You Are Here: AHRQ Home > About AHRQ > AHRQ 2008 Annual Conference > Slide Presentation (Text Version)

Explaining Racial and Ethnic Differences in Children's Use of Stimulant Medications

Slide Presentation from the AHRQ 2008 Annual Conference

On September 8, 2008, J.L. Hudson, G.E. Miller, and J.B. Kirby, made this presentation at the 2008 Annual Conference. Select to access the PowerPoint® presentation (695 KB).

Slide 1

Explaining Racial and Ethnic Differences in Children's Use of Stimulant Medications

J.L. Hudson, G.E. Miller, and J.B. Kirby
September 8, 2008

Slide 2

Published Research

This presentation is based on the results from the following published paper:
- Hudson, J., Miller, G.E. and Kirby, J.B. (2007). "Explaining Racial and Ethnic Differences in Children's Use of Stimulant Medications." Medical Care 45(11).

Slide 3

Motivation

Sharp increase in stimulant use by children in early 1990's.
Highlighed concerns with:
- Tolerance
- Dependence
- Side Effects
Case studies report over/under prescribing of stimulants.
Large difference across race/ethnicity.

Slide 4

Our Research

Important to understand factors that contribute to racial/ethnic differences in stimulant use among children.
Use of Medical Expenditure Panel Survey (MEPS) with secondary data sources to:
- Identify differences in characteristics across racial/ethnic groups
- Quantify the role these characteristics play in differential use of stimulants

Slide 5

Literature

Blacks and Hispanic differ from Whites on the following dimensions:
- Usual Source of Care
- RX Expenditures
Differential Use of Stimulants found in:
- Case Studies
- Medicaid Claims Data
- Nationally Representative Surveys

Slide 6

Data

All Children ages 5-17
Medical Expenditure Panel Survey 2000-2002:
- Stimulant Use
- Family characteristics
- Insurance Status
- Health Status
- Race (Hispanic, Non-Hispanic White, Non-Hispanic Black)
Medications identified using National Drug Code to link to Multum Lexicon database
Local Area Characteristics at Block Level from 2000 Decennial Census

Slide 7

Top Selling Stimulants among US Children 5-17, MEPS 2000-2002

The table lists the drug, its common brand name, and annual purchases in both total millions and Pct.

All stimulants: 13.9 million; 100 Pct
Methylphenidate: known as Concerta/Ritilin; 8.3 million; 59.6 Pct
Amphetamine-Dextroamphetamine: known as Adderall/Adderall XR; 4.8 million; 34.8 Pct
Dextroamphetamine: known as Dexedrine/Dexostat; 0.8 million; 5.5 Pct

Slide 8

Stimulant Use & Treatment for ADHD [attention deficit hyperactivity disorder] Children 5-17 MEPS 2000-2002

The table lists the stimulant use and treatment for ADHD and the percentage of use among various races

Any stimulant use: 4.2 (0.2) among all; 5.1 (0.3) among Whites; 2.8* (0.4) among Blacks; and 2.1* (0.3) among Hispanics
Any treatment for ADHD: 4.7 (0.3) among all; 5.8 (0.4) among Whites; 2.8* (0.4) among Blacks; 2.4* (0.3) among Hispanics
Note: *Statistically different from whites at 5% level

Slide 9

Oxaca-Blinder Decomposition

Regression based decomposition
Any differences in stimulant use across two groups must result from the following:
- Difference in characteristics of the groups (means)
- Difference in how characteristics affect stimulant use across the two groups (coefficients)

Slide 10

Oxaca-Blinder Wages

First used to study wage discrimination between men and women in 1970's
Consider education:
- Women were less likely to have college degree than men (mean)
- In wage regressions by gender—having a college degree provided a larger boost to wage for men (coefficient)
- Potential sign of discrimination because men and women with same level of education were not paid the same

Slide 11

Oxaca-Blinder

Using means and coefficients—can calculate what percent of gap in stimulant use is due to differences in a particular characteristic.
For a characteristic to explain difference in outcome—it MUST have a significant impact on the outcome in question:
- Consider eye color and Male-Female wages.
- If women were more likely to have brown eyes than men.
- But having brown eyes makes no difference in a wage regression.
- Differences in eye color cannot explain differences in wages.

Slide 12

Linear Probability Model

Sample —all children 5-17
Regressions run separately by race/ethnicity
Dependent Variable—Any Stimulant Use in the year
Independent Variables:
- Age
- Family Income as percent of poverty line
- Parental Education
- Insurance Status (private, public, uninsured)
- Family Structure
- Census Region
- Health related:
  - Usual Source of Care
  - Fair/Poor Health
  - Fair/Poor Mental Health
  - Child Limitation
  - Columbia Impairment Scale—behavioral health measure

Slide 13

Mean Characteristics by Race/Ethnicity—Family

The table lists the mean characteristics by race/ethnicity with regard to the family.

No high school degree: White-4.8; Black-15.8*; Hispanic-37.6*
Below 100% poverty: White-10.0; Black-32.5*; Hispanic-27.0*
Two parents: White-78.7; Black-39.8*; Hispanic-68.1*
Note: *Significantly different from whites at 5% level

Slide 14

Mean Characteristics by Race/Ethnicity—Insurance

The table lists the mean characteristics by race/ethnicity in regard to insurance.

Public insurance: White-13.4; Black-40.4*; Hispanic-36.1*
Private insurance: White-80.2; Black-52.0*; Hispanic-44.1*
Uninsured: White-6.44; Black-7.6; Hispanic-19.9*
Note: *Significantly different from whites at 5% level

Slide 15

Mean Characteristics by Race/Ethnicity—Health Status

The table lists the mean characteristics by race/ethnicity in regard to health status.

Fair/poor health: Whites-2.4; Blacks-3.5*; Hispanics-3.8*
Fair/poor mental health: Whites-2.8; Blacks-3.2; Hispanics-3.1
Child limitations: Whites-6.8; Blacks-6.0; Hispanics-4.7*
CIS behavioral: Whites-12.0; Blacks-11.4; Hispanics-19.2*
Note: *Significantly different from whites at 5% level

Slide 16

Coefficients by Race/Ethnicity—Family

The table lists the coefficients by race/ethnicity in regard to the family.

It was found not to be statistically significant.

Slide 17

Coefficients by Race/Ethnicity—Insurance

The table lists the coefficients by race/ethnicity in regard to insurance.

Public insurance: Whites-.029*; Blacks-.034*; Hispanics-not significant
Private insurance: Whites-.027*; Blacks-.014*; Hispanics-not significant
Uninsured: Base category

Note: *Significantly significant at 5% level

Slide 18

Coefficients by Race/Ethnicity—Health Status

The table lists the coefficients by race/ethnicity in regard to health status.

Fair/poor health: Whites-not significant; Blacks-not significant; Hispanics-(-.05*)
Fair/poor mental health: Whites-.14*; Blacks-not significant; Hispanics-.015*
Child limitation: Whites-.11*; Blacks-.07*; Hispanics-.07*
CIS behavioral: Whites-.07*; Blacks-.04*; Hispanics-.06*

Note: *Significantly significant at 5% level

Slide 19

Oxaca Blinder Calculations

Differences in mean characteristics explain:
- None of the gap for whites—blacks
- 25% of gap for whites—Hispanics
Blacks:
- Many of the differences between the groups had no significant impact on stimulant use (family, health status).
Hispanics:
- Differences can be explained by whites faring better in terms of Insurance Status and Health Status.

Slide 20

Comparative Means & Coefficients for Health Status

The table lists the comparative means and coefficients for health status.

Fair/poor mental health: means for Whites-2.8/Blacks-3.2; coefficients for Whites-.14*/Blacks-not significant
Child limitation: means for Whites-6.8/Blacks-6.0; coefficients for Whites-.11*/Blacks-.07*
CIS behavioral: means for Whites-12.0/Blacks-11.4; coefficients for Whites-.07*/Blacks-.04*
Note: *Significantly significant at 5% level

Slide 21

Race/Ethnicity Interacted Model Health Status Measures

Most or all of racial/ethnic differences are due to differences in the way these groups respond to the same characteristic in terms of stimulant use.
Run a new Linear Probability Regression that includes all children in a single model.
Model has interactions of race/ethnicity with the following Mental Health measures:
- Fair/Poor Mental Health
- Child Limitation
- CIS—Behavioral

Slide 22

Interacted Model Findings

Significant difference in how whites and blacks with the same mental health status use stimulants.
No significant difference between whites and Hispanics.
Black children reported to be in Fair/Poor Mental Health are 10% points less likely to use stimulants than white children in Fair/poor Mental Health.
Black children reported to have Behavioral Issues are 4% points less likely to use stimulants than whites with Behavioral Issues.

Slide 23

Discussion

Potential explanations for differences in "effect" of Mental Health characteristics on stimulant use:
- Cultural Differences across groups:
  - Response to behavioral cues
  - Trust of medical system
  - Beliefs—Willingness to "medicate"
- Environmental Differences across groups:
  - School Policies in reporting ADHD
  - Medical Treatment of ADHD

Slide 24

Conclusion

Our paper is the first to quantitatively explore differences in stimulant use by race/ethnicity.
Much of the difference is due to how racial/ethnic groups respond to characteristics.
Results are consistent with case studies suggesting cultural differences in treatment of mental health issues and corresponding use of medications.
Results are consistent with research from 90's finding Black families are reluctant to use medications to treat psychiatric disorders.

Current as of December 2008

Internet Citation:

Explaining Racial and Ethnic Differences in Children's Use of Stimulant Medications. Slide Presentation from the AHRQ 2008 Annual Conference (Text Version). January 2009. Agency for Healthcare Research and Quality, Rockville, MD. http://www.ahrq.gov/about/annualmtg08/090808slides/Hudson.htm