1. Evaluation of Newborn Metabolic Screening Forms as a Surveillance Tool for Perinatal Hepatitis B, New York City
Ruth Link-Gelles1, Julie E. Lazaroff, MPH2,
Christopher M. Zimmerman, MD, MPH2, Jane R. Zucker, MD, MSc2
National Immunization Conference
Atlanta, Georgia
April 19-22, 2010
1 Department of Epidemiology, Rollins School of Public Health, Emory University, Atlanta, GA
2 Bureau of Immunization, New York City Department of Health and Mental Hygiene
Hi, my name is Ruth Link-Gelles and I’m currently finishing my MPH at the Rollins School of Public Health at Emory University. This project was completed as part of a summer internship through the Health Research Training Program at the New York City Department of Health and Mental Hygiene.

2. Perinatal Hepatitis B
In 2008, there were 12,000 infants born to hepatitis B infected women in the U.S.
Without intervention, ~40% of such infants will become infected with hepatitis B at birth (perinatal Hep B)
Prevention of perinatal Hep B involves administration of post-exposure prophylaxis (PEP) with hepatitis B vaccination and hepatitis B immune globulin at birth
PEP is 90-95% effective at preventing perinatal Hep B
New York City’s Perinatal Hepatitis B Prevention Program (PHBPP) conducts surveillance for hepatitis B surface antigen positive (HBsAg+) pregnant women to prevent perinatal hep B
¼ of perinatally infected newborns will eventually die of chronic liver disease
- PEP consists of HB immune globulin and HB vaccine within 12 hours of birth, followed by 2 additional doses of HB vaccine and 1-2 months and 6 months of age
Vast majority of mother-to-infant transmissions occur at birth, rather than through the placenta
NYC has Second highest caseload in the U.S.

3. Perinatal Hepatitis B Surveillance, New York City
PHBPP identified 1,768 births to HBsAg+ pregnant women in NYC, in 2008
Reporting sources for surveillance:
Prenatal reporting
Prenatal providers
Laboratory
Postnatal reporting
Newborn nursery
Newborn Metabolic Screening Form (NMSF)
Only universal reporting method currently available in NYC
Prenatal Provider - 45%
Laboratory - 16%
Newborn Nursery - 32%
NSC - 7%

4. Newborn Metabolic Screening Form (NMSF)
Every newborn has a blood specimen collected on the NMSF
New York State DOH tests for >40 congenital diseases
NMSF has a field for recording Maternal HBsAg status
Maternal HBsAg status is transcribed from mother’s chart onto NMSF as:
“Negative”, “Positive”, “Unknown”, or missing
Requires access to maternal result and proper transcription of result onto form
-NMSFs: at birth, all newborns are tested for a number of metabolic disorders. Blood specimens are sent to the state lab. On the NMSF, the hospital writes the mother’s hep-b status, which isn’t taken from a test, but rather simply hand-copied from her chart. There’s a lot of room for error in this system: transcription error, data entry error, parental refusal, etc.
NMSFs first identify approximately 10% of cases in NYC.
In addition to helping identify such a large number of women, this is the last chance to identify infected women and their newborns before theyleave the healthcare system after birth. This is often the last opportunity for immunoprophylaxis for infants
From here on, I’ll talk about unknown/missing grouped together

5. NYC NMSFs All NYC NMSFs are reported to PHBPP
All positives investigated routinely
NYC for 2008, NMSF results:
97% negative
2% positive
1% missing/unknown

6. Objective
To assess completeness of NMSFs for identifying HBsAg+ pregnant women
Are all infants getting NMSFs?
How to interpret NMSFs with “unknown” status or missing data
To assess the accuracy of NMSFs relative to PHBPP data
Are positive NMSFs really positive? (sensitivity)
How often are negative NMSFs actually negative? (specificity)
Identify best practices for collection of NMSF data

7. Evaluations Birth certificate comparison
Chart reviews at select hospitals
Case match between NMSFs and PHBPP data
Site visits to assess best practices
- Reason to believe there is error, since there are so many steps
- Errors include transcription from lab report to maternal chart to NMSF, failing to report positives, failing to order a second test when first is inconclusive, failing to include a copy of the lab report in the file

8. Birth Certificate Comparison
Method
Compared number of birth certificates to NMSFs received for NYC births
City-wide and hospital to hospital comparison
Results
City-wide: 128,960 birth certificates and 128,521 NMSFs
439 fewer NMSFs than birth certificates, a 0.3% difference
Hospital-by-hospital
94% of hospitals had >99% concordance between birth certificates and NMSFs
- rationale: NMSFs are supposedly universal, while birth certificates are truly universal, so they were used as a gold standard, in a sense
High concordance at city and hospital level is an indication that NMSFs are basically universal
Differences are possibly caused by homebirths or parental refusals

9. Chart Reviews
Charts reviewed at the five hospitals with largest number of HBsAg+ deliveries in 2008
Hospitals selected as to bias sample for infected women
All unknown/missing NMSFs reviewed: 84 charts
90% found to be negative
1 chart had a HBsAg+ result
8% no HBsAg result found in chart
Random sample of “Negative” NMSFs reviewed: 166 charts
None found to be positive
One chart with no HBsAg result
NegativePositiveNo Result
Negatives16501
Missing/Unknowns7617
1 new HBsAg+ pregnant woman identified that was previously unreported to PHBPP

10. Chart Reviews II
Reviewed charts at 5 hospitals with the most NMSFs with unknown/missing status
All charts with an unknown/missing status on NMSF reviewed: 521 charts
11 HBsAg+ mothers (2%), of which 8 had not been previously identified by PHBPP
427 HBsAg- mothers (82%)
Unable to locate 83 charts (16%)
Emphasize greater likelihood that there would be a positive among the missing/unknown at lower prevalence hospitals – not so used to dealing with cases, might make mistakes – recommend following up on all reports of missing/unknown at all hospitals
-High number of charts could not be located

11. Case Match Results Newborn Metabolic Screening Forms: 120,195
All NMSFs matched against all available information in PHBPP database
Matched on infant date of birth, infant medical record number, and mother’s name and address
Newborn Metabolic Screening Forms: 120,195
False Positives: 181
Positives: 1,800
Non-matches: 9
Negative/missing/ unknown: 118,395
All NMSF matched against all available information PHBPP database (positive cases, and positive reports found to be negatives)
Negative NMSFs not found in PHBPP database taken as true negatives based on chart review results
Matching cases: 1,610
Non-matches: 6
False Negatives: 169
Confirmed HBsAg+ Cases by PHBPP: 1,785

12. Case Match Results PHBPP Case PHBPP Non-Case Total NMSF Positive 1,610
181
1,791
NMSF Negative/missing/unknown
169
118,226
118,395
1,779
118,407
120,186*
Sensitivity= (actual positives correctly identified)
Specificity = (actual negatives correctly identified)
Positive Pred. Value = (proportion that tested positive that were actually positive)
Negative Pred. Value = (proportion that tested negative that were actually negative)
High number of false positives (costs $$)
Caveat – did not check a random sample of negatives to prove that they were true negatives. Also, identified some cases that had not been previously found by PHBPP
Sensitivity= % CI=(89.05, 91.78)
Specificity= % CI=(99.82, 99.87)
* Does not include non-matches from previous slide

13. Site Visits
Visited two hospitals, one with high completion of maternal HBsAg status on the NMSF, and one with relatively low completion of the NMSF
Met with hematology lab and nurse manager in Newborn Nursery
Reviewed written policies and discussed where errors might occur
Differences noted between the two hospitals, leading to identification of best practices
High performing hospital – policy of re-running blood on any mother admitted to L&D without the actual lab result in the chart (i.e., transcribed result, missing result, etc.) – less room for transcription error
Low performing hospital – 3 no HBIG cases in less clear what the procedure was for women admitted without results. Did not require lab report.

14. Best Practices Placing a copy of lab report in maternal record
Training staff to interpret and transcribe lab report onto NMSF
Repeating tests for women without a copy of the lab report
Establishing lab protocols to ensure fast and accurate reporting

15. Summary NMSFs appear generally complete, but have limitations
>99% of infants get an NMSF
9% of NMSF-positives are actually negative
Almost 10% of PHBPP cases were identified as negative/missing/unknown on the NMSF
Small proportion of unknown/missing NMSFs represent HBsAg+ mothers not otherwise identified by PHBPP
Following best practices leads to more accurate NMSF data
Additional universal reporting method that involves less steps – electronic birth certificate – used in some states already. - would reduce transcription errors and allow for faster reporting and updating of incorrect results
- weakness: duplicates

16. Limitations
Birth certificates and NMSFs were not matched on individual level
Cannot determine reasons some children lacked NMSFs
Chart reviews conducted at select hospitals
Assumption that our selection would bias results to finding more cases
Medical record staff unable to provide all charts at time of review
Biased selection of hospitals may take away from the representativeness of the results

17. Recommendations
All hospitals should implement best practices for recording maternal HBsAg status on NMSFs to improve accuracy
PHBPP will be communicating best practices to all NYC birthing hospitals
In addition to positive NMSFs, all NMSFs with an unknown/missing status should be investigated

18. Acknowledgements Co-Authors: PHBPP Staff Julie Lazaroff, MPH
Jane R. Zucker, MD, MSc
Christopher M. Zimmerman, MD, MPH
PHBPP Staff
Aleruchi Mpi, MPH
Amy Kwok-Ye
Copley Kemp

19. References
Hou J, Liu Z, Gu F. Epidemiology and Prevention of Hepatitis B Virus Infection. Int J Med Sci. 2005; 2(1): 50–57.
CDC, US Department of Health and Human Resources. Perinatal Transmission. Atlanta, GA; ( (Accessed November 1, 2009).
CDC. Hepatitis B Virus: A Comprehensive Strategy for Eliminating Transmission in the United States Through Universal Childhood Vaccination: Recommendations of the Immunization Practices Advisory Committee (ACIP). MMWR. 40(RR-13);1-19.
Neil S. Silverman; Marilyn J. Darby; Sheila L. Ronkin; Ronald J. Wapner. Hepatitis B Prevalence in an Unregistered Prenatal Population: Implications for Neonatal Therapy. JAMA. 1991;266(20):

21. Case Match Results Positive NMSFs Match*
9% (1,610/1,800) confirmed
10% (181/1,800) false positive
Negative/unknown/missing NMSFs
0.1% (169/118,226) false negatives
Could not find match for
6 (0.3%) of positive PHBPP cases in NMSF
9 (0.5%) of positive NMSFs in PHBPP
Match*
True Positives
NMSF= Positive
PHBPP= Positive
n=1,610
NMSFs
False Positives
NMSF= Positive
PHBPP= Negative
120, 195
PHBPP
Cases
n=181
False Negatives
NMSF= Negative/
unknown/Missing
PHBPP= Positives
1,785
n=169
* Cases matched on maternal name, address, infant date of birth and MR#

22. Case Match Results NMSF Results PBHPP Result % Match (n) (N=120,195)
Interpretation
Positive
1.34% (1610)
True Pos
Negative
0.15% (181)
False Pos
0.13% (162)
False Neg
Unknown/missing
% (7)
Not Found
% (9)
Under investigation
Not found
98.36 (118,226)
True Neg
% (6)