The Effects of Attention-Deficit/Hyperactivity Disorder on Employment and Household Income

Joseph Biederman, MD; Stephen V. Faraone, PhD

In This Article

Study Design

In the first phase of this work, random digit-dialing probability sampling was used to contact subjects throughout the United States by RoperASW, a company that provides survey research services. With random digit-dialing sampling, telephone numbers are selected randomly and virtually every American household with a telephone has a probability of being included in the survey. This methodology yields samples that are nationally representative of households having telephones. All subjects were asked whether they had been diagnosed with ADHD. From this group, RoperASW randomly selected 500 subjects who reported that they had been diagnosed with ADHD and 501 who reported that they had not been diagnosed with ADHD. In the second phase, these subjects were contacted again and telephone interviews were conducted by professional interviewers experienced in the administration of telephone surveys. Each interview lasted approximately 25 minutes and subjects verbally consented to the interview before survey questions were asked. Data collection was carried out in conformance with the Code of Standards and Ethics for Survey Research of the Council of American Survey Research Organizations (CASRO). Approval for data analysis was sought from, and considered exempt by, the Institutional Review Board of the Massachusetts General Hospital.

Overall ADHD prevalence was estimated from respondents who were selected randomly from the US population. The survey interviewees were culled from a sample of adults from RoperASW's list of respondents who had identified themselves or a household member as having been diagnosed with ADHD during their adult life, and a gender- and age-matched group of respondents who had indicated that they had not been diagnosed with ADHD. The ADHD and control groups were compared using chi-square and t tests. For analyses of this survey, the maximum margin of error at the 95% confidence level is within ± 4 percentage points for sample sizes of 500.

Econometric Models

Econometric analyses were performed by the Lewin Group, Inc. (Falls Church, Virginia) on the basis of the survey results. The method used to estimate income loss was used previously in a study evaluating the economic impact of chronic fatigue syndrome[17]. It is based on the human capital theory, which suggests that ADHD is likely to have 3 possible effects that result in lower labor market earnings and incomes: lower-than-expected educational achievement, based on the individual's demographic characteristics; lower-than-expected employment rates, based on the individual's demographic characteristics and level of education; and poorer-than-expected labor market performance, based on the individual's demographic characteristics, educational achievement, and employment status. The impact of each of these effects was evaluated separately.

The approach starts with the underlying theory of labor market participation and productivity. It allows labor market earnings to differ both because of hours of work differences and differences in wages between the control group and those with the condition. Usually the effects are conditional on a given level of education. In the case of ADHD, we allowed for education levels between the cells to differ because of ADHD. By doing this, the impact of ADHD on lower education and resulting lower earnings and hours of work can be isolated.

Multivariate analysis was used to control for various individual and family characteristics unrelated to ADHD. The basic specifications controlled for sex, race, urban or rural living area, marital status, and age. Advance specification included the same controls as basic specifications, plus level of education and high school performance. Each of these specifications was further analyzed using the Household Income Model, an ordinary least-squares regression model that predicts household income on the basis of individual and family characteristics, and the Heckman Model, a 2-part model with a logistic equation for full-time employment and an ordinary least-squares regression for household income that is conditional on full-time employment, with the Heckman correction to control for selection bias.[18] Each model compared the household income of subjects in the ADHD group with that of subjects in the control group (without ADHD). An estimate of the income losses due to ADHD was obtained by comparing the income of subjects in the control group with that of subjects in the ADHD group.

We used Faraone and colleagues'[9] estimate of the prevalence of ADHD (2.9%), which was based on a random sample of the population, to project overall losses to the US population. The total income loss attributed to ADHD was calculated using the individual income loss due to ADHD times the number of subjects with ADHD, with the number of subjects with ADHD equal to the total US population times the prevalence of ADHD in the United States.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: