The strongest individual-level evidence

Most social-determinants claims in this area rest on ecological studies — county-level or state-level correlations, which cannot tell you about individuals. One study avoids that problem. The Mortality Disparities in American Communities Study linked roughly four million respondents to the 2008 American Community Survey to National Death Index records through 2015, and modeled opioid-overdose death at the individual level.

Characteristic	Adjusted hazard ratio for fatal opioid overdose (95% CI)
Disabled	2.80 (2.59–3.03)
Recently incarcerated	2.70 (1.91–3.81)
High school education only, vs a graduate degree	2.48 (2.00–3.06)
Unemployed	2.46 (2.17–2.79)
Widowed	2.44 (2.03–2.95)
Renting rather than owning	1.36 (1.25–1.48)
Income below the poverty line	1.36 (1.20–1.54)
Uninsured	1.30 (1.20–1.41)

Two features of that table deserve attention. The gradient is steeper for employment, disability and education than for income poverty as such — this is not simply about money. And two of the results run against the popular narrative: the study found higher risk in non-rural than rural areas (HR 1.46) and substantially higher risk among citizens than non-citizens (HR 4.62).

Childhood adversity

The Adverse Childhood Experiences study followed 8,613 adult members of a California health plan and found a graded relationship between the number of adverse childhood experience categories and later drug problems: compared with people reporting none, those reporting five or more had seven to ten times the odds of drug problems, addiction and parenteral drug use. The relationship held across four birth cohorts. The study is retrospective and self-reported, in an insured single-metropolitan population, and its outcome is illicit drug use broadly rather than opioid use disorder specifically.

Among people who already have opioid use disorder, adversity tracks severity. In 457 consecutive patients entering inpatient opioid detoxification, mean ACE score was 3.64, and each additional point was associated with beginning opioid use half a year earlier (−0.50 years, 95% CI −0.70 to −0.29) and higher odds of past-month injection and lifetime overdose.

Homelessness

A cohort of 60,092 adults served by a Boston homeless healthcare program, linked to state death records from 2003 to 2018, recorded 7,130 deaths of which 1,727 — one in four — were drug overdoses. Standardized overdose mortality was 278.9 per 100,000 person-years, roughly twelve times the Massachusetts adult rate, with opioids involved in 91.0%. Synthetic-opioid mortality in that population rose from 21.6 to 327.0 per 100,000 person-years between 2013 and 2018. It is a single city and a single program, and the magnitude is still hard to look at.

Where the evidence is genuinely contested

The “deaths of despair” framing — that rising mortality among middle-aged Americans reflects economic and social decline — is the most-cited and least-settled idea in this area. It deserves both sides.

• For: the original 2015 analysis documented rising midlife mortality among white non-Hispanic Americans driven by drugs, alcohol and suicide, against declining mortality in every comparable country.
• Against: subsequent work found that suicide and alcohol contributions were roughly flat for thirty years, and that drug deaths follow a period pattern — consistent with a changing drug supply — rather than the cohort pattern a distress explanation predicts.
• Quantified: one economist estimated that less than a tenth of the rise in drug fatality is explained by economic conditions, “quite possibly zero” once selection is allowed for — revising his own earlier finding in the process. His most recent work estimates that deteriorating mental health accounts for something like 9% to 29% of the mortality rise among prime-age white Americans.

So: the mortality trends are not in dispute. The despair mechanism is, and estimates of the economic contribution range from near zero to about a third depending on the design. Anyone presenting this as settled in either direction is ahead of the evidence.

The federal position is that this is an evidence gap rather than a body of findings. The National Academies’ 2017 report on pain management and the opioid epidemic recommended that NIDA and CDC fund research on the social determinants underpinning misuse and illicit use — a recommendation, not a conclusion. It is worth knowing that the chapter in that report titled “Risk Factors for Prescription Opioid Misuse and Overdose” is about product characteristics (compound, formulation, route), not social conditions. It is frequently mis-cited for the latter.

Why a program has to take account of this

Because it determines whether an intervention reaches anyone. A delivery program is worth little to someone with no stable address. A monitoring schedule assumes a phone and a place to charge it. A follow-up appointment assumes transport. A safe-storage requirement assumes a lockable space in a home the person controls. Screening for behavioral health comorbidity is not an add-on when the population is this one; it is the main event.

That is why social determinants verification is a program component rather than a note in a report, and why the federal Task Force put a biopsychosocial model at the head of its recommendations. See the Task Force recommendations.