Background According to the National Center for Health Statistics, Kentucky had the third highest drug overdose fatality rate in the nation in 2015 at 29.9 drug overdose fatalities per 100 000 population.
Objective The elevated drug overdose fatality rate necessitated the development and implementation of a comprehensive multisource drug overdose fatality surveillance system (DOFSS).
Methods DOFSS stakeholder work group members and data sources were identified, and memorandums of understanding were established. The following data sources were used to establish DOFSS: (1) death certificates; (2) autopsy reports; (3) toxicology result reports; (4) coroner reports; and (5) Kentucky All Schedule Prescription Electronic Reporting (KASPER) (prescription drug monitoring programme) data. Drug overdose poisonings were defined using Injury Surveillance Workgroup 7 definitions. Analyses were performed to investigate possible drug overdose-related health disparities for disabled drug overdose decedents and to characterise gabapentin in drug overdose deaths.
Results DOFSS identified 2106 drug overdose poisoning fatalities in Kentucky for 2013–2014. Identification of specific drugs involved in drug overdose deaths increased from 75.8% using a single data source to 97.5% using multiple data sources. Disabled drug overdose decedents were significantly more likely to have an active prescription for drugs identified in their system compared with the non-disabled drug overdose decedents. Toxicology data showed increased gabapentin involvement in drug overdose deaths from 2.9% in 2013 to 17% in 2014. Alprazolam was found most often in combination with gabapentin (41%), along with various other benzodiazepines and prescription opioids.
Conclusions A comprehensive multisource DOFSS improved drug overdose fatality surveillance by increasing completeness of data and data quality. DOFSS is a model that can be considered by other states to enhance their efforts in tracking drug overdose fatalities, identifying new and emerging trends, and informing policies and best practices, to address and reduce drug overdoses.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Drug overdoses have increased significantly over the last two decades resulting in a national epidemic and public health crisis. Kentucky’s drug overdose fatality rates have sextupled from 4.9 deaths per 100 000 residents in 1999 to 29.9 per 100 000 in 2015, putting Kentucky in the top five states with the highest age-adjusted drug overdose fatality rates.1 The elevated drug overdose fatality numbers and rates require targeted multipronged public health-based approaches (including improved opioid prescribing, prevention interventions and programmes, organisational and legislative policies, and community best practices) to address and reduce the current drug epidemic.2 One public health approach to inform drug overdose prevention programmes and practices is to establish a state-wide comprehensive multisource drug overdose fatality surveillance system (DOFSS) to identify new and emerging risk factors associated with drug overdose fatalities and inform policymaker and community practitioner decisions.
Potential barriers to the development and implementation of a DOFSS in Kentucky were identified. Coroners complete drug overdose death certificates with the underlying causes of death and contributing causes of death, including the listing of specific drug(s) involved in the death.3 In 2011, 30.5% of all Kentucky drug overdose death certificates did not list the specific drug(s) involved in the drug overdose fatality.4 The lack of identification of specific drugs on death certificates may hinder public health officials’ effectiveness in identifying accurate emerging trends of drug use, misuse and abuse.4–7 Also, drug overdose fatality investigations in Kentucky involve multiple governmental agencies due to a decentralised, hybrid death investigation system, in which county coroners refer cases to state medical examiners. County coroners are elected to office and have the legal responsibility to investigate drug overdose deaths.3 8 The state has four regional medical examiner offices (MEOs) that support the coroners with forensic pathology consultation, complete autopsy services, ancillary testing services (including postmortem toxicology services) and so on.9 With 120 coroners and four regional MEOs, case records (paper and electronic) are geographically dispersed throughout the state.
The complicated nature of drug overdose death certificate completion by coroners as well as the hybrid death investigation system necessitates strategies to improve drug overdose death certificate listing of specific drugs and drug combinations that contributed to or were the underlying cause of death, and strategies to improve the drug overdose death investigation process. The purpose of this study was to (1) identify the critical data sources necessary for the establishment of a comprehensive DOFSS; (2) evaluate the individual contribution of data sources in the identification of specific drugs involved in fatal drug overdoses; and (3) provide examples of the beneficial use of DOFSS to inform drug overdose prevention policies and community best practices.
DOFSS stakeholder workgroup identification
New and existing partners among state, regional and local agencies with expertise in the areas of drug abuse prevention, treatment, policy and enforcement were identified as potential DOFSS stakeholders. The following stakeholder agencies agreed to participate in the design and development of DOFSS: (1) MEO; (2) Office of Drug Control Policy; (3) Office of Vital Statistics (OVS); (4) Kentucky All Schedule Prescription Electronic Reporting (KASPER) (Kentucky’s prescription drug monitoring programme) system; (5) Coroner liaison to Chief Medical Examiner (ME); (6) Office of Inspector General; and (7) KASPER Advisory Council that included representation by Board of Medical Licensure, Board of Nursing, Board of Dentistry, substance abuse and mental health professionals, community mental health centres and pharmacists.
Analysis of the Kentucky drug overdose death investigation process and certification system
The drug overdose fatality investigation and certification process was reviewed and is described in figure 1. Kentucky law requires a coroner investigation, including toxicology testing, for all suspected drug overdose fatalities.3 Coroner investigations involve on-scene examination of decedents; documentation of scene findings (eg, documentation of drug paraphernalia and performance of on-scene pill counts of prescription medication); interviews of family, witnesses and law enforcement; review of medical records; documentation of drug history; and toxicology testing. Coroner and ME fatal drug overdose toxicology testing is paid by state funds through the MEO and is outsourced to a private accredited forensic laboratory through a state contract. A panel of the most commonly abused drugs has been established and is performed consistently on all toxicology specimens submitted. This panel consists of both controlled and non-controlled substances. Beyond this routine panel, additional drugs may be requested. The laboratory performs drug screenings to indicate the presence of a drug and then confirmatory testing is performed for quantification. At the discretion of the elected county coroner’s office investigating the suspected fatal drug overdose, the case also may be referred to the ME for autopsy and further examination. Cases autopsied by the MEO result in an autopsy report that provides the ME’s opinion on the cause and manner of death, and contributing and underlying factors. The ME also provides a death certificate template to the coroner with a suggested cause-of-death statement, and contributing causes of death. Per state-wide MEO policy and procedure, the Kentucky ME lists the specific drugs involved based on their opinion. After the coroner reviews the autopsy report and suggested causes of death (as well as other investigative information), the death certificate is completed with the cause of death, certified by the coroner and finalised by OVS. Additionally, Kentucky law mandates the reporting of all fatal drug overdoses by county coroner and medical examiners to multiple state-wide agencies.10 After the death certificate is finalised it is then sent to the National Center for Health Statistics for cause of death coding within the International Classification of Diseases 10 (ICD10) coding system. OVS then receives the coded death certificate for electronic filing and storage.
Drug poisoning definitions
Drug overdose definitions were based on Injury Surveillance Workgroup 7 ICD10 definitions for acute poisonings due to the effects of drugs. Drug overdose deaths were identified based on the following underlying cause of death ICD10 codes: X40–X44, X60–X64, X85 or Y10–Y14.11 DOFSS cases included (1) Kentucky residents who died in-state and out-of-state; and (2) non-Kentucky residents who died in Kentucky. Both resident and non-resident groups were included to ensure that all DOFSS stakeholders’ needs were met. Cases that did not meet these criteria were not input into DOFSS.
Identification of DOFSS data sources
The DOFSS stakeholder group identified the following five data sources necessary for the establishment of the comprehensive DOFSS: (1) electronic death certificate records; (2) coroner reports; (3) postmortem toxicology reports; (4) autopsy reports; and (5) KASPER records.
Memorandums of understanding
Memorandums of understanding (MOUs) were established to receive (1) monthly electronic death certificate record extracts; (2) drug-related autopsy and toxicology reports; and (3) KASPER patient prescription records. MOUs were not established with the 120 individual county coroners; coroners’ investigative reports were available on written request.
Establishment of DOFSS
DOFSS was built using Epi Info V.7 software, a public domain epidemiological statistical software tool created by the Centers for Disease Control and Prevention. Epi Info V.7 was chosen to establish DOFSS because of its ability to (1) easily create complex data collection forms; (2) automatically create a relational database that populates data variables from the five individual data sources; (3) perform statistical analysis for reporting; and (4) export data for sophisticated data analysis in statistical software platforms.12
DOFSS data elements
The DOFSS Epi Info data collection includes five data entry forms: (1) death certificate data; (2) coroner report data; (3) autopsy report data; (4) toxicology report data; and (5) KASPER report data (figure 2). All available data variables from each data source were included in DOFSS as individual data fields; duplicate fields were removed. Additional variables were developed to address known or suspected risk factors identified from a review of narrative fields on all original data sources. The DOFSS data fields include, but are not limited to, death certificate information (demographics, place of injury and death, causes of death literal text, significant contributing conditions text, ICD10-coded underlying and multiple causes of death); coroner reports (eg, drug paraphernalia found at the scene, history of drug abuse, chronic pain, mental illness or suicidal ideations, known medical history, compliance with prescribed medications, pill counts from the scene, coroner narrative text); autopsy reports (eg, body mass index, evidence of needle or track marks, internal organ weights, pills identified in the stomach, causes of death per ME’s opinions); postmortem toxicology results (eg, detected drugs and concentrations found in the decedent’s system in blood, urine and/or vitreous fluid at the time of death); and KASPER data (drug names, drug doses, date filled, pharmacy where prescription was filled, etc).
DOFSS data collection and entry
DOFSS was initiated in January 2015 with 2013 drug overdose fatality data, which was the first full year of data available. For the purpose of this study, data analysed are based on years 2013 and 2014; although, DOFSS data collection and entry is ongoing. The database and all source documents are stored on a secure server at the Kentucky Injury Prevention and Research Center (KIPRC) in Lexington, Kentucky.
Depending on the source document, data are manually entered by KIPRC data abstractors into DOFSS or are imported into Epi Info V.7 from an electronic file. Imported data quality control is performed, and each batch upload is verified. KASPER data were not directly imported into DOFSS due to electronic KASPER data set size. KASPER data were linked to decedents' DOFSS records using multiple fields (social security number, first and last names, street address, zip code, and date of birth). KASPER uses a probabilistic data matching method to aggregate individual prescription records into a cluster of prescriptions representing the decedent’s controlled substance prescription history.13 Data from toxicology reports, autopsy reports and coroner reports are manually abstracted. The current lag time for DOFSS data entry is under 1 year.
DOFSS data analysis
DOFSS data are routinely analysed. State-level and county-level reports are produced that evaluate changes in drug overdose trends and involvement of specific drugs, inform training and education on drug overdose data quality improvement, identify vulnerable populations and geographical areas at increased risk for drug overdoses, detect new and emerging drugs of misuse/abuse, and inform policymaking and evaluation. Community, local and state agencies and organisations can request aggregate data, within the guidelines of our established MOUs, to further assist with state drug overdose prevention efforts. Technical assistance is available for these organisations to assist with interpretation and use of the data requested.
For years 2013 and 2014, DOFSS identified 2106 drug overdose poisoning fatalities in Kentucky; 2008 (95.4%) of these drug overdose cases were Kentucky residents who died in Kentucky. Of the 4.6% whom were non-residents, 3% were border state residents. Approximately 61.6% of all decedents were male, 38.4% of the decedents were female. Of all decedents 95.4% were white and 3.7% were black (data not shown).
DOFSS data analysis to assess contribution of individual data sources in the identification of specific drugs involved in fatal drug overdoses
Statistical analysis of years 2013–2014 DOFSS drug overdose death certificates alone found that 75.8% of all drug overdose death certificates listed at least one specific drug or drug class involved in the overdose fatality (table 1).
Autopsy reports were available for 162 of 509 drug overdose death certificates that did not list any contributing drugs; when DOFSS statistical analysis included the autopsy reports, the percentage of drug overdose deaths with identified specific drug involvement increased to 83.5%. Postmortem toxicology reports identify specific drugs and concentrations in body fluids but do not assess their contribution to drug overdose deaths; inclusion of toxicology results in DOFSS increased identification of drugs involved in drug overdose deaths to 97.4%. The addition of coroner reports in DOFSS further increased identification of drug involvement in drug overdose deaths to 97.5% (table 1). Coroner reports also identified pills and/or prescriptions found at the scene in 23.8% of all 2013–2014 DOFSS cases. Three-fourths of the 53 cases in DOFSS from 2013 to 2014 without any specified drugs identified during the drug overdose death investigation were determined to be fatal drug overdoses certified by a physician in a hospital setting without coroner notification. Antemortem inhospital medical records and hospital drug screening results are not currently available for entry into DOFSS.
DOFSS data analysis to investigate possible health disparities in fatal drug overdoses
Drug overdose fatalities from 2013 to 2014 were categorised as disabled and non-disabled based on the usual (longest held) occupation listed on drug overdose death certificates. The ‘disabled’ occupation is a standard occupation description that is completed on death certificates; decedent’s usual occupation is listed by the certifier as disabled on the death certificate if the decedent did not work due to disability for the majority of their life.14 While definitive research on the sensitivity of the ‘disabled’ occupational variable has not been performed on death certificate to our knowledge, a study by Swanson et al 15 showed that when hospital medical records and death certificates were compared, the usual occupation and industry exactly matched in 76% of the death certificates analysed.15
Commonly identified drugs in the system at time of death on the toxicology report were examined in both disabled and non-disabled drug overdose fatalities. KASPER prescription history for drug overdose decedents who tested positive for oxycodone showed that a statistically significant higher percentage of disabled (68.8%) versus non-disabled (31.6%) drug overdose decedents had an active prescription (table 2).
DOFSS data analysis to inform legislative policymaking
Analysis of DOFSS 2013–2014 toxicology data showed increased gabapentin involvement in drug overdose deaths from 2.9% (n=30) in 2013 to 17.0% (n=185) in 2014. Gabapentin became a routinely tested drug on toxicology panels in Kentucky in July 2014. Table 3 shows the ability of DOFSS to identify the involvement of gabapentin before and after implementation of routine gabapentin testing. Prior to routine gabapentin toxicology testing, 70.1% of gabapentin involvement was identified by coroner reports. By adding gabapentin to the toxicology panel, the number of unique cases involving gabapentin that toxicology reporting identified increased from 43.3% to 95.6%. Additionally the number of death certificates that captured gabapentin’s involvement in a fatality doubled from 14.4% to 28.7% in only 6 months. With improved drug testing, of the 502 drug overdose deaths that occurred from August and December 2014, 168 (33.5%) deaths had positive gabapentin toxicology results. Gabapentin is not a scheduled controlled substance so there were no available corresponding gabapentin prescription data in KASPER data for analysis.13 The results from this analysis, along with pharmacists’ observations, informed discussion in the state on the necessity of scheduling gabapentin as a controlled substance for tracking in KASPER.
Without KASPER data on gabapentin researchers are limited; yet, the most common gabapentin drug combinations in toxicology were with prescription opioids and benzodiazepines, both drug classes that are tracked by prescription drug monitoring programmes (table 4). Alprazolam, clonazepam and diazepam were each identified in combination with gabapentin in over 20% of gabapentin-involved cases. Yet on average only a third of decedents had legal prescriptions for the benzodiazepines. When examining the top prescription opioids found in combination with gabapentin, just under half had a prescription for oxycodone or hydrocodone.
Implementation of a comprehensive surveillance system provides opportunities for enhanced surveillance data, data quality improvement, and intervention and policy implementation and evaluation that are not optimum using single source data. This is especially important in a hybrid death investigation state, such as Kentucky. The use of a single data source only identified 75.8% of all drug overdose fatalities with specific drug involvement. The inclusion of multiple data sources in DOFSS increased the identification of specific drug involvement in drug overdose fatalities by more than 20% to 97.5%. Each additional data source contributed to increased identification of specific drugs involved in fatal drug overdoses. A low percentage (2%) of cases was certified by a physician in a hospital setting without coroner notification. Kentucky Revised Statute 72.020 requires that any person, hospital or institution shall notify the county coroner of any drug overdose fatality.16 Our DOFSS data quality checks identify a need for increased training of attending physicians on drug overdose fatality referral to coroners with corresponding drug screen results.
DOFSS was built with well established mortality case definitions for acute drug overdose deaths.11 Coroners and medical examiners do not use standardised case definitions for acute drug overdoses on death certificates.17 Coroners and medical examiners often report drug cases using a ‘drug-related’ definition, which in Kentucky encompasses acute drug poisonings as the cause of death or contributory factor, substance abuse and use disorders, and chronic disease or infection due to drug use (personal communication, April 2017, Kentucky State Medical Examiner). Therefore, not all cases reported by coroners or medical examiners as drug-related will fit acute drug overdose poisoning definitions. Through our multiple data source approach we identified 262 possible 2013–2014 DOFSS cases that did not fit our current DOFSS case definition based on death certificate ICD10 codes, but fit our coroner/medical examiner definitions of drug-relatedness. These cases were individually reviewed by forensic pathologists to determine if any of these cases were acute drug overdose deaths missed by our death certificate-based definition due to vague language on the death certificate or miscoding. With forensic pathologist experts and review of multisource documentation, 145 of the 262 cases were considered appropriate for addition to DOFSS. Going forward on a continual bases, a forensic pathologist expert will screen coroner and medical examiner drug-related cases for acute drug poisonings missed by ICD10 definitions. These cases will be added to DOFSS and a review of potential reasons why these cases were missed will be performed to decrease these occurrences.
Additionally, the use of multiple data sources allows researchers to identify (1) new and emerging risk factors associated with drug overdoses; (2) trends in controlled substance prescribing practices; (3) specific drug use, misuse, abuse and diversion that would not have been discovered with a single data source; and (4) vulnerable populations at elevated risk for drug overdoses and abuse, to inform and improve drug prevention policies, interventions and practices.
Our DOFSS data analysis showed that disabled drug overdose decedents had similar toxicology levels of the most common prescription drugs identified in their systems at the time of death compared with non-disabled drug overdose decedents. Yet, the disabled population had a significantly higher proportion of active prescriptions for these same drugs at the time of death. For example, among drug overdose decedents with oxycodone identified in their system at the time of death, a significantly higher (p<0.0001) proportion of the disabled population had an active prescription for oxycodone (68.8%) compared with non-disabled decedents (31.6%). These results may be indicative of elevated use and prescribing of these medications for treatment of pain in the disabled population. This also indicates a lower risk of opioid drug diversion, or individuals using unlawfully obtained prescription drugs, in disabled individuals when looking at the complement rate of decedents without an active prescription.18–20 For example, 68.4% of non-disabled decedents with oxycodone identified did not have an active prescription for oxycodone at the time of death versus 31.2% for the disabled overdose decedents. Drug diversion may be possible in the disabled drug overdose decedent population with regards to the two most common benzodiazepines identified in their system at the time of death; 53.8% of disabled drug overdose decedents with alprazolam identified did not have an active prescription for alprazolam at the time of death, and 41.2% did not have an active prescription for clonazepam at the time of death. The use of (1) alternative opioid treatment therapies for chronic pain and alternative anxiety treatment therapies; (2) improved professional guidelines to ensure safer prescribing and dosing of controlled substances; and (3) increased controlled substance prescriber vigilance, may be warranted in this vulnerable population.2
Comprehensive drug overdose fatality surveillance informs implementation and evaluation of drug overdose prevention and control programmes at both local and state levels. Increased use of DOFSS data results can assist local and state public health departments in development of data-driven informed drug overdose and substance use disorder prevention, treatment and recovery programmes. With geographical analysis capabilities, drug overdose fatality surveillance data can be tailored specifically to individual communities’ needs. DOFSS data can also be used to inform new and improve existing legislative and organisational policy initiatives.
The convening of a multidisciplinary DOFSS stakeholder workgroup was crucial to the establishment of DOFSS. DOFSS stakeholders reviewed DOFSS data and analytical results, reported progress on ongoing initiatives, established state-wide drug overdose prevention priorities and provided avenues for broad dissemination of DOFSS data results to other stakeholder agencies, such as law enforcement, and substance use disorder treatment and recovery entities. The DOFSS stakeholder workgroup also informed new DOFSS data analyses and identified other stakeholders.
Gabapentin is a drug used to prevent and control seizures and to treat nerve pain; it also has a number of off-label uses.21 Prior to the addition of gabapentin to Kentucky’s standard drug of abuse toxicology panel in July 2014, gabapentin-involved cases averaged 5.4 cases per month. Afterwards that increased to an average of 25.9 gabapentin-involved cases per month. What is not known is if this drastic increase is fully due to improved testing or also due to an increase in use. Though the data that are available show a steady increase with approximately one in three drug overdose deaths from August 2014 through December 2014 involving gabapentin. With over a third of decedents testing positive for gabapentin in the time frame available after gabapentin became routinely tested for, may indicate that the use and prescribing of the drug should be further monitored.
DOFSS data show that gabapentin trends of use vary from non-gabapentin involved overdoses. Gabapentin was majorly used in conjunction with other prescription drugs. Currently, gabapentin is not a scheduled controlled substance so it is not tracked in KASPER limiting researchers’ ability to determine if the drug is being overprescribed or diverted.13 The occurrence of prescription with these drugs found in combination may be indicative of how the decedents also obtained the gabapentin.
Our gabapentin results were included in the evidence package for a new state law that schedules gabapentin as a controlled substance. Another piece of the evidence package included pharmacists’ concerns about increased prescribing of gabapentin and the potential of abuse. The KASPER Advisory Committee (including DOFSS stakeholder members) met in 2016 to address these rising concerns about gabapentin use. The committee used DOFSS results to validate their concerns and as a result of that discussion, a regulation was filed with the Kentucky General Assembly to add gabapentin as a schedule V controlled substance. This regulation was approved and will take effect from 1 July 2017.22 The inclusion of gabapentin in KASPER data will help stakeholders and researchers gain a better understanding of gabapentin prescribing practices and its involvement in drug overdoses.
DOFSS improved drug overdose fatality surveillance efforts, identified associated risk factors and contributed to a new state legislative policy. DOFSS is a model that can be considered by other states to enhance their efforts in tracking drug overdose fatalities, identifying new and emerging trends, and informing policies and best practices, to address and reduce drug overdoses and drug abuse.
What is already known on the subject
Drug overdose fatality surveillance efforts are widely based on single source data models.
Single source surveillance undercounts overdose deaths and underestimates the role of specific drugs in overdose deaths.
What this study adds
The value and methodology for the development of a multisource surveillance system for fatal drug overdose is described.
This study provides recommendations for the surveillance of fatal drug overdoses.
The authors thank the Kentucky Department for Public Health’s Office of Vital Statistics for access to death certificate records, and the Office of Inspector General for access to KASPER data, in the Cabinet for Health and Family Services. The authors also thank the Chief Medical Examiner’s Office for access to autopsy reports and postmortem toxicology reports in the Kentucky Justice and Public Safety Cabinet, and the county coroners for access to coroner reports.
Contributors All authors participated in the development and implementation of the Kentucky Drug Overdose Fatality Surveillance System (DOFSS). SLH collected data, performed data analysis, interpreted results and wrote the draft manuscript. SS performed data analysis, interpreted analytical results and co-wrote the draft manuscript. TLB established agency data use agreements, interpreted results and co-wrote the manuscript. DQ performed a legal review of relevant policies and manuscript content, and created figures 1 and 2. TC established agency data use agreements, interpreted analytical results and evaluated policy implications. WR interpreted analytical results and evaluated policy implications. MDS performed data analysis and interpreted analytical results. VI evaluated policy and practice implications.
Funding This publication was supported by the Grant or Cooperative Agreement Number, 1NU17CE924832-01, funded by the Centers for Disease Control and Prevention. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the Centers for Disease Control and Prevention or the Department of Health and Human Services. This grant was received by Kentucky Injury Prevention and Research Center as the bona fide for the Kentucky Department for Public Health.
Competing interests None declared.
Patient consent None.
Ethics approval The study was part of the Kentucky Drug Overdose Prevention Program and was approved by the University of Kentucky Institutional Review Board.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement No data from this study is available for sharing.
Correction notice This article has been corrected since it was published Online First. Three text encoding errors in the abstract and in the final footnotes have been corrected. The symbol related to the THC value (table 4, right column) has also been corrected.