JMIR Publications

ABSTRACT

Background:

Epicenters of drug-related harms are expanding to US rural areas, with rural young adults bearing a disproportionate burden. A large body of conceptual and empirical work suggests that place characteristics (e.g., spatial access to health services) shape vulnerability to drug-related harms among urban residents. Research on the role of place characteristics in shaping these harms among rural residents is nascent, as are related methods of gathering place-based data.

Objective:

To (1) analyze whether young adult rural people who use drugs (PWUD) answer self-administered web-based mapping items about locations where they engaged in risk behaviors; and (2) determine the precision of mapped locations.

Methods:

Eligible individuals had to report recent heroin or other opioid misuse; be aged 18-35; and live in the five-county rural Appalachian Kentucky study area. We used targeted outreach and peer referral methods to recruit participants. The self-administered web-based survey asked participants to “drop a pin” in interactive maps to mark where they completed the survey, and where they had slept most; misused drugs most; and had sex most in the past 6 months. Precision was assessed by (1) determining whether mapped locations were within 100 meters of a structure, and (2) calculating the Euclidean distance between the “pin-drop home location” and the street address where participants reported sleeping most often. Measures of central tendency and dispersion were calculated for all variables; distributions of missingness for mapping items and for the Euclidean distance variable were explored across participant characteristics.

Results:

Of the 151 participants, 88.9% completed all mapping items, and >92% dropped a pin at each of the four locations queried. Missingness did not vary across most participant characteristics, except that lower percentages of full-time workers and peer-recruited participants mapped some locations. Two-thirds of the pin-drop sex and drug use locations were <100 meters from a structure, as were 92% of pin-drop home locations. The median distance between the pin-drop and street-address home locations was 2.0 miles (25th percentile=0.8 miles; 75th percentile=5.5 miles); distances were shorter for high-school graduates, staff-recruited participants, and participants reporting no technical difficulties completing the survey.

Conclusions:

Missingness for mapping items was low and unlikely to introduce bias, given that it varied across few participant characteristics. Precision results were mixed. In a rural study area of 1,378 square miles, most pin-drop home addresses were near a structure; it is unsurprising that fewer drug and sex locations were near structures, since most participants reported engaging in these activities outside at times. The error in pin-drop home locations, however, might be too large for some purposes. We offer several recommendations to strengthen future research, including gathering metadata on the extent to which participants “zoom” in on each map and recruiting participants via trusted staff. Clinical Trial: N/A