Methodological approaches for the prediction of opioid use-related epidemics in the United States: a narrative review and cross-disciplinary call to action

Abstract

The opioid crisis in the United States has been defined by waves of drug- and locality-specific Opioid use-Related Epidemics (OREs) of overdose and bloodborne infections, among a range of health harms. The ability to identify localities at risk of such OREs, and better yet, to predict which ones will experience them, holds the potential to mitigate further morbidity and mortality. This narrative review was conducted to identify and describe quantitative approaches aimed at the "risk assessment," "detection" or "prediction" of OREs in the United States. We implemented a PubMed search composed of the: (1) objective (eg, prediction), (2) epidemiologic outcome (eg, outbreak), (3) underlying cause (ie, opioid use), (4) health outcome (eg, overdose, HIV), (5) location (ie, US). In total, 46 studies were included, and the following information extracted: discipline, objective, health outcome, drug/substance type, geographic region/unit of analysis, and data sources. Studies identified relied on clinical, epidemiological, behavioral and drug markets surveillance and applied a range of methods including statistical regression, geospatial analyses, dynamic modeling, phylogenetic analyses and machine learning. Studies for the prediction of overdose mortality at national/state/county and zip code level are rapidly emerging. Geospatial methods are increasingly used to identify hotspots of opioid use and overdose. In the context of infectious disease OREs, routine genetic sequencing of patient samples to identify growing transmission clusters via phylogenetic methods could increase early detection capacity. A coordinated implementation of multiple, complementary approaches would increase our ability to successfully anticipate outbreak risk and respond preemptively. We present a multi-disciplinary framework for the prediction of OREs in the US and reflect on challenges research teams will face in implementing such strategies along with good practices.

Figure 1.. Early detection, risk assessment and prediction and of opioid use related epidemics.

The black curve corresponds to the epidemic at baseline, while the dashed orange, blue and green curves correspond to the epidemic in the presence of interventions resulting from early detection, risk assessment and prediction, respectively. The time periods for each of these analyses are also colored in blue, green, and orange, respectively. We hypothesize that accurate prediction would have the strongest prevention impact because it would confer time to plan and implement an appropriate response, followed by risk assessment, which is less specific and therefore less informative, and by early detection, which is highly specific but occurs once the epidemic has started spreading. However, this impact will depend on how this evidence is used by decision makers.

Fig 2.. Multidisciplinary framework to enable the early detection, risk assessment and prediction of opioid use related epidemics (OREs).

The diagram should be read from the bottom upwards, with each layer corresponding to a different component determining the choice of method in a step by step manner: 1) health outcome of interest, 2) type(s) of surveillance data available, 3) characteristics of the collected data, 4) objective, 5) analytical method. First, the bottom layer refers to the surveillance and analytical infrastructure – these are pieces that must be in place to collect data and to analyze it. The role of surveillance, in the context of these studies, is often undertaken by public health agencies and institutions such as the CDC. As such, the first step is identifying available ORE-driven outcomes. These generally represent the outcome of focus for a given research project. Then, for identifying available measures of the outcome and potential predictors, researchers should ask which types of surveillance data are available to them. Clinical data (i.e., EMS, hospital records, death records), epidemiologic data (i.e., poison call centers, harm reduction services, 311 calls), behavioral data (i.e., observational studies, internet data), and drug market data (i.e. DEA, drug sample testing, wastewater sampling) represent four types of data of importance to consider. At this stage, depending on study purpose, we recommend that researchers aim to identify sources for each type of data. Next, after identifying potential data sources, data should be extracted. We have identified five types of data (traditional epi, internet data, genetic data, geospatial data, and social network data) – identifying which types of data are available can inform study objective. Prior to selecting the method to employ, we then recommend choosing an over-arching objective. Failing to do so can lead to confusion amongst the research team about the underlying purpose of a study. For example, certain approaches may be well-suited for risk assessment but not prediction, and a failure to explicitly identify study objective prior to choosing a method may result in choosing an inappropriate analytic approach. Finally, once the research team has identified their health outcome, the data that is available, and the overarching objective of their study, they can select the method(s) best suited to their data and objective. In addition, multiple methods can be used in parallel to increase the sensitivity and accuracy of findings.