Ease off the alarm bells: New data on ADHD diagnosis rates

The New York Times has a cover story today reporting on the estimated prevalence of Attention-Deficit/Hyperactivity Disorder from the 2011-2012 National Survey of Children’s Health (they don’t identify the survey by name).

The story is going to get a lot of people interested in what is happening to children — every new datapoint on ADHD is noteworthy because it allows journalists to reopen the black box on childhood behavioral health disorders, and to raise the perennial alarm bells about over-diagnosis of children.

All of the issues raised in the article are valid. Many children with very mild impairments are getting a diagnosis, and enterprising drug companies are increasing demand for their product by implying that ADHD medications are a cure for generalized social impairments.

But — and this is critical — we have little systematic population-level data to compare the reported prevalence of a diagnosis with underlying data on ADHD symptoms in children. The NSCH, for example, does not include a validated set of questions measuring ADHD symptoms (like the Strengths and Difficulties questionnaire). This is a major shortcoming, because it leaves us to speculate about how much the diagnosis data reflect real underlying changes in symptoms in children, versus more prescription-pad-happy diagnosis from doctors and therapists.

For example, the article quotes William Graf at Yale: “Mild symptoms are being diagnosed so readily, which goes well beyond the disorder and beyond the zone of ambiguity to pure enhancement of children who are otherwise healthy.”

Dr. Graf is undoubtedly right, but nobody can say with certainty how much the diagnosis patterns may also be driven by alarming changes in the profile of environmental toxicants, exposure to over-stimulating television for toddlers, or disruptive family environments (since 2007, we also went through four years of major economic turmoil, which is disruptive to children and their parents). Again, this is speculative, but we need to get better answers to these and other potential exposures.

Second, and also critical, the NSCH numbers reported in the NYT focus on lifetime diagnosis rather than current diagnosis. This is a less accurate way to measure changes across successive cohorts, because it tangles up two factors — how doctors are diagnosing now, and how they diagnosed at earlier periods in time. Perhaps not surprisingly, the largest growth in lifetime ADHD diagnosis occurred among high school aged, not preschool-aged, youth. High school kids today may be carrying around a diagnosis that was given to them when they were 8 or 10 years old, but may no longer have symptoms of ADHD.

Third, although the lifetime diagnosis rates are growing, the rate of growth is slowing. Check out the CDC report from the 2003 and 2007 data. The diagnosis grew about 21.8% between those two survey waves compared to the 16% reported between 2007 and 2011. Again, these are concerning numbers, but I would have expected the lifetime diagnosis rate to have grown even if the current diagnosis rate did not increase, because kids diagnosed during the peak period (elementary school) are still carrying their old diagnoses in high school age.

We need to work much more with these data to understand what is happening, but let’s drop the panic level for a moment while we do. We still don’t know whether the ADHD epidemic is still peaking or has begun to steady out, and we also don’t have a great handle on what biological and social exposures may be driving ADHD symptoms over time (and whether these exposures are rising or falling).

7 responses to “Ease off the alarm bells: New data on ADHD diagnosis rates”

  1. I’d be interested to know how what the socio-economic background of these children shows. How many are from “poor” families? I feel doctors are way too quick to diagnose ADHD and put children on meds when the issue is their learned behaviour from their families of origin.

  2. Thanks for that comment Angel. You can see a slight SES gradient in diagnosis rates if you check out that older CDC report that I reference in the blog post (for example, children in families below poverty have 1.3 times the rate of diagnosis as children in families with income >200% of the poverty line). The DSM requires that a child display impairment in multiple settings (school, home, etc.) in order to receive a diagnosis and these symptoms need to persist over an extended period of time, and so a skilled diagnostician should be able to identify kids that are acting out because of stress in the family. Of course, the reality is more like what you describe. Kids get diagnosed when it’s not clear if their behavior reflects family and social stress. It’s really complicated, however, because some of those factors may actually trigger neurological changes in young children that lead to the development of ADHD (and not just ADHD symptoms associated with other problems). As far as I know, nobody has a good grip on this issue (but I’d love to hear from people with more psych/neuro training).

  3. For those wanting to layer into this story, I encourage you to check out a website that will allow you to search specific information on ADHD prevalence across states and by many population subgroups (www.childhealthdata.org). The staff of this website can assist with further analysis as well.

    Critical to consider in this dialogue is the lower prevalence of ADHD we arrive at if ONLY children reported to “Currently Have ADHD” are counted in assessing prevalence. Here, prevalence for children 4-17 is 9.4% (compared to the 11.4% for “Ever Had”).

    If we further identify those who “currently have ADHD” and also meet criteria for having a special health care need (using a validated tool), the rate is further reduced to 8.2%. Using this filter is commonly done as a way to attenuate known variations and uncertainties in diagnosis.

    A few key facts to consider:

    1. Regardless of the methods used to estimate prevalence, the demographic profile at the national level appears to remain stable (more boys than girls, more older vs. younger children, common across household income levels, common across type of insurance –public vs. private sector). This means the populations to focus on are not altered by the prevalence estimate we select.

    2. Let’s compare! Group A: children/youth with “current ADHD” and who experience above routine health care needs (assessed using a validated tool to; Group B: those whose parents reported that they “ever had” ADHD but do not currently have ADHD or those who currently have ADHD but do not experience special/above routine health care needs.
    When we do this, not only does the prevalence estimate fall by 28% to 8.2% (from 11.4%), we also see dramatic differences in what these children need and experience in their day to day lives.

    Specifically, children/youth in Group A are over three times more likely to take medications and are much more likely to miss a great deal of school, also have another chronic condition, fail to be engaged in school and have had to repeat at least one grade in school and have parents reporting high levels of parenting stress.

    These few facts (and there is much more where this came from) lead me to conclude that a more complex estimate of prevalence is warranted and possible using the 2011-12 NSCH. While the full strengths and difficulties questionnaire is not included, MANY variables are included that allow us to arrive at a more reliable estimate that does identify a group of children and youth experiencing a great many difficulties indicative of ADHD.

    About the methods: The NSCH employs the standard method in the US to assess prevalence of health conditions in the US population (similar to the longstanding National Health Interview Survey).

    Again, when we enrich the method as described above, the prevalence rate of ADHD for children/youth age 4-17 is 8.2% –which is in fact higher than the same rate in the 2007 NSCH, which was 6.0%.

    Nonetheless, it is most important to identify children according to their health related needs and consequences, regardless of their diagnosis–especially given the vagaries of diagnosis and wacky incentives that can sometimes kick in to assign diagnoses! In this way we can be sure to focus on what children need and experience and not get overly focused on diagnoses–this is especially important given the majority of children with any one health condition (like current ADHD) also have at least one, if not more, chronic conditions and there is often as much variation among children with any one diagnosis as between children with different diagnoses in terms of health care needs and impacts!

    When we do this and identify all children with special health care needs (CSHCN), we find that about 1 in 5 US children are identified as CSHCN. About a third of all CSHCN have ADHD reported as one of the often multiple conditions the child experiences. Over 60% of CSHCN have more complex health service needs and all CSHCN are much more likely to have difficulties in daily life (social, emotional, activities) compared to non-CSHCN. I suggest we consider this method as well as diagnosis–or at least in combination with it!

    Christina Bethell
    Professor, Department of Pediatrics
    Oregon Health & Science University
    Director, The Child and Adolescent Health Measurement Initiative

  4. Hi Christina,

    I’m really grateful for your comment (I’ll follow up with you more offline). You make a few critical observations which I want to echo:

    1. The NSCH allows for a much more subtle analysis than the unadjusted prevalence of a lifetime ADHD diagnosis captures.
    2. Adding in the special needs qualification allows us to get a better sense of ADHD+serious impairment. That’s a critical group to consider, but from an epidemiological and policy perspective we should also be interested in the less severe population, because they are still exposed to psychiatric medications, and there may be some social consequences to their ADHD (think of all the high functioning kids that still meet ADHD criteria).
    3. I should note that NSCH asks parents to rate the severity of the child’s ADHD, but I don’t know how validated that measure is.
    4. If we want to estimate the incidence of ADHD, the NSCH is still not the best instrument to use. We really need to systematically compare successive birth cohorts. That is possible with the ECLS panels, but I have not seen those data yet.

    Anyway, I really appreciate your comment — you have a great deal of expertise on this topic — and it’s wonderful to see how much progress you’ve made with the NSCH data.



    • Thank you for this response and I look forward to continued dialogue. It is true that about 21% of those with current ADHD and who do not meet criteria for having a special health care need (as well as low evidence of daily impact) nonetheless that medications. Ofcourse that is a potential circular issue! Much to explore and mostly just want to be sure we focus on needs, consequences and outcomes–regardless of what we do around diagnoses! My guess is is we focus on health and the child and family, we will employ much more low-tech interventions and achieve better outcomes. Mind body methods are especially promising for ADHD–thanks for keeping the dialogue alive! Christy

  5. Having read this I believed it was very enlightening. I appreciate you taking the time and energy to put this article together.
    I once again find myself spending a significant amount of time both reading and leaving comments.

    But so what, it was still worthwhile!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: