Rochelle Kingsley, MPH Office of Program Decision Support Texas Department of State Health Services Noha H. Farag, MD, PhD CDC EIS Field Assignments Branch.

Rochelle Kingsley, MPHOffice of Program Decision Support

Texas Department of State Health Services

Noha H. Farag, MD, PhDCDC EIS Field Assignments Branch

Birth Defects Surveillance and Epidemiology, DSHS

How PRAMS Can Inform Healthy Texas Babies Initiative

Office of Surveillance, Epidemiology, and Laboratory Services

Scientific Education and Professional Development Program Office

Healthy Texas Babies (HTB)

Initiative to decrease infant mortality Goals:

Use evidence-based interventions

Provide local partnerships and coalitions with major roles in shaping programs in their communities

Decrease preterm birth rate by 8% over two years

Save $7.2 million in Medicaid costs over two years

Pregnancy Risk Assessment Monitoring System (PRAMS)

CDC and DSHS-funded, state-based complex survey

Monthly stratified random sample of moms pulled from birth certificate Stratified on birth weight and race/ethnicity

Moms surveyed 2-3 months after delivery Maternal behaviors and experiences before,

during, and after pregnancy Population estimates representative of all

women in Texas who recently delivered a live birth

Significance of Texas PRAMS Data

Source of detailed state-level information on risk factors relevant to birth outcomes Behavioral factors: smoking, alcohol use

Psychosocial factors: stress, social support

Medical conditions: diabetes, hypertension, pregnancy complications

Not only during pregnancy, but also preconception and postpartum

50% of Texas births are to Hispanics

PRECONCEPTION HEALTH INDICATORS

Early Prenatal Care is Too Late

First few weeks after conception are the most critical for fetal development

Many risk factors that can affect fetal development have greatest effect 17-56 days of pregnancy

Most women not aware they are pregnant until after this period

Important to deliver interventions before pregnancy to reduce risks of adverse outcomes: Preterm delivery, low birthweight, birth defects

Preconception Health and Health Care

Preconception Health refers to the health of women during their reproductive years Important for men too Everyone benefits, regardless of pregnancy intention

Preconception Care refers to interventions designed to lower preconception risks that contribute to adverse maternal and infant outcomes

PRAMS Data Analysis

Birth year 2002-2010 combined 15,386 respondents (weighted estimate:

3,292,432) Aged 13-47 years

Preconception Health Indicators

Health Behaviors Smoking Alcohol consumption Binge drinking Physical Activity Multivitamin use

Health Conditions Weight status (underweight, overweight, obese) Diabetes Hypertension Anemia

Indicators Broken Down By:

Health insurance before pregnancy Pregnancy intention Medicaid paid for delivery (proxy for

socioeconomic status) Race/ethnicity Age Education

Health Insurance Status, Pregnancy Intention,

2002–2010—Texas PRAMS

Prevalence (%)

No health insurance before pregnancy

48%

Unintended Pregnancy 46%

Medicaid paid for delivery 59%

No Daily Multivitamin*

2002–2010—Texas PRAMS

Overall White Black Hispanic0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0.75268%

80% 80%

*During the month before pregnancy did not take a multivitamin at all, or took multivitamins but not every day of the week.

Smoking Three Months Before Pregnancy

2002–2010—Texas PRAMS

Overall White Black Hispanic0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0.166

27%

15%9%

Prepregnancy Obesity* 2002–2010—Texas PRAMS

Overall White Black Hispanic0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0.216 18%27% 23%

BMI of 30 or higher.

Indicators by Pregnancy Intention and Insurance, 2002–2010—Texas

PRAMS

Unintended and Uninsured

Intended and Insured 0%

10%20%30%40%50%60%70%80%90%

100%88%

58%

No Daily Multivitamin

Indicators by Pregnancy Intention and Insurance, 2002–2010—Texas

PRAMS

Unintended and Uninsured

Intended and Insured 0%

10%20%30%40%50%60%70%80%90%

100%88%

58%

22%13%

No Daily Multivitamin Smoking

Indicators by Pregnancy Intention and Insurance, 2002–2010—Texas

PRAMS

Unintended and Uninsured

Intended and Insured 0%

20%

40%

60%

80%

100% 88%

58%

22%13%

23% 20%

No Daily Multivitamin SmokingObesity

Implications

Significant racial/ethnic disparities among all indicators

Even among those with intended pregnancy and health care coverage, rates could use improvement Missing link?

Take-Home Message

Preconception health of women in Texas is less than optimal

To accomplish HTB goal of reducing infant mortality by 8%, analysis of gaps in preconception care is important

This is just a small snapshot of the wealth of data available from PRAMS

PREDICTORS OF PRETERM BIRTH

Preterm Birth (PTB) in Texas

2006 2007 2008 2009 Healthy People 2020

March of Dimes 2020

0

2

4

6

8

10

12

14

1613.7% 13.6% 13.3% 13.1%

11.4%

9.6%

Pre

vale

nce

PTB: deliveries at < 37 weeks gestation

National Facts

PTB leading cause of neonatal mortality Disparities in PTB persistent public health

problem Low education and poverty associated with

higher PTB risk Blacks have ≥ 50% higher PTB risk

How Are We Doing in Texas?

PRAMS 2004–2008 Socioeconomic disparity Race/ethnic disparity

Medicaid Payment for Delivery 2004–2008—Texas PRAMS

White Black Hispanic0

10

20

30

40

50

60

70

80

90

36%

69%77%

Pre

vale

nce

Race/Ethnic Disparity in PTB 2004–2008—Texas PRAMS

Early PTB0

2

4

6

8

10

12

2.5%

5.4%

3.1%

WhiteBlackHispanic

Pre

vale

nce

< 34 weeks gestation

Race/Ethnic Disparity in PTB 2004–2008—Texas PRAMS

Early PTB Late PTB0

2

4

6

8

10

12

2.5%

8.5%

5.4%

10.5%

3.1%

7.5%

WhiteBlackHispanic

Pre

vale

nce

< 34 weeks gestation

34–36 weeks gestation

Reasons for Race/Ethnic Disparity

Theories Early-life programming Weathering Hypothesis Racism Stress

Facts Stress associated with increased PTB risk Stress higher in blacks

Original Research Question

Does stress explain the observed race/ethnic disparity in PTB in Texas?

Reported Stress by Race/Ethnicity

Financial Emotional Partner Traumatic0

10

20

30

40

50

60White

Black

Pre

vale

nce *

*

** *

*

Stress

Stress and PTB

Stress OR (95% CI)

Financial 1.3 (1.1–1.5)

Traumatic 1.3 (1.1–1.6)

Emotional 1.2 (0.9–1.4)

Partner-related 1.1 (0.9–1.4)

No difference by race/ethnicity

Selected Risk Factors for PTB

OR (95% CI)

Age ≥ 35 yrs 1.3 (1.03–1.7)

Unmarried 1.2 (1.03–1.4)

Medicaid paid for delivery 1.3 (1.1–1.5)

Obesity 1.4 (1.1–1.7)

Pregestational Diabetes 3.7 (2.4–5.7)

Preconception smoking 1.4 (1.1–1.7)

Infections 1.3 (1.1–1.6)

No difference by race/ethnicity

Beyond Traditional Risk Factors

Look further upstream at causal pathway Consider contextual factors (neighborhood

characteristics) Proportion of residents in census tract:

Poverty (income < 150% of federal poverty level) Race/ethnic composition (proportion black

residents)

Revised Research Question

Do neighborhood characteristics explain race/ethnic disparity in PTB in Texas?

Data Source for Neighborhood Characteristics

American Community Survey Component of census Provides updated population estimates Neighborhood factors:

Proportion less than high school education Proportion non-Hispanic black

Statistical Considerations

PRAMS data not random sample Need survey procedures SUDAAN or SAS survey procedures

Neighborhood data Individuals in same census tract have more in common

with one another than they do with those in others census tracts

Account for correlation using multilevel models

Combining neighborhood data with complex survey data problematic

Published methods do not account for both neighborhood characteristics and survey design

What Texas Did

Modified multilevel methods to account for design factors in PRAMS Existing multilevel models accounted for neighborhood

effects, NOT design factors

Accurately estimate associations between neighborhood characteristics, individual-level risk factors, and PTB

Effect of Revised Method on Conclusions

High Proportion Black

OR P

Published Method 0.7 0.001

Proportion black in neighborhood and PTB among blacks

Referent: black women living in predominantly white neighborhoods

Effect of Revised Method on Conclusions

High Proportion Black

OR P

Published Method 0.7 0.001

Revised Method 0.7 0.3

Proportion black in neighborhood and PTB among blacks

Referent: black women living in predominantly white neighborhoods

Effect of Proportion Black in Neighborhood on PTB Risk

White Black Hispanic White Black Hispanic0

1

2

3

High Proportion Black

Medium Proportion BlackReferent: women living in predominantly white neighborhood

Effect of Proportion Black in Neighborhood on PTB Risk

White Black Hispanic White Black Hispanic0

1

2

3

High Proportion Black

Medium Proportion BlackReferent: women living in predominantly white neighborhood

Effect of Proportion Black in Neighborhood on PTB Risk

White Black Hispanic White Black Hispanic0

1

2

3

High Proportion Black

Medium Proportion BlackReferent: women living in predominantly white neighborhood

Effect of Proportion Black in Neighborhood on PTB Risk

White Black Hispanic White Black Hispanic0

1

2

3

High Proportion Black

Medium Proportion BlackReferent: women living in predominantly white neighborhood

Effect of Neighborhood Education on PTB Risk

White Black Hispanic White Black Hispanic0

1

2

3

Low Education Medium EducationReferent: women living in predominantly white neighborhood

Effect of Neighborhood Education on PTB Risk

White Black Hispanic White Black Hispanic0

1

2

3

Low Education Medium EducationReferent: women living in predominantly white neighborhood

Effect of Neighborhood Education on PTB Risk

White Black Hispanic White Black Hispanic0

1

2

3

Low Education Medium EducationReferent: women living in predominantly white neighborhood

Summary

Neighborhood factors did not explain excess PTB risk among black women

However, they did have an effect on risk in Hispanic women

Non-response among black women problematic

Public Health Significance

First study of neighborhood effects among Hispanic women

Previous studies compared black and white women

In Texas, Hispanic women represented 50% of weighted sample

In 2010, three published PRAMS studies used analytic methods inappropriately

Revising statistical methods ensured public health policies based on sound statistical evidence

Future Directions

Develop neighborhood deprivation index Target communities and individuals within them with

highest PTB risk

Risk factors for early versus late PTB

PRAMS Can Inform Healthy Texas Babies

Funds for community interventions can target highest risk communities

Evaluate effectiveness of interventions On outcome, PTB

On risk factors, preconception indicators

Reducing PTB is not equal to reducing disparity

Acknowledgments

Office of Surveillance, Epidemiology, and Laboratory Services

Scientific Education and Professional Development Program Office

The findings and conclusions in this report are those of the authors and do not represent the official position of the Centers for Disease Control and Prevention

CDC/EFAB−Betsy Cadwell−Diana Bensyl−Julie Magri

CDC/PRAMS−Indu Ahluwalia

Texas DSHS −Mark Canfield

−Rebecca Martin−Chris Webb

Contact Information

Rochelle Kingsley, MPHPRAMS Coordinator

email: [email protected]

Noha H. Farag, MD, PhD EIS Officer

email: [email protected]