Birth Hospitals’ Role in Access to Early Intervention Services among Drug-Exposed Infants

Birth Hospitals’ Role in Access to Early Intervention

Services among Drug-Exposed Infants

Taletha Derrington, PhD & Milton Kotelchuck, PhD, MPH141st APHA Annual Meeting

November 4, 2013 ● Boston, MA

Policy Context

2003 Keeping Children and Families Safe Act (better known by it’s precursor law, CAPTA – Child Abuse Prevention and Treatment Act)

2004 Individuals with Disabilities Education Improvement Act (IDEA)

2

3

Study Question 1

What are the rates and trends of Early Intervention (EI) referrals by hospitals among drug-exposed infants (DEI) born from 1998-2005?

4

Study Question 2 Are any of the following predictors of referral?

• Neonatal abstinence syndrome (NAS) diagnosis• Toxicology screen results• Insurance type• Maternal race/ethnicity• Hospital maternity level of care• Birth hospital discharge status

5

Drug Exposed Infant Identification Algorithm(DEIIA)

Pregnancy to Early Live Longitudinal (PELL) Data System

624,269 live births from 1998-2005

Birth Certificate

Hospital Discharge Delivery (Mother)

Hospital Discharge Birth (Child)

CORE

Child post-birth records(to age 3)

Maternal prenatal records(DOB – Gestational Age)

Emergency Department

Observational Stays

Non-birth Hospital Discharge

Early Intervention Service Records 1998-2008

7,348 DEI(1.2% of births)

4,436 referrals(60.9% of DEI)

6

Analytic Methods

Hospital referral source Pre- to Post-Mandate differences in referral

• Chi squared • Time series

7

Analytic Methods

Predictors of referral• Generalized estimating equations (GEE) logistic

regression• Interaction analyses with “Ai-Norton” corrections for

NAS and toxicology screens• Difference in differences to model interaction effects

8

Hospital Referrals of DEI

DEI Births - All Referral Sources*

DEI Births - Hospital

Referral Source*

DEI Referrals - Hospital

Referral Source*

0%

50%

100%

59%

12% 21%

66%

17% 25%

Pre- to Post-Mandate Differences in DEI Births Referred to EI

Pre: 1998-2003Post: 2004-2005Pe

rcen

t

* Chi Squared P < .01

9Births Referrals Hospital Referrals

Jan-98 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03 Jan-04 Jan-05 Jan-060

20

40

60

80

100

120

140

Pre- and Post-Mandate Time Series of DEI Births, Referrals, & Hospital Referrals

Birth Month & Year

Coun

t

Mandate

Hospital Referrals of DEI

10

Predictors of Referral

Other vs. NHW

Asian/Pacific Islander vs. NHW**

Hispanic vs. NHW

Non-Hispanic Black vs. NHW

No vs. Private Ins*

Public vs. Private Ins***

Positive vs. Negative Tox**

NAS Diagnosis vs. None***

0.0 0.5 1.0 1.5 2.0 2.5

Adjusted Odds Ratio a

Ins = insurance; NAS = neonatal abstinence syndrome; NHW = Non-Hispanic White; Tox = Toxicology Screen

Good or expected outcomeDisparity for reference groupDisparity for comparison group

*** P < .001** P < .01* P < .05

11

Predictors of Referral

Home Health Discharge vs. PC***

Transferred vs. PC***

Special Care Nursery vs. NICU

Well-baby nursery vs. NICU*

0.0 0.5 1.0 1.5 2.0 2.5

Adjusted Odds Ratio a

NICU = Neonatal Intensive Care Unit; PC = Parental care

a Adjusted for: birth weight, gestational age, clinical risk factors for EI eligibility, conditions establishing EI eligibility (e.g., Down syndrome), maternal characteristics (age, education, and nativity), maternal custody of infant, region of residence, rural/urban residence, and neighborhood poverty

Good or expected outcomeDisparity for reference groupDisparity for comparison group

*** P < .001** P < .01* P < .05

NAS Diagnosis Interaction

12

None Public Private30%

40%

50%

60%

70%

80%

90%

55% 56%

35%

66%

80%

55%

Differences in Predicted % DEI Referred by Insurance

No NASNAS Dx

Pred

icte

d Pe

rcen

t Ref

erre

d

24.4

Difference in differences: None vs. Pvt. = -8.4 (P < .0001)Pub. vs. Pvt. = 4.8 (not significant)

11.2

19.6

NAS Diagnosis Interaction

13

Well-Baby Special Care

Neonatal Intensive Care

30%

40%

50%

60%

70%

80%

90%

47%55% 55%

65%

78% 81%

Differences in Predicted % DEI Referred by Birth Hospital Maternity Level

No NASNAS

Pred

icte

d Pe

rcen

t Ref

erre

d

23.3

Difference in differences: Well-baby vs. NICU = -7.6 (P < .05)Special Care vs. NICU = -2.5 (not significant)

25.8

18.2

Toxicology Screen Interaction

14

None Public Private30%35%40%45%50%55%60%65%70%75%80%

56%60%

40%

60%

77%

36%

Differences in Predicted % DEI Referred by Insurance

NegativePostive

Pred

icte

d Pe

rcen

t Ref

erre

d

16.8

Difference in differences: None vs. Pvt. = 8.0 (not significant)Pub. vs. Pvt. = 20.5 (P < .0001)

- 3.7

4.3

Toxicology Screen Interaction

15

Well-Baby Special Care

Neonatal Intensive Care

30%35%40%45%50%55%60%65%70%75%80%

48%

60% 60%62%

76%74%

Differences in Predicted % DEI Referred by Birth Hospital Maternity Level

NegativePostive

Pred

icte

d Pe

rcen

t Ref

erre

d

15.5

Difference in differences: Well-baby vs. NICU = -1.1 (P < .01)Special Care vs. NICU = 1.0 (P < .05)

14.4

13.4

16

Conclusions DEI access to EI is suboptimal

• 34% of post-mandate births not referred Hospitals could identify and refer most DEI

• Referred only 17% of post-mandate births• General program improvement for all birth

hospitals needed to accelerate the weak upward trend in referrals

Conclusions Referrals of DEI with NAS or positive

toxicology screens should not vary across non-clinical factors• All children with NAS or positive toxicology

screens should be referred• Type of insurance should not be related• Targeted program improvement needed for

well-baby hospitals

17

Limitations Potential under-ascertainment of referral

• EI linkage rates 84%, similar to other studies• DEI may have lower linkage rates due to greater adoption &

mobility Validity of key measures

• Referral source in EI data• Toxicology screen measure on birth certificate

18

19

Implications for Research & Policy

Birth hospitals as potential universal referral source• Encourage birth hospitals to refer – use DEIIA• DEIIA – feasible screening tool & should undergo

further validation studies as a research tool More longitudinally linked data systems are

needed for research to inform program improvement and policy

20

Implications for Research & Policy

Need additional research on EI referrals by hospitals• Why are DEI born to mothers with private

insurance are not being referred as often? • Why is private insurance related to different

referral patterns for children with NAS or positive toxicology screens?

21

AcknowledgementsThis study is dedicated to the memory of

Dr. Lorraine Vogel Klerman, an inspirational mentor and champion of students

Dissertation Committee Marji Erickson Warfield Jody Hoffer Gittell Dominic Hodgkin Milton Kotelchuck

Dissertation funding support Nancy Lurie Marks Institute on

Disability Policy Fellowship Grants from the Heller Alumni

Association and the Office of the Provost, Brandeis University

I have no financial interests or disclosures

Thank You!E-mail: taletha.derrington@sri.com Web: http://dasycenter.org

REFERENCES

Ai, C & Norton, EC. Interaction terms in logit and probit models. Economics Letters. 2003; 80(1):123-129.

Derrington, TM. Development of the Drug-Exposed Infant Identification Algorithm (DEIIA) and Its Application to Measuring Part C Early Intervention Referral and Eligibility in Massachusetts, 1998–2005. Maternal & Child Health Journal. 2012; 10.1007/s10995-012-1157-x

Norton, EC, Wang, H & Ai, C. Computing interaction effects and standard errors in logit and probit models. State Journal. 2004; 4(2): 154-167.

22