The primary objective of this project was to develop a continuum of care model for opioid use disorder that could be generalized to be used by other healthcare organizations, public health departments, and other jurisdictions, the way the HIV care continuum has helped create comparable metrics of program success at program and population levels across the US and the world. The pilot phase of developing the model was focused on opioid use disorder using data from the Denver Health organization. Once that model was established and validated for opioids, it will also be modified for other substances, such as cocaine, methamphetamines, alcohol, etc.
Other objectives were not only to quantify the extent of the opioid use disorder (OUD) problem and the number of individuals engaged in treatment, but also to identify the gaps along the continuum where engagement in treatment declines. Through analysis and evaluation of the continuum of care metrics, further investigation can identify areas where individuals become disengaged from treatment, and used to evaluate methods to improve retention in care. Thus, another important objective has been to improve the quality and success of treatment services and retention in care.
Early in the process a logic model was created to provide some structure and framework on the key information sources, activities, and short-term and long-term outputs. As the DPH team became more knowledgeable about the different clinical processes, the logic model continued to evolve and expand to address these additional areas. Definitions, data sources, activities, and desired outputs were adjusted and refined. A number of additional outputs desired were identified. However, the project scope needed to remain relatively constant so that the primary objective, creating the opioid continuum of care model, would be achieved. Many other desired outputs were identified for future enhancements once the continuum of care model was established.
The primary objective of creating the opioid continuum of care model was achieved. The model went through a number of early revisions to address subtle use cases which were identified through in-depth analysis of the data. One example, of these types of scenarios was categorizing patients who may have stopped treatment for a brief period and then reengaged. Through discussion with the key clinical stakeholders a consensus was reached and incorporated into the model. The periodic Evaluation and Knowledge Management team meetings were instrumental in maintaining open dialogue and developing a strong and adaptive continuum of care model.
Another objective, expanding the use of the model to other systems or jurisdictions, is currently being pursued. Once the model was completed, meetings were held with other organizations who desired to apply this continuum of care model to their opioid treatment services. The DPH team worked jointly with these new partners to apply for a grant where DPH staff would provide consultation services to these other healthcare organizations to utilize their data to develop the opioid continuum of care model for their patient population.
The expansion of the continuum of care model to other substances is also being developed. Programs have been written to extract those patients who have been identified as using methamphetamines and also patients who have alcohol use disorder. After having the initial experience with opioids, the process for defining how patients are referred, engaged in treatment, and retained in care will take less time and be enhanced. The development of these continuum of care models is currently being actively pursued.
As mentioned previously, the use of only the ICD-10 diagnosis codes was inadequate in identifying the patients with OUD. The data source for identifying patients with OUD is the electronic health record (EHR), but it required extracting data from various elements. The criteria includes: ICD-10 codes, a Clinical Opioid Withdrawal Score (COWS) assessment, medication assisted treatment (methadone or buprenorphine dosing used for addiction treatment), laboratory test results indicating positive heroin use, opioid notes in the medical history, and provider and substance counselor notes with specific key words pertaining to opioid use. These definitions, especially utilizing key words in provider notes, required careful scrutiny and exclusion terms so that false positives were not included in the final identification of those patients with OUD.
The number of patients receiving medication assisted therapy is based upon two data sources. The EHR contains prescription information on buprenorphine and some methadone. In addition, there is a separate database that contains daily dispensing information of methadone. These two data sources are merged together to identify those initially receiving MAT. These data are also utilized, along with counselor visit information, to identify the number of patients continuing on MAT after 90 days and those who are retained in treatment for a year or more.
One of the first analyses used the continuum of care model for inpatients discharged during 2017. Following that analysis the definitions and criteria were further refined to ensure that all patients within the Denver Health system, both inpatient and outpatient, were appropriately included. The first validated and complete version was for the first six months of 2018. The data were examined and various scenarios and use cases were reviewed to ensure that they were addressed appropriately.
Once the model was validated, opioid continuum of care statistics were created for the years of 2017 and 2018. Comparison of these two years helped assess what impact various interventions have had on the number of patients receiving treatment and their retention in care over time, including the operationalization of the Center for Addiction Medicine in October 2018. It has been instructive to see objective metrics of the increasing numbers of patients receiving treatment during the start-up of the Center.
The results of these two opioid continuum of care models are the following:
Year 2017
Estimated opioid use disorder, opioid misuse, or opioid poisoning in Denver County 6688
Identified opioid use disorder, opioid misuse, or opioid poisoning at Denver Health 3930
Received medication assisted treatment at Denver Health 1392
Retained in care for greater than or equal 90 days of MAT 939
Retained in care for greater than or equal one year of MAT 653
Year 2018
Estimated opioid use disorder, opioid misuse, or opioid poisoning in Denver County 7010
Identified opioid use disorder, opioid misuse, or opioid poisoning at Denver Health 4593
Received medication assisted treatment at Denver Health 1747
Retained in care for greater than or equal 90 days of MAT 1023
Retained in care for greater than or equal one year of MAT 692
These results demonstrate that more patients were enrolled in care during 2018 and that the number of patients retained in care has increased. This provides evidence that improved access to expanded service capacity is helping more people affected by the opioid epidemic.
These two years of analysis also provide information to address one of the other objectives: identify the gaps along the continuum where engagement in treatment declines. The first gap to examine was the gap between the estimated number of opioid use disorder, opioid misuse, or opioid poisoning in Denver County versus those identified with opioid use disorder, opioid misuse, or opioid poisoning at Denver Health. The 2017 gap was 41.2% (6688-3930/6688) and the 2018 gap was 34.5% (7010-4593/7010). This indicates that an increasing proportion of the county's population with opioid use problems are finding their way into the Denver Health System, hopefully, in part, because of a welcoming treatment system.
The next gap to evaluate was the gap between Denver Health patients identified with opioid use issues versus those receiving MAT. In 2017 the gap was 64.6% (3930-1392/3930) and in 2018 the gap was 62.0% (4593-1747/4593). This is also a positive trend showing that the gap is decreasing, which suggests more individuals are being referred and engaging in treatment.
The third gap evaluated was the number receiving MAT versus those continuing to receive MAT after 90 days. In 2017 the gap was 32.5% (1392-939/1392) versus the 2018 gap of 41.4% (1747-1023/1747). This suggests that there has been a decline in those continuing on treatment. Discussion with OBHS staff and care providers in the emergency psychiatric department identified that one contributor to this decline was referring patients to treatment facilities outside the Denver Health system when those facilities were more convenient for patients, which increased in 2018. Since the continuum only currently measures Denver Health treatment, these cases are unaccounted for, illustrating the importance of extending the continuum of care model from a single isolated health system to data from across the entire referral network. Efforts are being made to pursue data sharing with these outside treatment facilities to better represent accurate retention in care metrics for those being referred elsewhere. Information is currently being extracted on who were referred to outside organizations but getting information on care retention at these facilities is not currently possible.
The final gap evaluated is the number receiving MAT versus those being retained in care and receiving MAT after one year. The 2017 gap is 53.1% (1392-653/1392) and the gap in 2018 is 60.4% (1747-692/1747). This decline is similar to the previous gap and can be explained by the significant number of referrals to outside treatment facilities during 2018.
The opioid continuum of care model has been extremely useful in identifying the success of enrolling patients into care and their retention in care over time. This model has been presented at the highest levels of Denver Health leadership and is a key tool for measuring and evaluating the effectiveness of the opioid treatment services.
Another evaluation activity once the continuum of care model was established was to look at newly identified patients with OUD. The 2017 and 2018 numbers represented earlier included patients who may have had OUD for a number of years. An evaluation was conducted to identify only patients who have no previous history of OUD and develop a continuum of care model for that population. The programs were modified to review the past medical history of patients to identify only those who were newly identified during the time period of January June 2018. The continuum of care model was then calculated for this unique population. The results show:
Identified opioid use disorder, opioid misuse, or opioid poisoning at Denver Health 907
Received medication assisted treatment at Denver Health 218
Retained in care for greater than or equal 90 days of MAT 116
Retained in care for greater than or equal one year of MAT 50
This version of the model will help identify if newly diagnosed patients with OUD are being referred and engaged in treatment more often and if their retention in care is improving over time. It will also provide information about whether the number of newly identified patients with OUD is increasing or decreasing over time.