Provider and Site-Level Determinants of Testosterone Prescribing in the Veterans Healthcare System

Development and Validation of Quality Measures for Testosterone Prescribing

Context

Accurate measures to assess appropriateness of testosterone prescribing are needed to improve prescribing practices.

Objective

This work aimed to develop and validate quality measures around the initiation and monitoring of testosterone prescribing.

Methods

This retrospective cohort study comprised a national cohort of male patients receiving care in the Veterans Health Administration who initiated testosterone during January or February 2020. Using laboratory data and diagnostic codes, we developed 9 initiation and 7 monitoring measures. These were based on the current Endocrine Society guidelines supplemented by expert opinion and prior work. We chose measures that could be operationalized using national VA electronic health record (EHR) data. We assessed criterion validity for these 16 measures by manual review of 142 charts. Main outcome measures included positive and negative predictive values (PPVs, NPVs), overall accuracy (OA), and Matthews Correlation Coefficients (MCCs).

Results

We found high PPVs (>78%), NPVs (>98%), OA (≥94%), and MCCs (>0.85) for the 10 measures based on laboratory data (5 initiation and 5 monitoring). For the 6 measures relying on diagnostic codes, we similarly found high NPVs (100%) and OAs (≥98%). However, PPVs for measures of acute conditions occurring before testosterone initiation (ie, acute myocardial infarction or stroke) or new conditions occurring after initiation (ie, prostate or breast cancer) PPVs were much lower (0% to 50%) due to few or no cases.

Conclusion

We developed several valid EHR-based quality measures for assessing testosterone-prescribing practices. Deployment of these measures in health care systems can facilitate identification of quality gaps in testosterone-prescribing and improve care of men with hypogonadism.

Physicians' clinical experience and its association with healthcare quality: a systematised review

BACKGROUND AND PURPOSE

There is conflicting evidence regarding whether physicians' clinical experience affects healthcare quality. Knowing whether an association exists and which dimensions of quality might be affected can help healthcare services close quality gaps by tailoring improvement initiatives according to physicians' clinical experience. Here, we present a systematised review that aims to assess the potential association between physicians' clinical experience and different dimensions of healthcare quality.

METHODS

We conducted a systematised literature review, including the databases MEDLINE, Embase, PsycINFO and PubMed. The search strategy involved combining predefined terms that describe physicians' clinical experience with terms that describe different dimensions of healthcare quality (ie, safety, clinical effectiveness, patient-centredness, timeliness, efficiency and equity). We included relevant, original research published from June 2004 to November 2020.

RESULTS

Fifty-two studies reporting 63 evaluations of the association between physicians' clinical experience and healthcare quality were included in the final analysis. Overall, 27 (43%) evaluations found a positive or partially positive association between physicians' clinical experience and healthcare quality; 22 (35%) found no association; and 14 (22%) evaluations reported a negative or partially negative association. We found a proportional association between physicians' clinical experience and quality regarding outcome measures that reflect safety, particularly in the surgical fields. For other dimensions of quality, no firm evidence was found.

CONCLUSION

We found no clear evidence of an association between measures of physicians' clinical experience and overall healthcare quality. For outcome measures related to safety, we found that physicians' clinical experience was proportional with safer care, particularly in surgical fields. Our findings support efforts to secure adequate training and supervision for early-career physicians regarding safety outcomes. Further research is needed to reveal the potential subgroups in which gaps in quality due to physicians' clinical experience might exist.

Polysubstance use and association with opioid use disorder treatment in the US Veterans Health Administration

AIMS

To understand the role of comorbid substance use disorders (SUDs), or polysubstance use, in the treatment of opioid use disorder (OUD), this study compared patients with OUD only to those with additional SUDs and examined association with OUD treatment receipt.

DESIGN, SETTING AND PARTICIPANTS

Retrospective national cohort study of Veterans diagnosed with OUD (n = 65 741) receiving care from the US Veterans Health Administration (VHA) in fiscal year (FY) 2017.

MEASUREMENTS

Patient characteristics were compared among those diagnosed with OUD only versus those with one other SUD (OUD + 1 SUD) and with multiple SUDs (OUD + ≥ 2 SUDs). The study examined the relationship between comorbid SUDs and receipt of buprenorphine, methadone and SUD outpatient treatment during 1-year follow-up, adjusting for patient demographic characteristics and clinical conditions.

FINDINGS

Among the 65 741 Veterans with OUD in FY 2017, 41.2% had OUD only, 22.9% had OUD + 1 SUD and 35.9% had OUD + ≥ 2 SUDs. Common comorbid SUDs included alcohol use disorder (41.3%), cocaine/stimulant use disorder (30.0%) and cannabis use disorder (22.4%). Adjusting for patient characteristics, patients with OUD + 1 SUD [adjusted odds ratio (aOR) = 0.87, 95% confidence interval (CI) = 0.82-0.93] and patients with OUD +≥ 2 SUDs (aOR = 0.65, 95% CI = 0.61-0.69) had lower odds of receiving buprenorphine compared with OUD only patients. There were also lower odds of receiving methadone for patients with OUD + 1 SUD (aOR = 0.91, 95% CI = 0.86-0.97)and for those with OUD + ≥2 SUDs (aOR = 0.79, 95% CI = 0.74-0.84). Patients with OUD + 1 SUD (aOR = 1.85, 95% CI = 1.77-1.93) and patients with OUD + ≥2 SUDs (aOR = 3.25, 95% CI = 3.103.41) were much more likely to have a SUD clinic visit.

CONCLUSIONS

The majority of Veterans in the US Veterans Health Administration diagnosed with opioid use disorder appeared to have at least one comorbid substance use disorder and many have multiple substance use disorders. Despite the higher likelihood of a substance use disorder clinic visit, having a non-opioid substance use disorder is associated with lower likelihood of buprenorphine treatment, suggesting the importance of addressing polysubstance use within efforts to expand treatment for opioid use disorder.

Understanding the Context of High- and Low-Testosterone Prescribing Facilities in the Veterans Health Administration (VHA): a Qualitative Study

BACKGROUND

Inappropriate testosterone use and variations in testosterone prescribing patterns exist in the Veterans Health Administration (VHA) despite the presence of clinical guidelines.

OBJECTIVE

We examined system and clinician factors that contribute to patterns of potentially inappropriate testosterone prescribing in VHA.

DESIGN

Qualitative study using a positive deviance approach to understand practice variation in high- and low-testosterone prescribing sites.

PARTICIPANTS

Twenty-two interview participants included primary care and specialty clinicians, key opinion leaders, and pharmacists at 3 high- and 3 low-testosterone prescribing sites.

APPROACH

Semi-structured phone interviews were conducted, transcribed, and coded using a priori theoretical constructs and emergent themes. Case studies were developed for each site and a cross-case matrix was created to evaluate variation across high- and low-prescribing sites.

KEY RESULTS

We identified four system-level domains related to variation in testosterone prescribing: organizational structures and processes specific to testosterone prescribing, availability of local guidance on testosterone prescribing, well-defined dissemination process for local testosterone polices, and engagement in best practices related to testosterone prescribing. Two clinician-level domains were also identified, specifically, structured initial testosterone prescribing process and specified follow-up testosterone prescribing process. High- and low-testosterone prescribing sites systematically varied in the four system-level domains, while the clinician-level domains looked similar across all sites. The third high-prescribing site was unusual in that it exhibited the four domains similar to the 3 low-prescribing sites at the time of our visit. This site had greatly reduced its prescribing of testosterone in the interim.

CONCLUSIONS

Findings suggest that local organizational factors play an important role in influencing prescribing. Sites have the potential to transform their utilization patterns by providing access to specialty care expertise, an electronic health record-based system to facilitate guideline-concordant prescribing, well-defined dissemination processes for information, guidance from multiple sources, and clarity regarding best practices for prescribing.

Use of testosterone in men infected with human immunodeficiency virus in the veterans healthcare system

Testosterone supplementation has been widely used in those infected with human immunodeficiency virus (HIV) for hypogonadism, and wasting. But with effective antiretroviral therapy and increasing recognition of atherosclerotic disease and adults infected with HIV, the risks of inappropriate testosterone use in HIV-infected patients are far better recognized than previously. Testosterone use has expanded among U.S. males, but few studies have examined prescribing in those infected with HIV. In a national cohort of males with at least one outpatient prescription in the Veterans Health Administration (VHA), we examined 9475 HIV-infected males, including 2484 who had received testosterone and a randomly selected 6991 who had not. For comparison, we identified 1,387,241 uninfected males (189,369 had received testosterone and a randomly selected 1,197,872 had not). We determined rates of new and prevalent testosterone use, and also examined the adequacy of the diagnostic evaluation that had preceded testosterone initiation among our HIV-infected and uninfected testosterone groups. Our main results were as follows. HIV-infected men had higher rates of initiation (0.8% vs. 0.4% in FY09; p < 0.001) and prevalence of testosterone use (2.2% vs. 0.8% in FY08; p < 0.001) compared to the uninfected men across the entire period. Trends of prescribing for both groups followed a similar pattern, rising from FY08, reaching a peak in FY13, and then dipping in FY 14. Only 1.1% of HIV-infected patients had a fully guideline-concordant workup before starting testosterone therapy, compared to 3.5% of uninfected patients (p < 0.001). In conclusion, testosterone use among HIV-infected patients in the VHA system rose to a peak in FY13 and has decreased somewhat since. Only a small minority of HIV-infected patients who receive testosterone therapy from VHA have undergone an appropriate workup before starting therapy, suggesting an opportunity for improvement.