Survey on the Treatment of Opioid Use Disorders

Supporting Statement B
For

Survey on the Treatment of Opioid Use Disorders
OFFICE OF NATIONAL DRUG CONTROL POLICY (ONDCP)
EXECUTIVE OFFICE OF THE PRESIDENT
March 10, 2020

Jayme Delano
Deputy Director, National HIDTA Program Office
Office of National Drug Control Policy
Executive Office of the President
Street address
Washington, DC ZIP
Phone: 202-395-6794
Email:
[email protected]

B.1 RESPONDENT UNIVERSE AND SAMPLING METHODS
The survey instrument, piloting procedures, sampling and analysis plan, and methods to reduce
nonresponse were developed for the Office of National Drug Control Policy (ONDCP) by the
National Association of Drug Court Professionals (NADCP) and its partner, Carnevale
Associates, LLC (CALLC). The proposed survey, the Survey on the Treatment of Opioid Use
Disorders (the survey), seeks to determine the current status of the availability and accessibility
of medication-assisted treatment (MAT) for opioid use disorders (OUD) among adult drug court
participants. The potential respondent universe for the survey are the drug courts of all 3,147
counties and county-equivalents in the 50 U.S. states and the District of Columbia. This potential
universe will be reduced to a more limited sample following the same sampling methods utilized
in a prior survey, which involves targeting those counties in the contiguous United States hardest
hit by the opioid epidemic. Simply stated, the criteria used to define those areas was determined
in two parts. One was to include counties where number of opioid deaths was at or exceeded the
U.S. average; the other was to include counites where the opioid death rate was at or above the
U.S. average. These methods achieved a sampling of the 269 counties and county-equivalents
across the U.S. that are the most highly impacted by the opioid epidemic. This 269 county
sample was assembled across a two-strata system; Strata 1 were the 107 counties that produced
50 percent of all fatal opioid-involved overdoses in the U.S., and Strata 2 were the 162 counties
that had a fatal opioid-involved overdose per capita rate above the U.S. average of 15.1 per
100,000 individuals. By using these two strata, the sampled counites included both urban and
rural areas where epidemic had the most impact. Counties that met the criteria for both Strata
were included only in Strata 1. It is anticipated that repeating these methods with more recent
data will achieve a similar sample size.
As was previously discussed, because of the lack of available county-level opioid use disorder
data, county-level fatal opioid-involved overdose data were used as the primary measure to
defined areas most affected by the opioid epidemic. Fatal opioid-involved overdose data are
available for over 99 percent of U.S. counties and county-equivalents, making it the most
widespread metric available. These overdose data were collected from the Center for Disease
Control and Prevention’s (CDC) Wide-ranging ONline Data for Epidemiological Research
(WONDER) system, which draws from the National Vital Statistics System. The 2017 multiple
cause of death (MCD) data were the most recent at the time of the prior survey, whereas the
Survey on the Treatment of Opioid Use Disorders will use 2018 data. The search syntax from the
prior survey will be used (see Table 1 in Section B.2.). This search syntax is consistent with the
CDC search procedures for its Morbidity and Mortality Reports.
The expected survey response rate is 90%. This estimate is based on another survey CALLC is
working on using a sample of state department of corrections. Our experience with working
with the NADCP is its strong linkage with local drug court facilitators will ensure that our
response rate will be, at a minimum, 80%. NADCP is in constant contact with local drug court
coordinators who will be key in ensuring a high response rate.

B.2. PROCEDURES FOR THE COLLECTION OF INFORMATION
Table B.2.1 shows the prior search syntax that will be replicated for this survey. To explain what
it shows in plain English, the National Vital Statistics System collects cause-of-death data from
all 50 states and the District of Columbia and their localities. However, due to a lack of
standardization in data reporting across and within states, it is not possible to simply report the
number of deaths from any cause, including opioid-involved overdoses. Instead, a combination
of Underlying Cause-of-Death (UCD) and Multiple Cause-of-Death (MCD) values must be
assembled to attempt to approximate the true number.
In this search, consistent with standard CDC search procedures, we identified all death records
which had any drug overdose listed as one of the underlying causes of death. This was then
subset to only those death records where an opioid was identified as one of the multiple causes of
death. This two-step process is necessary because of the variance across counties and states in
how opioid-involved overdoses are reported. If we only looked at UCD codes for opioid
overdoses, then any overdose deaths where an opioid was involved but the medical examiner or
corner did not determine it to be the reason for death would be excluded. And if we only looked
at MCD codes where an opioid was involved, then any deaths where an opioid was involved but
there was not an overdose would be included.
Table B 2.1: Search Syntax for Obtaining Fatal Opioid-Involved Overdose Records
The MCD Data was subset to records that met the following conditions:
1. Group Results by= County
AND
2. Location= All
AND
3. Demographics= All
AND
4. Year and Month= 2017 (All dates)
AND
5. Weekday, Autopsy, and Time of Death= All
AND
6. Underlying cause of death (UCD) Codes= Any of:
a. X42 unintentional narcotic overdoses
b. X62 intentional narcotic overdoses
c. Y12 poisoning or exposure to narcotics
d. X40-X44, X60-X64, and Y10-Y14 other drug overdose codes
AND
7. Multiple cause of death (MCD) Codes= Any of:
a. T40.0 opium
b. T40.1 heroin
c. T40.2 other opioids
d. T40.3 methadone
e. T40.4 other synthetic narcotics
f. T40.6 other and unspecified narcotics.

The data obtained from this search revealed the relative concentration of the opioid epidemic
among counties and county-equivalents; 701 counties (22% of the 3,147 total) accounted for 81

percent of fatal opioid-involved overdoses in 2017 (37,526 of 46,262 deaths), and the top 1
percent of counties accounted for 27 percent of fatal opioid-involved overdoses. At the same
time, the data also showed that certain low-population counties did not produce high raw totals
of fatal opioid-involved overdoses but did have per capita rates far above the national average.
As such, a two-strata definition of “highly impacted” was created:
1. Counties with the highest numbers of fatal opioid-involved overdoses
2. Counties with the highest per capita rates of fatal opioid-involved overdoses

This two-strata system was deemed appropriate both because it encompasses the counties most
impacted by the opioid epidemic and it would create a more diversified sample; with Strata 1
likely to encompass higher population, urban counties and Strata 2 likely to encompass lower
population, rural counties. Strata 1 was created by organizing counties by raw number of fatal
opioid-involved overdoses and including counties in the Strata until the total number of
overdoses reached 50 percent of the U.S. total. This produced a list of 107 counties (3.4% of the
3,147 total). Strata 2 was created by organizing counties by their per capita rate of fatal opioidinvolved overdoses and including every county where the lower bound of the 95 percent
confidence interval of their per capita rate was above the upper bound of the 95 percent
confidence interval for the nationwide per capita rate; which was 15.1 per 100,000 individuals.
This produced a list of 162 counties (5.1% of the 3,147). Counties that qualified for both Strata
were included only in Strata 1. The combined sample of 269 counties account for 8.5 percent of
all counties and encompass 66.2 percent of the U.S. population.
It is important to note that because the CDC systems suppress data for counties that have less
than 10 deaths from any given search syntax, it is possible that some counties that would qualify
for Strata 2 were excluded because their data were not available. However, these are likely to be
extremely low population, rural counties that are unlikely to have their own drug courts that
could respond to the survey.
The identification process for matching counties identified by the two-strata system with
counties with adult drug courts or adult hybrid courts (i.e. drug/DWI courts) involved two stages.
First, county-level drug court data was harvested in January 2019 from the National Drug Court
Research Center (NDCRC) managed by a cooperative agreement between American
University’s Justice Programs Office and the U.S. Department of Justice, Bureau of Justice
Assistance (BJA). The NDCRC database revealed 141 (52.4%) counties of the 269 sampled had
an adult drug court. NADCP reviewed the NDCRC data and determined that the database of
adult drug courts was not comprehensive. The second stage involved contacting state drug court
coordinators in relevant states to verify the existence of adult drug courts in each county of the
sample included in their state. Data was harvested by NADCP following email communication
with state drug court coordinators and completion of a Microsoft Excel-based tracking sheet. The
second stage of the drug court identification process took place in March 2019.
The survey will be administered electronically, where the 269 potential respondents will receive
an email inviting them to participate in the survey and a link to the survey’s web address.
Respondents can complete the survey at their convenience. Survey responses will be stored on
the cloud-based server of the survey software company (Cvent) and downloaded by the
NADCP/CALLC team once all data are collected. Once downloaded, data will be stored in
Microsoft Excel on the NADCP/CALLC team network. All analyses will be conducted in SPSS

and Tableau will be used for data visualization. This is a one-time data collection, so there is no
need to modify data collection cycles to reduce burden.
There are no complex analytical techniques that will be used for the survey data analyses.
Descriptive statistics and frequency distributions will be used to answer questions including:
•
•
•
•
•
•

How many adult drug courts have MAT available for their clients?
To what extent is MAT used in these courts?
What specific medications are available to drug court participants?
What are the factors that limit MAT use?
How many adult drug courts have received training that improved the delivery of MAT?
What are adult drug court practices regarding naloxone to reverse opioid overdose and
death?

When applicable, answers to these questions will be compared to previous survey results.
However, due to the comprehensive nature of the proposed survey, many of the questions are
being asked of drug courts for the first time. The survey results will demonstrate if changes in the
following have occurred: accessibility and availability of MAT; factors limiting use;
circumstances where MAT is permitted; and, practitioner attitudes toward MAT. Survey results
will also be used to determine if ONDCP-funded trainings and knowledge dissemination efforts
have been effective in promoting MAT as the evidence-based standard of care for OUD. Finally,
the survey will uncover where there are gaps in knowledge and practice so future training efforts
can be modified and resources appropriately allocated.
B.3. METHODS TO MAXIMIZE RESPONSE RATES AND DEAL WITH NON-RESPONSE
The survey project will employ different procedures to maximize response rates and to address
cases of nonresponse. The deployment plan for the survey will include methods and procedures
to make it easy and appealing to comply with the request for information. survey participants
will receive prenotification of survey disbursal, an explanation of the different survey elements,
instructions for completion, and requests to offer any related questions. The NADCP/CALLC
team will be responsive to the queries of survey participants.
To further aid completion, the NADCP/CALLC team made brevity and ease-of-completion a
priority in survey development. Questions are as simple and brief as possible. The survey
instrument is designed to minimize burden on respondents. For example, there are binary skip
patterns throughout the survey and at the beginning of each section to determine relevance. If a
section is not relevant to the respondent, they are able to skip each particular section. This would
especially ease any burden for respondents from states who do not permit the use of MAT.
Where respondents do answer, questions are primarily structured so respondents are not required
to enter exact numbers or percentages. In addition, open-ended questions do not feature
prominently in the survey. Beyond the construction of survey question types and logic, survey
language will minimize the use of technical language and inaccessible jargon. The survey
disbursement platform is easy to use across mobile, tablet, and conventional settings.

Although no personally identifiable information (PII) will be collected from respondents, the
NADCP/CALLC team will know what county the respondent represents. This will permit the
team to engage in standard follow-up techniques to reduce nonresponse. Upon disbursement, the
NADCP/CALLC team will employ several types of follow-up with those who did not respond
during the first response period. The communication preferences of each survey respondent will
be collected and will help guide targeted follow-up efforts. The NADCP/CALLC team will
dedicate staff to review responses once a week and send reminder emails out to non-responders.
The NADCP/CALLC team will conduct direct email and phone communication with survey
non-respondents to offer reminders, answer questions, resend the survey, and assist in
completion over the telephone. Within the three-month survey response period, there will be no
limit on the number of follow-up attempts.
NADCP has a strong relationship with local drug court coordinators across the country and has
extensive experience working alongside and partnering with drug courts in multifarious settings.
NADCP knows who to contact and how to communicate with coordinators to achieve results.
NADCP benefits from a sterling reputation as the national convening body for drug court
training, technical assistance, and collaboration. These relationships and reputational strengths
will assist in bolstering the interest, uptake, and completion of the survey.

B.4. TESTS OF PROCEDURE OR METHODS TO BE UNDERTAKEN
The survey instrument was piloted with eight individuals in July and August of 2019. These
individuals were a combination of ONDCP staff, former state and local drug court coordinators,
and consultants to NADCP. Pilot respondents were sent the link to the survey and asked to
record the length of time to take the survey. The average length of time to complete the survey
was calculated and used to determine time burden estimates in the Federal Register notices and
Supporting Statement A. The pilot test also confirmed that the skip patterns work for different
answer combinations. Feedback from the pilot respondents was incorporated into subsequent
versions of the survey instrument and confirmed that the survey is clear, user friendly, and easy
to understand.
B.5. INDIVIDUALS TO BE CONSULTED ON STATISTICAL ASPECTS AND
INDIVIDUALS COLLECTING AND/OR ANALAYZING DATA
NADCP has three individuals who will be consulted on statistical aspects of data analysis. In
addition, Carnevale Associates will also have four staff members reviewing the data analysis
process. Those staff are listed below:
NAME/TITLE
Doug Marlowe, J.D., Ph.D.
Senior Scientific Consultant

Jacqueline VanWomer, Ph.D.
Senior Subject Matter Expert

ADDRESS
National Association of Drug Court
Professionals
625 N. Washington Street, # 212
Alexandria, VA 22314
National Association of Drug Court
Professionals
625 N. Washington Street, # 212
Alexandria, VA 22314

CONTACT INFORMATION
Phone: 610.388.8606
[email protected]

Phone: 703.575.9400
[email protected]

Carolyn Hardin, M.A.
Project Director

John Carnevale, Ph.D.
Senior Subject Matter Expert

Erika Ostlie, M.A.
Research Director
David Theiss, M.S.
Senior Data Analyst
Lauren Pappacena, M.A.
Research Associate

National Association of Drug Court
Professionals
625 N. Washington Street, # 212
Alexandria, VA 22314
Carnevale Associates, LLC
P.O. Box 84085
Gaithersburg, MD 20883

Phone: 703.575.9400
[email protected]

Carnevale Associates, LLC
P.O. Box 84085
Gaithersburg, MD 20883
Carnevale Associates, LLC
P.O. Box 84085
Gaithersburg, MD 20883
Carnevale Associates, LLC
P.O. Box 84085
Gaithersburg, MD 20883

Phone: 301.977.3600
[email protected]

Phone: 301.977.3600
[email protected]

Phone: 301.977.3600
[email protected]
Phone: 301.977.3600
[email protected]