Recent searches
Search options
Administered by:
New paper: An integrated abundance model for estimating county-level prevalence of opioid misuse in Ohio
Work co-led by Staci Hepler & I #WakeForest with Lance Waller #Emory, Andrea Bonny & Erin McKnight #NationwideChildrens
Our research developed an indirect method for estimating population size by integrating multiple aggregate data sources on different levels of spatio-temporal support.
https://doi.org/10.1093/jrsssa/qnac013
#Statistics #Epidemiology #PublicHealth #Opioid
/1
#Opioid misuse is a major #PublicHealth challenge, but direct information on local prevalence is lacking. This is important for allocation of resources and interventions so they are proportionate to the size of the problem.
In Massachusetts, @jabarocas and colleagues indirectly estimated prevalence of opioid misuse using multiple individually linked data sets and capture-recapture methods. In general, access to individually linked data is limited.
/2
We build on methods for #Bayesian abundance modeling, which is a technique used in ecology for estimating hidden population size without identifying individuals and is conceptually similar to capture-recapture. We extend this technique for indirectly estimating population size in a #PublicHealth application and overcome model identification challenges through integration of multiple data sources rather than through replication as is common in ecology.
/3
Our study uses annual county-level counts of #opioid involved #overdose deaths and OUD treatment admissions as imperfect markers of opioid misuse. That is, they reflect a partial count of the hidden population of people who misuse opioids. We also incorporate multi-year NSDUH survey estimates of state-level prevalence to help with identification of model parameters. To characterize the social environment, we use 1- and 5-year county-level estimates from the American Community Survey.
/4
Using our approach, we estimate prevalence of opioid misuse for each of Ohio’s 88 counties annually from 2008-2019. At the peak, prevalence of opioid misuse was estimated to be nearly 13% in the most affected counties. These counties tended to in southern Ohio and the peaks tend to be around 2010-2011. We also see that most counties with prevalence estimates above the state average are in southern Ohio. Prevalence generally decreased over the last 5 years studied.
/5
An added bonus is that we are able to estimate rates of treatment admissions & overdose deaths among those who misuse opioids. Studies typically estimate these rates relative to the total population.
For example, overdose deaths can increase if misuse prevalence goes up & the death rate is constant or if prevalence is constant & the death rate increases. With a total population denominator, we can’t untangle which situation is occurring and that may have policy implications.
/6
Estimated prevalence is not increasing over the last 5 years of the study, but the death rate among people who misuse is increasing - suggesting observed increases in overdose deaths are tied to a more deadly supply & not increasing misuse.
While prevention policies may have helped reduce prevalence, expanded harm reduction policies are needed to try to reduce the death rate among people who misuse opioids. This is particularly true in the state’s urban areas.
/7
Looking at the treatment rates among those who misuse, we see increasing rates over time, particularly in southern Ohio. We also note slight changes in the trends associated with the allocation of state funds to expand treatment in 2010 and in 2014 when the state expanded Medicaid. However, these are anecdotal observations that were not causally evaluated in this work.
/end