- There is no local or national source that accurately quantifies drugs entering or circulating in terms of absolute volume. Overdose mortality data are a useful metric because it's the outcome that matters most to most people. But it is at best a proxy for the drug supply.
- Specific substances identified in mortality data are limited by what is assayed for and how they are named in death records.
- Death records in many, but not all, states are public information.
- Causal emphasis on what substance was involved in an overdose can vary by coroner or medical examiner, a longstanding problem that has made considerable strides.
- Death certificates are limited by what is measured and recorded. They often contain little/no mental health and social antecedents.
- We cannot systematically distinguish intentional vs. accidental polysubstance use from mortality data.
- Most models do not account for if people moved or died of other causes.
- ICD-10 coding using T-codes lumps illicitly manufactured fentanyl with pharmaceutical fentanyl.
- Most data dashboards include unintentional deaths and suicides.
- Presence of a substance in post-mortem body fluids doesn't automatically mean that that substance was involved in the fatal event.
- Stimulant-related toxicity has a lot of variation in how it is ascribed in a death.
- There is a 4-18 month delay in cause of death investigations, varying by state resources. Not all states have the same resources for these investigations.
- Coroner-based systems (in contrast to medical examiners) may have less specific causes of death. Some states have coroners, others do not.
- Incomplete death records are an issue, and missingness is not distributed uniformly.
- There are know differences between urban and rural areas, so a state's average is inherently heterogeneous. Same goes for age, sex, and race.
- Spikes in overdose happen, often based on street drug potency (at the current time). Single month comparisons need to be interpreted in this context.
- Declines in opioid overdose may be offset by increases in other types of substances.