMS Data Quality Improvement ProjectGoal to assess sensitivity and specificity of selected 2003 birth certificate and PRAMS items using the medical record as the gold standard.This presentation will focus on the birth certificate items that were included in the project.Slide3
Previous Birth Certificate Evaluations
Evaluations of the 1989 versionExcellent sensitivity of birth weight, sex, mode deliveryModerate to low sensitivity for medical complicationsExcellent specificity 2003 revised birth certificated includes Maternal height, pre-pregnancy weightMaternal weight at deliveryAugmentation/induction laborGestational diabetes, gestational hypertensionSlide4
PRAMS Data Quality Improvement Project
In FY 2011, CDC’s Division of Reproductive Health and CDC’s Division of HIV Prevention co-funded:New York City Department of Mental Health and HygieneVermont Department of HealthSlide5
Sample
New York CityAll PRAMS respondents who had delivered at a city hospital January-June 4, 2009N=41 hospitalsN=603 respondentsVermontAll PRAMS respondents who delivered at a Vermont hospital or at a New Hampshire hospital close to Vermont’s state border during January-August 2009N=13 hospitalsN=664 respondentsSlide6
Abstraction of Data
Abstracted information from parent worksheet, prenatal care and hospital delivery records onto a standardized abstraction formData abstractors were nurse midwives or trained medical abstractors Slide7
Abstraction of Data cont.
Abstractors trained on use of abstraction formTo evaluate reliability of record abstraction, about 25 records in New York City and Vermont were re-abstractedErrors in abstraction were noted, reviewed, and resolved Slide8
Analysis
Weighted analysesStratified by siteMissing values ranged from NYC: BMI 33%VT: BMI 13% Calculated Sensitivity, Specificity, 95% Confidence Intervals (CI) for categorical variablesPearson correlations for continuous variables Slide9
Analysis
Sensitivity and SpecificityHigh (>90%)Moderate (70-90%)Low (<70%)Continuous variablesPearson Correlations – assesses linear relationship1 is perfect correlation Slide10
Analysis
Medical record gold standard for most itemsNo gold standard for height and pre-pregnancy weight Based on self-report captured on parent worksheetExamined correlations between data abstracted Slide11
Pre-pregnancy Items
Mother’s height Mother’s weightBody mass indexPrevious live birth Previous C-sectionPrevious preterm Slide12
During Pregnancy Items
Gestational diabetes Gestational hypertensionWeight at delivery Slide13
Labor and Delivery Items
Augmentation InductionProlonged laborPremature rupture of the membranesVaginal deliveryC-section Slide14
Infant Items
Neonatal intensive care unit admissionGestational age – obstetric estimateGestational age – LMP estimate Slide15
Preliminary ResultsSlide16
Demographics
Demographic Characteristics NYCVermont<12 Years Education259
WIC
5746
Unmarried
46
37
White-NH
22
95
Black-NH
20
0.6
Hispanic
42
1
Other
15
3Slide17
Pre-pregnancy Items - Correlations
Worksheet vs. BCNYCVermontHeight.71.78Pre-pregnancy weight
.82
.99BMI.80
.97Slide18
Sensitivity for Pre-pregnancy ItemsSlide19
During Pregnancy Items - Correlations
NYCVermont Weight at delivery.87.94Slide20
Sensitivity for Pregnancy ItemsSlide21
Sensitivity for Labor and Delivery ItemsSlide22
Sensitivity for Labor and Delivery ItemsSlide23
Sensitivity for Infant ItemsSlide24
Sensitivity-Summary
High>90%Moderate 70-90%Low<70%Previous
LB
GDMPrevious pretermVaginal
delivery
GA
<37 LMP
Prolonged labor
C-section
NICU
Gest.
hypertension
PROM
Previous
C-section
Augmentation
GA <37 OB
Induction
Augmentation
GA <37 OB
Induction
Previous
C-section
Both
VT Only
NYC OnlySlide25
Specificity
>90% for all items Slide26
Summary
Consistent with previous studiesHigh specificity overallHigh sensitivity for mode of deliveryModerate to low sensitivity for pre-pregnancy and pregnancy items, complicationsNew itemsGood correlations for pre-pregnancy BMI - VTModerate to low sensitivity for induction and augmentation – varied by siteSlide27
Limitations
Possible data abstraction errorsConducted in only 2 sitesLarge percent of missing values pre-pregnancy BMIWide confidence intervals for rare conditions Slide28
Implications
New item, pre-pregnancy BMI, collected from mother good additionDifferences between sites suggests medical record documentation, hospitals procedures for birth certificate reporting affects sensitivityItems with poor data quality for both sites suggest poor documentation in medical recordLink items with low sensitivity to other data sources Possibly drop in next revision Slide29
Acknowledgments
New York City Department of Health and Mental Hygiene:Candace Mulready-Ward, MPHJudith Sackoff, PhDMelissa Gambatese, MPHRegina Zimmerman, PhDAnn Madsen, PhDSarah Rubinstein, MPHJahidah Reid, PAOdessa Fynn, CM, MS
Vermont Department of Health:
Peggy Brozicevic, BSJohn Gauthier, MSBrennan Martin, MPHMichael Nyland-Funke, MAJoan Mongeon, MSKathleen Keleher, RN, MPH, FACNM
Cynthia Hooley, BS
CDC:
Jennifer Bombard, MSPH
Lucinda England, MD, MSPH
Leslie Harrison, MPH
Tonya Stancil, PhD
Allan Taylor, MD, MPH
Lauren Fitzharris, MPHSlide30
CDC Disclaimer
The findings and conclusions in this presentation are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and PreventionSlide31
Patricia M. Dietz,
DrPHTeam LeadMaternal and Infant Health BranchDivision of Reproductive Healthpdietz@cdc.gov; 770.488.5128
Evaluating the Accuracy of Birth Certificate Data:
Preliminary Findings from the PRAMS Data Quality
Improvement Project (DQIP)Slide32
Analysis
Condition (according to medical record)
Yes
No
Data SourceYes
True Positive
False Positive
No
False Negative
True Negative
Sensitivity=
Specificity=
True
Positive
True
Negative
Condition (Yes)
Condition
(No)Slide33
Sensitivity for Pre-pregnancy ItemsSlide34
Analysis
Condition (according to medical record)
Yes
No
Data SourceYes
True Positive
False Positive
No
False Negative
True Negative
Sensitivity=
Specificity=
True
Positive
True
Negative
Condition (Yes)
Condition
(No)Slide35
Analysis
Condition (according to medical record)
Yes
No
Data SourceYes
True Positive
False Positive
No
False Negative
True Negative
Sensitivity=
Specificity=
True
Positive
True
Negative
Condition (Yes)
Condition
(No)Slide36
2003 Birth CertificateParent Worksheet
Pre-pregnancy weightHeightFacility WorksheetMedical complicationsPregnancy historyLabor and Delivery complicationsInfant itemsSlide37
Availability of recordsNew York City
Prenatal record (50.8% full, 43.6% partial)Hospital record (99.4%)VermontPrenatal record (89.3% full)Hospital record (99.8%)Slide38
Analysis
Sensitivity% of people who have a condition that are correctly classified with the condition on data sourceSpecificity% of people who do not have a condition that are correctly classified without condition on data source Slide39
Strengths
Representative of live births in each localityTwo different locations: one urban and one rural