A Note on a Debate about the Opioid Epidemic and Adolescent Mental Health
Jean Twenge offers good insights into parental drug abuse matters but does not fully invalidate the plausibility of substantial impacts on adolescent mental health.
Recent estimates of the number of U.S. children who lost a parent to drug overdose have reinvigorated attention to the role of adult drug abuse in the rise of mental health problems among adolescents. In particular, critics of Jon Haidt and Jean Twenge have pointed to the opioid epidemic as one of the alternative culprits to smartphones and social media.
Psychologist Jean Twenge addressed a few of these critics yesterday, and while her response is illuminating and persuasive in some respects, it is also problematic and tenuous in other aspects of the issue.
Before I start with my critique, let me note that Jean Twenge as well as Jon Haidt have been exemplary in the amount of time and effort they invest in answering their critics.
This is seldom seen in social sciences.
The Wrong Question
The first problem with the response by Twenge is already present in her title:
Parent drug overdoses: The true cause of the adolescent mental health crisis?
Do any credible psychologists assert that fatal overdose increases caused orders of magnitude larger increases in depression?
Twenge easily disposes of the question posed by herself with Figure 1, designed to illustrate the massive gap between the frequencies of overdoses and depression.
Primary vs. Substantial Causes
The problem with the title is indicative of a tendency with Twenge to focus only on ‘primary causes’ instead of substantial causes. As I explained last year (see The Fallacy of Requiring a Single Cause), there may be no single factor that explains the majority of the rise in depression — it could be a confluence of three to four major factors whose comparative impacts fluctuate from year to year.
To be fair to Twenge, she does continue on to explore more plausible mechanisms in her post, though she does not clearly articulate these theories.
Parental Drug Abuse
The key argument by Twenge is that use of illicit drugs other than marijuana has barely budged among 35- to 49-year-old parents, per NSDUH self-reports by parents. Twenge should add younger and older age groups to present the full picture, and if the pattern holds, than it will be compelling evidence. This still leaves the issue, however, of a rise in legal prescription pills abuse.
Twenge is less persuasive when she declares:
The image below is the trend in the estimated number of children impacted by parental drug overdose by race and ethnicity in the Jones paper. That’s followed by Figure 3 showing the increase in teen depression by race and ethnicity using the same colors. The pattern is completely different.
I do not see these patterns as being much different at all. Twenge needs to provide robust trend calculations instead of presuming that everyone agrees on visual impressions.
Twenge points to Native Americans, but that is a group too small and exceptional to use as the basis for dismissal of theories about the entire U.S. population.
Suicide versus Depression
Twenge conflates two related but fundamentally distinct theories:
Adult drug abuse rise was a substantial cause of adolescent depression rise.
Adult overdoses were a substantial cause of adolescent suicide rise.
This conflation occurs, for example, when Twenge examines regional differences.
Regional Differences
Twenge argues:
Adolescent suicide rates have increased across all three types of states. There is variation across regions, but most of that divide is around red vs. blue rather than low vs. high increases in adult mortality.
First of all, Twenge does not quantify the differences between low vs. high states — if they are small, why should this have much of an impact?
Also, the OH/IN/KT population is fairly small (~20 mil), not highly robust statistically.
Furthermore, this may be relevant to suicide, but for depression one needs drug abuse in general, not overdoses.
Note that overdoses may be higher in rural areas despite similar rates of drug abuse — in urban population health emergencies may be noticed by others sooner and ER is closer.
Adult Depression and Stress
Twenge also addresses speculation by Chris Ferguson that adolescent depression rose due to the deterioration of parental mental health by showing that middle-age depression remained flat and stress did not increase until the pandemic.
Twenge is persuasive in her dismissal of Ferguson’s theory but this is an issue distinct from drug abuse and perhaps deserves a dedicated article.
Childhood Trauma
Twenge needs to consider cumulative influence of both drug abuse and, especially, of drug overdoses. The percentage of teens affected by parental overdose during childhood is an order of magnitude larger than the percentage affected by it in any particular year.
Furthermore, the childhood trauma trend might be quite different from the year-to-year trend.
As I’ve shown back in 2020 (see Youth Suicide Rise), year-to-year adult suicide is a poor predictor of concurrent teen suicide rates, but a 7-year cumulative trend in adult suicide is a very good predictor. In other words increases in parental suicide during childhood might produce delayed increases in adolescent suicide risk.
I’ve not expanded my analysis to overdoses, but I plan to do so in the future.
If childhood exposed to parental overdose helps explain teen suicide, then it would fit into a larger ‘childhood trauma’ explanation of adolescent suicide trends and, specifically, support the notion that ‘abandonment deaths’ (mainly suicides and overdoses) have a particularly strong influence.
Miscellaneous Issues
Twenge needs to provide more robust and complete evidence regarding parental drug abuse and regional differences.
Twenge needs to give specific links to data sources, not merely refer to ‘Behavioral Risk Factor Surveillance System’ or ‘NSDUH’.
Comparing relative rises between low and high rates visually on the same graph is difficult; Twenge needs to provide robust calculations such as percentage diffs between 3-year intervals.
Conclusion
Twenge provides some good insights and persuasive rebuttals, but she needs to address distinctions between suicide and depression theories, between primary and major causes, and she needs to strengthen her evidence and clarify her arguments.
David Stein raises good points and makes big mistakes. The worst: “Do any credible psychologists assert that overdose increases caused orders of magnitude larger increases in depression? Twenge easily disposes of the question posed by herself with Figure 1, designed to illustrate the massive gap between the frequencies of overdoses and depression.”
That’s wrong. Twenge’s oranges-apples method error comparing general self-reported attitudes versus specified vital statistics deaths is obvious. For overdose frequency, she cites only fatal overdoses afflicting parents – an arbitrarily small numerator. Why doesn’t she cite the 5.3 million drug-related hospital ER cases in 2022 among ages 26-64 that SAMHSA estimated? Wouldn’t that massive, rising number (tripling since 2010) depress more teens?
Or, what about the 25% of teens, some 10 million, estimated to live with a drug/alcohol-abusing parent? Or soaring drug/alcohol abuse by non-parent grownups who strongly affects teens – like parents’ partners, other household adults, close relatives, teachers, coaches, etc.?
As Stein points out (very important), adult and teen suicide trends need not precisely coincide for adult trends to influence teen trends. From 2000 to 2012, crucial growing-up years for Gen Z, age-30-59 suicides surged from 16,000 to 23,000 per year, and drug overdose deaths leaped from 13,000 to 30,000. Yet, no one mentions that.
If we adopt Twenge’s logic, “teen suicide” is too trivial even to think about. Of 10 million girls ages 10-14, just 240 committed suicide in the worst year, 0.000024 by proportion, far below older ages’ rates. How, then, could any credible psychologist assert that young girls’ below-negligible suicide is of any importance? (And if it is, why are the 2,500 to 3,000 annual suicides in each of Haidt’s, Stein’s, Murtha's, and my older-male cohorts ignored?)
Of course, deaths can be measurable iceberg-tips of larger crises, like middle-aged drug abuse levels 100 to 200 times more widespread – and affecting massively more teens – than just overdose fatalities.
Twenge and Haidt admit social-media screen time’s raw correlation with girls’ depression is small (r-value of 0.10 to 0.15 at most). That means social media is associated with just 1% to 2% (R-squared value) of girls’ depression and zero effect on girls’ suicide – too trivial to bother with, and certainly no cause of the 12-point increase in teen depression. Yet, in this case, Twenge argues for legislative obsession with tiny effects.
Further, why do Twenge, Haidt, et al even care – amid today’s mammoth drug crisis most afflicting millions of middle agers – that middle-agers check boxes on pencil-and-paper surveys declaring they feel just fine mentally and don’t do drugs? Mental illness and addiction are deeply stigmatized in American culture. Twenge knows that.
We don’t even know what teens mean by “depression.” We never ask them, despite serious contradictions. Liberal teens must mean something different by their greater “depression” than conservatives, as manifest by the much higher teen suicide rates and worse teen suicide trends in conservative America (Republican-voting counties in Republican-run states) than in liberal America (Democrat-voting counties in Democrat-run states) even after race and gender are controlled.
Frequently online teens report more depression but less suicide and self-harm than do teens who are rarely/never online – which makes the anti-social-media crusade hyping “teen suicide” downright fraudulent.
I appreciate Jean Twenge exploring some troublesome areas, but devastating contradictions require much more explanation. If interested in my larger critique, see: https://mikemales.substack.com/p/how-jean-twenge-et-al-get-the-middle