688534
research-article2017
JADXXX10.1177/1087054716688534Journal of Attention DisordersFairman et al.
Article
Diagnosis and Treatment of ADHD in the United States: Update by Gender and Race
Journal of Attention Disorders 1–10 © The Author(s) 2017 Reprints and permissions: sagepub.com/journalsPermissions.nav https://doi.org/10.1177/1087054716688534 DOI: 10.1177/1087054716688534 journals.sagepub.com/home/jad
Kathleen A. Fairman1, Alyssa M. Peckham1, and David A. Sclar1
Abstract Objective: The aim of this article is to update ADHD diagnosis/treatment trends by age, gender, and race. Method: National Ambulatory Medical Care Survey data were obtained for 2008-2009 to 2012-2013. Physician office visits including ADHD diagnosis and pharmacotherapy were measured per 1,000 population and per 1,000 office visits overall, and by demographic group. Logistic regression models controlled for demographics, psychiatric comorbidities, insurance type, and time period. Interactions of time, demographics, comorbidities, and insurance type were tested. Results: Diagnoses of ADHD increased by 36% in adults and 18% in youth, and diagnosis + drug by 29% in female and 10% in male youths. ADHD diagnosis was 77% less likely among Black than White adults but 24% more likely among Black than White youths in 2012-2013. Conduct disorder (CD) in youths multiplied odds of diagnosis + drug by 3.31; interaction of Black race × CD by 3.78. Conclusion: Upward trends in ADHD diagnosis and treatment have continued but vary markedly by group. Studies of undertreatment/overtreatment are needed. (J. of Att. Dis. XXXX; XX(X) XX-XX) Keywords ADHD, gender, epidemiology, race, prescription prevalence
Introduction Over a period of about two decades beginning in 1990, substantial upward trends in the rates of diagnosis and treatment for ADHD were observed in nationally representative samples of U.S. physician office visits obtained through the National Ambulatory Medical Care Survey (NAMCS; Olfson, Blanco, Wang, & Greenhill, 2013; Robison, Sclar, & Skaer, 2005; Robison, Skaer, & Sclar, 2004; Robison, Skaer, Sclar, & Galin, 2002; Sclar, Robison, Bowen, et al., 2012; Sclar, Robison, Castillo, et al., 2012). Since the most recent time periods studied, 2006-2009 for adults (Olfson et al., 2013) and 2007-2008 for children and youth (Sclar, Robison, Bowen, et al., 2012), several important developments have taken place. These include the approval of several new ADHD drugs and formulations (CenterWatch, n.d.); the formation of the American Psychiatric Association (APA) Diagnostic and Statistical Manual of Mental Disorders–5 Task Force in 2007, culminating in the publication of Diagnostic and Statistical Manual of Mental Disorders (DSM-5; 5th ed.; APA, 2013, n.d.); and updated guidelines, from the American Academy of Pediatrics (AAP) in 2011 (Wolraich et al., 2011) and from the American Academy of Family Practice in 2012 (adults) and 2014 (children and teens; Felt, Biermann, Christiner, Kochhar, & Harrison, 2014; Post & Kurlansik, 2012).
These changes may have increased the number of cases diagnosed and treated for ADHD in several ways. First, the age criteria for diagnosis broadened, both in the DSM-5 (symptoms no later than age 12 years vs. age 6 years in the Diagnostic and Statistical Manual of Mental Disorders [4th ed.; DSM-IV]; APA, 1994) and in the AAP guidelines (age 4-18 years in 2011 vs. 6-12 years in 2001; APA, 2013; Wolraich et al., 2011). Second, the DSM-5 guidelines require adults and adolescents to display only five ADHD symptoms, rather than the six required for a diagnosis in children (APA, 2013). Finally, a description of ADHD symptomatology in those aged 17 years or older was added to the DSM-5 for clarity (APA, 2013). Given this expansion of diagnostic criteria and available pharmacotherapies for ADHD, the present study was conducted to provide an update and reassessment of three previously observed demographic trends: 1
Midwestern University, Glendale, AZ, USA
Corresponding Author: Alyssa M. Peckham, Department of Pharmacy Practice, College of Pharmacy-Glendale, Midwestern University, 19555 N. 59th Avenue, Glendale, AZ 85308, USA. Email: [email protected]
2
Journal of Attention Disorders 1. Large increases in ADHD diagnosis and treatment among adults. From 1995-1996 to 2007-2008, the number of office visits at which an ADHD diagnosis was made increased from 3.1 to 14.5 per 1,000 U.S. adults (aged 20 years or older; Sclar, Robison, Castillo, et al., 2012). For the same time span and group, office visits including both ADHD diagnosis and medication increased from 1.9 to 11.4 per 1,000 (Sclar, Robison, Castillo, et al., 2012). Similarly, the percentage of visits in which stimulants were prescribed to adults (aged 18 years or older) diagnosed with ADHD increased from 52% in 1994-1997 to 67% in 2006-2009 (Olfson et al., 2013). 2. Greater rates of increase in ADHD diagnosis and treatment among female than male children and teens (aged 5 to 18 years). From 1991-1992 to 2007-2008, rates of ADHD diagnosis increased 3.7fold for boys (from 39.5 to 144.6 per 1,000 population) and 5.6-fold for girls (from 12.3 to 68.5 per 1,000 population; Sclar, Robison, Bowen, et al., 2012). In the same time period, rates of ADHD diagnosis coupled with drug treatment increased 4.4fold for boys and 6.5-fold for girls (Sclar, Robison, Bowen, et al., 2012). 3. Lower rates of diagnosis and treatment for nonWhite than White populations. Several studies conducted in the 1990s and early 2000s found that non-White youths were less likely than White youths to receive health care services for mental disorders, including ADHD (Coker et al., 2016; Kataoka, Zhang, & Wells, 2002; Zimmerman, 2005). Similarly in 2006-2009, 0.68% of office visits made by White adults, compared with 0.36% for non-White adults, included the prescribing of a stimulant medication (Olfson et al., 2013).
Method Data Source Study data were drawn from the NAMCS, a cross-sectional, nationally representative survey of U.S. physician office visits that is conducted by the National Center for Health Statistics (NCHS) and has been used extensively in studies of ADHD treatment trends (Olfson et al., 2013; Robison et al., 2005; Robison et al., 2004; Robison et al., 2002; Sclar, Robison, Bowen, et al., 2012; Sclar, Robison, Castillo, et al., 2012). Survey data are collected annually by the U.S. Bureau of the Census using a three-stage probability sampling procedure that has been described in detail elsewhere (U.S. Centers for Disease Control and Prevention, [NCHS], 2010a, 2010b). Briefly, the NAMCS unit of analysis is an individual office visit. Sampling is conducted first by primary sampling units (PSUs), which are geographic
areas comprising counties, county groups, or geographic equivalents; then within PSU, by physician name based on master lists maintained by the American Medical Association and the American Osteopathic Association, stratified by physician specialty. Each selected physician is randomly assigned to 1 of the 52 weeks of data collection within the survey year, and within each reporting week, a systematic random sample of office visits is made. Visits represent office-based patient care, and contacts made outside the office (e.g., by telephone, in-home) or for administrative purposes only (e.g., medication refill) are excluded. For each recorded office visit, the NAMCS data file includes a sampling weight that adjusts for the multistage (i.e., clustered) sampling design and for survey nonresponse. Application of the weights to the data yields nationally representative information for all community-based physician office visits in the United States. In accordance with NAMCS recommendations for use of the data and with the methodology of previous studies using this data set, visits were grouped into multiyear increments for the present study analyses to increase the reliability of the weighted estimates (Olfson et al., 2013; Robison et al., 2005; Robison et al., 2004; Robison et al., 2002; Sclar, Robison, Bowen, et al., 2012; Sclar, Robison, Castillo, et al., 2012). Three time periods were studied: 2008-2009, 2010-2011, and 2012-2013.
Data Elements Counts of visits in which a diagnosis of ADHD was made were based on International Classification of Diseases– Ninth Revision (ICD-9) codes of 314.0x to 314.9x in any of the three diagnosis fields available in the NAMCS data set (Olfson et al., 2013). Identification of the prescribing of ADHD medication was made using the NAMCS drug identifiers, available for up to eight drugs continued or newly prescribed during the office visit, for generic and brand formulations of stimulant and nonstimulant medications to treat ADHD: amphetamines and amphetamine salt combinations, atomoxetine, dextroamphetamine, dexmethylphenidate, extended-release guanfacine or clonidine, lisdexamfetamine, and methylphenidate. Estimates measured whether an ADHD diagnosis had been made and whether it was accompanied by a prescription for ADHD medication (diagnosis + drug). Although sometimes used to treat ADHD in adults, especially those with comorbid psychiatric disorders, antidepressants were not counted as ADHD pharmacotherapy because they are not considered first-line treatment for ADHD (Post & Kurlansik, 2012).
Calculations Prevalence estimates were made using two different denominators, both of which have been used in previous research
3
Fairman et al. using the NAMCS (Olfson et al., 2013; Robison et al., 2005; Robison et al., 2004; Robison et al., 2002; Sclar, Robison, Bowen, et al., 2012; Sclar, Robison, Castillo, et al., 2012). First, overall and by sex, separately for adults (aged 20 years or older) and youth (aged 19 years or younger), estimates were calculated as annual number of visits per U.S. population, with annual averages calculated as the total number of visits for each 2-year period divided by 2 (Sclar, Robison, Bowen, et al., 2012; Sclar, Robison, Castillo, et al., 2012). Use of this method allowed for comparison of the present study’s results with those of previously published work using the same method, to produce updated trend estimates. Population estimates for each of the three time periods were obtained from U.S. Bureau of the Census figures for 2008, 2010, and 2012, respectively (“Age and sex composition”) (U.S. Department of Commerce, U.S. Census Bureau, n.d.). Second, estimates of the percentage of visits resulting in an ADHD diagnosis were calculated overall and by factors associated with ADHD diagnosis and treatment in previous work: sex, age group, race, and insurance type (Olfson et al., 2013). Sensitivity analyses compared trends determined using each denominator. Two changes from previous calculations made for samples of youth are notable. First is that, although children younger than 5 years were excluded from some previous samples, they were included in the present study sample, and their rates of drug therapy were assessed separately from those of older children. This change was made because recent guidelines recommend behavioral therapy as firstline treatment in this group (Felt et al., 2014; Wolraich et al., 2011). Second, teens aged 19 years were included in the youth cohort because of the discrepancy in definitions of an adult across different previous studies; specifically, those 19 years of age have been excluded from some studies of youth and were defined as adults in one previous study (Olfson et al., 2013; Sclar, Robison, Castillo, et al., 2012). Finally, because demographic differences in ADHD diagnosis and treatment may be confounded by relevant clinical characteristics, binary logistic regression models of diagnosis and of diagnosis + drug were estimated separately for youth and adults. Predictors included the aforementioned demographic characteristics, time period, and several mental disorders commonly associated with ADHD such as anxiety disorders (ICD-9 codes: 293.84, 300.0x, 300.2x, 300.3x, 300.7x, and 300.8x), mood disorders (293.83, 296.xx, 300.4x, 301.1x, and 311), substance addiction/abuse (291.xx, 292.xx, 303.xx, 304.xx, and 305.xx), conduct disorders (312.xx), and psychoses (293.81, 293.82, 295.xx, 297.xx, 298.xx, and 301.2x) (Felt et al., 2014; Olfson et al., 2013; Wolraich et al., 2011). For each outcome and group, modeling was performed using backward stepwise regression (minimum F probability for entry = .05, maximum F probability for removal = .10) in
three stages: (a) Test first-order (single factor) terms for age, sex, time period, race, insurance type, and the mental disorders. (b) To the resulting model, add and test terms representing the interaction of time period with each demographic group term. (3) To the resulting model, add and test terms representing interactions of race and sex with comorbid disorders and with insurance type. All analyses were performed at alpha of .05 using SPSS Version 23.0 (IBM SPSS, Armonk, New York).
Results The data set comprised 254,326 individual office visit records, representing approximate annual averages of 194 million visits for youth up to age 19 years and 780 million visits for adults aged 20 years or older from 2008 through 2013. A diagnosis of ADHD, with or without prescribed medication, was made at about 4.0 million adult and 7.7 million youth visits in 2008-2009; these numbers grew to 5.7 million and 9.1 million, respectively, in 2012-2013 (Table 1). Expressed as visits per 1,000 population, diagnoses of adult ADHD increased over the 6-year study period by 36%, and diagnoses coupled with pharmacotherapy (diagnosis + drug) increased by 21% (Table 1). These rates continued trends observed in previous studies using the same metrics (Figure 1). Rates of increase for youth similarly trended upward, albeit more slowly than for adults, at 18% and 16%, respectively (Table 1). Growth trends by gender differed markedly for the two age groups (Table 1). Among adults, rates of increase in diagnosis and diagnosis + drug for males were more than double those for females, respectively (52% vs. 23% for diagnosis, 30% vs. 13% for diagnosis + drug). In contrast, rates of diagnosis increased somewhat more for female than male youths (22% vs. 17%, respectively), and the rate of increase in diagnosis + drug for female youths was nearly triple than that for males (29% vs. 10%, respectively). Increases overall and by sex displayed similar trends whether measured per 1,000 population or per 1,000 office visits. However, the magnitudes of upward trend were generally somewhat greater in the latter analysis (Table 2). For each 1,000 office visits made by adults in 20122013, 7.62 included a diagnosis of ADHD and 5.48 included a diagnosis + drug (Table 2). Among youth, rates were about 7 times those of adults: 52.05 and 40.55 per 1,000 office visits for diagnosis and diagnosis + drug, respectively. For both adults and youth, absolute rates on both metrics were consistently lower for females than males. However, as in the analysis per 1,000 population, rates of growth were generally higher for male than female adults, especially for diagnosis, and higher for female than male youth, especially for diagnosis + drug.
4
Journal of Attention Disorders
Table 1. Number of ADHD Diagnosis and Pharmacotherapy Visits Per 1,000 Population by Sex and Age Group. ADHD diagnosis Adults aged ≥20 years Total unweighted number of visits Total weighted number of visitsa Rates per 1,000 population All adults Female adults Male adults Youth aged ≤19 years Total unweighted number of visits Total weighted number of visitsa Rates per 1,000 population All youth Female youth Male youth
ADHD diagnosis + drug
2008-2009
2010-2011
2012-2013
Change T1 to T3 (%)
2008-2009
2010-2011
2012-2013
311 4,019,154
369 4,886,182
867 5,723,614
— —
245 3,264,301
284 3,835,189
628 4,117,705
18.5 19.2 17.8
22.1 19.4 25.0
25.3 23.7 27.0
36.4 23.2 51.6
15.1 15.6 14.4
17.4 14.8 20.1
18.2 17.6 18.8
542 7,657,410
582 8,286,634
1,432 9,085,860
— —
428 6,115,333
454 6,484,622
1,108 7,079,461
93.1 56.3 128.3
99.6 56.9 140.6
110.3 68.5 150.3
18.5 21.7 17.1
74.3 45.5 102.0
77.9 41.4 113.0
86.0 58.5 112.2
Change T1 to T3 (%) — — 20.8 12.5 30.4 — — 15.6 28.8 10.0
Note. T = time period (T1 = 2008-2009; T3 = 2012-2013). a Annual averages, calculated as the total number of visits for each 2-year period divided by 2.
30
Present study findings
Findings of Sclar, Robinson, Castillo, et al. (2012)1 25
20
15
10
ADHD Diagnosis ADHD Diagnosis and Drug
5
0 1995-1996
1997-1998
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2008-2009
2010-2011
2012-2013
Figure 1. ADHD diagnosis and pharmacotherapy per 1,000 population: U.S. office visits for adults aged ≥20 years.
For older and non-White adults, rates of diagnosis and diagnosis + drug were notably low, despite a sharply increased trend over time (Table 2). Compared with visits made by White adults, those made by Black adults were 86% less likely to include an ADHD diagnosis and 85% less likely to include a diagnosis + drug in 2008-2009. Growth trends in these metrics over time were much greater for Black than for White adults; however, a large racial difference persisted in 2012-2013, when Blacks were 77% less likely to receive a diagnosis or diagnosis + drug. For youths, trends by race differed considerably from those observed for adults (Table 2). In 2008-2009, visits made by Black youths were somewhat less likely than
those of White youths to include an ADHD diagnosis or drug. However, over the 6-year study period, rates of growth in both metrics for Blacks were approximately 3 times those of Whites. Thus, in 2012-2013, visits made by Black youths were 24% more likely to include an ADHD diagnosis and 19% more likely to include a diagnosis + drug than were visits made by White youths. Youths of other race, predominantly Asian/Pacific Islander, were persistently less likely than White youths to receive an ADHD diagnosis or drug, despite growth rates far surpassing those for youths of other races. In logistic regression analyses, the factors most strongly and consistently predictive of ADHD diagnosis or diagnosis
5
Fairman et al. Table 2. Rates of ADHD Diagnosis and Pharmacotherapy Per 1,000 Office Visits by Demographic Group. ADHD diagnosis Adults aged ≥20 years All Female Male Age group (years) 20-29 30-39 40-49 50-59 60-64 65 or older Race White Black Other Payment source Private pay Medicare Medicaid Other Youth Aged ≤19 Years All Female Male Age group (years) Younger than 5 5-9 10-14 15-19 Race White Black Other Payment source Private pay Medicare Medicaid Other
2008-2009
2010-2011
2012-2013
4.99 4.32 6.08
6.24 4.67 8.62
7.62 6.16 9.81
17.86 9.36 8.37 2.78 1.65 0.20
29.06 8.12 6.25 4.57 2.90 0.44
5.70 0.81 1.45
ADHD diagnosis + drug Change T1 to T3 (%)
2010-2011
52.6 42.6 61.3
4.06 3.52 4.93
4.90 3.57 6.91
5.48 4.58 6.84
35.2 30.2 38.8
23.59 15.89 10.62 6.45 2.87 0.87
32.1 69.8 26.9 131.9 73.8 347.8
15.37 7.29 6.43 2.32 1.31 0.17
22.80 6.54 4.93 3.50 2.50 0.26
17.67 11.51 7.14 4.97 1.85 0.49
15.0 58.0 11.1 114.5 41.4 187.4
6.93 2.16 3.60
8.40 1.94 4.90
47.4 140.0 238.3
4.61 0.68 1.45
5.48 1.29 2.96
6.02 1.36 4.03
30.6 100.4 178.3
5.44 0.86 10.43 13.02
7.32 0.99 6.96 18.15
10.25 1.63 4.52 16.28
88.4 89.9 −56.7 25.0
4.56 0.59 8.38 9.74
5.85 0.72 5.02 13.97
7.53 1.00 2.62 11.69
65.1 69.6 −68.7 20.0
39.90 24.07 55.16
38.61 22.33 53.81
52.05 31.93 71.79
30.4 32.7 30.1
31.87 19.43 43.86
30.21 16.24 43.26
40.55 27.28 53.59
27.3 40.4 22.2
1.85 64.60 97.22 43.33
0.92 69.91 80.03 43.58
1.42 78.64 113.00 66.30
−23.7 21.7 16.2 53.0
0.75 51.67 79.41 34.45
0.52 55.23 66.45 30.16
0.37 61.39 92.85 48.51
−51.2 18.8 16.9 40.8
42.05 38.11 14.17
39.03 46.85 18.80
51.72 64.34 33.78
23.0 68.8 138.4
33.83 31.22 6.27
30.43 37.49 14.79
40.70 48.55 23.80
20.3 55.5 279.6
33.07 39.33 44.17 98.17
33.17 33.50 49.22 53.84
47.90 45.72 48.33 127.94
44.8 16.3 9.4 30.3
26.75 24.98 35.52 72.93
25.59 22.00 40.14 38.49
35.95 25.24 39.53 102.64
34.4 1.0 11.3 40.7
+ drug were psychiatric comorbidities, particularly mood and anxiety disorders (Table 3). Conduct disorder was not significantly associated with diagnosis or treatment for adults, but among youths, multiplied the odds of diagnosis by a factor of 4.53 (95% confidence interval [CI] = [3.07, 6.68]) and diagnosis + drug by a factor of 3.31 (95% CI = [2.2, 4.98]). For both adults and children, the interaction of a mood or anxiety disorder with female sex increased the odds of ADHD diagnosis and treatment by factors ranging from
2012-2013
Change T1 to T3 (%)
2008-2009
approximately 1.7 to 2.5. Additional significant interactions for adults included male sex with Medicaid enrollment (odds ratio [OR] for diagnosis = 2.09, 95% CI = [1.42, 3.08]) and Black race with anxiety/depression (OR for diagnosis = 1.98, 95% CI = [1.11, 3.55]). Among youths, the interaction of Black race with a conduct disorder increased the odds of diagnosis + drug (OR = 3.78, 95% CI = [1.39, 10.32]). For ADHD diagnosis, this interaction effect trended in the positive direction but was not statistically significant.
6
Journal of Attention Disorders
Table 3. Logistic Regressions of ADHD Diagnosis and Treatment on Demographic and Clinical Predictors. Youth, ADHD diagnosis
Adult, ADHD diagnosis
Youth, ADHD diagnosis + drug
Adult, ADHD diagnosis + drug
45,572 3,960*** .246 0.841
195,366 3,507*** .210 0.883
45,572 3,093*** .226 0.840
195,366 2,579*** .196 0.884
Number of casesa Model chi-square Nagelkerke R2 C statistic Predictors
Exp(B)
95% CI
Exp(B)
95% CI
Exp(B)
95% CI
Exp(B)
95% CI
2008-2011b 2012-2013 Age
research-article2017
JADXXX10.1177/1087054716688534Journal of Attention DisordersFairman et al.
Article
Diagnosis and Treatment of ADHD in the United States: Update by Gender and Race
Journal of Attention Disorders 1–10 © The Author(s) 2017 Reprints and permissions: sagepub.com/journalsPermissions.nav https://doi.org/10.1177/1087054716688534 DOI: 10.1177/1087054716688534 journals.sagepub.com/home/jad
Kathleen A. Fairman1, Alyssa M. Peckham1, and David A. Sclar1
Abstract Objective: The aim of this article is to update ADHD diagnosis/treatment trends by age, gender, and race. Method: National Ambulatory Medical Care Survey data were obtained for 2008-2009 to 2012-2013. Physician office visits including ADHD diagnosis and pharmacotherapy were measured per 1,000 population and per 1,000 office visits overall, and by demographic group. Logistic regression models controlled for demographics, psychiatric comorbidities, insurance type, and time period. Interactions of time, demographics, comorbidities, and insurance type were tested. Results: Diagnoses of ADHD increased by 36% in adults and 18% in youth, and diagnosis + drug by 29% in female and 10% in male youths. ADHD diagnosis was 77% less likely among Black than White adults but 24% more likely among Black than White youths in 2012-2013. Conduct disorder (CD) in youths multiplied odds of diagnosis + drug by 3.31; interaction of Black race × CD by 3.78. Conclusion: Upward trends in ADHD diagnosis and treatment have continued but vary markedly by group. Studies of undertreatment/overtreatment are needed. (J. of Att. Dis. XXXX; XX(X) XX-XX) Keywords ADHD, gender, epidemiology, race, prescription prevalence
Introduction Over a period of about two decades beginning in 1990, substantial upward trends in the rates of diagnosis and treatment for ADHD were observed in nationally representative samples of U.S. physician office visits obtained through the National Ambulatory Medical Care Survey (NAMCS; Olfson, Blanco, Wang, & Greenhill, 2013; Robison, Sclar, & Skaer, 2005; Robison, Skaer, & Sclar, 2004; Robison, Skaer, Sclar, & Galin, 2002; Sclar, Robison, Bowen, et al., 2012; Sclar, Robison, Castillo, et al., 2012). Since the most recent time periods studied, 2006-2009 for adults (Olfson et al., 2013) and 2007-2008 for children and youth (Sclar, Robison, Bowen, et al., 2012), several important developments have taken place. These include the approval of several new ADHD drugs and formulations (CenterWatch, n.d.); the formation of the American Psychiatric Association (APA) Diagnostic and Statistical Manual of Mental Disorders–5 Task Force in 2007, culminating in the publication of Diagnostic and Statistical Manual of Mental Disorders (DSM-5; 5th ed.; APA, 2013, n.d.); and updated guidelines, from the American Academy of Pediatrics (AAP) in 2011 (Wolraich et al., 2011) and from the American Academy of Family Practice in 2012 (adults) and 2014 (children and teens; Felt, Biermann, Christiner, Kochhar, & Harrison, 2014; Post & Kurlansik, 2012).
These changes may have increased the number of cases diagnosed and treated for ADHD in several ways. First, the age criteria for diagnosis broadened, both in the DSM-5 (symptoms no later than age 12 years vs. age 6 years in the Diagnostic and Statistical Manual of Mental Disorders [4th ed.; DSM-IV]; APA, 1994) and in the AAP guidelines (age 4-18 years in 2011 vs. 6-12 years in 2001; APA, 2013; Wolraich et al., 2011). Second, the DSM-5 guidelines require adults and adolescents to display only five ADHD symptoms, rather than the six required for a diagnosis in children (APA, 2013). Finally, a description of ADHD symptomatology in those aged 17 years or older was added to the DSM-5 for clarity (APA, 2013). Given this expansion of diagnostic criteria and available pharmacotherapies for ADHD, the present study was conducted to provide an update and reassessment of three previously observed demographic trends: 1
Midwestern University, Glendale, AZ, USA
Corresponding Author: Alyssa M. Peckham, Department of Pharmacy Practice, College of Pharmacy-Glendale, Midwestern University, 19555 N. 59th Avenue, Glendale, AZ 85308, USA. Email: [email protected]
2
Journal of Attention Disorders 1. Large increases in ADHD diagnosis and treatment among adults. From 1995-1996 to 2007-2008, the number of office visits at which an ADHD diagnosis was made increased from 3.1 to 14.5 per 1,000 U.S. adults (aged 20 years or older; Sclar, Robison, Castillo, et al., 2012). For the same time span and group, office visits including both ADHD diagnosis and medication increased from 1.9 to 11.4 per 1,000 (Sclar, Robison, Castillo, et al., 2012). Similarly, the percentage of visits in which stimulants were prescribed to adults (aged 18 years or older) diagnosed with ADHD increased from 52% in 1994-1997 to 67% in 2006-2009 (Olfson et al., 2013). 2. Greater rates of increase in ADHD diagnosis and treatment among female than male children and teens (aged 5 to 18 years). From 1991-1992 to 2007-2008, rates of ADHD diagnosis increased 3.7fold for boys (from 39.5 to 144.6 per 1,000 population) and 5.6-fold for girls (from 12.3 to 68.5 per 1,000 population; Sclar, Robison, Bowen, et al., 2012). In the same time period, rates of ADHD diagnosis coupled with drug treatment increased 4.4fold for boys and 6.5-fold for girls (Sclar, Robison, Bowen, et al., 2012). 3. Lower rates of diagnosis and treatment for nonWhite than White populations. Several studies conducted in the 1990s and early 2000s found that non-White youths were less likely than White youths to receive health care services for mental disorders, including ADHD (Coker et al., 2016; Kataoka, Zhang, & Wells, 2002; Zimmerman, 2005). Similarly in 2006-2009, 0.68% of office visits made by White adults, compared with 0.36% for non-White adults, included the prescribing of a stimulant medication (Olfson et al., 2013).
Method Data Source Study data were drawn from the NAMCS, a cross-sectional, nationally representative survey of U.S. physician office visits that is conducted by the National Center for Health Statistics (NCHS) and has been used extensively in studies of ADHD treatment trends (Olfson et al., 2013; Robison et al., 2005; Robison et al., 2004; Robison et al., 2002; Sclar, Robison, Bowen, et al., 2012; Sclar, Robison, Castillo, et al., 2012). Survey data are collected annually by the U.S. Bureau of the Census using a three-stage probability sampling procedure that has been described in detail elsewhere (U.S. Centers for Disease Control and Prevention, [NCHS], 2010a, 2010b). Briefly, the NAMCS unit of analysis is an individual office visit. Sampling is conducted first by primary sampling units (PSUs), which are geographic
areas comprising counties, county groups, or geographic equivalents; then within PSU, by physician name based on master lists maintained by the American Medical Association and the American Osteopathic Association, stratified by physician specialty. Each selected physician is randomly assigned to 1 of the 52 weeks of data collection within the survey year, and within each reporting week, a systematic random sample of office visits is made. Visits represent office-based patient care, and contacts made outside the office (e.g., by telephone, in-home) or for administrative purposes only (e.g., medication refill) are excluded. For each recorded office visit, the NAMCS data file includes a sampling weight that adjusts for the multistage (i.e., clustered) sampling design and for survey nonresponse. Application of the weights to the data yields nationally representative information for all community-based physician office visits in the United States. In accordance with NAMCS recommendations for use of the data and with the methodology of previous studies using this data set, visits were grouped into multiyear increments for the present study analyses to increase the reliability of the weighted estimates (Olfson et al., 2013; Robison et al., 2005; Robison et al., 2004; Robison et al., 2002; Sclar, Robison, Bowen, et al., 2012; Sclar, Robison, Castillo, et al., 2012). Three time periods were studied: 2008-2009, 2010-2011, and 2012-2013.
Data Elements Counts of visits in which a diagnosis of ADHD was made were based on International Classification of Diseases– Ninth Revision (ICD-9) codes of 314.0x to 314.9x in any of the three diagnosis fields available in the NAMCS data set (Olfson et al., 2013). Identification of the prescribing of ADHD medication was made using the NAMCS drug identifiers, available for up to eight drugs continued or newly prescribed during the office visit, for generic and brand formulations of stimulant and nonstimulant medications to treat ADHD: amphetamines and amphetamine salt combinations, atomoxetine, dextroamphetamine, dexmethylphenidate, extended-release guanfacine or clonidine, lisdexamfetamine, and methylphenidate. Estimates measured whether an ADHD diagnosis had been made and whether it was accompanied by a prescription for ADHD medication (diagnosis + drug). Although sometimes used to treat ADHD in adults, especially those with comorbid psychiatric disorders, antidepressants were not counted as ADHD pharmacotherapy because they are not considered first-line treatment for ADHD (Post & Kurlansik, 2012).
Calculations Prevalence estimates were made using two different denominators, both of which have been used in previous research
3
Fairman et al. using the NAMCS (Olfson et al., 2013; Robison et al., 2005; Robison et al., 2004; Robison et al., 2002; Sclar, Robison, Bowen, et al., 2012; Sclar, Robison, Castillo, et al., 2012). First, overall and by sex, separately for adults (aged 20 years or older) and youth (aged 19 years or younger), estimates were calculated as annual number of visits per U.S. population, with annual averages calculated as the total number of visits for each 2-year period divided by 2 (Sclar, Robison, Bowen, et al., 2012; Sclar, Robison, Castillo, et al., 2012). Use of this method allowed for comparison of the present study’s results with those of previously published work using the same method, to produce updated trend estimates. Population estimates for each of the three time periods were obtained from U.S. Bureau of the Census figures for 2008, 2010, and 2012, respectively (“Age and sex composition”) (U.S. Department of Commerce, U.S. Census Bureau, n.d.). Second, estimates of the percentage of visits resulting in an ADHD diagnosis were calculated overall and by factors associated with ADHD diagnosis and treatment in previous work: sex, age group, race, and insurance type (Olfson et al., 2013). Sensitivity analyses compared trends determined using each denominator. Two changes from previous calculations made for samples of youth are notable. First is that, although children younger than 5 years were excluded from some previous samples, they were included in the present study sample, and their rates of drug therapy were assessed separately from those of older children. This change was made because recent guidelines recommend behavioral therapy as firstline treatment in this group (Felt et al., 2014; Wolraich et al., 2011). Second, teens aged 19 years were included in the youth cohort because of the discrepancy in definitions of an adult across different previous studies; specifically, those 19 years of age have been excluded from some studies of youth and were defined as adults in one previous study (Olfson et al., 2013; Sclar, Robison, Castillo, et al., 2012). Finally, because demographic differences in ADHD diagnosis and treatment may be confounded by relevant clinical characteristics, binary logistic regression models of diagnosis and of diagnosis + drug were estimated separately for youth and adults. Predictors included the aforementioned demographic characteristics, time period, and several mental disorders commonly associated with ADHD such as anxiety disorders (ICD-9 codes: 293.84, 300.0x, 300.2x, 300.3x, 300.7x, and 300.8x), mood disorders (293.83, 296.xx, 300.4x, 301.1x, and 311), substance addiction/abuse (291.xx, 292.xx, 303.xx, 304.xx, and 305.xx), conduct disorders (312.xx), and psychoses (293.81, 293.82, 295.xx, 297.xx, 298.xx, and 301.2x) (Felt et al., 2014; Olfson et al., 2013; Wolraich et al., 2011). For each outcome and group, modeling was performed using backward stepwise regression (minimum F probability for entry = .05, maximum F probability for removal = .10) in
three stages: (a) Test first-order (single factor) terms for age, sex, time period, race, insurance type, and the mental disorders. (b) To the resulting model, add and test terms representing the interaction of time period with each demographic group term. (3) To the resulting model, add and test terms representing interactions of race and sex with comorbid disorders and with insurance type. All analyses were performed at alpha of .05 using SPSS Version 23.0 (IBM SPSS, Armonk, New York).
Results The data set comprised 254,326 individual office visit records, representing approximate annual averages of 194 million visits for youth up to age 19 years and 780 million visits for adults aged 20 years or older from 2008 through 2013. A diagnosis of ADHD, with or without prescribed medication, was made at about 4.0 million adult and 7.7 million youth visits in 2008-2009; these numbers grew to 5.7 million and 9.1 million, respectively, in 2012-2013 (Table 1). Expressed as visits per 1,000 population, diagnoses of adult ADHD increased over the 6-year study period by 36%, and diagnoses coupled with pharmacotherapy (diagnosis + drug) increased by 21% (Table 1). These rates continued trends observed in previous studies using the same metrics (Figure 1). Rates of increase for youth similarly trended upward, albeit more slowly than for adults, at 18% and 16%, respectively (Table 1). Growth trends by gender differed markedly for the two age groups (Table 1). Among adults, rates of increase in diagnosis and diagnosis + drug for males were more than double those for females, respectively (52% vs. 23% for diagnosis, 30% vs. 13% for diagnosis + drug). In contrast, rates of diagnosis increased somewhat more for female than male youths (22% vs. 17%, respectively), and the rate of increase in diagnosis + drug for female youths was nearly triple than that for males (29% vs. 10%, respectively). Increases overall and by sex displayed similar trends whether measured per 1,000 population or per 1,000 office visits. However, the magnitudes of upward trend were generally somewhat greater in the latter analysis (Table 2). For each 1,000 office visits made by adults in 20122013, 7.62 included a diagnosis of ADHD and 5.48 included a diagnosis + drug (Table 2). Among youth, rates were about 7 times those of adults: 52.05 and 40.55 per 1,000 office visits for diagnosis and diagnosis + drug, respectively. For both adults and youth, absolute rates on both metrics were consistently lower for females than males. However, as in the analysis per 1,000 population, rates of growth were generally higher for male than female adults, especially for diagnosis, and higher for female than male youth, especially for diagnosis + drug.
4
Journal of Attention Disorders
Table 1. Number of ADHD Diagnosis and Pharmacotherapy Visits Per 1,000 Population by Sex and Age Group. ADHD diagnosis Adults aged ≥20 years Total unweighted number of visits Total weighted number of visitsa Rates per 1,000 population All adults Female adults Male adults Youth aged ≤19 years Total unweighted number of visits Total weighted number of visitsa Rates per 1,000 population All youth Female youth Male youth
ADHD diagnosis + drug
2008-2009
2010-2011
2012-2013
Change T1 to T3 (%)
2008-2009
2010-2011
2012-2013
311 4,019,154
369 4,886,182
867 5,723,614
— —
245 3,264,301
284 3,835,189
628 4,117,705
18.5 19.2 17.8
22.1 19.4 25.0
25.3 23.7 27.0
36.4 23.2 51.6
15.1 15.6 14.4
17.4 14.8 20.1
18.2 17.6 18.8
542 7,657,410
582 8,286,634
1,432 9,085,860
— —
428 6,115,333
454 6,484,622
1,108 7,079,461
93.1 56.3 128.3
99.6 56.9 140.6
110.3 68.5 150.3
18.5 21.7 17.1
74.3 45.5 102.0
77.9 41.4 113.0
86.0 58.5 112.2
Change T1 to T3 (%) — — 20.8 12.5 30.4 — — 15.6 28.8 10.0
Note. T = time period (T1 = 2008-2009; T3 = 2012-2013). a Annual averages, calculated as the total number of visits for each 2-year period divided by 2.
30
Present study findings
Findings of Sclar, Robinson, Castillo, et al. (2012)1 25
20
15
10
ADHD Diagnosis ADHD Diagnosis and Drug
5
0 1995-1996
1997-1998
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2008-2009
2010-2011
2012-2013
Figure 1. ADHD diagnosis and pharmacotherapy per 1,000 population: U.S. office visits for adults aged ≥20 years.
For older and non-White adults, rates of diagnosis and diagnosis + drug were notably low, despite a sharply increased trend over time (Table 2). Compared with visits made by White adults, those made by Black adults were 86% less likely to include an ADHD diagnosis and 85% less likely to include a diagnosis + drug in 2008-2009. Growth trends in these metrics over time were much greater for Black than for White adults; however, a large racial difference persisted in 2012-2013, when Blacks were 77% less likely to receive a diagnosis or diagnosis + drug. For youths, trends by race differed considerably from those observed for adults (Table 2). In 2008-2009, visits made by Black youths were somewhat less likely than
those of White youths to include an ADHD diagnosis or drug. However, over the 6-year study period, rates of growth in both metrics for Blacks were approximately 3 times those of Whites. Thus, in 2012-2013, visits made by Black youths were 24% more likely to include an ADHD diagnosis and 19% more likely to include a diagnosis + drug than were visits made by White youths. Youths of other race, predominantly Asian/Pacific Islander, were persistently less likely than White youths to receive an ADHD diagnosis or drug, despite growth rates far surpassing those for youths of other races. In logistic regression analyses, the factors most strongly and consistently predictive of ADHD diagnosis or diagnosis
5
Fairman et al. Table 2. Rates of ADHD Diagnosis and Pharmacotherapy Per 1,000 Office Visits by Demographic Group. ADHD diagnosis Adults aged ≥20 years All Female Male Age group (years) 20-29 30-39 40-49 50-59 60-64 65 or older Race White Black Other Payment source Private pay Medicare Medicaid Other Youth Aged ≤19 Years All Female Male Age group (years) Younger than 5 5-9 10-14 15-19 Race White Black Other Payment source Private pay Medicare Medicaid Other
2008-2009
2010-2011
2012-2013
4.99 4.32 6.08
6.24 4.67 8.62
7.62 6.16 9.81
17.86 9.36 8.37 2.78 1.65 0.20
29.06 8.12 6.25 4.57 2.90 0.44
5.70 0.81 1.45
ADHD diagnosis + drug Change T1 to T3 (%)
2010-2011
52.6 42.6 61.3
4.06 3.52 4.93
4.90 3.57 6.91
5.48 4.58 6.84
35.2 30.2 38.8
23.59 15.89 10.62 6.45 2.87 0.87
32.1 69.8 26.9 131.9 73.8 347.8
15.37 7.29 6.43 2.32 1.31 0.17
22.80 6.54 4.93 3.50 2.50 0.26
17.67 11.51 7.14 4.97 1.85 0.49
15.0 58.0 11.1 114.5 41.4 187.4
6.93 2.16 3.60
8.40 1.94 4.90
47.4 140.0 238.3
4.61 0.68 1.45
5.48 1.29 2.96
6.02 1.36 4.03
30.6 100.4 178.3
5.44 0.86 10.43 13.02
7.32 0.99 6.96 18.15
10.25 1.63 4.52 16.28
88.4 89.9 −56.7 25.0
4.56 0.59 8.38 9.74
5.85 0.72 5.02 13.97
7.53 1.00 2.62 11.69
65.1 69.6 −68.7 20.0
39.90 24.07 55.16
38.61 22.33 53.81
52.05 31.93 71.79
30.4 32.7 30.1
31.87 19.43 43.86
30.21 16.24 43.26
40.55 27.28 53.59
27.3 40.4 22.2
1.85 64.60 97.22 43.33
0.92 69.91 80.03 43.58
1.42 78.64 113.00 66.30
−23.7 21.7 16.2 53.0
0.75 51.67 79.41 34.45
0.52 55.23 66.45 30.16
0.37 61.39 92.85 48.51
−51.2 18.8 16.9 40.8
42.05 38.11 14.17
39.03 46.85 18.80
51.72 64.34 33.78
23.0 68.8 138.4
33.83 31.22 6.27
30.43 37.49 14.79
40.70 48.55 23.80
20.3 55.5 279.6
33.07 39.33 44.17 98.17
33.17 33.50 49.22 53.84
47.90 45.72 48.33 127.94
44.8 16.3 9.4 30.3
26.75 24.98 35.52 72.93
25.59 22.00 40.14 38.49
35.95 25.24 39.53 102.64
34.4 1.0 11.3 40.7
+ drug were psychiatric comorbidities, particularly mood and anxiety disorders (Table 3). Conduct disorder was not significantly associated with diagnosis or treatment for adults, but among youths, multiplied the odds of diagnosis by a factor of 4.53 (95% confidence interval [CI] = [3.07, 6.68]) and diagnosis + drug by a factor of 3.31 (95% CI = [2.2, 4.98]). For both adults and children, the interaction of a mood or anxiety disorder with female sex increased the odds of ADHD diagnosis and treatment by factors ranging from
2012-2013
Change T1 to T3 (%)
2008-2009
approximately 1.7 to 2.5. Additional significant interactions for adults included male sex with Medicaid enrollment (odds ratio [OR] for diagnosis = 2.09, 95% CI = [1.42, 3.08]) and Black race with anxiety/depression (OR for diagnosis = 1.98, 95% CI = [1.11, 3.55]). Among youths, the interaction of Black race with a conduct disorder increased the odds of diagnosis + drug (OR = 3.78, 95% CI = [1.39, 10.32]). For ADHD diagnosis, this interaction effect trended in the positive direction but was not statistically significant.
6
Journal of Attention Disorders
Table 3. Logistic Regressions of ADHD Diagnosis and Treatment on Demographic and Clinical Predictors. Youth, ADHD diagnosis
Adult, ADHD diagnosis
Youth, ADHD diagnosis + drug
Adult, ADHD diagnosis + drug
45,572 3,960*** .246 0.841
195,366 3,507*** .210 0.883
45,572 3,093*** .226 0.840
195,366 2,579*** .196 0.884
Number of casesa Model chi-square Nagelkerke R2 C statistic Predictors
Exp(B)
95% CI
Exp(B)
95% CI
Exp(B)
95% CI
Exp(B)
95% CI
2008-2011b 2012-2013 Age
