Federal Practitioner

Urine drug screen (UDS) monitoring is a common risk-mitigation strategy tool for prescribing controlled substances.^1-3 Not only is UDS monitoring highlighted by clinical practice guidelines for opioid prescribing for chronic pain,^1,2 it has also been suggested as best practice for benzodiazepines³ and a consideration for other controlled substances. Monitoring UDSs helps confirm adherence to the prescribed treatment regimen while also screening for substance use that may increase patient risk.

UDS results can be complex and have profound implications for the patient’s treatment plan. Drug metabolites for opioids are particularly complicated; for example, synthetic and semisynthetic opioids are not detected on routine opiate immunoassays.⁴ This may lead a clinician to falsely assume the patient is not taking their fentanyl or tramadol medication as directed—or potentially even diverting—in the face of a negative opiate result.⁵ Routine UDSs are also subject to the pitfall of false-positive results due to coprescribed medications; for example, bupropion can lead to a false-positive amphetamine result, whereas sertraline can lead to a false-positive benzodiazepine result.⁶ Retrospective reviews of clinician behavior surrounding UDS interpretation have demonstrated knowledge gaps and inconsistent communication practices with patients.^7,8

Given the complexity of UDS interpretation and its close relationship with medications, pharmacists are positioned to play an important role in the process. Pharmacists are embedded in pain-management teams and involved in prescription drug monitoring programs (PDMPs) for many health systems. The Veterans Health Administration (VHA) has supported the hiring of pain management, opioid safety, and PDMP coordinators (PMOP) at its facilities to provide clinical pain-management guidance, support national initiatives, and uphold legislative requirements.⁹ In many facilities, a pharmacist is hired specifically for these positions.

Clinical dashboards have been used by pharmacists in a variety of settings.^10-13 They allow clinicians at a broad level to target interventions needed across a patient population, then produce a list of actionable patients to facilitate delivery of that intervention on an individual level.¹³ Between 2021 and 2022, a clinical dashboard to review potentially discrepant UDS results was made available for use at US Department of Veterans Affairs (VA) medical centers. Evidence exists in primary and specialty care settings that implementation of an opioid-prescribing clinical dashboard improves completion rates of risk-mitigation strategies such as UDS and opioid treatment agreements.^14,15 To our knowledge there is no published research on the use and outcomes of a clinical dashboard that allows users to efficiently review discrepant UDS results when compared to a list of currently prescribed medications.

Given the availability of the UDS dashboard at the VA Black Hills Health Care System (VABHHCS) in South Dakota and the hiring of a PMOP coordinator pharmacist, the aim of this quality improvement project was 2-fold: to implement a pharmacist-led process to monitor the UDS dashboard for potentially discrepant results and to describe the quantity and types of interventions made by the clinical pharmacist leading this process.

Quality Improvement Project

A clinical UDS dashboard was created by the VA Northwest Health Network and made available for use at VHA sites between 2021 and 2022. The UDS dashboard is housed on a secure, Power BI Report Server (Microsoft), with access restricted to only those with patient health data privileges. The dashboard identifies all local patients with a UDS that returned with a potential discrepancy, defined as an unexpected positive result (eg, a detected substance not recently prescribed or documented on the patient’s medication list) and/or an unexpected negative result (eg, a prescribed substance not detected). The UDS dashboard identifies these discrepancies by comparing the patient’s current medication list (both VHA and non-VHA) to their UDS results.

The UDS dashboard displays a summary of UDSs performed, unexpected negative results, unexpected positive results, and potential discrepancies. The user may also specify the laboratory type and time frame of interest to limit displayed results. The user can then view patient-specific data for any category. Among the data are the patient’s UDS results and the completion date, detected (or nondetected) substance(s), ordering clinician, associated medication(s) with last fill date and days’ supply, and whether a confirmatory test has been performed in the past year.

VABHHCS uses an extended UDS immunoassay (PROFILE-V, MEDTOX Diagnostics) that reports on 11 substances: opiates, oxycodone, buprenorphine, methadone, amphetamines, methamphetamine, barbiturates, benzodiazepines, cocaine metabolites, cannabinoids (tetrahydrocannabinol [THC]), and phencyclidine. These substances appear on the UDS dashboard. The project protocol initially included monitoring for tramadol but that was later removed because it was not available with this UDS immunoassay.

Pharmacist Process

Either the PMOP coordinator or pharmacy resident monitored the UDS dashboard weekly. Any patients identified as having a potential discrepancy were reviewed. If the discrepancy was determined to be significant, the PMOP coordinator or pharmacy resident would review the patient electronic health record. If warranted, the patient was contacted and asked about newly prescribed medications, missed and recent medication doses, and illicit substance use. Potential interventions during in-depth review included: (1) discussing future actions with the primary care clinician and/or prescriber of the controlled substance; (2) ordering a confirmatory test on the original urine sample; (3) evaluating for sources of potential false-positive results; (4) completing an updated PDMP if not performed within the past year; (5) referring patients for substance use disorder treatment or counseling; or (6) consulting the local narcotics review committee. A progress note was entered into the electronic health record with the findings and any actions taken, and an alert for the primary care clinician and/or prescriber of the controlled substance.

Implementation and Analysis

This quality improvement project spanned 16 weeks from June 2022 through September 2022. Any patient with a UDS that returned with a significant discrepancy was reviewed. The primary outcome was interventions made by the PMOP coordinator or pharmacy resident, as well as time taken to perform the in-depth review of each patient. Patient demographics were also collected. The protocol for this project was approved by the VABHHCS pharmacy and therapeutics committee and was determined to meet guidelines for a nonresearch quality improvement project.

Results

From June 2022 through September 2022, 700 UDSs were performed at VABHHCS with 278 (39.7%) patients identified as having a potential discrepancy based on UDS results. Sixty patients (8.6%) had significant discrepancies that warranted in-depth review. The most common reasons for determining whether a potential discrepancy was not significant included unexpected negatives due to documented non-VA medications no longer being prescribed, unexpected positives due to recent expiration of a controlled substance prescription the patient was still taking, or unexpected positives due to the detection of a substance for which the clinician was already aware. During the 16-week study period, the mean number of patients warranting in-depth review was 4 per week.

The patients were predominantly male with a mean age of 61 years, and most (87%) were prescribed at least 1 controlled substance (mean, 1.1), primarily opioids for pain management (Table 1). Most patients had recent substance risk mitigation with UDS (56%) and PDMP (65%) checks within the past year. Of the 60 patients reviewed with significant UDS discrepancies, 50% had a history of discrepant UDS results. Of the 60 UDS discrepancies, there were 37 unexpected positive results (62%), 17 unexpected negative results (28%), and 10 patients with both positive and negative results (17%). THC was the most frequently detected substance, followed by opiates, benzodiazepines, and amphetamines (Table 2).

Each in-depth review with interventions by the PMOP coordinator or pharmacy resident lasted a mean of 14 minutes (Table 3). Five patients were successfully contacted for an interview and 7 patients could not be contacted. The ordering clinician of the UDS sometimes had contacted these patients prior to the PMOP coordinator or pharmacy resident reviewing the UDS dashboard, eliminating the need for additional follow-up.

The most common pharmacist intervention was discussing future actions with the primary care clinician and/or prescriber of the controlled substance (n = 39; 65%). These conversations resulted in actions such as ordering a repeat UDS with confirmatory testing at a future date or agreeing that the clinician would discuss the results and subsequent actions with the patient at an upcoming visit. Pharmacist interventions also included 25 PDMP queries (42%) and 9 orders of confirmatory UDS on the original urine sample (15%). Only 1 patient was evaluated by the narcotics review committee, which resulted in a controlled substance flag being placed on their profile. No patients were referred to substance use disorder treatment or counseling. It was offered to and declined by 1 patient, and 3 patients were already engaged in these services.

Medication therapies that could contribute to false-positive results were also evaluated. Fourteen patients who tested positive for THC had a prescription for a nonsteroidal anti-inflammatory drug or proton-pump inhibitor, which could have created a false-positive result.⁶ One patient who tested positive for amphetamines had a prescription for phentermine.¹⁶ No other potential false-positive results were identified.

 

 

Discussion

Findings of this project illustrate that the use of a clinical pharmacist to monitor a dashboard of discrepant UDS results created opportunities for collaboration with clinicians and impacted confirmatory testing and PDMP monitoring practices.

At the local level, the process had numerous benefits. First, it was a reasonable amount of workload to generate pharmacist interventions: the PMOP coordinator conducted an average of 4 in-depth reviews weekly, each lasting about 14 minutes. Thus, the UDS dashboard allowed the PMOP coordinator to actively surveil all incoming UDS results for potential discrepancies in about 1 hour each week. Pairing the automation of the UDS dashboard with the clinical judgment of the PMOP coordinator seemed to maximize efficiency. VABHHCS provides primary and secondary medical and surgical care to a rural population of approximately 20,000 patients across 5 states; the time required at facilities that serve a higher volume of patients may be greater.

Second, the project served as an opportunity for the PMOP coordinator to provide case-specific clinician education on UDS monitoring. As medication experts, pharmacists can apply their medication-related knowledge to UDS interpretation. This includes understanding drug metabolism and classification and how they apply to UDS results, as well as recognizing medication therapies that could contribute to false-positive UDS results. Research suggests that clinicians may have gaps in their knowledge and may welcome pharmacist assistance in interpreting UDS results.^7,8

Third, the project helped improve rates of confirmatory testing for those with unexpected positive UDS results. Confirmatory testing should be strongly considered if positive results would have significant implications on the future course of treatment.⁴ The PMOP coordinator ordered a confirmatory test on 9 patients using the same urine sample used to conduct the initial UDS, minimizing the burden on the patient and laboratory staff. Confirmatory testing was limited by the laboratory’s sample retention period; if the need for confirmatory testing was not recognized soon enough, the sample would no longer be available for retesting. Health systems may consider the use of reflexive confirmatory testing with UDS as an alternative approach, although this may come at an additional cost and may not be warranted in many cases (eg, only 39.7% of all potential discrepancies were deemed as significant within our project).

There were notable incidental findings in our quality improvement project. Among patients with a significant discrepancy on UDS, 50% had a history of ≥ 1 discrepant UDS result. This further emphasizes the importance of appropriate use and interpretation of UDS monitoring for all clinicians, as this may prevent prolonged and potentially inappropriate treatment regimens. Secondly, rates of mental health diagnoses among those with a significant UDS discrepancy seemed relatively high compared to population-level data. For example, among veterans, the overall lifetime prevalence of posttraumatic stress disorder is estimated to be 8.0%; in our project, 35% of patients with a significant UDS discrepancy had a posttraumatic stress disorder diagnosis.¹⁷ This relationship may be an area of further study.

Lastly, it was surprising that the overall rates of UDS and PDMP checks within the past year were 56% and 65%, respectively. VABHHCS requires veterans on controlled substances to have these risk-mitigation strategies performed annually, so our suspicion is that many were falling out due to having been most recently evaluated 12 to 16 months prior. This may represent a limitation of our data-collection method, which reviewed only the previous 12 months.

Limitations

This project was carried out over a period of only 4 months. As a result, only 60 patients received an in-depth review from the PMOP coordinator. Second, the timeliness of the intervention seemed crucial, as delayed in-depth reviews resulted in fewer opportunities to order confirmatory tests or collaborate with clinicians prior to devising an updated plan. Additionally, our process called for UDS dashboard monitoring once a week. Given that the laboratory held samples for only 48 hours, twice- or thrice-weekly review of the UDS dashboard would have allowed for more confirmatory testing, along with more immediate clinician collaboration. Most importantly, the outcomes of this project are only presented via descriptive statistics and without the results of any comparison group, making it impossible to draw firm conclusions about this approach compared to standard-care processes.

Conclusions

This quality improvement project has proven to be valuable at VABHHCS and we intend to continue this pharmacist-led process to monitor the UDS dashboard. VABHHCS leadership are also discussing UDS practices more broadly to further enhance patient management. Within the VA, the PMOP coordinator—charged with being the local coordinator of appropriate pain management and opioid safety practices—is well positioned to assume these responsibilities. Outside of the VA, a pain-management clinical pharmacist or any pharmacist embedded within primary care could similarly perform these duties. Previous literature regarding the implementation of clinical dashboards suggests that with the appropriate software engineering teams and infrastructure, this tool could also be feasibly developed and implemented at other health systems relatively quickly.¹⁴

Overall, a pharmacist-led process to efficiently monitor a dashboard of discrepant UDS results led to opportunities for collaboration with clinicians and positively impacted confirmatory testing and PDMP monitoring at a rural VA health system.

Acknowledgments

The authors express their gratitude to Patrick Spoutz, PharmD, BCPS, VISN 20 Pharmacist Executive, for introducing and sharing the UDS dashboard with our team.

Urine drug screen (UDS) monitoring is a common risk-mitigation strategy tool for prescribing controlled substances.^1-3 Not only is UDS monitoring highlighted by clinical practice guidelines for opioid prescribing for chronic pain,^1,2 it has also been suggested as best practice for benzodiazepines³ and a consideration for other controlled substances. Monitoring UDSs helps confirm adherence to the prescribed treatment regimen while also screening for substance use that may increase patient risk.

UDS results can be complex and have profound implications for the patient’s treatment plan. Drug metabolites for opioids are particularly complicated; for example, synthetic and semisynthetic opioids are not detected on routine opiate immunoassays.⁴ This may lead a clinician to falsely assume the patient is not taking their fentanyl or tramadol medication as directed—or potentially even diverting—in the face of a negative opiate result.⁵ Routine UDSs are also subject to the pitfall of false-positive results due to coprescribed medications; for example, bupropion can lead to a false-positive amphetamine result, whereas sertraline can lead to a false-positive benzodiazepine result.⁶ Retrospective reviews of clinician behavior surrounding UDS interpretation have demonstrated knowledge gaps and inconsistent communication practices with patients.^7,8

Given the complexity of UDS interpretation and its close relationship with medications, pharmacists are positioned to play an important role in the process. Pharmacists are embedded in pain-management teams and involved in prescription drug monitoring programs (PDMPs) for many health systems. The Veterans Health Administration (VHA) has supported the hiring of pain management, opioid safety, and PDMP coordinators (PMOP) at its facilities to provide clinical pain-management guidance, support national initiatives, and uphold legislative requirements.⁹ In many facilities, a pharmacist is hired specifically for these positions.

Clinical dashboards have been used by pharmacists in a variety of settings.^10-13 They allow clinicians at a broad level to target interventions needed across a patient population, then produce a list of actionable patients to facilitate delivery of that intervention on an individual level.¹³ Between 2021 and 2022, a clinical dashboard to review potentially discrepant UDS results was made available for use at US Department of Veterans Affairs (VA) medical centers. Evidence exists in primary and specialty care settings that implementation of an opioid-prescribing clinical dashboard improves completion rates of risk-mitigation strategies such as UDS and opioid treatment agreements.^14,15 To our knowledge there is no published research on the use and outcomes of a clinical dashboard that allows users to efficiently review discrepant UDS results when compared to a list of currently prescribed medications.

Given the availability of the UDS dashboard at the VA Black Hills Health Care System (VABHHCS) in South Dakota and the hiring of a PMOP coordinator pharmacist, the aim of this quality improvement project was 2-fold: to implement a pharmacist-led process to monitor the UDS dashboard for potentially discrepant results and to describe the quantity and types of interventions made by the clinical pharmacist leading this process.

Quality Improvement Project

A clinical UDS dashboard was created by the VA Northwest Health Network and made available for use at VHA sites between 2021 and 2022. The UDS dashboard is housed on a secure, Power BI Report Server (Microsoft), with access restricted to only those with patient health data privileges. The dashboard identifies all local patients with a UDS that returned with a potential discrepancy, defined as an unexpected positive result (eg, a detected substance not recently prescribed or documented on the patient’s medication list) and/or an unexpected negative result (eg, a prescribed substance not detected). The UDS dashboard identifies these discrepancies by comparing the patient’s current medication list (both VHA and non-VHA) to their UDS results.

The UDS dashboard displays a summary of UDSs performed, unexpected negative results, unexpected positive results, and potential discrepancies. The user may also specify the laboratory type and time frame of interest to limit displayed results. The user can then view patient-specific data for any category. Among the data are the patient’s UDS results and the completion date, detected (or nondetected) substance(s), ordering clinician, associated medication(s) with last fill date and days’ supply, and whether a confirmatory test has been performed in the past year.

VABHHCS uses an extended UDS immunoassay (PROFILE-V, MEDTOX Diagnostics) that reports on 11 substances: opiates, oxycodone, buprenorphine, methadone, amphetamines, methamphetamine, barbiturates, benzodiazepines, cocaine metabolites, cannabinoids (tetrahydrocannabinol [THC]), and phencyclidine. These substances appear on the UDS dashboard. The project protocol initially included monitoring for tramadol but that was later removed because it was not available with this UDS immunoassay.

Pharmacist Process

Either the PMOP coordinator or pharmacy resident monitored the UDS dashboard weekly. Any patients identified as having a potential discrepancy were reviewed. If the discrepancy was determined to be significant, the PMOP coordinator or pharmacy resident would review the patient electronic health record. If warranted, the patient was contacted and asked about newly prescribed medications, missed and recent medication doses, and illicit substance use. Potential interventions during in-depth review included: (1) discussing future actions with the primary care clinician and/or prescriber of the controlled substance; (2) ordering a confirmatory test on the original urine sample; (3) evaluating for sources of potential false-positive results; (4) completing an updated PDMP if not performed within the past year; (5) referring patients for substance use disorder treatment or counseling; or (6) consulting the local narcotics review committee. A progress note was entered into the electronic health record with the findings and any actions taken, and an alert for the primary care clinician and/or prescriber of the controlled substance.

Implementation and Analysis

This quality improvement project spanned 16 weeks from June 2022 through September 2022. Any patient with a UDS that returned with a significant discrepancy was reviewed. The primary outcome was interventions made by the PMOP coordinator or pharmacy resident, as well as time taken to perform the in-depth review of each patient. Patient demographics were also collected. The protocol for this project was approved by the VABHHCS pharmacy and therapeutics committee and was determined to meet guidelines for a nonresearch quality improvement project.

Results

From June 2022 through September 2022, 700 UDSs were performed at VABHHCS with 278 (39.7%) patients identified as having a potential discrepancy based on UDS results. Sixty patients (8.6%) had significant discrepancies that warranted in-depth review. The most common reasons for determining whether a potential discrepancy was not significant included unexpected negatives due to documented non-VA medications no longer being prescribed, unexpected positives due to recent expiration of a controlled substance prescription the patient was still taking, or unexpected positives due to the detection of a substance for which the clinician was already aware. During the 16-week study period, the mean number of patients warranting in-depth review was 4 per week.

The patients were predominantly male with a mean age of 61 years, and most (87%) were prescribed at least 1 controlled substance (mean, 1.1), primarily opioids for pain management (Table 1). Most patients had recent substance risk mitigation with UDS (56%) and PDMP (65%) checks within the past year. Of the 60 patients reviewed with significant UDS discrepancies, 50% had a history of discrepant UDS results. Of the 60 UDS discrepancies, there were 37 unexpected positive results (62%), 17 unexpected negative results (28%), and 10 patients with both positive and negative results (17%). THC was the most frequently detected substance, followed by opiates, benzodiazepines, and amphetamines (Table 2).

Each in-depth review with interventions by the PMOP coordinator or pharmacy resident lasted a mean of 14 minutes (Table 3). Five patients were successfully contacted for an interview and 7 patients could not be contacted. The ordering clinician of the UDS sometimes had contacted these patients prior to the PMOP coordinator or pharmacy resident reviewing the UDS dashboard, eliminating the need for additional follow-up.

The most common pharmacist intervention was discussing future actions with the primary care clinician and/or prescriber of the controlled substance (n = 39; 65%). These conversations resulted in actions such as ordering a repeat UDS with confirmatory testing at a future date or agreeing that the clinician would discuss the results and subsequent actions with the patient at an upcoming visit. Pharmacist interventions also included 25 PDMP queries (42%) and 9 orders of confirmatory UDS on the original urine sample (15%). Only 1 patient was evaluated by the narcotics review committee, which resulted in a controlled substance flag being placed on their profile. No patients were referred to substance use disorder treatment or counseling. It was offered to and declined by 1 patient, and 3 patients were already engaged in these services.

Medication therapies that could contribute to false-positive results were also evaluated. Fourteen patients who tested positive for THC had a prescription for a nonsteroidal anti-inflammatory drug or proton-pump inhibitor, which could have created a false-positive result.⁶ One patient who tested positive for amphetamines had a prescription for phentermine.¹⁶ No other potential false-positive results were identified.

 

 

Discussion

Findings of this project illustrate that the use of a clinical pharmacist to monitor a dashboard of discrepant UDS results created opportunities for collaboration with clinicians and impacted confirmatory testing and PDMP monitoring practices.

At the local level, the process had numerous benefits. First, it was a reasonable amount of workload to generate pharmacist interventions: the PMOP coordinator conducted an average of 4 in-depth reviews weekly, each lasting about 14 minutes. Thus, the UDS dashboard allowed the PMOP coordinator to actively surveil all incoming UDS results for potential discrepancies in about 1 hour each week. Pairing the automation of the UDS dashboard with the clinical judgment of the PMOP coordinator seemed to maximize efficiency. VABHHCS provides primary and secondary medical and surgical care to a rural population of approximately 20,000 patients across 5 states; the time required at facilities that serve a higher volume of patients may be greater.

Second, the project served as an opportunity for the PMOP coordinator to provide case-specific clinician education on UDS monitoring. As medication experts, pharmacists can apply their medication-related knowledge to UDS interpretation. This includes understanding drug metabolism and classification and how they apply to UDS results, as well as recognizing medication therapies that could contribute to false-positive UDS results. Research suggests that clinicians may have gaps in their knowledge and may welcome pharmacist assistance in interpreting UDS results.^7,8

Third, the project helped improve rates of confirmatory testing for those with unexpected positive UDS results. Confirmatory testing should be strongly considered if positive results would have significant implications on the future course of treatment.⁴ The PMOP coordinator ordered a confirmatory test on 9 patients using the same urine sample used to conduct the initial UDS, minimizing the burden on the patient and laboratory staff. Confirmatory testing was limited by the laboratory’s sample retention period; if the need for confirmatory testing was not recognized soon enough, the sample would no longer be available for retesting. Health systems may consider the use of reflexive confirmatory testing with UDS as an alternative approach, although this may come at an additional cost and may not be warranted in many cases (eg, only 39.7% of all potential discrepancies were deemed as significant within our project).

There were notable incidental findings in our quality improvement project. Among patients with a significant discrepancy on UDS, 50% had a history of ≥ 1 discrepant UDS result. This further emphasizes the importance of appropriate use and interpretation of UDS monitoring for all clinicians, as this may prevent prolonged and potentially inappropriate treatment regimens. Secondly, rates of mental health diagnoses among those with a significant UDS discrepancy seemed relatively high compared to population-level data. For example, among veterans, the overall lifetime prevalence of posttraumatic stress disorder is estimated to be 8.0%; in our project, 35% of patients with a significant UDS discrepancy had a posttraumatic stress disorder diagnosis.¹⁷ This relationship may be an area of further study.

Lastly, it was surprising that the overall rates of UDS and PDMP checks within the past year were 56% and 65%, respectively. VABHHCS requires veterans on controlled substances to have these risk-mitigation strategies performed annually, so our suspicion is that many were falling out due to having been most recently evaluated 12 to 16 months prior. This may represent a limitation of our data-collection method, which reviewed only the previous 12 months.

Limitations

This project was carried out over a period of only 4 months. As a result, only 60 patients received an in-depth review from the PMOP coordinator. Second, the timeliness of the intervention seemed crucial, as delayed in-depth reviews resulted in fewer opportunities to order confirmatory tests or collaborate with clinicians prior to devising an updated plan. Additionally, our process called for UDS dashboard monitoring once a week. Given that the laboratory held samples for only 48 hours, twice- or thrice-weekly review of the UDS dashboard would have allowed for more confirmatory testing, along with more immediate clinician collaboration. Most importantly, the outcomes of this project are only presented via descriptive statistics and without the results of any comparison group, making it impossible to draw firm conclusions about this approach compared to standard-care processes.

Conclusions

This quality improvement project has proven to be valuable at VABHHCS and we intend to continue this pharmacist-led process to monitor the UDS dashboard. VABHHCS leadership are also discussing UDS practices more broadly to further enhance patient management. Within the VA, the PMOP coordinator—charged with being the local coordinator of appropriate pain management and opioid safety practices—is well positioned to assume these responsibilities. Outside of the VA, a pain-management clinical pharmacist or any pharmacist embedded within primary care could similarly perform these duties. Previous literature regarding the implementation of clinical dashboards suggests that with the appropriate software engineering teams and infrastructure, this tool could also be feasibly developed and implemented at other health systems relatively quickly.¹⁴

Overall, a pharmacist-led process to efficiently monitor a dashboard of discrepant UDS results led to opportunities for collaboration with clinicians and positively impacted confirmatory testing and PDMP monitoring at a rural VA health system.

Acknowledgments

The authors express their gratitude to Patrick Spoutz, PharmD, BCPS, VISN 20 Pharmacist Executive, for introducing and sharing the UDS dashboard with our team.

Urine drug screen (UDS) monitoring is a common risk-mitigation strategy tool for prescribing controlled substances.^1-3 Not only is UDS monitoring highlighted by clinical practice guidelines for opioid prescribing for chronic pain,^1,2 it has also been suggested as best practice for benzodiazepines³ and a consideration for other controlled substances. Monitoring UDSs helps confirm adherence to the prescribed treatment regimen while also screening for substance use that may increase patient risk.

UDS results can be complex and have profound implications for the patient’s treatment plan. Drug metabolites for opioids are particularly complicated; for example, synthetic and semisynthetic opioids are not detected on routine opiate immunoassays.⁴ This may lead a clinician to falsely assume the patient is not taking their fentanyl or tramadol medication as directed—or potentially even diverting—in the face of a negative opiate result.⁵ Routine UDSs are also subject to the pitfall of false-positive results due to coprescribed medications; for example, bupropion can lead to a false-positive amphetamine result, whereas sertraline can lead to a false-positive benzodiazepine result.⁶ Retrospective reviews of clinician behavior surrounding UDS interpretation have demonstrated knowledge gaps and inconsistent communication practices with patients.^7,8

Given the complexity of UDS interpretation and its close relationship with medications, pharmacists are positioned to play an important role in the process. Pharmacists are embedded in pain-management teams and involved in prescription drug monitoring programs (PDMPs) for many health systems. The Veterans Health Administration (VHA) has supported the hiring of pain management, opioid safety, and PDMP coordinators (PMOP) at its facilities to provide clinical pain-management guidance, support national initiatives, and uphold legislative requirements.⁹ In many facilities, a pharmacist is hired specifically for these positions.

Clinical dashboards have been used by pharmacists in a variety of settings.^10-13 They allow clinicians at a broad level to target interventions needed across a patient population, then produce a list of actionable patients to facilitate delivery of that intervention on an individual level.¹³ Between 2021 and 2022, a clinical dashboard to review potentially discrepant UDS results was made available for use at US Department of Veterans Affairs (VA) medical centers. Evidence exists in primary and specialty care settings that implementation of an opioid-prescribing clinical dashboard improves completion rates of risk-mitigation strategies such as UDS and opioid treatment agreements.^14,15 To our knowledge there is no published research on the use and outcomes of a clinical dashboard that allows users to efficiently review discrepant UDS results when compared to a list of currently prescribed medications.

Given the availability of the UDS dashboard at the VA Black Hills Health Care System (VABHHCS) in South Dakota and the hiring of a PMOP coordinator pharmacist, the aim of this quality improvement project was 2-fold: to implement a pharmacist-led process to monitor the UDS dashboard for potentially discrepant results and to describe the quantity and types of interventions made by the clinical pharmacist leading this process.

Quality Improvement Project

A clinical UDS dashboard was created by the VA Northwest Health Network and made available for use at VHA sites between 2021 and 2022. The UDS dashboard is housed on a secure, Power BI Report Server (Microsoft), with access restricted to only those with patient health data privileges. The dashboard identifies all local patients with a UDS that returned with a potential discrepancy, defined as an unexpected positive result (eg, a detected substance not recently prescribed or documented on the patient’s medication list) and/or an unexpected negative result (eg, a prescribed substance not detected). The UDS dashboard identifies these discrepancies by comparing the patient’s current medication list (both VHA and non-VHA) to their UDS results.

The UDS dashboard displays a summary of UDSs performed, unexpected negative results, unexpected positive results, and potential discrepancies. The user may also specify the laboratory type and time frame of interest to limit displayed results. The user can then view patient-specific data for any category. Among the data are the patient’s UDS results and the completion date, detected (or nondetected) substance(s), ordering clinician, associated medication(s) with last fill date and days’ supply, and whether a confirmatory test has been performed in the past year.

VABHHCS uses an extended UDS immunoassay (PROFILE-V, MEDTOX Diagnostics) that reports on 11 substances: opiates, oxycodone, buprenorphine, methadone, amphetamines, methamphetamine, barbiturates, benzodiazepines, cocaine metabolites, cannabinoids (tetrahydrocannabinol [THC]), and phencyclidine. These substances appear on the UDS dashboard. The project protocol initially included monitoring for tramadol but that was later removed because it was not available with this UDS immunoassay.

Pharmacist Process

Either the PMOP coordinator or pharmacy resident monitored the UDS dashboard weekly. Any patients identified as having a potential discrepancy were reviewed. If the discrepancy was determined to be significant, the PMOP coordinator or pharmacy resident would review the patient electronic health record. If warranted, the patient was contacted and asked about newly prescribed medications, missed and recent medication doses, and illicit substance use. Potential interventions during in-depth review included: (1) discussing future actions with the primary care clinician and/or prescriber of the controlled substance; (2) ordering a confirmatory test on the original urine sample; (3) evaluating for sources of potential false-positive results; (4) completing an updated PDMP if not performed within the past year; (5) referring patients for substance use disorder treatment or counseling; or (6) consulting the local narcotics review committee. A progress note was entered into the electronic health record with the findings and any actions taken, and an alert for the primary care clinician and/or prescriber of the controlled substance.

Implementation and Analysis

This quality improvement project spanned 16 weeks from June 2022 through September 2022. Any patient with a UDS that returned with a significant discrepancy was reviewed. The primary outcome was interventions made by the PMOP coordinator or pharmacy resident, as well as time taken to perform the in-depth review of each patient. Patient demographics were also collected. The protocol for this project was approved by the VABHHCS pharmacy and therapeutics committee and was determined to meet guidelines for a nonresearch quality improvement project.

Results

From June 2022 through September 2022, 700 UDSs were performed at VABHHCS with 278 (39.7%) patients identified as having a potential discrepancy based on UDS results. Sixty patients (8.6%) had significant discrepancies that warranted in-depth review. The most common reasons for determining whether a potential discrepancy was not significant included unexpected negatives due to documented non-VA medications no longer being prescribed, unexpected positives due to recent expiration of a controlled substance prescription the patient was still taking, or unexpected positives due to the detection of a substance for which the clinician was already aware. During the 16-week study period, the mean number of patients warranting in-depth review was 4 per week.

The patients were predominantly male with a mean age of 61 years, and most (87%) were prescribed at least 1 controlled substance (mean, 1.1), primarily opioids for pain management (Table 1). Most patients had recent substance risk mitigation with UDS (56%) and PDMP (65%) checks within the past year. Of the 60 patients reviewed with significant UDS discrepancies, 50% had a history of discrepant UDS results. Of the 60 UDS discrepancies, there were 37 unexpected positive results (62%), 17 unexpected negative results (28%), and 10 patients with both positive and negative results (17%). THC was the most frequently detected substance, followed by opiates, benzodiazepines, and amphetamines (Table 2).

Each in-depth review with interventions by the PMOP coordinator or pharmacy resident lasted a mean of 14 minutes (Table 3). Five patients were successfully contacted for an interview and 7 patients could not be contacted. The ordering clinician of the UDS sometimes had contacted these patients prior to the PMOP coordinator or pharmacy resident reviewing the UDS dashboard, eliminating the need for additional follow-up.

The most common pharmacist intervention was discussing future actions with the primary care clinician and/or prescriber of the controlled substance (n = 39; 65%). These conversations resulted in actions such as ordering a repeat UDS with confirmatory testing at a future date or agreeing that the clinician would discuss the results and subsequent actions with the patient at an upcoming visit. Pharmacist interventions also included 25 PDMP queries (42%) and 9 orders of confirmatory UDS on the original urine sample (15%). Only 1 patient was evaluated by the narcotics review committee, which resulted in a controlled substance flag being placed on their profile. No patients were referred to substance use disorder treatment or counseling. It was offered to and declined by 1 patient, and 3 patients were already engaged in these services.

Medication therapies that could contribute to false-positive results were also evaluated. Fourteen patients who tested positive for THC had a prescription for a nonsteroidal anti-inflammatory drug or proton-pump inhibitor, which could have created a false-positive result.⁶ One patient who tested positive for amphetamines had a prescription for phentermine.¹⁶ No other potential false-positive results were identified.

 

 

Discussion

Findings of this project illustrate that the use of a clinical pharmacist to monitor a dashboard of discrepant UDS results created opportunities for collaboration with clinicians and impacted confirmatory testing and PDMP monitoring practices.

At the local level, the process had numerous benefits. First, it was a reasonable amount of workload to generate pharmacist interventions: the PMOP coordinator conducted an average of 4 in-depth reviews weekly, each lasting about 14 minutes. Thus, the UDS dashboard allowed the PMOP coordinator to actively surveil all incoming UDS results for potential discrepancies in about 1 hour each week. Pairing the automation of the UDS dashboard with the clinical judgment of the PMOP coordinator seemed to maximize efficiency. VABHHCS provides primary and secondary medical and surgical care to a rural population of approximately 20,000 patients across 5 states; the time required at facilities that serve a higher volume of patients may be greater.

Second, the project served as an opportunity for the PMOP coordinator to provide case-specific clinician education on UDS monitoring. As medication experts, pharmacists can apply their medication-related knowledge to UDS interpretation. This includes understanding drug metabolism and classification and how they apply to UDS results, as well as recognizing medication therapies that could contribute to false-positive UDS results. Research suggests that clinicians may have gaps in their knowledge and may welcome pharmacist assistance in interpreting UDS results.^7,8

Third, the project helped improve rates of confirmatory testing for those with unexpected positive UDS results. Confirmatory testing should be strongly considered if positive results would have significant implications on the future course of treatment.⁴ The PMOP coordinator ordered a confirmatory test on 9 patients using the same urine sample used to conduct the initial UDS, minimizing the burden on the patient and laboratory staff. Confirmatory testing was limited by the laboratory’s sample retention period; if the need for confirmatory testing was not recognized soon enough, the sample would no longer be available for retesting. Health systems may consider the use of reflexive confirmatory testing with UDS as an alternative approach, although this may come at an additional cost and may not be warranted in many cases (eg, only 39.7% of all potential discrepancies were deemed as significant within our project).

There were notable incidental findings in our quality improvement project. Among patients with a significant discrepancy on UDS, 50% had a history of ≥ 1 discrepant UDS result. This further emphasizes the importance of appropriate use and interpretation of UDS monitoring for all clinicians, as this may prevent prolonged and potentially inappropriate treatment regimens. Secondly, rates of mental health diagnoses among those with a significant UDS discrepancy seemed relatively high compared to population-level data. For example, among veterans, the overall lifetime prevalence of posttraumatic stress disorder is estimated to be 8.0%; in our project, 35% of patients with a significant UDS discrepancy had a posttraumatic stress disorder diagnosis.¹⁷ This relationship may be an area of further study.

Lastly, it was surprising that the overall rates of UDS and PDMP checks within the past year were 56% and 65%, respectively. VABHHCS requires veterans on controlled substances to have these risk-mitigation strategies performed annually, so our suspicion is that many were falling out due to having been most recently evaluated 12 to 16 months prior. This may represent a limitation of our data-collection method, which reviewed only the previous 12 months.

Limitations

This project was carried out over a period of only 4 months. As a result, only 60 patients received an in-depth review from the PMOP coordinator. Second, the timeliness of the intervention seemed crucial, as delayed in-depth reviews resulted in fewer opportunities to order confirmatory tests or collaborate with clinicians prior to devising an updated plan. Additionally, our process called for UDS dashboard monitoring once a week. Given that the laboratory held samples for only 48 hours, twice- or thrice-weekly review of the UDS dashboard would have allowed for more confirmatory testing, along with more immediate clinician collaboration. Most importantly, the outcomes of this project are only presented via descriptive statistics and without the results of any comparison group, making it impossible to draw firm conclusions about this approach compared to standard-care processes.

Conclusions

This quality improvement project has proven to be valuable at VABHHCS and we intend to continue this pharmacist-led process to monitor the UDS dashboard. VABHHCS leadership are also discussing UDS practices more broadly to further enhance patient management. Within the VA, the PMOP coordinator—charged with being the local coordinator of appropriate pain management and opioid safety practices—is well positioned to assume these responsibilities. Outside of the VA, a pain-management clinical pharmacist or any pharmacist embedded within primary care could similarly perform these duties. Previous literature regarding the implementation of clinical dashboards suggests that with the appropriate software engineering teams and infrastructure, this tool could also be feasibly developed and implemented at other health systems relatively quickly.¹⁴

Overall, a pharmacist-led process to efficiently monitor a dashboard of discrepant UDS results led to opportunities for collaboration with clinicians and positively impacted confirmatory testing and PDMP monitoring at a rural VA health system.

Acknowledgments

The authors express their gratitude to Patrick Spoutz, PharmD, BCPS, VISN 20 Pharmacist Executive, for introducing and sharing the UDS dashboard with our team.

Although 2.4 million adults in the United States have been diagnosed with hepatitis C virus (HCV) infection, it remains underdiagnosed and undertreated, particularly among difficult to reach populations, such as persons who inject drugs, marginally housed individuals, correctional populations, and pregnant women.¹ Though the US Preventive Services Task Force (USPSTF) broadened HCV screening recommendations to include individuals aged 18 to 79 years, rates of new HCV prescriptions sharply declined during the COVID-19 pandemic.^2,3

During the pandemic, many health care systems adopted virtual health care modalities. Within the Veteran Health Administration (VHA), there was an 11-fold increase in virtual encounters. However, veterans aged > 45 years, homeless, and had other insurance were less likely to utilize virtual care.^4,5 As health care delivery continues to evolve, health systems must adapt and test innovative models for the treatment of HCV.

There is limited understanding of HCV treatments when exclusively conducted virtually. The aim of this study was to evaluate the effects of the HCV treatment program at the Veterans Affairs Greater Los Angeles Healthcare System (VAGLAHS) during the early phase of the COVID-19 pandemic, when telehealth modalities and mail-order prescriptions were used for HCV diagnosis and treatment. The secondary aim of this study was to understand patient factors associated with treatment initiation and discontinuation for patients using telehealth.

Methods

The VHA is the largest provider of HCV care in the US.⁶ At VAGLAHS, veterans with HCV are referred for evaluation to a viral hepatitis clinic staffed by gastroenterologists and infectious disease specialists. Veterans with detectable HCV on an HCV RNA test have an additional workup ordered if necessary and are referred to an HCV-specialist pharmacist or physician’s assistant to start treatment. In March 2020, all HCV evaluations and treatment initiation in the viral hepatitis clinic started being conducted exclusively via telehealth. This was the primary modality of HCV evaluations and treatment initiation until COVID-19 restrictions were lifted to permit in-person evaluations. Prescriptions were delivered by mail to patients following treatment initiation appointments.

We retrospectively reviewed electronic health records of veterans referred to start treatment March 1, 2020, through September 30, 2020. The endpoint of the reviewed records was set because during this specific time frame, VAGLAHS used an exclusively telehealth-based model for HCV evaluation and treatment. Patients were followed until June 15, 2021. Due to evolving COVID-19 restrictions at the time, and despite requests received, treatment initiations by the pharmacy team were suspended in March 2020 but HCV treatments resumed in May. Data collected included baseline demographics (age, sex, race, ethnicity, housing status, distance to VAGLAHS), comorbidities (cirrhosis, hepatitis B virus coinfection, HIV coinfection), psychiatric conditions (mood or psychotic disorder, alcohol use disorder [AUD], opioid use disorder), and treatment characteristics (HCV genotype, HCV treatment regimen, baseline viral load). Distance from the patient’s home to VAGLAHS was calculated using CDXZipStream software. Comorbidities and psychiatric conditions were identified by the presence of the appropriate diagnosis via International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes in the health record and confirmed by review of clinician notes. Active AUD was defined as: (1) the presence of AUD diagnosis code; (2) AUD Identification Test-Consumption (AUDIT-C) score of high or severe risk based on established cutoffs; and (3) active alcohol use noted in the electronic health record. All patients had an AUDIT-C score completed within 1 year of initiating treatment. Opioid use disorder was defined by the presence of diagnostic codes for opioid dependence or opioid abuse.

The reasons for treatment noninitiation and discontinuation were each captured. We calculated descriptive statistics to analyze the frequency distributions of all variables. Independent t tests were used to analyze continuous data and Pearson χ² test was used to analyze categorical data. Statistical significance was set as P < .05.

Results

From March 1, 2020, through September 30, 2020, 73 veterans were referred to the HCV clinical pharmacist for treatment (Figure). Forty-three veterans (59%) initiated HCV treatment and 34 (79%) completed the full treatment course (Table 1). Twenty-five patients (65%) had their sustained virologic response at 12 weeks (SVR12) testing and 22 patients achieved SVR12 (88%; 30% of total sample). One patient did not achieve SVR, and 2 patients died (variceal hemorrhage and progression of cerebral amyloidosis/function decline) before the completion of laboratory testing. From March 2020 to May 2020, HCV treatments requests were paused as new COVID-19 policies were being introduced; 33 patients were referred during this time and 21 initiated treatment.

Veterans that did not start HCV treatment had a significantly higher rate of active AUD when compared with those that initiated treatment: 30% vs 9% (P = .02). Of the patients who started and discontinued treatment, none had active AUD. Other baseline demographics, clinical characteristics, and treatment characteristics were similar between the groups. No patient demographic characteristics were significantly associated with HCV treatment discontinuation. We did not observe any major health disparities in initiation or discontinuation by sex, race, ethnicity, or geography. Eleven patients (37%) could not be contacted, which was the most common reason veterans did not initiate treatment (Table 2). Of the 9 patients that did not complete SVR12, 5 patients could not be contacted for follow-up, which was the most common reason veterans discontinued treatment.

 

 

Discussion

This study highlights the experience of treating patients with HCV with an exclusively telehealth model in the months following implementation of stay-at-home orders from March 19, 2020, to September 30, 2020, during the COVID-19 pandemic at VAGLAHS. We were able to successfully complete treatment for 34 veterans (47%) and achieved SVR rates of 88%. We found that AUD was associated with unsuccessful treatment initiation. There were no statistically significant patient characteristic findings for treatment discontinuation in our study (Table 3). Unhealthy alcohol use and AUD are highly prevalent among veterans with HCV and prior to the pandemic, studies have demonstrated AUD as a barrier to HCV treatment.⁷

Since worse hepatic outcomes have been observed in veterans with HCV and AUD and increased harmful patterns of drinking occurred during the pandemic, a renewed interest in treating AUD in these veterans during the era of telehealth is critical.⁸ While we were unable to ascertain whether alcohol misuse in our cohort increased during the pandemic or whether changes in drinking patterns affected HCV treatment outcomes before and after the pandemic, such an association should reinforce the need for clinicians to expeditiously link patients to substance use care. It should also stimulate further considerations of addressing social determinants of health not captured in this study.

During the pandemic, veterans with posttraumatic stress disorder, a history of serving in combat roles, and experiencing related financial stressors had higher risk of AUD.^9,10 For veterans with AUD who initiated HCV treatment, none discontinued their therapy, aligning with other studies showed that patients with AUD were able to achieve high rates of SVR and emphasizing that veterans should be treated irrespective of an AUD diagnosis.¹¹ However, more innovative engagement initiatives for veterans with AUD should be explored as we continue to adapt more telehealth-based care for HCV direct-acting antiviral treatments. A more in-depth understanding of how alcohol use relates to treatment noninitiation is warranted, as this may stem from behavioral patterns that could not be captured in the present study.

The inability to reach veterans by telephone was a major reason for noninitiation and discontinuation of treatment. While the expansion of telehealth services has been noted across the VHA, there is still room for improving methods of engaging veterans in health care postpandemic.¹² Prior studies in veteran populations that were successful in increasing uptake of HCV treatment have employed telehealth strategies that further emphasizes its integral role in HCV elimination.¹³ Although our study did not show mental health comorbidities and housing status as statistically significant, it is important to note that 20% of patients referred for HCV treatment had an incomplete evaluation which can lead to potentially unobserved indicators not captured by our study such as quality of linkage to care. It is imperative to stress the best practices for HCV initiation by integrating a multidisciplinary team to address patients’ psychosocial comorbidities.¹⁴ Finally, we did not observe any major disparities in treating veterans with HCV during the pandemic. This observation is reassuring and consistent with other VHA data given the heightened recognition of health disparities seen in health care sectors across the country, especially evident during the COVID-19 pandemic and the current era of increased adaptation of telehealth.

Limitations

Limitations to this study include its retrospective nature, small sample size, and short study time frame as a proportion of veterans have yet to complete HCV treatment which can potentially explain how larger studies were able to find other statistically significant patient-related factors impacting treatment initiation compared to ours. Given the lack of universal standardized diagnostic criterion of AUD, this can limit how our study can be compared to others in similar populations. Additionally, this study was conducted at a single facility with a predominantly older male veteran population, which may not be generalizable to other populations.

Conclusions

Treating HCV during the COVID-19 pandemic with telehealth and mail-out medications was feasible and led to high SVR rates, but unhealthy alcohol use and an inability to contact veterans were predominant barriers to success. Future quality improvement efforts should focus on addressing these barriers and exploring the relationship between alcohol use and HCV treatment initiation.

Although 2.4 million adults in the United States have been diagnosed with hepatitis C virus (HCV) infection, it remains underdiagnosed and undertreated, particularly among difficult to reach populations, such as persons who inject drugs, marginally housed individuals, correctional populations, and pregnant women.¹ Though the US Preventive Services Task Force (USPSTF) broadened HCV screening recommendations to include individuals aged 18 to 79 years, rates of new HCV prescriptions sharply declined during the COVID-19 pandemic.^2,3

During the pandemic, many health care systems adopted virtual health care modalities. Within the Veteran Health Administration (VHA), there was an 11-fold increase in virtual encounters. However, veterans aged > 45 years, homeless, and had other insurance were less likely to utilize virtual care.^4,5 As health care delivery continues to evolve, health systems must adapt and test innovative models for the treatment of HCV.

There is limited understanding of HCV treatments when exclusively conducted virtually. The aim of this study was to evaluate the effects of the HCV treatment program at the Veterans Affairs Greater Los Angeles Healthcare System (VAGLAHS) during the early phase of the COVID-19 pandemic, when telehealth modalities and mail-order prescriptions were used for HCV diagnosis and treatment. The secondary aim of this study was to understand patient factors associated with treatment initiation and discontinuation for patients using telehealth.

Methods

The VHA is the largest provider of HCV care in the US.⁶ At VAGLAHS, veterans with HCV are referred for evaluation to a viral hepatitis clinic staffed by gastroenterologists and infectious disease specialists. Veterans with detectable HCV on an HCV RNA test have an additional workup ordered if necessary and are referred to an HCV-specialist pharmacist or physician’s assistant to start treatment. In March 2020, all HCV evaluations and treatment initiation in the viral hepatitis clinic started being conducted exclusively via telehealth. This was the primary modality of HCV evaluations and treatment initiation until COVID-19 restrictions were lifted to permit in-person evaluations. Prescriptions were delivered by mail to patients following treatment initiation appointments.

We retrospectively reviewed electronic health records of veterans referred to start treatment March 1, 2020, through September 30, 2020. The endpoint of the reviewed records was set because during this specific time frame, VAGLAHS used an exclusively telehealth-based model for HCV evaluation and treatment. Patients were followed until June 15, 2021. Due to evolving COVID-19 restrictions at the time, and despite requests received, treatment initiations by the pharmacy team were suspended in March 2020 but HCV treatments resumed in May. Data collected included baseline demographics (age, sex, race, ethnicity, housing status, distance to VAGLAHS), comorbidities (cirrhosis, hepatitis B virus coinfection, HIV coinfection), psychiatric conditions (mood or psychotic disorder, alcohol use disorder [AUD], opioid use disorder), and treatment characteristics (HCV genotype, HCV treatment regimen, baseline viral load). Distance from the patient’s home to VAGLAHS was calculated using CDXZipStream software. Comorbidities and psychiatric conditions were identified by the presence of the appropriate diagnosis via International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes in the health record and confirmed by review of clinician notes. Active AUD was defined as: (1) the presence of AUD diagnosis code; (2) AUD Identification Test-Consumption (AUDIT-C) score of high or severe risk based on established cutoffs; and (3) active alcohol use noted in the electronic health record. All patients had an AUDIT-C score completed within 1 year of initiating treatment. Opioid use disorder was defined by the presence of diagnostic codes for opioid dependence or opioid abuse.

The reasons for treatment noninitiation and discontinuation were each captured. We calculated descriptive statistics to analyze the frequency distributions of all variables. Independent t tests were used to analyze continuous data and Pearson χ² test was used to analyze categorical data. Statistical significance was set as P < .05.

Results

From March 1, 2020, through September 30, 2020, 73 veterans were referred to the HCV clinical pharmacist for treatment (Figure). Forty-three veterans (59%) initiated HCV treatment and 34 (79%) completed the full treatment course (Table 1). Twenty-five patients (65%) had their sustained virologic response at 12 weeks (SVR12) testing and 22 patients achieved SVR12 (88%; 30% of total sample). One patient did not achieve SVR, and 2 patients died (variceal hemorrhage and progression of cerebral amyloidosis/function decline) before the completion of laboratory testing. From March 2020 to May 2020, HCV treatments requests were paused as new COVID-19 policies were being introduced; 33 patients were referred during this time and 21 initiated treatment.

Veterans that did not start HCV treatment had a significantly higher rate of active AUD when compared with those that initiated treatment: 30% vs 9% (P = .02). Of the patients who started and discontinued treatment, none had active AUD. Other baseline demographics, clinical characteristics, and treatment characteristics were similar between the groups. No patient demographic characteristics were significantly associated with HCV treatment discontinuation. We did not observe any major health disparities in initiation or discontinuation by sex, race, ethnicity, or geography. Eleven patients (37%) could not be contacted, which was the most common reason veterans did not initiate treatment (Table 2). Of the 9 patients that did not complete SVR12, 5 patients could not be contacted for follow-up, which was the most common reason veterans discontinued treatment.

 

 

Discussion

This study highlights the experience of treating patients with HCV with an exclusively telehealth model in the months following implementation of stay-at-home orders from March 19, 2020, to September 30, 2020, during the COVID-19 pandemic at VAGLAHS. We were able to successfully complete treatment for 34 veterans (47%) and achieved SVR rates of 88%. We found that AUD was associated with unsuccessful treatment initiation. There were no statistically significant patient characteristic findings for treatment discontinuation in our study (Table 3). Unhealthy alcohol use and AUD are highly prevalent among veterans with HCV and prior to the pandemic, studies have demonstrated AUD as a barrier to HCV treatment.⁷

Since worse hepatic outcomes have been observed in veterans with HCV and AUD and increased harmful patterns of drinking occurred during the pandemic, a renewed interest in treating AUD in these veterans during the era of telehealth is critical.⁸ While we were unable to ascertain whether alcohol misuse in our cohort increased during the pandemic or whether changes in drinking patterns affected HCV treatment outcomes before and after the pandemic, such an association should reinforce the need for clinicians to expeditiously link patients to substance use care. It should also stimulate further considerations of addressing social determinants of health not captured in this study.

During the pandemic, veterans with posttraumatic stress disorder, a history of serving in combat roles, and experiencing related financial stressors had higher risk of AUD.^9,10 For veterans with AUD who initiated HCV treatment, none discontinued their therapy, aligning with other studies showed that patients with AUD were able to achieve high rates of SVR and emphasizing that veterans should be treated irrespective of an AUD diagnosis.¹¹ However, more innovative engagement initiatives for veterans with AUD should be explored as we continue to adapt more telehealth-based care for HCV direct-acting antiviral treatments. A more in-depth understanding of how alcohol use relates to treatment noninitiation is warranted, as this may stem from behavioral patterns that could not be captured in the present study.

The inability to reach veterans by telephone was a major reason for noninitiation and discontinuation of treatment. While the expansion of telehealth services has been noted across the VHA, there is still room for improving methods of engaging veterans in health care postpandemic.¹² Prior studies in veteran populations that were successful in increasing uptake of HCV treatment have employed telehealth strategies that further emphasizes its integral role in HCV elimination.¹³ Although our study did not show mental health comorbidities and housing status as statistically significant, it is important to note that 20% of patients referred for HCV treatment had an incomplete evaluation which can lead to potentially unobserved indicators not captured by our study such as quality of linkage to care. It is imperative to stress the best practices for HCV initiation by integrating a multidisciplinary team to address patients’ psychosocial comorbidities.¹⁴ Finally, we did not observe any major disparities in treating veterans with HCV during the pandemic. This observation is reassuring and consistent with other VHA data given the heightened recognition of health disparities seen in health care sectors across the country, especially evident during the COVID-19 pandemic and the current era of increased adaptation of telehealth.

Limitations

Limitations to this study include its retrospective nature, small sample size, and short study time frame as a proportion of veterans have yet to complete HCV treatment which can potentially explain how larger studies were able to find other statistically significant patient-related factors impacting treatment initiation compared to ours. Given the lack of universal standardized diagnostic criterion of AUD, this can limit how our study can be compared to others in similar populations. Additionally, this study was conducted at a single facility with a predominantly older male veteran population, which may not be generalizable to other populations.

Conclusions

Treating HCV during the COVID-19 pandemic with telehealth and mail-out medications was feasible and led to high SVR rates, but unhealthy alcohol use and an inability to contact veterans were predominant barriers to success. Future quality improvement efforts should focus on addressing these barriers and exploring the relationship between alcohol use and HCV treatment initiation.

Although 2.4 million adults in the United States have been diagnosed with hepatitis C virus (HCV) infection, it remains underdiagnosed and undertreated, particularly among difficult to reach populations, such as persons who inject drugs, marginally housed individuals, correctional populations, and pregnant women.¹ Though the US Preventive Services Task Force (USPSTF) broadened HCV screening recommendations to include individuals aged 18 to 79 years, rates of new HCV prescriptions sharply declined during the COVID-19 pandemic.^2,3

During the pandemic, many health care systems adopted virtual health care modalities. Within the Veteran Health Administration (VHA), there was an 11-fold increase in virtual encounters. However, veterans aged > 45 years, homeless, and had other insurance were less likely to utilize virtual care.^4,5 As health care delivery continues to evolve, health systems must adapt and test innovative models for the treatment of HCV.

There is limited understanding of HCV treatments when exclusively conducted virtually. The aim of this study was to evaluate the effects of the HCV treatment program at the Veterans Affairs Greater Los Angeles Healthcare System (VAGLAHS) during the early phase of the COVID-19 pandemic, when telehealth modalities and mail-order prescriptions were used for HCV diagnosis and treatment. The secondary aim of this study was to understand patient factors associated with treatment initiation and discontinuation for patients using telehealth.

Methods

The VHA is the largest provider of HCV care in the US.⁶ At VAGLAHS, veterans with HCV are referred for evaluation to a viral hepatitis clinic staffed by gastroenterologists and infectious disease specialists. Veterans with detectable HCV on an HCV RNA test have an additional workup ordered if necessary and are referred to an HCV-specialist pharmacist or physician’s assistant to start treatment. In March 2020, all HCV evaluations and treatment initiation in the viral hepatitis clinic started being conducted exclusively via telehealth. This was the primary modality of HCV evaluations and treatment initiation until COVID-19 restrictions were lifted to permit in-person evaluations. Prescriptions were delivered by mail to patients following treatment initiation appointments.

We retrospectively reviewed electronic health records of veterans referred to start treatment March 1, 2020, through September 30, 2020. The endpoint of the reviewed records was set because during this specific time frame, VAGLAHS used an exclusively telehealth-based model for HCV evaluation and treatment. Patients were followed until June 15, 2021. Due to evolving COVID-19 restrictions at the time, and despite requests received, treatment initiations by the pharmacy team were suspended in March 2020 but HCV treatments resumed in May. Data collected included baseline demographics (age, sex, race, ethnicity, housing status, distance to VAGLAHS), comorbidities (cirrhosis, hepatitis B virus coinfection, HIV coinfection), psychiatric conditions (mood or psychotic disorder, alcohol use disorder [AUD], opioid use disorder), and treatment characteristics (HCV genotype, HCV treatment regimen, baseline viral load). Distance from the patient’s home to VAGLAHS was calculated using CDXZipStream software. Comorbidities and psychiatric conditions were identified by the presence of the appropriate diagnosis via International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes in the health record and confirmed by review of clinician notes. Active AUD was defined as: (1) the presence of AUD diagnosis code; (2) AUD Identification Test-Consumption (AUDIT-C) score of high or severe risk based on established cutoffs; and (3) active alcohol use noted in the electronic health record. All patients had an AUDIT-C score completed within 1 year of initiating treatment. Opioid use disorder was defined by the presence of diagnostic codes for opioid dependence or opioid abuse.

The reasons for treatment noninitiation and discontinuation were each captured. We calculated descriptive statistics to analyze the frequency distributions of all variables. Independent t tests were used to analyze continuous data and Pearson χ² test was used to analyze categorical data. Statistical significance was set as P < .05.

Results

From March 1, 2020, through September 30, 2020, 73 veterans were referred to the HCV clinical pharmacist for treatment (Figure). Forty-three veterans (59%) initiated HCV treatment and 34 (79%) completed the full treatment course (Table 1). Twenty-five patients (65%) had their sustained virologic response at 12 weeks (SVR12) testing and 22 patients achieved SVR12 (88%; 30% of total sample). One patient did not achieve SVR, and 2 patients died (variceal hemorrhage and progression of cerebral amyloidosis/function decline) before the completion of laboratory testing. From March 2020 to May 2020, HCV treatments requests were paused as new COVID-19 policies were being introduced; 33 patients were referred during this time and 21 initiated treatment.

Veterans that did not start HCV treatment had a significantly higher rate of active AUD when compared with those that initiated treatment: 30% vs 9% (P = .02). Of the patients who started and discontinued treatment, none had active AUD. Other baseline demographics, clinical characteristics, and treatment characteristics were similar between the groups. No patient demographic characteristics were significantly associated with HCV treatment discontinuation. We did not observe any major health disparities in initiation or discontinuation by sex, race, ethnicity, or geography. Eleven patients (37%) could not be contacted, which was the most common reason veterans did not initiate treatment (Table 2). Of the 9 patients that did not complete SVR12, 5 patients could not be contacted for follow-up, which was the most common reason veterans discontinued treatment.

 

 

Discussion

This study highlights the experience of treating patients with HCV with an exclusively telehealth model in the months following implementation of stay-at-home orders from March 19, 2020, to September 30, 2020, during the COVID-19 pandemic at VAGLAHS. We were able to successfully complete treatment for 34 veterans (47%) and achieved SVR rates of 88%. We found that AUD was associated with unsuccessful treatment initiation. There were no statistically significant patient characteristic findings for treatment discontinuation in our study (Table 3). Unhealthy alcohol use and AUD are highly prevalent among veterans with HCV and prior to the pandemic, studies have demonstrated AUD as a barrier to HCV treatment.⁷

Since worse hepatic outcomes have been observed in veterans with HCV and AUD and increased harmful patterns of drinking occurred during the pandemic, a renewed interest in treating AUD in these veterans during the era of telehealth is critical.⁸ While we were unable to ascertain whether alcohol misuse in our cohort increased during the pandemic or whether changes in drinking patterns affected HCV treatment outcomes before and after the pandemic, such an association should reinforce the need for clinicians to expeditiously link patients to substance use care. It should also stimulate further considerations of addressing social determinants of health not captured in this study.

During the pandemic, veterans with posttraumatic stress disorder, a history of serving in combat roles, and experiencing related financial stressors had higher risk of AUD.^9,10 For veterans with AUD who initiated HCV treatment, none discontinued their therapy, aligning with other studies showed that patients with AUD were able to achieve high rates of SVR and emphasizing that veterans should be treated irrespective of an AUD diagnosis.¹¹ However, more innovative engagement initiatives for veterans with AUD should be explored as we continue to adapt more telehealth-based care for HCV direct-acting antiviral treatments. A more in-depth understanding of how alcohol use relates to treatment noninitiation is warranted, as this may stem from behavioral patterns that could not be captured in the present study.

The inability to reach veterans by telephone was a major reason for noninitiation and discontinuation of treatment. While the expansion of telehealth services has been noted across the VHA, there is still room for improving methods of engaging veterans in health care postpandemic.¹² Prior studies in veteran populations that were successful in increasing uptake of HCV treatment have employed telehealth strategies that further emphasizes its integral role in HCV elimination.¹³ Although our study did not show mental health comorbidities and housing status as statistically significant, it is important to note that 20% of patients referred for HCV treatment had an incomplete evaluation which can lead to potentially unobserved indicators not captured by our study such as quality of linkage to care. It is imperative to stress the best practices for HCV initiation by integrating a multidisciplinary team to address patients’ psychosocial comorbidities.¹⁴ Finally, we did not observe any major disparities in treating veterans with HCV during the pandemic. This observation is reassuring and consistent with other VHA data given the heightened recognition of health disparities seen in health care sectors across the country, especially evident during the COVID-19 pandemic and the current era of increased adaptation of telehealth.

Limitations

Limitations to this study include its retrospective nature, small sample size, and short study time frame as a proportion of veterans have yet to complete HCV treatment which can potentially explain how larger studies were able to find other statistically significant patient-related factors impacting treatment initiation compared to ours. Given the lack of universal standardized diagnostic criterion of AUD, this can limit how our study can be compared to others in similar populations. Additionally, this study was conducted at a single facility with a predominantly older male veteran population, which may not be generalizable to other populations.

Conclusions

Treating HCV during the COVID-19 pandemic with telehealth and mail-out medications was feasible and led to high SVR rates, but unhealthy alcohol use and an inability to contact veterans were predominant barriers to success. Future quality improvement efforts should focus on addressing these barriers and exploring the relationship between alcohol use and HCV treatment initiation.

Nonadherence to medications is an issue across health care. In endocrinology, hypothyroidism, a deficiency of thyroid hormones, is most often treated with levothyroxine and if left untreated can lead to myxedema coma, which can lead to death due to multiorgan dysfunction.¹ Therefore, adherence to levothyroxine is very important in preventing fatal complications.

We present the case of a patient with persistent primary hypothyroidism who was suspected to be nonadherent to levothyroxine, although the patient consistently claimed adherence. The patient’s plasma thyrotropin (TSH) level improved to reference range after 6 weeks of weekly primary care clinic visits. After stopping the visits, his plasma TSH level increased again, so 9 more weeks of visits resumed, which again helped bring down his plasma TSH levels.

Case Presentation

A male patient aged 67 years presented to the Dayton Veterans Affairs Medical Center (VAMC) endocrinology clinic for evaluation of thyroid nodules. The patient reported no history of neck irradiation and a physical examination was unremarkable. At that time, laboratory results showed a slightly elevated plasma TSH level of 4.35 uIU/mL (reference range, 0.35-4.00 uIU/mL) and normal free thyroxine (T4) of 1.00 ng/dL (reference range, 0.74-1.46 ng/dL). Later that year, the patient underwent a total thyroidectomy at the Cincinnati VAMC for Hurthle cell variant papillary thyroid carcinoma that was noted on biopsy at the Dayton VAMC. After surgical pathology results were available, the patient started levothyroxine 200 mcg daily, although 224 mcg would have been more appropriate based on his 142 kg weight. Due to a history of arrhythmia, the goal plasma TSH level was 0.10 to 0.50 uIU/mL. The patient subsequently underwent radioactive iodine ablation. After levothyroxine dose adjustments, the patient’s plasma TSH level was noted to be within his target range at 0.28 uIU/mL 3 months postablation.

Over the next 5 years the patient had regular laboratory tests during which his plasma TSH level rose and were typically high despite adjusting levothyroxine doses between 200 mcg and 325 mcg. The patient received counseling on taking the medication in the morning on an empty stomach and waiting at least 1 hour before consuming anything, and he went to many follow-up visits at the Dayton VAMC endocrinology clinic. He reported no vomiting or diarrhea but endorsed weight gain once. The patient also had high free T4 at times and did not take extra levothyroxine before undergoing laboratory tests.

Nonadherence to levothyroxine was suspected, but the patient insisted he was adherent. He received the medication in the mail regularly, generally had 90-day refills unless a dose change was made, used a pill box, and had social support from his son, but he did not use a phone alarm to remind him to take it. A home care nurse made weekly visits to make sure the remaining levothyroxine pill counts were correct; however, the patient continued to have difficulty maintaining daily adherence at home as indicated by the nurse’s pill counts not aligning with the number of pills which should have been left if the patient was talking the pills daily.

The patient was asked to visit a local community-based outpatient clinic (CBOC) weekly (to avoid patient travel time to Dayton VAMC > 1 hour) to check pill counts and assess adherence. The patient went to the CBOC clinic for these visits, during which pill counts indicated much better but not 100% adherence. After 6 weeks of clinic visits, his plasma TSH decreased to 1.01 uIU/mL, which was within the reference range, and the patient stopped coming to the weekly clinic visits (Table). Four months later, the patient's plasma TSH levels increased to 80.72 uIU/mL. Nonadherence to levothyroxine was suspected again. He was asked to resume weekly clinic visits, and the life-threatening effects of hypothyroidism and not taking levothyroxine were discussed with the patient and his son. The patient made CBOC clinic visits for 9 weeks, after which his plasma TSH level was low at 0.23 uIU/mL.

 

 

Discussion

There are multiple important causes to consider in patients with persistent hypothyroidism. One is medication nonadherence, which was most likely seen in the patient in this case. Missing even 1 day of levothyroxine can affect TSH and thyroid hormone levels for several days due to the long half-life of the medication.² Hepp and colleagues found that patients with hypothyroidism were significantly more likely to be nonadherent to levothyroxine if they had comorbid conditions such as type 2 diabetes or were obese.³ Another study of levothyroxine adherence found that the most common reason for missing doses was forgetfulness.⁴ However, memory and cognition impairments can also be symptoms of hypothyroidism itself; Haskard-Zolnierek and colleagues found a significant association between nonadherence to levothyroxine and self-reported brain fog in patients with hypothyroidism.⁵

Another cause of persistent hypothyroidism is malabsorption. Absorption of levothyroxine can be affected by intestinal malabsorption due to inflammatory bowel disease, lactose intolerance, or gastrointestinal infection, as well as several foods, drinks (eg, coffee), medications, vitamins, and supplements (eg, proton-pump inhibitors and calcium).^2,6 Levothyroxine is absorbed mainly at the jejunum and upper ileum, so any pathologies or ingested items that would directly or indirectly affect absorption at those sites can affect levothyroxine absorption.²

A liquid levothyroxine formulation can help with malabsorption.² Alternatively, weight gain may lead to a need for increasing the dosage of levothyroxine.^2,6 Other factors that can affect TSH levels include Addison disease, dysregulation of the hypothalamic-pituitary-thyroid axis, and TSH heterophile antibodies.²

Research describes methods that have effectively treated hypothyroidism in patients struggling with levothyroxine adherence. Two case reports describe weekly visits for levothyroxine administration successfully treating uncontrolled hypothyroidism.^7,8 A meta-analysis found that while weekly levothyroxine tablets led to a higher mean TSH level than daily use, weekly use still led to reference-range TSH levels, suggesting that weekly levothyroxine may be a helpful alternative for nonadherent patients.⁹ Alternatively, patients taking levothyroxine tablets have been shown to forget to take their medication more frequently compared to those taking the liquid formulation.^10,11 Additionally, a study by El Helou and colleagues found that adherence to levothyroxine was significantly improved when patients had endocrinology visits once a month and when the endocrinologist provided information about hypothyroidism.¹²

Another method that may improve adherence to levothyroxine is telehealth visits. This would be especially helpful for patients who live far from the clinic or do not have the time, transportation, or financial means to visit the clinic for weekly visits to assess medication adherence. Additionally, patients may be afraid of admitting to a health care professional that they are nonadherent. Clinicians must be tactful when asking about adherence to make the patient feel comfortable with admitting to nonadherence if their cognition is not impaired. Then, a patient-led conversation can occur regarding realistic ways the patient feels they can work toward adherence.

To our knowledge, the patient in this case report had no symptoms of intestinal malabsorption, and weight gain was not thought to be the issue, as levothyroxine dosage was adjusted multiple times. His plasma TSH levels returned to reference range after weekly pill count visits for 6 weeks and after weekly pill count visits for 9 weeks. Therefore, nonadherence to levothyroxine was suspected to be the cause of frequently elevated plasma TSH levels despite the patient’s insistence on adherence. While the patient did not report memory issues, cognitive impairments due to hypothyroidism may have been contributing to his probable nonadherence. Additionally, he had comorbidities, such as type 2 diabetes mellitus and obesity, which may have made adherence more difficult.

Levothyroxine was also only prescribed in daily tablet form, so the frequency and formulation may have also contributed to nonadherence. While the home nurse was originally sent to assess the patient’s adherence, the care team could have had the nurse start giving the patient weekly levothyroxine once nonadherence was determined to be a likely issue. The patient’s adherence only improved when he went to the clinic for pill counts but not when the home nurse came to his house weekly; this could be because the patient knew he had to invest the time to physically go to clinic visits for pill checks, motivating him to increase adherence.

Conclusions

This case reports a patient with frequently high plasma TSH levels achieving normalization of plasma TSH levels after weekly medication adherence checks at a primary care clinic. Weekly visits to a clinic seem impractical compared to weekly dosing with a visiting nurse; however, after review of the literature, this may be an approach to consider in the future. This strategy may especially help in cases of persistent abnormal plasma TSH levels in which no etiology can be found other than suspected medication nonadherence. Knowing their medication use will be checked at weekly clinic visits may motivate patients to be adherent.

Nonadherence to medications is an issue across health care. In endocrinology, hypothyroidism, a deficiency of thyroid hormones, is most often treated with levothyroxine and if left untreated can lead to myxedema coma, which can lead to death due to multiorgan dysfunction.¹ Therefore, adherence to levothyroxine is very important in preventing fatal complications.

We present the case of a patient with persistent primary hypothyroidism who was suspected to be nonadherent to levothyroxine, although the patient consistently claimed adherence. The patient’s plasma thyrotropin (TSH) level improved to reference range after 6 weeks of weekly primary care clinic visits. After stopping the visits, his plasma TSH level increased again, so 9 more weeks of visits resumed, which again helped bring down his plasma TSH levels.

Case Presentation

A male patient aged 67 years presented to the Dayton Veterans Affairs Medical Center (VAMC) endocrinology clinic for evaluation of thyroid nodules. The patient reported no history of neck irradiation and a physical examination was unremarkable. At that time, laboratory results showed a slightly elevated plasma TSH level of 4.35 uIU/mL (reference range, 0.35-4.00 uIU/mL) and normal free thyroxine (T4) of 1.00 ng/dL (reference range, 0.74-1.46 ng/dL). Later that year, the patient underwent a total thyroidectomy at the Cincinnati VAMC for Hurthle cell variant papillary thyroid carcinoma that was noted on biopsy at the Dayton VAMC. After surgical pathology results were available, the patient started levothyroxine 200 mcg daily, although 224 mcg would have been more appropriate based on his 142 kg weight. Due to a history of arrhythmia, the goal plasma TSH level was 0.10 to 0.50 uIU/mL. The patient subsequently underwent radioactive iodine ablation. After levothyroxine dose adjustments, the patient’s plasma TSH level was noted to be within his target range at 0.28 uIU/mL 3 months postablation.

Over the next 5 years the patient had regular laboratory tests during which his plasma TSH level rose and were typically high despite adjusting levothyroxine doses between 200 mcg and 325 mcg. The patient received counseling on taking the medication in the morning on an empty stomach and waiting at least 1 hour before consuming anything, and he went to many follow-up visits at the Dayton VAMC endocrinology clinic. He reported no vomiting or diarrhea but endorsed weight gain once. The patient also had high free T4 at times and did not take extra levothyroxine before undergoing laboratory tests.

Nonadherence to levothyroxine was suspected, but the patient insisted he was adherent. He received the medication in the mail regularly, generally had 90-day refills unless a dose change was made, used a pill box, and had social support from his son, but he did not use a phone alarm to remind him to take it. A home care nurse made weekly visits to make sure the remaining levothyroxine pill counts were correct; however, the patient continued to have difficulty maintaining daily adherence at home as indicated by the nurse’s pill counts not aligning with the number of pills which should have been left if the patient was talking the pills daily.

The patient was asked to visit a local community-based outpatient clinic (CBOC) weekly (to avoid patient travel time to Dayton VAMC > 1 hour) to check pill counts and assess adherence. The patient went to the CBOC clinic for these visits, during which pill counts indicated much better but not 100% adherence. After 6 weeks of clinic visits, his plasma TSH decreased to 1.01 uIU/mL, which was within the reference range, and the patient stopped coming to the weekly clinic visits (Table). Four months later, the patient's plasma TSH levels increased to 80.72 uIU/mL. Nonadherence to levothyroxine was suspected again. He was asked to resume weekly clinic visits, and the life-threatening effects of hypothyroidism and not taking levothyroxine were discussed with the patient and his son. The patient made CBOC clinic visits for 9 weeks, after which his plasma TSH level was low at 0.23 uIU/mL.

 

 

Discussion

There are multiple important causes to consider in patients with persistent hypothyroidism. One is medication nonadherence, which was most likely seen in the patient in this case. Missing even 1 day of levothyroxine can affect TSH and thyroid hormone levels for several days due to the long half-life of the medication.² Hepp and colleagues found that patients with hypothyroidism were significantly more likely to be nonadherent to levothyroxine if they had comorbid conditions such as type 2 diabetes or were obese.³ Another study of levothyroxine adherence found that the most common reason for missing doses was forgetfulness.⁴ However, memory and cognition impairments can also be symptoms of hypothyroidism itself; Haskard-Zolnierek and colleagues found a significant association between nonadherence to levothyroxine and self-reported brain fog in patients with hypothyroidism.⁵

Another cause of persistent hypothyroidism is malabsorption. Absorption of levothyroxine can be affected by intestinal malabsorption due to inflammatory bowel disease, lactose intolerance, or gastrointestinal infection, as well as several foods, drinks (eg, coffee), medications, vitamins, and supplements (eg, proton-pump inhibitors and calcium).^2,6 Levothyroxine is absorbed mainly at the jejunum and upper ileum, so any pathologies or ingested items that would directly or indirectly affect absorption at those sites can affect levothyroxine absorption.²

A liquid levothyroxine formulation can help with malabsorption.² Alternatively, weight gain may lead to a need for increasing the dosage of levothyroxine.^2,6 Other factors that can affect TSH levels include Addison disease, dysregulation of the hypothalamic-pituitary-thyroid axis, and TSH heterophile antibodies.²

Research describes methods that have effectively treated hypothyroidism in patients struggling with levothyroxine adherence. Two case reports describe weekly visits for levothyroxine administration successfully treating uncontrolled hypothyroidism.^7,8 A meta-analysis found that while weekly levothyroxine tablets led to a higher mean TSH level than daily use, weekly use still led to reference-range TSH levels, suggesting that weekly levothyroxine may be a helpful alternative for nonadherent patients.⁹ Alternatively, patients taking levothyroxine tablets have been shown to forget to take their medication more frequently compared to those taking the liquid formulation.^10,11 Additionally, a study by El Helou and colleagues found that adherence to levothyroxine was significantly improved when patients had endocrinology visits once a month and when the endocrinologist provided information about hypothyroidism.¹²

Another method that may improve adherence to levothyroxine is telehealth visits. This would be especially helpful for patients who live far from the clinic or do not have the time, transportation, or financial means to visit the clinic for weekly visits to assess medication adherence. Additionally, patients may be afraid of admitting to a health care professional that they are nonadherent. Clinicians must be tactful when asking about adherence to make the patient feel comfortable with admitting to nonadherence if their cognition is not impaired. Then, a patient-led conversation can occur regarding realistic ways the patient feels they can work toward adherence.

To our knowledge, the patient in this case report had no symptoms of intestinal malabsorption, and weight gain was not thought to be the issue, as levothyroxine dosage was adjusted multiple times. His plasma TSH levels returned to reference range after weekly pill count visits for 6 weeks and after weekly pill count visits for 9 weeks. Therefore, nonadherence to levothyroxine was suspected to be the cause of frequently elevated plasma TSH levels despite the patient’s insistence on adherence. While the patient did not report memory issues, cognitive impairments due to hypothyroidism may have been contributing to his probable nonadherence. Additionally, he had comorbidities, such as type 2 diabetes mellitus and obesity, which may have made adherence more difficult.

Levothyroxine was also only prescribed in daily tablet form, so the frequency and formulation may have also contributed to nonadherence. While the home nurse was originally sent to assess the patient’s adherence, the care team could have had the nurse start giving the patient weekly levothyroxine once nonadherence was determined to be a likely issue. The patient’s adherence only improved when he went to the clinic for pill counts but not when the home nurse came to his house weekly; this could be because the patient knew he had to invest the time to physically go to clinic visits for pill checks, motivating him to increase adherence.

Conclusions

This case reports a patient with frequently high plasma TSH levels achieving normalization of plasma TSH levels after weekly medication adherence checks at a primary care clinic. Weekly visits to a clinic seem impractical compared to weekly dosing with a visiting nurse; however, after review of the literature, this may be an approach to consider in the future. This strategy may especially help in cases of persistent abnormal plasma TSH levels in which no etiology can be found other than suspected medication nonadherence. Knowing their medication use will be checked at weekly clinic visits may motivate patients to be adherent.

Nonadherence to medications is an issue across health care. In endocrinology, hypothyroidism, a deficiency of thyroid hormones, is most often treated with levothyroxine and if left untreated can lead to myxedema coma, which can lead to death due to multiorgan dysfunction.¹ Therefore, adherence to levothyroxine is very important in preventing fatal complications.

We present the case of a patient with persistent primary hypothyroidism who was suspected to be nonadherent to levothyroxine, although the patient consistently claimed adherence. The patient’s plasma thyrotropin (TSH) level improved to reference range after 6 weeks of weekly primary care clinic visits. After stopping the visits, his plasma TSH level increased again, so 9 more weeks of visits resumed, which again helped bring down his plasma TSH levels.

Case Presentation

A male patient aged 67 years presented to the Dayton Veterans Affairs Medical Center (VAMC) endocrinology clinic for evaluation of thyroid nodules. The patient reported no history of neck irradiation and a physical examination was unremarkable. At that time, laboratory results showed a slightly elevated plasma TSH level of 4.35 uIU/mL (reference range, 0.35-4.00 uIU/mL) and normal free thyroxine (T4) of 1.00 ng/dL (reference range, 0.74-1.46 ng/dL). Later that year, the patient underwent a total thyroidectomy at the Cincinnati VAMC for Hurthle cell variant papillary thyroid carcinoma that was noted on biopsy at the Dayton VAMC. After surgical pathology results were available, the patient started levothyroxine 200 mcg daily, although 224 mcg would have been more appropriate based on his 142 kg weight. Due to a history of arrhythmia, the goal plasma TSH level was 0.10 to 0.50 uIU/mL. The patient subsequently underwent radioactive iodine ablation. After levothyroxine dose adjustments, the patient’s plasma TSH level was noted to be within his target range at 0.28 uIU/mL 3 months postablation.

Over the next 5 years the patient had regular laboratory tests during which his plasma TSH level rose and were typically high despite adjusting levothyroxine doses between 200 mcg and 325 mcg. The patient received counseling on taking the medication in the morning on an empty stomach and waiting at least 1 hour before consuming anything, and he went to many follow-up visits at the Dayton VAMC endocrinology clinic. He reported no vomiting or diarrhea but endorsed weight gain once. The patient also had high free T4 at times and did not take extra levothyroxine before undergoing laboratory tests.

Nonadherence to levothyroxine was suspected, but the patient insisted he was adherent. He received the medication in the mail regularly, generally had 90-day refills unless a dose change was made, used a pill box, and had social support from his son, but he did not use a phone alarm to remind him to take it. A home care nurse made weekly visits to make sure the remaining levothyroxine pill counts were correct; however, the patient continued to have difficulty maintaining daily adherence at home as indicated by the nurse’s pill counts not aligning with the number of pills which should have been left if the patient was talking the pills daily.

The patient was asked to visit a local community-based outpatient clinic (CBOC) weekly (to avoid patient travel time to Dayton VAMC > 1 hour) to check pill counts and assess adherence. The patient went to the CBOC clinic for these visits, during which pill counts indicated much better but not 100% adherence. After 6 weeks of clinic visits, his plasma TSH decreased to 1.01 uIU/mL, which was within the reference range, and the patient stopped coming to the weekly clinic visits (Table). Four months later, the patient's plasma TSH levels increased to 80.72 uIU/mL. Nonadherence to levothyroxine was suspected again. He was asked to resume weekly clinic visits, and the life-threatening effects of hypothyroidism and not taking levothyroxine were discussed with the patient and his son. The patient made CBOC clinic visits for 9 weeks, after which his plasma TSH level was low at 0.23 uIU/mL.

 

 

Discussion

There are multiple important causes to consider in patients with persistent hypothyroidism. One is medication nonadherence, which was most likely seen in the patient in this case. Missing even 1 day of levothyroxine can affect TSH and thyroid hormone levels for several days due to the long half-life of the medication.² Hepp and colleagues found that patients with hypothyroidism were significantly more likely to be nonadherent to levothyroxine if they had comorbid conditions such as type 2 diabetes or were obese.³ Another study of levothyroxine adherence found that the most common reason for missing doses was forgetfulness.⁴ However, memory and cognition impairments can also be symptoms of hypothyroidism itself; Haskard-Zolnierek and colleagues found a significant association between nonadherence to levothyroxine and self-reported brain fog in patients with hypothyroidism.⁵

Another cause of persistent hypothyroidism is malabsorption. Absorption of levothyroxine can be affected by intestinal malabsorption due to inflammatory bowel disease, lactose intolerance, or gastrointestinal infection, as well as several foods, drinks (eg, coffee), medications, vitamins, and supplements (eg, proton-pump inhibitors and calcium).^2,6 Levothyroxine is absorbed mainly at the jejunum and upper ileum, so any pathologies or ingested items that would directly or indirectly affect absorption at those sites can affect levothyroxine absorption.²

A liquid levothyroxine formulation can help with malabsorption.² Alternatively, weight gain may lead to a need for increasing the dosage of levothyroxine.^2,6 Other factors that can affect TSH levels include Addison disease, dysregulation of the hypothalamic-pituitary-thyroid axis, and TSH heterophile antibodies.²

Research describes methods that have effectively treated hypothyroidism in patients struggling with levothyroxine adherence. Two case reports describe weekly visits for levothyroxine administration successfully treating uncontrolled hypothyroidism.^7,8 A meta-analysis found that while weekly levothyroxine tablets led to a higher mean TSH level than daily use, weekly use still led to reference-range TSH levels, suggesting that weekly levothyroxine may be a helpful alternative for nonadherent patients.⁹ Alternatively, patients taking levothyroxine tablets have been shown to forget to take their medication more frequently compared to those taking the liquid formulation.^10,11 Additionally, a study by El Helou and colleagues found that adherence to levothyroxine was significantly improved when patients had endocrinology visits once a month and when the endocrinologist provided information about hypothyroidism.¹²

Another method that may improve adherence to levothyroxine is telehealth visits. This would be especially helpful for patients who live far from the clinic or do not have the time, transportation, or financial means to visit the clinic for weekly visits to assess medication adherence. Additionally, patients may be afraid of admitting to a health care professional that they are nonadherent. Clinicians must be tactful when asking about adherence to make the patient feel comfortable with admitting to nonadherence if their cognition is not impaired. Then, a patient-led conversation can occur regarding realistic ways the patient feels they can work toward adherence.

To our knowledge, the patient in this case report had no symptoms of intestinal malabsorption, and weight gain was not thought to be the issue, as levothyroxine dosage was adjusted multiple times. His plasma TSH levels returned to reference range after weekly pill count visits for 6 weeks and after weekly pill count visits for 9 weeks. Therefore, nonadherence to levothyroxine was suspected to be the cause of frequently elevated plasma TSH levels despite the patient’s insistence on adherence. While the patient did not report memory issues, cognitive impairments due to hypothyroidism may have been contributing to his probable nonadherence. Additionally, he had comorbidities, such as type 2 diabetes mellitus and obesity, which may have made adherence more difficult.

Levothyroxine was also only prescribed in daily tablet form, so the frequency and formulation may have also contributed to nonadherence. While the home nurse was originally sent to assess the patient’s adherence, the care team could have had the nurse start giving the patient weekly levothyroxine once nonadherence was determined to be a likely issue. The patient’s adherence only improved when he went to the clinic for pill counts but not when the home nurse came to his house weekly; this could be because the patient knew he had to invest the time to physically go to clinic visits for pill checks, motivating him to increase adherence.

Conclusions

This case reports a patient with frequently high plasma TSH levels achieving normalization of plasma TSH levels after weekly medication adherence checks at a primary care clinic. Weekly visits to a clinic seem impractical compared to weekly dosing with a visiting nurse; however, after review of the literature, this may be an approach to consider in the future. This strategy may especially help in cases of persistent abnormal plasma TSH levels in which no etiology can be found other than suspected medication nonadherence. Knowing their medication use will be checked at weekly clinic visits may motivate patients to be adherent.

Chronic obstructive pulmonary disease (COPD) is a respiratory disorder associated with slowly progressive systemic inflammation. It includes emphysema, chronic bronchitis, and small airway disease. Patients with COPD have an incomplete reversibility of airway obstruction, the key differentiating factor between it and asthma.¹

The Global Initiative for Chronic Obstructive Lung Disease (GOLD) guidelines recommend a combination inhaler consisting of a long-acting β-2 agonist (LABA) and inhaled corticosteroid (ICS) for patients with a history of COPD exacerbations.² Blood eosinophil count is another marker for the initiation of an ICS in patients with COPD. According to the 2023 GOLD Report, ICS therapy is appropriate for patients who experience frequent exacerbations and have a blood eosinophil count > 100 cells/μL, while on maximum tolerated inhaler therapy.³ A 2019 meta-analysis found an overall reduction in the risk of exacerbations in patients with blood eosinophil counts ≥ 100 cells/µL after initiating an ICS.⁴

Common ICS-LABA inhalers include the combination of budesonide/formoterol as well as fluticasone/salmeterol. Though these combinations are within the same therapeutic class, they have different delivery systems: budesonide/formoterol is a metered dose inhaler, while fluticasone/salmeterol is a dry powder inhaler. The PATHOS study compared the exacerbation rates for the 2 inhalers in primary care patients with COPD. Patients treated long-term with the budesonide/formoterol inhaler were significantly less likely to experience a COPD exacerbation than those treated with the fluticasone/salmeterol inhaler.⁵

In 2021, The Veteran Health Administration transitioned patients from budesonide/formoterol inhalers to fluticasone/salmeterol inhalers through a formulary conversion. The purpose of this study was to examine the outcomes for patients undergoing the transition.

Methods

A retrospective chart review was conducted on patients at the Hershel “Woody” Williams Veterans Affairs Medical Center in Huntington, West Virginia, with COPD and prescriptions for both budesonide/formoterol and fluticasone/salmeterol inhalers between February 1, 2021, and May 30, 2022. In 2018, the prevalence of COPD in West Virginia was 13.9%, highest in the US.⁶ Data was obtained through the US Department of Veteran Affairs (VA) Corporate Data Warehouse and stored on a VA Informatics and Computing Infrastructure server. Patients were randomly selected from this cohort and included if they were aged 18 to 89 years, prescribed both inhalers, and had a confirmed COPD diagnosis. Patients were excluded if they also had an asthma diagnosis, if they had an interstitial lung disease, or any tracheostomy tubes. The date of transition from a budesonide/formoterol inhaler to a fluticasone/salmeterol inhaler was collected to establish a timeline of 6 months before and 6 months after the transition.

The primary endpoint was to assess clinical outcomes such as the number of COPD exacerbations and hospitalizations within 6 months of the transition for patients affected by the formulary conversion. Secondary outcomes included the incidence of adverse effects (AEs), treatment failure, tobacco use, and systemic corticosteroid/antimicrobial utilization.

Statistical analyses were performed using STATA v.15. Numerical data was analyzed using a Wilcoxon signed rank test. Categorical data was analyzed by a logistic regression analysis.

 

 

Results

Of 1497 included patients who transitioned from budesonide/formoterol to fluticasone/salmeterol inhalers, 165 were randomly selected and 100 patients were included in this analysis. Of the 100 patients, 99 were male with a mean (SEM) age of 71 (0.69) years (range, 54-87) (Table).

The transition from budesonide/formoterol to fluticasone/salmeterol inhalers did not have a statistically significant impact on exacerbations (P = .56). Thirty patients had ≥ 1 exacerbation: 12 had an exacerbation before the transition, 10 had an exacerbation after the transition, and 8 had exacerbations before and after the transition. In the 6 months prior to the transition while on a budesonide/formoterol inhaler, there were 24 exacerbations among 20 patients. Five patients had > 1 exacerbation, accounting for 11 of the 24 exacerbations. There were 29 exacerbations among 19 patients while on a fluticasone/salmeterol inhaler in the 6 months after the transition. Four of these patients had > 1 exacerbation, accounting for 14 of 29 exacerbations (Figure).

Secondary endpoints showed 3 patients experienced an AE related to fluticasone/salmeterol, including thrush, coughing and throat irritation, and dyspnea. Eighteen fluticasone/salmeterol therapeutic failures were indicated by related prior authorization medication requests in the electronic health record. Twelve of 18 patients experienced no difference in exacerbations before vs after the transition to budesonide/formoterol. Twenty-three patients transitioned from fluticasone/salmeterol to a different ICS-LABA therapy; 20 of those 23 patients transitioned back to a budesonide/formoterol inhaler.

There were 48 documented active tobacco users in the study. There was no statistically significant correlation (P = .52) when comparing tobacco use at time of conversion and exacerbation frequency, although the coefficient showed a negative correlation of -0.387. In the 6 months prior to the transition, there were 17 prescriptions for systemic corticosteroids and 24 for antibiotics to treat COPD exacerbations. Following the transition, there were only 12 prescriptions for systemic corticosteroids and 23 for antibiotics. Fifty-two patients had an active prescription for a fluticasone/salmeterol inhaler at the time of the data review (November to December 2022); of the 48 patients who did not, 10 were no longer active due to patient death between the study period and data retrieval.

Discussion

Patients who transitioned from budesonide/formoterol to fluticasone/salmeterol inhalers did not show a significant difference in clinical COPD outcomes. While the total number of exacerbations increased after switching to the fluticasone/salmeterol inhaler, fewer patients had exacerbations during fluticasone/salmeterol therapy when compared with budesonide/fluticasone therapy. The number of patients receiving systemic corticosteroids and antibiotics to treat exacerbations before and after the transition were similar.

The frequency of treatment failures and AEs to the fluticasone/salmeterol inhaler could be due to the change of the inhaler delivery systems. Budesonide/formoterol is a metered dose inhaler (MDI). It is equipped with a pressurized canister that allows a spacer to be used to maximize benefit. Spacers can assist in preventing oral candidiasis by reducing the amount of medication that touches the back of the throat. Spacers are an option for patients, but not all use them for their MDIs, which can result in a less effective administered dose. Fluticasone/salmeterol is a dry powder inhaler, which requires a deep, fast breath to maximize the benefit, and spacers cannot be used with them. MDIs have been shown to be responsible for a negative impact on climate change, which can be reduced by switching to a dry powder inhaler.⁷

Tobacco cessation is very important in limiting the progression of COPD. As shown with the negative coefficient correlation, not being an active tobacco user at the time of transition correlated (although not significantly) with less frequent exacerbations. When comparing this study to similar research, such as the PATHOS study, several differences are observed.⁵ The PATHOS study compared long term treatment (> 1 year) of budesonide/formoterol or fluticasone/salmeterol, a longer period than this study. It regarded similar outcomes for the definition of an exacerbation, such as antibiotic/steroid use or hospital admission. While the current study showed no significant difference between the 2 inhalers and their effect on exacerbations, the PATHOS study found that those treated with a budesonide/formoterol inhaler were less likely to experience COPD-related exacerbations than those treated with the fluticasone/salmeterol inhaler. The PATHOS study had a larger mainly Scandinavian sample (N = 5500). This population could exhibit baseline differences from a study of US veterans.⁵ A similar Canadian matched cohort study of 2262 patients compared the 2 inhalers to assess their relative effectiveness. It found that COPD exacerbations did not differ between the 2 groups, but the budesonide/formoterol group was significantly less likely to have an emergency department visit compared to the fluticasone salmeterol group.⁸ Like the PATHOS study, the Canadian study had a larger sample size and longer timeframe than did our study.

 

 

Limitations

There are various limitations to this study. It was a retrospective, single-center study and the patient population was relatively homogenous, with only 1 female and a mean age of 71 years. As a study conducted in a veteran population in West Virginia, the findings may not be representative of the general population with COPD, which includes more women and more racial diversity.⁹ The American Lung Association discusses how environmental exposures to hazardous conditions increase the risks of pulmonary diseases for veterans.¹⁰ It has been reported that the prevalence of COPD is higher among veterans compared to the general population, but it is not different in terms of disease manifestation.¹⁰

Another limitation is the short time frame. Clinical guidelines, including the GOLD Report, typically track the number of exacerbations for 1 year to escalate therapy.³ Six months was a relatively short time frame, and it is possible that more exacerbations may have occurred beyond the study time frame. Ten patients in the sample died between the end of the study period and data retrieval, which might have been caught by a longer study period. An additional limitation was the inability to measure adherence. As this was a formulary conversion, many patients had been mailed a 30- or 90-day prescription of the budesonide/formoterol inhaler when transitioned to the fluticasone/salmeterol inhaler. There was no way to accurately determine when the patient made the switch to the fluticasone/salmeterol inhaler. This study also had a small sample group (a pre-post analysis of the same group), a limitation when evaluating the impact of this formulary change on a small percentage of the population transitioned.

This formulary conversion occurred during the COVID-19 pandemic, and some exacerbations could have been the result of a misdiagnosed COVID-19 infection. Respiratory infections, including COVID-19, are common causes of exacerbations. It is also possible that some patients elected not to receive medical care for symptoms of an exacerbation during the pandemic.¹¹

Conclusions

Switching from the budesonide/formoterol inhaler to the fluticasone/salmeterol inhaler through formulary conversion did not have a significant impact on the clinical outcomes in patients with COPD. This study found that although the inhalers contain different active ingredients, products within the same therapeutic class yielded nonsignificant changes. When conducting formulary conversions, intolerances and treatment failures should be expected when switching from different inhaler delivery systems. This study further justifies the ability to be cost effective by making formulary conversions within the same therapeutic class within a veterans population.

Acknowledgments

The authors would like to acknowledge James Brown, PharmD, PhD.

Chronic obstructive pulmonary disease (COPD) is a respiratory disorder associated with slowly progressive systemic inflammation. It includes emphysema, chronic bronchitis, and small airway disease. Patients with COPD have an incomplete reversibility of airway obstruction, the key differentiating factor between it and asthma.¹

The Global Initiative for Chronic Obstructive Lung Disease (GOLD) guidelines recommend a combination inhaler consisting of a long-acting β-2 agonist (LABA) and inhaled corticosteroid (ICS) for patients with a history of COPD exacerbations.² Blood eosinophil count is another marker for the initiation of an ICS in patients with COPD. According to the 2023 GOLD Report, ICS therapy is appropriate for patients who experience frequent exacerbations and have a blood eosinophil count > 100 cells/μL, while on maximum tolerated inhaler therapy.³ A 2019 meta-analysis found an overall reduction in the risk of exacerbations in patients with blood eosinophil counts ≥ 100 cells/µL after initiating an ICS.⁴

Common ICS-LABA inhalers include the combination of budesonide/formoterol as well as fluticasone/salmeterol. Though these combinations are within the same therapeutic class, they have different delivery systems: budesonide/formoterol is a metered dose inhaler, while fluticasone/salmeterol is a dry powder inhaler. The PATHOS study compared the exacerbation rates for the 2 inhalers in primary care patients with COPD. Patients treated long-term with the budesonide/formoterol inhaler were significantly less likely to experience a COPD exacerbation than those treated with the fluticasone/salmeterol inhaler.⁵

In 2021, The Veteran Health Administration transitioned patients from budesonide/formoterol inhalers to fluticasone/salmeterol inhalers through a formulary conversion. The purpose of this study was to examine the outcomes for patients undergoing the transition.

Methods

A retrospective chart review was conducted on patients at the Hershel “Woody” Williams Veterans Affairs Medical Center in Huntington, West Virginia, with COPD and prescriptions for both budesonide/formoterol and fluticasone/salmeterol inhalers between February 1, 2021, and May 30, 2022. In 2018, the prevalence of COPD in West Virginia was 13.9%, highest in the US.⁶ Data was obtained through the US Department of Veteran Affairs (VA) Corporate Data Warehouse and stored on a VA Informatics and Computing Infrastructure server. Patients were randomly selected from this cohort and included if they were aged 18 to 89 years, prescribed both inhalers, and had a confirmed COPD diagnosis. Patients were excluded if they also had an asthma diagnosis, if they had an interstitial lung disease, or any tracheostomy tubes. The date of transition from a budesonide/formoterol inhaler to a fluticasone/salmeterol inhaler was collected to establish a timeline of 6 months before and 6 months after the transition.

The primary endpoint was to assess clinical outcomes such as the number of COPD exacerbations and hospitalizations within 6 months of the transition for patients affected by the formulary conversion. Secondary outcomes included the incidence of adverse effects (AEs), treatment failure, tobacco use, and systemic corticosteroid/antimicrobial utilization.

Statistical analyses were performed using STATA v.15. Numerical data was analyzed using a Wilcoxon signed rank test. Categorical data was analyzed by a logistic regression analysis.

 

 

Results

Of 1497 included patients who transitioned from budesonide/formoterol to fluticasone/salmeterol inhalers, 165 were randomly selected and 100 patients were included in this analysis. Of the 100 patients, 99 were male with a mean (SEM) age of 71 (0.69) years (range, 54-87) (Table).

The transition from budesonide/formoterol to fluticasone/salmeterol inhalers did not have a statistically significant impact on exacerbations (P = .56). Thirty patients had ≥ 1 exacerbation: 12 had an exacerbation before the transition, 10 had an exacerbation after the transition, and 8 had exacerbations before and after the transition. In the 6 months prior to the transition while on a budesonide/formoterol inhaler, there were 24 exacerbations among 20 patients. Five patients had > 1 exacerbation, accounting for 11 of the 24 exacerbations. There were 29 exacerbations among 19 patients while on a fluticasone/salmeterol inhaler in the 6 months after the transition. Four of these patients had > 1 exacerbation, accounting for 14 of 29 exacerbations (Figure).

Secondary endpoints showed 3 patients experienced an AE related to fluticasone/salmeterol, including thrush, coughing and throat irritation, and dyspnea. Eighteen fluticasone/salmeterol therapeutic failures were indicated by related prior authorization medication requests in the electronic health record. Twelve of 18 patients experienced no difference in exacerbations before vs after the transition to budesonide/formoterol. Twenty-three patients transitioned from fluticasone/salmeterol to a different ICS-LABA therapy; 20 of those 23 patients transitioned back to a budesonide/formoterol inhaler.

There were 48 documented active tobacco users in the study. There was no statistically significant correlation (P = .52) when comparing tobacco use at time of conversion and exacerbation frequency, although the coefficient showed a negative correlation of -0.387. In the 6 months prior to the transition, there were 17 prescriptions for systemic corticosteroids and 24 for antibiotics to treat COPD exacerbations. Following the transition, there were only 12 prescriptions for systemic corticosteroids and 23 for antibiotics. Fifty-two patients had an active prescription for a fluticasone/salmeterol inhaler at the time of the data review (November to December 2022); of the 48 patients who did not, 10 were no longer active due to patient death between the study period and data retrieval.

Discussion

Patients who transitioned from budesonide/formoterol to fluticasone/salmeterol inhalers did not show a significant difference in clinical COPD outcomes. While the total number of exacerbations increased after switching to the fluticasone/salmeterol inhaler, fewer patients had exacerbations during fluticasone/salmeterol therapy when compared with budesonide/fluticasone therapy. The number of patients receiving systemic corticosteroids and antibiotics to treat exacerbations before and after the transition were similar.

The frequency of treatment failures and AEs to the fluticasone/salmeterol inhaler could be due to the change of the inhaler delivery systems. Budesonide/formoterol is a metered dose inhaler (MDI). It is equipped with a pressurized canister that allows a spacer to be used to maximize benefit. Spacers can assist in preventing oral candidiasis by reducing the amount of medication that touches the back of the throat. Spacers are an option for patients, but not all use them for their MDIs, which can result in a less effective administered dose. Fluticasone/salmeterol is a dry powder inhaler, which requires a deep, fast breath to maximize the benefit, and spacers cannot be used with them. MDIs have been shown to be responsible for a negative impact on climate change, which can be reduced by switching to a dry powder inhaler.⁷

Tobacco cessation is very important in limiting the progression of COPD. As shown with the negative coefficient correlation, not being an active tobacco user at the time of transition correlated (although not significantly) with less frequent exacerbations. When comparing this study to similar research, such as the PATHOS study, several differences are observed.⁵ The PATHOS study compared long term treatment (> 1 year) of budesonide/formoterol or fluticasone/salmeterol, a longer period than this study. It regarded similar outcomes for the definition of an exacerbation, such as antibiotic/steroid use or hospital admission. While the current study showed no significant difference between the 2 inhalers and their effect on exacerbations, the PATHOS study found that those treated with a budesonide/formoterol inhaler were less likely to experience COPD-related exacerbations than those treated with the fluticasone/salmeterol inhaler. The PATHOS study had a larger mainly Scandinavian sample (N = 5500). This population could exhibit baseline differences from a study of US veterans.⁵ A similar Canadian matched cohort study of 2262 patients compared the 2 inhalers to assess their relative effectiveness. It found that COPD exacerbations did not differ between the 2 groups, but the budesonide/formoterol group was significantly less likely to have an emergency department visit compared to the fluticasone salmeterol group.⁸ Like the PATHOS study, the Canadian study had a larger sample size and longer timeframe than did our study.

 

 

Limitations

There are various limitations to this study. It was a retrospective, single-center study and the patient population was relatively homogenous, with only 1 female and a mean age of 71 years. As a study conducted in a veteran population in West Virginia, the findings may not be representative of the general population with COPD, which includes more women and more racial diversity.⁹ The American Lung Association discusses how environmental exposures to hazardous conditions increase the risks of pulmonary diseases for veterans.¹⁰ It has been reported that the prevalence of COPD is higher among veterans compared to the general population, but it is not different in terms of disease manifestation.¹⁰

Another limitation is the short time frame. Clinical guidelines, including the GOLD Report, typically track the number of exacerbations for 1 year to escalate therapy.³ Six months was a relatively short time frame, and it is possible that more exacerbations may have occurred beyond the study time frame. Ten patients in the sample died between the end of the study period and data retrieval, which might have been caught by a longer study period. An additional limitation was the inability to measure adherence. As this was a formulary conversion, many patients had been mailed a 30- or 90-day prescription of the budesonide/formoterol inhaler when transitioned to the fluticasone/salmeterol inhaler. There was no way to accurately determine when the patient made the switch to the fluticasone/salmeterol inhaler. This study also had a small sample group (a pre-post analysis of the same group), a limitation when evaluating the impact of this formulary change on a small percentage of the population transitioned.

This formulary conversion occurred during the COVID-19 pandemic, and some exacerbations could have been the result of a misdiagnosed COVID-19 infection. Respiratory infections, including COVID-19, are common causes of exacerbations. It is also possible that some patients elected not to receive medical care for symptoms of an exacerbation during the pandemic.¹¹

Conclusions

Switching from the budesonide/formoterol inhaler to the fluticasone/salmeterol inhaler through formulary conversion did not have a significant impact on the clinical outcomes in patients with COPD. This study found that although the inhalers contain different active ingredients, products within the same therapeutic class yielded nonsignificant changes. When conducting formulary conversions, intolerances and treatment failures should be expected when switching from different inhaler delivery systems. This study further justifies the ability to be cost effective by making formulary conversions within the same therapeutic class within a veterans population.

Acknowledgments

The authors would like to acknowledge James Brown, PharmD, PhD.

Chronic obstructive pulmonary disease (COPD) is a respiratory disorder associated with slowly progressive systemic inflammation. It includes emphysema, chronic bronchitis, and small airway disease. Patients with COPD have an incomplete reversibility of airway obstruction, the key differentiating factor between it and asthma.¹

The Global Initiative for Chronic Obstructive Lung Disease (GOLD) guidelines recommend a combination inhaler consisting of a long-acting β-2 agonist (LABA) and inhaled corticosteroid (ICS) for patients with a history of COPD exacerbations.² Blood eosinophil count is another marker for the initiation of an ICS in patients with COPD. According to the 2023 GOLD Report, ICS therapy is appropriate for patients who experience frequent exacerbations and have a blood eosinophil count > 100 cells/μL, while on maximum tolerated inhaler therapy.³ A 2019 meta-analysis found an overall reduction in the risk of exacerbations in patients with blood eosinophil counts ≥ 100 cells/µL after initiating an ICS.⁴

Common ICS-LABA inhalers include the combination of budesonide/formoterol as well as fluticasone/salmeterol. Though these combinations are within the same therapeutic class, they have different delivery systems: budesonide/formoterol is a metered dose inhaler, while fluticasone/salmeterol is a dry powder inhaler. The PATHOS study compared the exacerbation rates for the 2 inhalers in primary care patients with COPD. Patients treated long-term with the budesonide/formoterol inhaler were significantly less likely to experience a COPD exacerbation than those treated with the fluticasone/salmeterol inhaler.⁵

In 2021, The Veteran Health Administration transitioned patients from budesonide/formoterol inhalers to fluticasone/salmeterol inhalers through a formulary conversion. The purpose of this study was to examine the outcomes for patients undergoing the transition.

Methods

A retrospective chart review was conducted on patients at the Hershel “Woody” Williams Veterans Affairs Medical Center in Huntington, West Virginia, with COPD and prescriptions for both budesonide/formoterol and fluticasone/salmeterol inhalers between February 1, 2021, and May 30, 2022. In 2018, the prevalence of COPD in West Virginia was 13.9%, highest in the US.⁶ Data was obtained through the US Department of Veteran Affairs (VA) Corporate Data Warehouse and stored on a VA Informatics and Computing Infrastructure server. Patients were randomly selected from this cohort and included if they were aged 18 to 89 years, prescribed both inhalers, and had a confirmed COPD diagnosis. Patients were excluded if they also had an asthma diagnosis, if they had an interstitial lung disease, or any tracheostomy tubes. The date of transition from a budesonide/formoterol inhaler to a fluticasone/salmeterol inhaler was collected to establish a timeline of 6 months before and 6 months after the transition.

The primary endpoint was to assess clinical outcomes such as the number of COPD exacerbations and hospitalizations within 6 months of the transition for patients affected by the formulary conversion. Secondary outcomes included the incidence of adverse effects (AEs), treatment failure, tobacco use, and systemic corticosteroid/antimicrobial utilization.

Statistical analyses were performed using STATA v.15. Numerical data was analyzed using a Wilcoxon signed rank test. Categorical data was analyzed by a logistic regression analysis.

 

 

Results

Of 1497 included patients who transitioned from budesonide/formoterol to fluticasone/salmeterol inhalers, 165 were randomly selected and 100 patients were included in this analysis. Of the 100 patients, 99 were male with a mean (SEM) age of 71 (0.69) years (range, 54-87) (Table).

The transition from budesonide/formoterol to fluticasone/salmeterol inhalers did not have a statistically significant impact on exacerbations (P = .56). Thirty patients had ≥ 1 exacerbation: 12 had an exacerbation before the transition, 10 had an exacerbation after the transition, and 8 had exacerbations before and after the transition. In the 6 months prior to the transition while on a budesonide/formoterol inhaler, there were 24 exacerbations among 20 patients. Five patients had > 1 exacerbation, accounting for 11 of the 24 exacerbations. There were 29 exacerbations among 19 patients while on a fluticasone/salmeterol inhaler in the 6 months after the transition. Four of these patients had > 1 exacerbation, accounting for 14 of 29 exacerbations (Figure).

Secondary endpoints showed 3 patients experienced an AE related to fluticasone/salmeterol, including thrush, coughing and throat irritation, and dyspnea. Eighteen fluticasone/salmeterol therapeutic failures were indicated by related prior authorization medication requests in the electronic health record. Twelve of 18 patients experienced no difference in exacerbations before vs after the transition to budesonide/formoterol. Twenty-three patients transitioned from fluticasone/salmeterol to a different ICS-LABA therapy; 20 of those 23 patients transitioned back to a budesonide/formoterol inhaler.

There were 48 documented active tobacco users in the study. There was no statistically significant correlation (P = .52) when comparing tobacco use at time of conversion and exacerbation frequency, although the coefficient showed a negative correlation of -0.387. In the 6 months prior to the transition, there were 17 prescriptions for systemic corticosteroids and 24 for antibiotics to treat COPD exacerbations. Following the transition, there were only 12 prescriptions for systemic corticosteroids and 23 for antibiotics. Fifty-two patients had an active prescription for a fluticasone/salmeterol inhaler at the time of the data review (November to December 2022); of the 48 patients who did not, 10 were no longer active due to patient death between the study period and data retrieval.

Discussion

Patients who transitioned from budesonide/formoterol to fluticasone/salmeterol inhalers did not show a significant difference in clinical COPD outcomes. While the total number of exacerbations increased after switching to the fluticasone/salmeterol inhaler, fewer patients had exacerbations during fluticasone/salmeterol therapy when compared with budesonide/fluticasone therapy. The number of patients receiving systemic corticosteroids and antibiotics to treat exacerbations before and after the transition were similar.

The frequency of treatment failures and AEs to the fluticasone/salmeterol inhaler could be due to the change of the inhaler delivery systems. Budesonide/formoterol is a metered dose inhaler (MDI). It is equipped with a pressurized canister that allows a spacer to be used to maximize benefit. Spacers can assist in preventing oral candidiasis by reducing the amount of medication that touches the back of the throat. Spacers are an option for patients, but not all use them for their MDIs, which can result in a less effective administered dose. Fluticasone/salmeterol is a dry powder inhaler, which requires a deep, fast breath to maximize the benefit, and spacers cannot be used with them. MDIs have been shown to be responsible for a negative impact on climate change, which can be reduced by switching to a dry powder inhaler.⁷

Tobacco cessation is very important in limiting the progression of COPD. As shown with the negative coefficient correlation, not being an active tobacco user at the time of transition correlated (although not significantly) with less frequent exacerbations. When comparing this study to similar research, such as the PATHOS study, several differences are observed.⁵ The PATHOS study compared long term treatment (> 1 year) of budesonide/formoterol or fluticasone/salmeterol, a longer period than this study. It regarded similar outcomes for the definition of an exacerbation, such as antibiotic/steroid use or hospital admission. While the current study showed no significant difference between the 2 inhalers and their effect on exacerbations, the PATHOS study found that those treated with a budesonide/formoterol inhaler were less likely to experience COPD-related exacerbations than those treated with the fluticasone/salmeterol inhaler. The PATHOS study had a larger mainly Scandinavian sample (N = 5500). This population could exhibit baseline differences from a study of US veterans.⁵ A similar Canadian matched cohort study of 2262 patients compared the 2 inhalers to assess their relative effectiveness. It found that COPD exacerbations did not differ between the 2 groups, but the budesonide/formoterol group was significantly less likely to have an emergency department visit compared to the fluticasone salmeterol group.⁸ Like the PATHOS study, the Canadian study had a larger sample size and longer timeframe than did our study.

 

 

Limitations

There are various limitations to this study. It was a retrospective, single-center study and the patient population was relatively homogenous, with only 1 female and a mean age of 71 years. As a study conducted in a veteran population in West Virginia, the findings may not be representative of the general population with COPD, which includes more women and more racial diversity.⁹ The American Lung Association discusses how environmental exposures to hazardous conditions increase the risks of pulmonary diseases for veterans.¹⁰ It has been reported that the prevalence of COPD is higher among veterans compared to the general population, but it is not different in terms of disease manifestation.¹⁰

Another limitation is the short time frame. Clinical guidelines, including the GOLD Report, typically track the number of exacerbations for 1 year to escalate therapy.³ Six months was a relatively short time frame, and it is possible that more exacerbations may have occurred beyond the study time frame. Ten patients in the sample died between the end of the study period and data retrieval, which might have been caught by a longer study period. An additional limitation was the inability to measure adherence. As this was a formulary conversion, many patients had been mailed a 30- or 90-day prescription of the budesonide/formoterol inhaler when transitioned to the fluticasone/salmeterol inhaler. There was no way to accurately determine when the patient made the switch to the fluticasone/salmeterol inhaler. This study also had a small sample group (a pre-post analysis of the same group), a limitation when evaluating the impact of this formulary change on a small percentage of the population transitioned.

This formulary conversion occurred during the COVID-19 pandemic, and some exacerbations could have been the result of a misdiagnosed COVID-19 infection. Respiratory infections, including COVID-19, are common causes of exacerbations. It is also possible that some patients elected not to receive medical care for symptoms of an exacerbation during the pandemic.¹¹

Conclusions

Switching from the budesonide/formoterol inhaler to the fluticasone/salmeterol inhaler through formulary conversion did not have a significant impact on the clinical outcomes in patients with COPD. This study found that although the inhalers contain different active ingredients, products within the same therapeutic class yielded nonsignificant changes. When conducting formulary conversions, intolerances and treatment failures should be expected when switching from different inhaler delivery systems. This study further justifies the ability to be cost effective by making formulary conversions within the same therapeutic class within a veterans population.

Acknowledgments

The authors would like to acknowledge James Brown, PharmD, PhD.

More than 37 million Americans (11.3%) have diabetes mellitus (DM), and 90% to 95% are diagnosed with type 2 DM, including nearly 1 in 4 veterans receiving Veterans Health Administration (VHA) care.^1,2 DM is associated with serious negative health outcomes, including cardiovascular disease and subsequent complications as well as significant health care system utilization and cost.^1,3

Group interventions have been identified as a possible method of improving DM outcomes. For example, shared medical appointments (SMAs) have been identified by the VHA as holding promise for improving care and efficiency for DM and other common health conditions.⁴ Although the precise structure and SMA process for managing DM has been noted to be heterogeneous, the appointment is typically led by an interdisciplinary health care team and includes individualized assessment including medication review and adjustment, group education, and troubleshooting challenges with management in a group format.⁵ Research suggests that DM SMAs are a worthwhile treatment approach.⁵ Several studies have found that SMAs were associated with decreased hemoglobin A_1c (Hb A_1c) levels and improvement in overall disease complications and severity.⁶

The high degree of SMA heterogeneity and lack of detailed description of structure and process of SMAs studied has made meta-analysis and other synthesis of the literature difficult.⁵ Consequently, there is inadequate empirically supported guidance for clinicians and health care organizations on how to best implement SMAs and similar group-based treatments. Edelman and colleagues recommended that future research should focus on more consistent and standardized intervention structures and real-world patient- and staff-centered outcomes to address gaps in the literature.⁵ They noted that a mental health professional was utilized in only a minority of SMAs studied.⁵ Additionally, we noted a paucity of studies examining patient satisfaction with SMAs.

Another group-based intervention found to be effective in improving DM outcomes is the 6-session Stanford Diabetes Self-Management Program (DSMP), a workshop led in part by trained peers with DM. The sessions focus on educating patients on DM care and self-management tools. The workshop encourages active practice in building DM self-management skills and confidence. DSMP participation has been associated with improvement in DM-related outcomes, including Hb A_1c levels, amount of exercise, and medication adherence.⁷

While SMAs and DSMP have been shown to enhance clinical outcomes, they provide differing types of patient support. SMAs allow for frequent interaction with a health care professional (HCP) and less emphasis on behavioral health interventions. DSMPs include behavioral health professionals and peer leaders and emphasize higher levels of psychosocial support, but do not offer access to clinicians. It is possible that combining these interventions could result in better outcomes than what either could provide on their own.

In 2018, the Cincinnati Veterans Affairs Medical Center (VAMC) in Ohio offered Diabetes Basic Training, a structured DM intervention. Patients enrolled in the program participated in a 9-week intervention that included 3 SMAs and 6 DSMP sessions. During the SMAs, a clinical psychologist or psychology postdoctoral fellow skilled in motivational interviewing facilitated the group to enhance patient engagement and empowerment for improved self-management. In addition, patients participated in structured DSMP groups with an emphasis on action-planning, often surrounding nutrition, physical activity, and other health behavior change information reviewed during the SMAs.

Design and Referral

Self-management programs for chronic health conditions are often underutilized. Although HCPs may wish to connect veterans with available programs, time constraints may limit opportunities for detailed discussions with patients about specific aspects of each program. To simplify this process, a 2-hour orientation program was offered that explained individual and group DM self-management options (Figure). During this initial visit, patients met with an interdisciplinary care team (registered dietician, diabetes nurse practitioner, and behavioral health specialist) and were informed about Diabetes Basic Training, DM clinical care practices, and other related resources available at the Cincinnati VAMC (eg, cooking classes, food pantry). Patients received individualized referral recommendations and were urged to consult with their primary care practitioner to finalize their treatment plan.

Shared Medical Appointments

Diabetes Basic Training interventions had an average of 6 to 8 veterans participating in the weekly groups. The first, fifth, and final weeks were SMAs in which an interdisciplinary team collaboratively provided group-based health care for DM. The team consisted of a registered nurse, a prescriber (eg, nurse practitioner), a moderator (eg, psychologist), and a content expert (eg, nutritionist). Before each SMA began, the nurse checked-in patients in the SMA room and collected heart rate and blood pressure, and performed a diabetic foot check. Each SMA consisted of introductions, group-driven discussions (facilitated by an HCP) and troubleshooting DM self-management challenges. During group discussions, the prescriber initiated a 1-on-1 discussion with each patient in a private office regarding their recent laboratory results, medication regimen, and other aspects of DM care. The patient’s medications were refilled and/or adjusted as needed and other orders and referrals were submitted. If a patient had a medical question, the prescriber and moderator engaged the entire group so all individuals could benefit from generating and hearing answers. When discussion slowed, education was provided on topics generated by the group. Frequent topics included challenges managing DM, concerns, how DM impacted daily life and relationships, and sharing successes. As needed, HCPs spoke individually with patients following the SMA. Patients were sometimes asked, but never required, to do homework consistent with standard DM care (eg, recording what they eat or blood sugar levels). Each SMA session lasted about 2 hours.

Diabetes Self-Management Program

The second, third, fourth, sixth, seventh, and eighth weeks of the program were devoted to the DSMP. These sessions were delivered primarily by veteran peers who received appropriate training, observation, and certification. Each 2-hour educational program provided ample practice in many fundamental self-management skills, such as decision making, problem solving, and action planning. Patients were asked, but never required, to practice related skills during the sessions and to create weekly action plans to be completed between sessions that typically involved increasing exercise or improving diet. Patients were encouraged to follow up with HCPs at SMAs when they had questions requiring HCP expertise. If participants had more immediate concerns regarding their treatment plan and/or medications, they contacted their primary care practitioner prior to the next SMA.

As a part of participation in the program, psychosocial and health data and Hb A_1c levels at baseline (the closest level to 90 days prior to start) and follow-up (the closest level to 90 days after the final session) were collected.⁸ In addition, Problem Areas in Diabetes (PAID), Patient Activation Measure (PAM)-13, and Diabetes Self-Management Questionnaire (DSMQ) were administered at 3 points: during the orientation, in the first week, and in the ninth week of the program.

PAID, a 20-item self-report questionnaire designed to capture emotional distress related to having DM, is a valid and reliable scale able to detect changes over time when used in intervention studies.^9,10 PAM-13 is a 13-item measure designed to assess patient knowledge, skill, and confidence in the self-management of health or chronic conditions based on the original PAM.^11,12 Scores fall into 1 of 4 activation levels, ranging from low levels of confidence and knowledge of health management to high levels of being proactive with one’s care. The PAM-13 has been widely used within health psychology, including research among adults with multiple chronic conditions, individuals with DM or osteoarthritis, and within primary care.^13-15 The DSMQ is a statistically reliable and valid instrument that allows for user-friendly assessment of self-care behaviors associated with glycemic control.^16-18

 

 

Observations

All measures were collected as part of traditional clinical care, and we present initial program evaluation data to demonstrate potential effectiveness of the clinic model. Paired samples t tests were used to examine differences between baseline and follow-up measures for the 24 veteran participants. The age of participants who completed the program ranged from 42 to 74 years (mean, 68 years); 29% of participants were Black veterans and 12% were female. Examination of clinical outcomes indicated that veterans reported significant increases in activation levels for managing their health increasing from a baseline mean (SD) 62.1 (12.3) to 68.4 (14.5) at follow up (t[23] = 2.15, P = .04). Hb A_1c levels trended downward from a mean (SD) 8.6% (1.3) at baseline to 8.2% (1.2) at 90-day follow up (t[21] 1.05, P = .30). Similar nonsignificant trends in PAID scores were seen for pre- and postprogram reductions in emotional distress related to having DM from a mean (SD) 7.9 (5.0) at baseline to 6.3 (5.1) (t[18] = 11.51, P = .15), and enhanced self-management of glucose with a mean (SD) 6.5 (1.5) at baseline to 6.8 (1.3) at follow up (t[19] = 0.52, P = .61). The trends found in this study show promising outcomes for this pilot group-based DM treatment, though the small sample size (N = 24) limits statistical power. These findings support further exploration and expansion of interdisciplinary health programs supporting veteran self-management.

Discussion

DM is a condition of epidemic proportions that causes substantial negative health outcomes and costs at a national level. Current standards of DM care do not appear to be reversing these trends. Wider implementation of group-based treatment for DM could improve efficiency of care, increase access to quality care, and reduce burden on individual HCPs.

The VHA continues the transformation of its care system, which shifts toward a patient-centered, proactive focus on veteran well-being. This new whole health approach integrates conventional medical treatment with veteran self-empowerment in the pursuit of health goals based on individual veteran’s identified values.¹⁹ This approach emphasizes peer-led explorations of veterans’ aspirations, purpose, and individual mission, personalized health planning, and use of whole health coaches and well-being programs, with both allopathic and complementary and integrative clinical care centered around veterans’ identified goals and priorities.²⁰

Including a program like Diabetes Basic Training as a part of whole health programming could offer several benefits. Diabetes Basic Training is unique in its integration of more traditional SMA structure with psychosocial interventions including values identification and motivational interviewing strategies to enhance patient engagement. Veterans can learn from each other’s experiences and concerns, leading to better DM management knowledge and skills. The group nature of the sessions enhances opportunities for emotional support and reduced isolation, as well as peer accountability for maintaining medication adherence.

By meeting with HCPs from multiple disciplines, veterans are exposed to different perspectives on self-management techniques, including behavioral approaches for overcoming barriers to behavior change. Clinicians have more time to engage with patients, building stronger relationships and trust. SMAs are cost-efficient and time efficient, allowing HCPs to see multiple patients at once, reducing wait times and increasing the number of patients treated in a given time frame.

The COVID-19 pandemic temporarily impacted the ongoing expansion of the program, when so many services were shifted from in-person to virtual classes. Due to staffing and other logistic issues, our pilot program was suspended during that time, but plans to resume the program by early 2024 are moving forward.

CONCLUSIONS

The Diabetes Basic Training program serves as a successful model for implementation within a VAMC. Although the number of veterans with complete data available for analysis was small, the trends exhibited in the preliminary outcome data are promising. We encourage other VAMCs to replicate this program with a larger participant base and evaluate its impact on veteran health outcomes. Next steps include comparing the clinical data from treatment as usual with outcomes from DM group participants. As the program resumes, we will reinitiate recruitment efforts to increase HCP referrals to this program.

More than 37 million Americans (11.3%) have diabetes mellitus (DM), and 90% to 95% are diagnosed with type 2 DM, including nearly 1 in 4 veterans receiving Veterans Health Administration (VHA) care.^1,2 DM is associated with serious negative health outcomes, including cardiovascular disease and subsequent complications as well as significant health care system utilization and cost.^1,3

Group interventions have been identified as a possible method of improving DM outcomes. For example, shared medical appointments (SMAs) have been identified by the VHA as holding promise for improving care and efficiency for DM and other common health conditions.⁴ Although the precise structure and SMA process for managing DM has been noted to be heterogeneous, the appointment is typically led by an interdisciplinary health care team and includes individualized assessment including medication review and adjustment, group education, and troubleshooting challenges with management in a group format.⁵ Research suggests that DM SMAs are a worthwhile treatment approach.⁵ Several studies have found that SMAs were associated with decreased hemoglobin A_1c (Hb A_1c) levels and improvement in overall disease complications and severity.⁶

The high degree of SMA heterogeneity and lack of detailed description of structure and process of SMAs studied has made meta-analysis and other synthesis of the literature difficult.⁵ Consequently, there is inadequate empirically supported guidance for clinicians and health care organizations on how to best implement SMAs and similar group-based treatments. Edelman and colleagues recommended that future research should focus on more consistent and standardized intervention structures and real-world patient- and staff-centered outcomes to address gaps in the literature.⁵ They noted that a mental health professional was utilized in only a minority of SMAs studied.⁵ Additionally, we noted a paucity of studies examining patient satisfaction with SMAs.

Another group-based intervention found to be effective in improving DM outcomes is the 6-session Stanford Diabetes Self-Management Program (DSMP), a workshop led in part by trained peers with DM. The sessions focus on educating patients on DM care and self-management tools. The workshop encourages active practice in building DM self-management skills and confidence. DSMP participation has been associated with improvement in DM-related outcomes, including Hb A_1c levels, amount of exercise, and medication adherence.⁷

While SMAs and DSMP have been shown to enhance clinical outcomes, they provide differing types of patient support. SMAs allow for frequent interaction with a health care professional (HCP) and less emphasis on behavioral health interventions. DSMPs include behavioral health professionals and peer leaders and emphasize higher levels of psychosocial support, but do not offer access to clinicians. It is possible that combining these interventions could result in better outcomes than what either could provide on their own.

In 2018, the Cincinnati Veterans Affairs Medical Center (VAMC) in Ohio offered Diabetes Basic Training, a structured DM intervention. Patients enrolled in the program participated in a 9-week intervention that included 3 SMAs and 6 DSMP sessions. During the SMAs, a clinical psychologist or psychology postdoctoral fellow skilled in motivational interviewing facilitated the group to enhance patient engagement and empowerment for improved self-management. In addition, patients participated in structured DSMP groups with an emphasis on action-planning, often surrounding nutrition, physical activity, and other health behavior change information reviewed during the SMAs.

Design and Referral

Self-management programs for chronic health conditions are often underutilized. Although HCPs may wish to connect veterans with available programs, time constraints may limit opportunities for detailed discussions with patients about specific aspects of each program. To simplify this process, a 2-hour orientation program was offered that explained individual and group DM self-management options (Figure). During this initial visit, patients met with an interdisciplinary care team (registered dietician, diabetes nurse practitioner, and behavioral health specialist) and were informed about Diabetes Basic Training, DM clinical care practices, and other related resources available at the Cincinnati VAMC (eg, cooking classes, food pantry). Patients received individualized referral recommendations and were urged to consult with their primary care practitioner to finalize their treatment plan.

Shared Medical Appointments

Diabetes Basic Training interventions had an average of 6 to 8 veterans participating in the weekly groups. The first, fifth, and final weeks were SMAs in which an interdisciplinary team collaboratively provided group-based health care for DM. The team consisted of a registered nurse, a prescriber (eg, nurse practitioner), a moderator (eg, psychologist), and a content expert (eg, nutritionist). Before each SMA began, the nurse checked-in patients in the SMA room and collected heart rate and blood pressure, and performed a diabetic foot check. Each SMA consisted of introductions, group-driven discussions (facilitated by an HCP) and troubleshooting DM self-management challenges. During group discussions, the prescriber initiated a 1-on-1 discussion with each patient in a private office regarding their recent laboratory results, medication regimen, and other aspects of DM care. The patient’s medications were refilled and/or adjusted as needed and other orders and referrals were submitted. If a patient had a medical question, the prescriber and moderator engaged the entire group so all individuals could benefit from generating and hearing answers. When discussion slowed, education was provided on topics generated by the group. Frequent topics included challenges managing DM, concerns, how DM impacted daily life and relationships, and sharing successes. As needed, HCPs spoke individually with patients following the SMA. Patients were sometimes asked, but never required, to do homework consistent with standard DM care (eg, recording what they eat or blood sugar levels). Each SMA session lasted about 2 hours.

Diabetes Self-Management Program

The second, third, fourth, sixth, seventh, and eighth weeks of the program were devoted to the DSMP. These sessions were delivered primarily by veteran peers who received appropriate training, observation, and certification. Each 2-hour educational program provided ample practice in many fundamental self-management skills, such as decision making, problem solving, and action planning. Patients were asked, but never required, to practice related skills during the sessions and to create weekly action plans to be completed between sessions that typically involved increasing exercise or improving diet. Patients were encouraged to follow up with HCPs at SMAs when they had questions requiring HCP expertise. If participants had more immediate concerns regarding their treatment plan and/or medications, they contacted their primary care practitioner prior to the next SMA.

As a part of participation in the program, psychosocial and health data and Hb A_1c levels at baseline (the closest level to 90 days prior to start) and follow-up (the closest level to 90 days after the final session) were collected.⁸ In addition, Problem Areas in Diabetes (PAID), Patient Activation Measure (PAM)-13, and Diabetes Self-Management Questionnaire (DSMQ) were administered at 3 points: during the orientation, in the first week, and in the ninth week of the program.

PAID, a 20-item self-report questionnaire designed to capture emotional distress related to having DM, is a valid and reliable scale able to detect changes over time when used in intervention studies.^9,10 PAM-13 is a 13-item measure designed to assess patient knowledge, skill, and confidence in the self-management of health or chronic conditions based on the original PAM.^11,12 Scores fall into 1 of 4 activation levels, ranging from low levels of confidence and knowledge of health management to high levels of being proactive with one’s care. The PAM-13 has been widely used within health psychology, including research among adults with multiple chronic conditions, individuals with DM or osteoarthritis, and within primary care.^13-15 The DSMQ is a statistically reliable and valid instrument that allows for user-friendly assessment of self-care behaviors associated with glycemic control.^16-18

 

 

Observations

All measures were collected as part of traditional clinical care, and we present initial program evaluation data to demonstrate potential effectiveness of the clinic model. Paired samples t tests were used to examine differences between baseline and follow-up measures for the 24 veteran participants. The age of participants who completed the program ranged from 42 to 74 years (mean, 68 years); 29% of participants were Black veterans and 12% were female. Examination of clinical outcomes indicated that veterans reported significant increases in activation levels for managing their health increasing from a baseline mean (SD) 62.1 (12.3) to 68.4 (14.5) at follow up (t[23] = 2.15, P = .04). Hb A_1c levels trended downward from a mean (SD) 8.6% (1.3) at baseline to 8.2% (1.2) at 90-day follow up (t[21] 1.05, P = .30). Similar nonsignificant trends in PAID scores were seen for pre- and postprogram reductions in emotional distress related to having DM from a mean (SD) 7.9 (5.0) at baseline to 6.3 (5.1) (t[18] = 11.51, P = .15), and enhanced self-management of glucose with a mean (SD) 6.5 (1.5) at baseline to 6.8 (1.3) at follow up (t[19] = 0.52, P = .61). The trends found in this study show promising outcomes for this pilot group-based DM treatment, though the small sample size (N = 24) limits statistical power. These findings support further exploration and expansion of interdisciplinary health programs supporting veteran self-management.

Discussion

DM is a condition of epidemic proportions that causes substantial negative health outcomes and costs at a national level. Current standards of DM care do not appear to be reversing these trends. Wider implementation of group-based treatment for DM could improve efficiency of care, increase access to quality care, and reduce burden on individual HCPs.

The VHA continues the transformation of its care system, which shifts toward a patient-centered, proactive focus on veteran well-being. This new whole health approach integrates conventional medical treatment with veteran self-empowerment in the pursuit of health goals based on individual veteran’s identified values.¹⁹ This approach emphasizes peer-led explorations of veterans’ aspirations, purpose, and individual mission, personalized health planning, and use of whole health coaches and well-being programs, with both allopathic and complementary and integrative clinical care centered around veterans’ identified goals and priorities.²⁰

Including a program like Diabetes Basic Training as a part of whole health programming could offer several benefits. Diabetes Basic Training is unique in its integration of more traditional SMA structure with psychosocial interventions including values identification and motivational interviewing strategies to enhance patient engagement. Veterans can learn from each other’s experiences and concerns, leading to better DM management knowledge and skills. The group nature of the sessions enhances opportunities for emotional support and reduced isolation, as well as peer accountability for maintaining medication adherence.

By meeting with HCPs from multiple disciplines, veterans are exposed to different perspectives on self-management techniques, including behavioral approaches for overcoming barriers to behavior change. Clinicians have more time to engage with patients, building stronger relationships and trust. SMAs are cost-efficient and time efficient, allowing HCPs to see multiple patients at once, reducing wait times and increasing the number of patients treated in a given time frame.

The COVID-19 pandemic temporarily impacted the ongoing expansion of the program, when so many services were shifted from in-person to virtual classes. Due to staffing and other logistic issues, our pilot program was suspended during that time, but plans to resume the program by early 2024 are moving forward.

CONCLUSIONS

The Diabetes Basic Training program serves as a successful model for implementation within a VAMC. Although the number of veterans with complete data available for analysis was small, the trends exhibited in the preliminary outcome data are promising. We encourage other VAMCs to replicate this program with a larger participant base and evaluate its impact on veteran health outcomes. Next steps include comparing the clinical data from treatment as usual with outcomes from DM group participants. As the program resumes, we will reinitiate recruitment efforts to increase HCP referrals to this program.

More than 37 million Americans (11.3%) have diabetes mellitus (DM), and 90% to 95% are diagnosed with type 2 DM, including nearly 1 in 4 veterans receiving Veterans Health Administration (VHA) care.^1,2 DM is associated with serious negative health outcomes, including cardiovascular disease and subsequent complications as well as significant health care system utilization and cost.^1,3

Group interventions have been identified as a possible method of improving DM outcomes. For example, shared medical appointments (SMAs) have been identified by the VHA as holding promise for improving care and efficiency for DM and other common health conditions.⁴ Although the precise structure and SMA process for managing DM has been noted to be heterogeneous, the appointment is typically led by an interdisciplinary health care team and includes individualized assessment including medication review and adjustment, group education, and troubleshooting challenges with management in a group format.⁵ Research suggests that DM SMAs are a worthwhile treatment approach.⁵ Several studies have found that SMAs were associated with decreased hemoglobin A_1c (Hb A_1c) levels and improvement in overall disease complications and severity.⁶

The high degree of SMA heterogeneity and lack of detailed description of structure and process of SMAs studied has made meta-analysis and other synthesis of the literature difficult.⁵ Consequently, there is inadequate empirically supported guidance for clinicians and health care organizations on how to best implement SMAs and similar group-based treatments. Edelman and colleagues recommended that future research should focus on more consistent and standardized intervention structures and real-world patient- and staff-centered outcomes to address gaps in the literature.⁵ They noted that a mental health professional was utilized in only a minority of SMAs studied.⁵ Additionally, we noted a paucity of studies examining patient satisfaction with SMAs.

Another group-based intervention found to be effective in improving DM outcomes is the 6-session Stanford Diabetes Self-Management Program (DSMP), a workshop led in part by trained peers with DM. The sessions focus on educating patients on DM care and self-management tools. The workshop encourages active practice in building DM self-management skills and confidence. DSMP participation has been associated with improvement in DM-related outcomes, including Hb A_1c levels, amount of exercise, and medication adherence.⁷

While SMAs and DSMP have been shown to enhance clinical outcomes, they provide differing types of patient support. SMAs allow for frequent interaction with a health care professional (HCP) and less emphasis on behavioral health interventions. DSMPs include behavioral health professionals and peer leaders and emphasize higher levels of psychosocial support, but do not offer access to clinicians. It is possible that combining these interventions could result in better outcomes than what either could provide on their own.

In 2018, the Cincinnati Veterans Affairs Medical Center (VAMC) in Ohio offered Diabetes Basic Training, a structured DM intervention. Patients enrolled in the program participated in a 9-week intervention that included 3 SMAs and 6 DSMP sessions. During the SMAs, a clinical psychologist or psychology postdoctoral fellow skilled in motivational interviewing facilitated the group to enhance patient engagement and empowerment for improved self-management. In addition, patients participated in structured DSMP groups with an emphasis on action-planning, often surrounding nutrition, physical activity, and other health behavior change information reviewed during the SMAs.

Design and Referral

Self-management programs for chronic health conditions are often underutilized. Although HCPs may wish to connect veterans with available programs, time constraints may limit opportunities for detailed discussions with patients about specific aspects of each program. To simplify this process, a 2-hour orientation program was offered that explained individual and group DM self-management options (Figure). During this initial visit, patients met with an interdisciplinary care team (registered dietician, diabetes nurse practitioner, and behavioral health specialist) and were informed about Diabetes Basic Training, DM clinical care practices, and other related resources available at the Cincinnati VAMC (eg, cooking classes, food pantry). Patients received individualized referral recommendations and were urged to consult with their primary care practitioner to finalize their treatment plan.

Shared Medical Appointments

Diabetes Basic Training interventions had an average of 6 to 8 veterans participating in the weekly groups. The first, fifth, and final weeks were SMAs in which an interdisciplinary team collaboratively provided group-based health care for DM. The team consisted of a registered nurse, a prescriber (eg, nurse practitioner), a moderator (eg, psychologist), and a content expert (eg, nutritionist). Before each SMA began, the nurse checked-in patients in the SMA room and collected heart rate and blood pressure, and performed a diabetic foot check. Each SMA consisted of introductions, group-driven discussions (facilitated by an HCP) and troubleshooting DM self-management challenges. During group discussions, the prescriber initiated a 1-on-1 discussion with each patient in a private office regarding their recent laboratory results, medication regimen, and other aspects of DM care. The patient’s medications were refilled and/or adjusted as needed and other orders and referrals were submitted. If a patient had a medical question, the prescriber and moderator engaged the entire group so all individuals could benefit from generating and hearing answers. When discussion slowed, education was provided on topics generated by the group. Frequent topics included challenges managing DM, concerns, how DM impacted daily life and relationships, and sharing successes. As needed, HCPs spoke individually with patients following the SMA. Patients were sometimes asked, but never required, to do homework consistent with standard DM care (eg, recording what they eat or blood sugar levels). Each SMA session lasted about 2 hours.

Diabetes Self-Management Program

The second, third, fourth, sixth, seventh, and eighth weeks of the program were devoted to the DSMP. These sessions were delivered primarily by veteran peers who received appropriate training, observation, and certification. Each 2-hour educational program provided ample practice in many fundamental self-management skills, such as decision making, problem solving, and action planning. Patients were asked, but never required, to practice related skills during the sessions and to create weekly action plans to be completed between sessions that typically involved increasing exercise or improving diet. Patients were encouraged to follow up with HCPs at SMAs when they had questions requiring HCP expertise. If participants had more immediate concerns regarding their treatment plan and/or medications, they contacted their primary care practitioner prior to the next SMA.

As a part of participation in the program, psychosocial and health data and Hb A_1c levels at baseline (the closest level to 90 days prior to start) and follow-up (the closest level to 90 days after the final session) were collected.⁸ In addition, Problem Areas in Diabetes (PAID), Patient Activation Measure (PAM)-13, and Diabetes Self-Management Questionnaire (DSMQ) were administered at 3 points: during the orientation, in the first week, and in the ninth week of the program.

PAID, a 20-item self-report questionnaire designed to capture emotional distress related to having DM, is a valid and reliable scale able to detect changes over time when used in intervention studies.^9,10 PAM-13 is a 13-item measure designed to assess patient knowledge, skill, and confidence in the self-management of health or chronic conditions based on the original PAM.^11,12 Scores fall into 1 of 4 activation levels, ranging from low levels of confidence and knowledge of health management to high levels of being proactive with one’s care. The PAM-13 has been widely used within health psychology, including research among adults with multiple chronic conditions, individuals with DM or osteoarthritis, and within primary care.^13-15 The DSMQ is a statistically reliable and valid instrument that allows for user-friendly assessment of self-care behaviors associated with glycemic control.^16-18

 

 

Observations

All measures were collected as part of traditional clinical care, and we present initial program evaluation data to demonstrate potential effectiveness of the clinic model. Paired samples t tests were used to examine differences between baseline and follow-up measures for the 24 veteran participants. The age of participants who completed the program ranged from 42 to 74 years (mean, 68 years); 29% of participants were Black veterans and 12% were female. Examination of clinical outcomes indicated that veterans reported significant increases in activation levels for managing their health increasing from a baseline mean (SD) 62.1 (12.3) to 68.4 (14.5) at follow up (t[23] = 2.15, P = .04). Hb A_1c levels trended downward from a mean (SD) 8.6% (1.3) at baseline to 8.2% (1.2) at 90-day follow up (t[21] 1.05, P = .30). Similar nonsignificant trends in PAID scores were seen for pre- and postprogram reductions in emotional distress related to having DM from a mean (SD) 7.9 (5.0) at baseline to 6.3 (5.1) (t[18] = 11.51, P = .15), and enhanced self-management of glucose with a mean (SD) 6.5 (1.5) at baseline to 6.8 (1.3) at follow up (t[19] = 0.52, P = .61). The trends found in this study show promising outcomes for this pilot group-based DM treatment, though the small sample size (N = 24) limits statistical power. These findings support further exploration and expansion of interdisciplinary health programs supporting veteran self-management.

Discussion

DM is a condition of epidemic proportions that causes substantial negative health outcomes and costs at a national level. Current standards of DM care do not appear to be reversing these trends. Wider implementation of group-based treatment for DM could improve efficiency of care, increase access to quality care, and reduce burden on individual HCPs.

The VHA continues the transformation of its care system, which shifts toward a patient-centered, proactive focus on veteran well-being. This new whole health approach integrates conventional medical treatment with veteran self-empowerment in the pursuit of health goals based on individual veteran’s identified values.¹⁹ This approach emphasizes peer-led explorations of veterans’ aspirations, purpose, and individual mission, personalized health planning, and use of whole health coaches and well-being programs, with both allopathic and complementary and integrative clinical care centered around veterans’ identified goals and priorities.²⁰

Including a program like Diabetes Basic Training as a part of whole health programming could offer several benefits. Diabetes Basic Training is unique in its integration of more traditional SMA structure with psychosocial interventions including values identification and motivational interviewing strategies to enhance patient engagement. Veterans can learn from each other’s experiences and concerns, leading to better DM management knowledge and skills. The group nature of the sessions enhances opportunities for emotional support and reduced isolation, as well as peer accountability for maintaining medication adherence.

By meeting with HCPs from multiple disciplines, veterans are exposed to different perspectives on self-management techniques, including behavioral approaches for overcoming barriers to behavior change. Clinicians have more time to engage with patients, building stronger relationships and trust. SMAs are cost-efficient and time efficient, allowing HCPs to see multiple patients at once, reducing wait times and increasing the number of patients treated in a given time frame.

The COVID-19 pandemic temporarily impacted the ongoing expansion of the program, when so many services were shifted from in-person to virtual classes. Due to staffing and other logistic issues, our pilot program was suspended during that time, but plans to resume the program by early 2024 are moving forward.

CONCLUSIONS

The Diabetes Basic Training program serves as a successful model for implementation within a VAMC. Although the number of veterans with complete data available for analysis was small, the trends exhibited in the preliminary outcome data are promising. We encourage other VAMCs to replicate this program with a larger participant base and evaluate its impact on veteran health outcomes. Next steps include comparing the clinical data from treatment as usual with outcomes from DM group participants. As the program resumes, we will reinitiate recruitment efforts to increase HCP referrals to this program.