News · Brain & Mental Health
Schools that kept a health class had less teen substance use
A JAMA Network Open study of 1.9 million California student surveys found schools sustaining a standalone health course had lower reported vaping, drinking and drug use.
Based on a peer-reviewed cohort study in JAMA Network Open
- Researchers linked California Healthy Kids Survey responses to university-approved course lists, publishing in JAMA Network Open.
- The sample covered 1,942,640 student responses from 915 public high schools across 345 districts.
- Students were in grades 9 and 11, surveyed between the 2017-18 and 2023-24 school years.
- The exposure was a school offering at least one standalone approved health course, counted as sustained when present two years running.
- Seven substance-use outcomes were tracked, all self-reported for the past 30 days.
- They covered alcohol, binge drinking, cigarettes, drugs generally, marijuana, prescription drug misuse and vaping.
- The design compares schools with each other before and after they gain or lose a course, which is stronger than a simple cross-section.
- Observational. Schools that sustain a health course differ from schools that do not in funding, staffing and priorities.
Health class is the lesson most often squeezed out of a crowded timetable. A study in JAMA Network Open, published by the American Medical Association, matched nearly two million student surveys against which California schools actually kept one.
Evidence is limited on whether sustained availability of dedicated school-based health coursework is associated with lower substance use prevalence at scale. This is the study that went looking.
Why the question had no answer
The problem it addresses is not in dispute. Adolescent substance use remains a leading preventable cause of morbidity and mortality in the US.
What was missing was evidence at the level decisions get made. Individual prevention programmes have been trialled repeatedly; whether simply having a health course on the timetable, year after year, shows up in the numbers had not been tested across a whole state.
Linking two datasets nobody usually joins
The design pairs a student survey with an administrative record. The study included California public high schools that participated in the California Healthy Kids Survey and were linked to annual University of California Office of the President course lists.
That linkage is the trick. One dataset says what students report doing; the other says, independently, whether their school ran an approved health course that year.
The exposure was school year availability of at least 1 standalone health course approved by the University of California Office of the President, counted as sustained when present for at least 2 consecutive school years.
The scale, and what was measured
The analytic sample included 1,942,640 student survey responses from 915 public high schools across 345 districts, covering students in grades 9 and 11 from the 2017-18 school year through 2023-24.
The outcomes were deliberately concrete: seven binary indicators of any past-30-day substance use, covering alcohol use, binge drinking, cigarette smoking, drug use, marijuana use, prescription drug misuse, and vaping.
Seven separate outcomes rather than one composite means a result cannot be manufactured by burying a null in an index.
The comparison that does the work
A simple correlation here would be worthless: schools with health courses differ from schools without them in every way that matters.
So the analysis used difference-in-differences with 2-way fixed effects and a prespecified event-study approach. Each school is compared against itself before and after the course appears or disappears, and against other schools over the same period.
Prespecified matters too. The analysis plan was fixed before the results were seen, which limits the room to search for a flattering specification.
What school-level data cannot show
Nobody randomized a curriculum. A district that sustains a standalone health course is a district with the money, staffing and inclination to do so, and those things predict student outcomes on their own. Fixed effects remove the stable part of that; they do not remove whatever else changed when the course arrived.
The outcomes are also self-reported by teenagers, with all the under-reporting that implies.
What the study offers is a policy-level association measured at a scale that policy actually operates on, which is rarer than it should be.
People also ask
What is a difference-in-differences design?
It compares how an outcome changes over time in schools that gained a health course against how it changed in schools that did not. Because each school acts partly as its own control, anything stable about a school - its neighbourhood, its intake, its culture - cancels out. What it cannot handle is something that changed at the same time as the course arrived.
Why does 'sustained' matter?
A course offered once may reach one cohort and vanish. The study defined availability as sustained when present for at least 2 consecutive school years, which is closer to what a school policy actually looks like and less likely to be a scheduling accident. It also makes the exposure something a district can decide to do.
Are self-reported figures reliable for teenagers?
Imperfectly, and that is inherent to this kind of survey. Students may under-report, and the direction of that error can vary by school climate. The saving grace is scale and consistency: the same anonymous instrument across 915 schools and seven years means bias would have to change systematically alongside course availability to manufacture the result.
Does this prove health classes reduce substance use?
No. Schools that sustain a standalone health course tend to have more funding, more staffing slack and different priorities, all of which independently predict student outcomes. The design removes the stable part of that, not the part that changes when a district decides to invest. It is evidence for the policy being worth testing properly, not evidence that it works.
Why does this matter now?
Because health courses compete for timetable space and are often the first thing cut. The authors note that adolescent substance use remains a leading preventable cause of morbidity and mortality in the US, and that evidence at this scale on whether school coursework helps has been limited. This is the kind of study a district can actually act on.