Cardiology

A 59-year-old man with a history of coronary artery bypass grafting (CABG), ischemic cardiomyopathy (ejection fraction, 15%-20%) with implantable cardioverter-defibrillator, recurrent paroxysmal ventricular tachycardia on amiodarone and mexiletine, and heart failure requiring left ventricular assist device (LVAD) placement presented for recurrent cellulitis and infection of the LVAD driveline exit site. He was initiated on minocycline 100 mg twice daily in combination with cefadroxil 500 mg twice daily. At his 8-week follow-up, the driveline site appeared improved with minimal erythema and no drainage. However, the patient developed a well-demarcated, linear, hyperpigmented patch along the length of the saphenous vein CABG harvest site and a few hyperpigmented macules medial to the harvest site (Figure).

Discussion

Hyperpigmentation presenting within scar tissue, as seen in this patient undergoing minocycline therapy, is a classic presentation of minocycline-induced hyperpigmentation (MIH) type I.

MIH is an uncommon, potentially cosmetically disfiguring adverse effect associated with systemic minocycline use. MIH can affect skin, teeth, nails, oral mucosa, sclera, and internal organs. The cumulative incidence of MIH in patients receiving minocycline over prolonged periods of time has been estimated from 2% to 15% in patients with acne and rosacea, to approximately 50% over 5 years in orthopedic patient populations.^1-3 The risk for developing MIH increases with vitamin D deficiency, liver disease, concurrent use with other medications that can induce hyperpigmentation, and higher cumulative doses (> 70-100 g; more important for MIH types II and III).^3,4 

There are 3 distinct types of MIH. Type I MIH is characterized by blue-black macules and patches at sites of inflammation or prior scarring, most commonly described in facial acne scars.^1,2,4 Type II is typified by blue-grey pigmentation on normal-appearing skin, most commonly on the shins, but also on sun-exposed sites.³ Biopsies of type I and II MIH demonstrate pigmented granules within macrophages or within the dermis.^4,5 Both Perls iron stain and Fontana-Masson melanin stain are positive in type I and II MIH.⁵ Type III MIH presents as diffuse brownish hyperpigmentation on normal skin in chronically sun-exposed sites.³ Histopathology of type III MIH can be distinguished by increased melanin noted inside basal keratinocytes as well as dermal melanophages that stain positive for only Fontana-Masson.⁵ The current case exemplifies a unique presentation of type I MIH along the length of the saphenous vein CABG harvest site. The concomitant use of amiodarone with minocycline may have contributed to the presentation.

The differential diagnosis for MIH depends on the type of MIH. Blue-grey pigmentation within scars is fairly unique to minocycline but has been reported with other medications, including vandetanib.⁶ The differential diagnosis for diffuse blue-grey or brown hyperpigmentation in predominately sun-exposed sites is broader, including endocrine disorders (ie, Addison disease), heavy metal poisoning (ie, argyria), inherited conditions (ie, alkaptonuria, Wilson disease, and hemochromatosis), medication-induced hyperpigmentation (ie, antipsychotics, anticonvulsant, antimalarials, amiodarone, and cytotoxic drugs), as well as inflammatory dermatoses, such as erythema dyschromicum perstans.⁷

MIH typically fades over months to years following minocycline discontinuation, so prompt recognition and discontinuation is recommended. Unfortunately, some cases persist or only partially fade over time. While MIH is benign, it can be of aesthetic concern, cause anxiety, and impact patients’ quality of life.^3,8 Persistent MIH is typically recalcitrant to topical hydroquinone.⁹ However, persistent MIH has been shown to improve with Q-switched, nanosecond lasers such as the 694 nm ruby, 755 nm alexandrite, and 1064 nm neodymium-doped yttrium aluminum garnet neodymium (Nd:YAG) lasers, as well as the 755 nm picosecond alexandrite laser.^4,9,10

In our patient, minocycline therapy was discontinued and replaced with doxycycline 100 mg twice daily monotherapy. At a subsequent visit 12 weeks later, the hyperpigmentation remained unchanged.

Conclusions

Though uncommon, we hope to encourage clinician awareness of MIH through our case, as prompt diagnosis and the discontinuation of minocycline are preferred to improve patient outcomes.

A 59-year-old man with a history of coronary artery bypass grafting (CABG), ischemic cardiomyopathy (ejection fraction, 15%-20%) with implantable cardioverter-defibrillator, recurrent paroxysmal ventricular tachycardia on amiodarone and mexiletine, and heart failure requiring left ventricular assist device (LVAD) placement presented for recurrent cellulitis and infection of the LVAD driveline exit site. He was initiated on minocycline 100 mg twice daily in combination with cefadroxil 500 mg twice daily. At his 8-week follow-up, the driveline site appeared improved with minimal erythema and no drainage. However, the patient developed a well-demarcated, linear, hyperpigmented patch along the length of the saphenous vein CABG harvest site and a few hyperpigmented macules medial to the harvest site (Figure).

Discussion

Hyperpigmentation presenting within scar tissue, as seen in this patient undergoing minocycline therapy, is a classic presentation of minocycline-induced hyperpigmentation (MIH) type I.

MIH is an uncommon, potentially cosmetically disfiguring adverse effect associated with systemic minocycline use. MIH can affect skin, teeth, nails, oral mucosa, sclera, and internal organs. The cumulative incidence of MIH in patients receiving minocycline over prolonged periods of time has been estimated from 2% to 15% in patients with acne and rosacea, to approximately 50% over 5 years in orthopedic patient populations.^1-3 The risk for developing MIH increases with vitamin D deficiency, liver disease, concurrent use with other medications that can induce hyperpigmentation, and higher cumulative doses (> 70-100 g; more important for MIH types II and III).^3,4 

There are 3 distinct types of MIH. Type I MIH is characterized by blue-black macules and patches at sites of inflammation or prior scarring, most commonly described in facial acne scars.^1,2,4 Type II is typified by blue-grey pigmentation on normal-appearing skin, most commonly on the shins, but also on sun-exposed sites.³ Biopsies of type I and II MIH demonstrate pigmented granules within macrophages or within the dermis.^4,5 Both Perls iron stain and Fontana-Masson melanin stain are positive in type I and II MIH.⁵ Type III MIH presents as diffuse brownish hyperpigmentation on normal skin in chronically sun-exposed sites.³ Histopathology of type III MIH can be distinguished by increased melanin noted inside basal keratinocytes as well as dermal melanophages that stain positive for only Fontana-Masson.⁵ The current case exemplifies a unique presentation of type I MIH along the length of the saphenous vein CABG harvest site. The concomitant use of amiodarone with minocycline may have contributed to the presentation.

The differential diagnosis for MIH depends on the type of MIH. Blue-grey pigmentation within scars is fairly unique to minocycline but has been reported with other medications, including vandetanib.⁶ The differential diagnosis for diffuse blue-grey or brown hyperpigmentation in predominately sun-exposed sites is broader, including endocrine disorders (ie, Addison disease), heavy metal poisoning (ie, argyria), inherited conditions (ie, alkaptonuria, Wilson disease, and hemochromatosis), medication-induced hyperpigmentation (ie, antipsychotics, anticonvulsant, antimalarials, amiodarone, and cytotoxic drugs), as well as inflammatory dermatoses, such as erythema dyschromicum perstans.⁷

MIH typically fades over months to years following minocycline discontinuation, so prompt recognition and discontinuation is recommended. Unfortunately, some cases persist or only partially fade over time. While MIH is benign, it can be of aesthetic concern, cause anxiety, and impact patients’ quality of life.^3,8 Persistent MIH is typically recalcitrant to topical hydroquinone.⁹ However, persistent MIH has been shown to improve with Q-switched, nanosecond lasers such as the 694 nm ruby, 755 nm alexandrite, and 1064 nm neodymium-doped yttrium aluminum garnet neodymium (Nd:YAG) lasers, as well as the 755 nm picosecond alexandrite laser.^4,9,10

In our patient, minocycline therapy was discontinued and replaced with doxycycline 100 mg twice daily monotherapy. At a subsequent visit 12 weeks later, the hyperpigmentation remained unchanged.

Conclusions

Though uncommon, we hope to encourage clinician awareness of MIH through our case, as prompt diagnosis and the discontinuation of minocycline are preferred to improve patient outcomes.

A 59-year-old man with a history of coronary artery bypass grafting (CABG), ischemic cardiomyopathy (ejection fraction, 15%-20%) with implantable cardioverter-defibrillator, recurrent paroxysmal ventricular tachycardia on amiodarone and mexiletine, and heart failure requiring left ventricular assist device (LVAD) placement presented for recurrent cellulitis and infection of the LVAD driveline exit site. He was initiated on minocycline 100 mg twice daily in combination with cefadroxil 500 mg twice daily. At his 8-week follow-up, the driveline site appeared improved with minimal erythema and no drainage. However, the patient developed a well-demarcated, linear, hyperpigmented patch along the length of the saphenous vein CABG harvest site and a few hyperpigmented macules medial to the harvest site (Figure).

Discussion

Hyperpigmentation presenting within scar tissue, as seen in this patient undergoing minocycline therapy, is a classic presentation of minocycline-induced hyperpigmentation (MIH) type I.

MIH is an uncommon, potentially cosmetically disfiguring adverse effect associated with systemic minocycline use. MIH can affect skin, teeth, nails, oral mucosa, sclera, and internal organs. The cumulative incidence of MIH in patients receiving minocycline over prolonged periods of time has been estimated from 2% to 15% in patients with acne and rosacea, to approximately 50% over 5 years in orthopedic patient populations.^1-3 The risk for developing MIH increases with vitamin D deficiency, liver disease, concurrent use with other medications that can induce hyperpigmentation, and higher cumulative doses (> 70-100 g; more important for MIH types II and III).^3,4 

There are 3 distinct types of MIH. Type I MIH is characterized by blue-black macules and patches at sites of inflammation or prior scarring, most commonly described in facial acne scars.^1,2,4 Type II is typified by blue-grey pigmentation on normal-appearing skin, most commonly on the shins, but also on sun-exposed sites.³ Biopsies of type I and II MIH demonstrate pigmented granules within macrophages or within the dermis.^4,5 Both Perls iron stain and Fontana-Masson melanin stain are positive in type I and II MIH.⁵ Type III MIH presents as diffuse brownish hyperpigmentation on normal skin in chronically sun-exposed sites.³ Histopathology of type III MIH can be distinguished by increased melanin noted inside basal keratinocytes as well as dermal melanophages that stain positive for only Fontana-Masson.⁵ The current case exemplifies a unique presentation of type I MIH along the length of the saphenous vein CABG harvest site. The concomitant use of amiodarone with minocycline may have contributed to the presentation.

The differential diagnosis for MIH depends on the type of MIH. Blue-grey pigmentation within scars is fairly unique to minocycline but has been reported with other medications, including vandetanib.⁶ The differential diagnosis for diffuse blue-grey or brown hyperpigmentation in predominately sun-exposed sites is broader, including endocrine disorders (ie, Addison disease), heavy metal poisoning (ie, argyria), inherited conditions (ie, alkaptonuria, Wilson disease, and hemochromatosis), medication-induced hyperpigmentation (ie, antipsychotics, anticonvulsant, antimalarials, amiodarone, and cytotoxic drugs), as well as inflammatory dermatoses, such as erythema dyschromicum perstans.⁷

MIH typically fades over months to years following minocycline discontinuation, so prompt recognition and discontinuation is recommended. Unfortunately, some cases persist or only partially fade over time. While MIH is benign, it can be of aesthetic concern, cause anxiety, and impact patients’ quality of life.^3,8 Persistent MIH is typically recalcitrant to topical hydroquinone.⁹ However, persistent MIH has been shown to improve with Q-switched, nanosecond lasers such as the 694 nm ruby, 755 nm alexandrite, and 1064 nm neodymium-doped yttrium aluminum garnet neodymium (Nd:YAG) lasers, as well as the 755 nm picosecond alexandrite laser.^4,9,10

In our patient, minocycline therapy was discontinued and replaced with doxycycline 100 mg twice daily monotherapy. At a subsequent visit 12 weeks later, the hyperpigmentation remained unchanged.

Conclusions

Though uncommon, we hope to encourage clinician awareness of MIH through our case, as prompt diagnosis and the discontinuation of minocycline are preferred to improve patient outcomes.

Hypertension is one of the most common cardiovascular disease (CVD) states, affecting nearly half of all adults in the United States.¹ Numerous classes of antihypertensives are available for blood pressure (BP) management, including thiazide diuretics, which contain both thiazide and thiazide-like agents. Thiazide diuretics available in the US include hydrochlorothiazide (HCTZ), chlorthalidone, metolazone, and indapamide. These agents are commonly used and recommended as first-line treatment in the current 2017 American College of Cardiology/American Heart Association (ACC/AHA) guideline for the prevention, detection, evaluation, and management of high BP in adults.²

The ACC/AHA guideline recommends chlorthalidone as the preferred thiazide diuretic.² This recommendation is based on its prolonged half-life compared with other thiazide agents, as well as the reduction of CVD seen with chlorthalidone in previous trials. The main evidence supporting chlorthalidone use comes from the ALLHAT trial, which compared chlorthalidone, amlodipine, and lisinopril in patients with hypertension. The primary composite outcome of fatal coronary artery disease or nonfatal myocardial infarction was not significantly different between groups. However, when looking at the incidence of heart failure, chlorthalidone was superior to both amlodipine and lisinopril.³ In the TOMHS trial, chlorthalidone was more effective in reducing left ventricular hypertrophy than amlodipine, enalapril, doxazosin, or acebutolol.⁴ Furthermore, both a systematic review and a retrospective cohort analysis suggested that chlorthalidone may be associated with improved CVD outcomes compared with HCTZ.^5,6 However, prospective randomized trial data is needed to confirm the superiority of chlorthalidone over other thiazide diuretics.

HCTZ has historically been the most common thiazide diuretic.⁷ However, with the available evidence and 2017 ACC/AHA BP guideline recommendations, it is unclear whether this trend continues and what impact it may have on CVD outcomes. It is unclear which thiazide diuretic is most commonly used in the US Department of Veterans Affairs (VA) health care system. The purpose of this project was to evaluate current thiazide diuretic utilization within the VA.

Methods

This retrospective, observational study evaluated the prescribing pattern of thiazide diuretics from all VA health care systems from January 1, 2016, to January 21, 2022. Thiazide diuretic agents included in this study were HCTZ, chlorthalidone, indapamide, and any combination antihypertensive products that included these 3 thiazide diuretics. Metolazone was excluded as it is commonly used in the setting of diuretic resistance with heart failure. Data was obtained from the VA Corporate Data Warehouse (CDW) and divided into 2 cohorts: the active and historic cohorts. The active cohort was of primary interest and included any active VA thiazide diuretic prescriptions on January 21, 2022. The historic cohort included thiazide prescriptions assessed at yearly intervals from January 1, 2016, to December 31, 2021. This date range was selected to assess what impact the 2017 ACC/AHA BP guideline had on clinician preferences and thiazide diuretic prescribing rates.

Within the active cohort, demographic data, vital information, and concomitant potassium or magnesium supplementation were collected. Baseline characteristics included were age, sex, race and ethnicity, and BP. Patients with > 1 race or ethnicity reported were categorized as other. The first BP reading documented after the active thiazide diuretic initiation date was included for analysis to capture on-therapy BPs while limiting confounding factors due to other potential antihypertensive changes. This project was ruled exempt from institutional review board review by the West Palm Beach VA Healthcare System Research and Development Committee.

The primary outcome was the evaluation of utilization rates of each thiazide in the active cohort, reported as a proportion of overall thiazide class utilization within the VA. Secondary outcomes in the active thiazide cohort included concomitant potassium or magnesium supplement utilization rates in each of the thiazide groups, BP values, and BP control rates. BP control was defined as a systolic BP < 130 mm Hg and a diastolic BP < 80 mm Hg. Finally, the change in thiazide diuretic utilization patterns from January 1, 2016, to December 31, 2021, was evaluated in the historic cohort.

Statistical Analysis

Data collection and analysis were completed using the CDW analyzed with Microsoft SQL Server Management Studio 18 and Microsoft Excel. All exported data to Microsoft Excel was kept in a secure network drive that was only accessible to the authors. Protected health information remained confidential per VA policy and the Health Insurance Portability and Accountability Act.

Baseline demographics were evaluated across thiazide arms using descriptive statistics. The primary outcome was assessed and a χ² test with a single comparison α level of 0.05 with Bonferroni correction to adjust for multiple comparisons when appropriate. For the secondary outcomes, analysis of continuous data was assessed using analysis of variance (ANOVA), and nominal data were assessed with a χ² test with a single comparison α level of 0.05 and Bonferroni correction to adjust for multiple comparisons where appropriate. When comparing all 3 thiazide groups, after the Bonferroni correction, P < .01667 was considered statistically significant to avoid a type 1 error in a family of statistical tests.

Results

As of January 21, 2022, the active thiazide cohort yielded 628,994 thiazide prescriptions within the VA nationwide. Most patients were male, with female patients representing 8.4%, 6.6%, and 5.6% of the HCTZ, chlorthalidone, and indapamide arms, respectively (Table 1). Utilization rates were significantly different between thiazide groups (P < .001). HCTZ was the most prescribed thiazide diuretic (84.6%) followed by chlorthalidone (14.9%) and indapamide (0.5%) (Table 2).

BP values documented after prescription initiation date were available for few individuals in the HCTZ, chlorthalidone, and indapamide groups (0.3%, 0.2%, and 0.5%, respectively). Overall, the mean BP values were similar among thiazide groups: 135/79 mm Hg for HCTZ, 137/78 mm Hg for chlorthalidone, and 133/79 mm Hg for indapamide (P = .32). BP control was also similar with control rates of 26.0%, 27.1%, and 33.3% for those on HCTZ, chlorthalidone, and indapamide, respectively (P = .75). The use of concomitant potassium or magnesium supplementation was significantly different between thiazide groups with rates of 12.4%, 22.6%, and 27.1% for HCTZ, chlorthalidone, and indapamide, respectively (P < .001). When comparing chlorthalidone to HCTZ, there was a significantly higher rate of concomitant supplementation with chlorthalidone (P < .001) (Table 3).

In the historic cohort, HCTZ utilization decreased from 90.2% to 83.5% (P < .001) and chlorthalidone utilization increased significantly from 9.3% to 16.0% (P < .001) (Figure). There was no significant change in the use of indapamide during this period (P = .73). Yearly trends from 2016 to 2021 are listed in Table 4.

Discussion

The findings of our evaluation demonstrate that despite the 2017 ACC/AHA BP guideline recommendations for using chlorthalidone, HCTZ predominates as the most prescribed thiazide diuretic within the VA. However, since the publication of this guideline, there has been an increase in chlorthalidone prescribing and a decrease in HCTZ prescribing within the VA.

A 2010 study by Ernst and colleagues revealed a similar trend to what was seen in our study. At that time, HCTZ was the most prescribed thiazide encompassing 95% of total thiazide utilization; however, chlorthalidone utilization increased from 1.1% in 2003 to 2.4% in 2008.⁸ In comparing our chlorthalidone utilization rates with these results, 9.3% in 2016 and 16.0% in 2021, the change in chlorthalidone prescribing from 2003 to 2016 represents a more than linear increase. This trend continued in our study from 2016 to 2021; the expected chlorthalidone utilization would be 21.2% in 2021 if it followed the 2003 to 2016 rate of change. Thus the trend in increasing chlorthalidone use predated the 2017 guideline recommendation. Nonetheless, this change in the thiazide prescribing pattern represents a positive shift in practice.

Our evaluation found a significantly higher rate of concomitant potassium or magnesium supplementation with chlorthalidone and indapamide compared with HCTZ in the active cohort. Electrolyte abnormalities are well documented adverse effects associated with thiazide diuretic use.⁹ A cross-sectional analysis by Ravioli and colleagues revealed thiazide diuretic use was an independent predictor of both hyponatremia (22.1% incidence) and hypokalemia (19% incidence) and that chlorthalidone was associated with the highest risk of electrolyte abnormalities whereas HCTZ was associated with the lowest risk. Their study also found these electrolyte abnormalities to have a dose-dependent relationship with the thiazide diuretic prescribed.¹⁰

While Ravioli and colleagues did not address the incidence of hypomagnesemia with thiazide diuretic use, a cross-sectional analysis by Kieboom and colleagues reported a significant increase in hypomagnesemia in patients prescribed thiazide diuretics.¹¹ Although rates of electrolyte abnormalities are reported in the literature, the rates of concomitant supplementation are unclear, especially when compared across thiazide agents. Our study provides insight into the use of concomitant potassium and magnesium supplementation compared between HCTZ, chlorthalidone, and indapamide. In our active cohort, potassium was more commonly prescribed than magnesium. Interestingly, magnesium supplementation accounted for 25.9% of the total supplement use for HCTZ compared with rates of 22.4% and 21.0% for chlorthalidone and indapamide, respectively. It is unclear if this trend highlights a greater incidence of hypomagnesemia with HCTZ or greater clinician awareness to monitor this agent, but this finding may warrant further investigation. In addition, when considering the overall lower rate of supplementation seen with HCTZ in our study, the use of potassium-sparing diuretics should be considered. These agents, including triamterene, amiloride, eplerenone, and spironolactone, can be supplement-sparing and are available in combination products only with HCTZ.

Low chlorthalidone utilization rates are concerning especially given the literature demonstrating CVD benefit with chlorthalidone and the lack of compelling outcomes data to support HCTZ as the preferred agent.^3,4 There are several reasons why HCTZ use may be higher in practice. First is clinical inertia, which is defined as a lack of treatment intensification or lack of changing practice patterns, despite evidence-based goals of care.¹² HCTZ has been the most widely prescribed thiazide diuretic for years.⁷ As a result, converting HCTZ to chlorthalidone for a patient with suboptimal BP control may not be considered and instead clinicians may add on another antihypertensive or titrate doses of current antihypertensives.

There is also a consideration for patient adherence. HCTZ has many more combination products available than chlorthalidone and indapamide. If switching a patient from an HCTZ-containing combination product to chlorthalidone, adherence and patient willingness to take another capsule or tablet must be considered. Finally, there may be clinical controversy and questions around switching patients from HCTZ to chlorthalidone. Although the guidelines do not explicitly recommend switching to chlorthalidone, it may be reasonable in most patients unless they have or are at significant risk of electrolyte or metabolic disturbances that may be exacerbated or triggered with conversion.

When converting from HCTZ to chlorthalidone, it is important to consider dosing. Previous studies have demonstrated that chlorthalidone is 1.5 to 2 times more potent than HCTZ.^13,14 Therefore, the conversion from HCTZ to chlorthalidone is not 1:1, but instead 50 mg of HCTZ is approximately equal to25 to 37.5 mg of chlorthalidone.¹⁴

Limitations

This study was limited by its retrospective design, gaps in data, duplicate active prescription data, and the assessment of concomitant electrolyte supplementation. As with any retrospective study, there is a potential for confounding and a concern for information bias with missing information. This study relied on proper documentation of prescription and demographic information in the Veterans Health Information Systems and Technology Architecture (VistA), as the CDW compiles information from this electronic health record. Strengths of the VistA include ease in clinical functions, documentation, and the ability for records to be updated from any VA facility nationally. However, there is always the possibility of user error and information to be omitted.

In our study, the documentation of BP values and subsequent analysis of overall BP control were limited. For BP values to be included in this study, they had to be recorded after the active thiazide prescription was written and from an in-person encounter documented in VistA. The COVID-19 pandemic shifted the clinical landscape and many primary care appointments during the active cohort evaluation period were conducted virtually. Therefore, patients may not have had formal vitals recorded. There may also be an aspect of selection bias regarding the chlorthalidone group. Although rates of thiazide switching were not assessed, some patients may have been switched from HCTZ or indapamide to chlorthalidone to achieve additional BP control. Thus, patients receiving chlorthalidone may represent a more difficult-to-control hypertensive population, making a finding of similar BP control rates between HCTZ and chlorthalidone an actual positive finding regarding chlorthalidone. Finally, this study did not assess adherence to medications. As the intent of the study was to analyze prescribing patterns, it is impossible to know if the patient was actively taking the medication at the time of assessment. When considering the rates of BP control, there were limited BP values, a potential for selection bias, and neither adherence nor patient self-reported home BP values were assessed. Therefore, the interpretation of overall BP control must be done with caution.

Additionally, duplicate prescriptions were noted in the active cohort. Rates of duplication were 0.2%, 0.08%, and 0.09% for HCTZ, chlorthalidone, and indapamide, respectively. With these small percentages, we felt this would not have a significant impact on the overall thiazide use trends seen in our study. Patients can receive prescriptions from multiple VA facilities and may have > 1 active prescriptions. This has been mitigated in recent years with the introduction of the OneVA program, allowing pharmacists to access any prescription on file from any VA facility and refill if needed (except controlled substance prescriptions). However, there are certain instances in which duplicate prescriptions may be necessary. These include patients enrolled and receiving care at another VA facility (eg, traveling for part of a year) and patients hospitalized at a different facility and given medications on discharge.

With the overall low rate of duplication prescriptions seen in each thiazide group, we determined that this was not large enough to cause substantial variation in the results of this evaluation and was unlikely to alter the results. This study also does not inform on the incidence of switching between thiazide diuretics. If a patient was switched from HCTZ to chlorthalidone in 2017, for example, a prescription for HCTZ and chlorthalidone would have been reported in this study. We felt that the change in chlorthalidone prescribing from January 1, 2016, to December 31, 2021, would reflect overall utilization rates, which may include switching from HCTZ or indapamide to chlorthalidone in addition to new chlorthalidone prescriptions.

Finally, there are confounders and trends in concomitant potassium or magnesium supplementation that were not accounted for in our study. These include concomitant loop diuretics or other medications that may cause electrolyte abnormalities and the dose-dependent relationship between thiazide diuretics and electrolyte abnormalities.¹⁰ Actual laboratory values were not included in this analysis and thus we cannot assess whether supplementation or management of electrolyte disturbances was clinically appropriate.

Conclusions

Thiazide utilization patterns have shifted possibly due to the 2017 ACC/AHA BP guideline recommendations. However, HCTZ continues to be the most widely prescribed thiazide diuretic within the VA. There is a need for future projects and clinician education to increase the implementation of guideline-recommended therapy within the VA, particularly regarding chlorthalidone use.

Hypertension is one of the most common cardiovascular disease (CVD) states, affecting nearly half of all adults in the United States.¹ Numerous classes of antihypertensives are available for blood pressure (BP) management, including thiazide diuretics, which contain both thiazide and thiazide-like agents. Thiazide diuretics available in the US include hydrochlorothiazide (HCTZ), chlorthalidone, metolazone, and indapamide. These agents are commonly used and recommended as first-line treatment in the current 2017 American College of Cardiology/American Heart Association (ACC/AHA) guideline for the prevention, detection, evaluation, and management of high BP in adults.²

The ACC/AHA guideline recommends chlorthalidone as the preferred thiazide diuretic.² This recommendation is based on its prolonged half-life compared with other thiazide agents, as well as the reduction of CVD seen with chlorthalidone in previous trials. The main evidence supporting chlorthalidone use comes from the ALLHAT trial, which compared chlorthalidone, amlodipine, and lisinopril in patients with hypertension. The primary composite outcome of fatal coronary artery disease or nonfatal myocardial infarction was not significantly different between groups. However, when looking at the incidence of heart failure, chlorthalidone was superior to both amlodipine and lisinopril.³ In the TOMHS trial, chlorthalidone was more effective in reducing left ventricular hypertrophy than amlodipine, enalapril, doxazosin, or acebutolol.⁴ Furthermore, both a systematic review and a retrospective cohort analysis suggested that chlorthalidone may be associated with improved CVD outcomes compared with HCTZ.^5,6 However, prospective randomized trial data is needed to confirm the superiority of chlorthalidone over other thiazide diuretics.

HCTZ has historically been the most common thiazide diuretic.⁷ However, with the available evidence and 2017 ACC/AHA BP guideline recommendations, it is unclear whether this trend continues and what impact it may have on CVD outcomes. It is unclear which thiazide diuretic is most commonly used in the US Department of Veterans Affairs (VA) health care system. The purpose of this project was to evaluate current thiazide diuretic utilization within the VA.

Methods

This retrospective, observational study evaluated the prescribing pattern of thiazide diuretics from all VA health care systems from January 1, 2016, to January 21, 2022. Thiazide diuretic agents included in this study were HCTZ, chlorthalidone, indapamide, and any combination antihypertensive products that included these 3 thiazide diuretics. Metolazone was excluded as it is commonly used in the setting of diuretic resistance with heart failure. Data was obtained from the VA Corporate Data Warehouse (CDW) and divided into 2 cohorts: the active and historic cohorts. The active cohort was of primary interest and included any active VA thiazide diuretic prescriptions on January 21, 2022. The historic cohort included thiazide prescriptions assessed at yearly intervals from January 1, 2016, to December 31, 2021. This date range was selected to assess what impact the 2017 ACC/AHA BP guideline had on clinician preferences and thiazide diuretic prescribing rates.

Within the active cohort, demographic data, vital information, and concomitant potassium or magnesium supplementation were collected. Baseline characteristics included were age, sex, race and ethnicity, and BP. Patients with > 1 race or ethnicity reported were categorized as other. The first BP reading documented after the active thiazide diuretic initiation date was included for analysis to capture on-therapy BPs while limiting confounding factors due to other potential antihypertensive changes. This project was ruled exempt from institutional review board review by the West Palm Beach VA Healthcare System Research and Development Committee.

The primary outcome was the evaluation of utilization rates of each thiazide in the active cohort, reported as a proportion of overall thiazide class utilization within the VA. Secondary outcomes in the active thiazide cohort included concomitant potassium or magnesium supplement utilization rates in each of the thiazide groups, BP values, and BP control rates. BP control was defined as a systolic BP < 130 mm Hg and a diastolic BP < 80 mm Hg. Finally, the change in thiazide diuretic utilization patterns from January 1, 2016, to December 31, 2021, was evaluated in the historic cohort.

Statistical Analysis

Data collection and analysis were completed using the CDW analyzed with Microsoft SQL Server Management Studio 18 and Microsoft Excel. All exported data to Microsoft Excel was kept in a secure network drive that was only accessible to the authors. Protected health information remained confidential per VA policy and the Health Insurance Portability and Accountability Act.

Baseline demographics were evaluated across thiazide arms using descriptive statistics. The primary outcome was assessed and a χ² test with a single comparison α level of 0.05 with Bonferroni correction to adjust for multiple comparisons when appropriate. For the secondary outcomes, analysis of continuous data was assessed using analysis of variance (ANOVA), and nominal data were assessed with a χ² test with a single comparison α level of 0.05 and Bonferroni correction to adjust for multiple comparisons where appropriate. When comparing all 3 thiazide groups, after the Bonferroni correction, P < .01667 was considered statistically significant to avoid a type 1 error in a family of statistical tests.

Results

As of January 21, 2022, the active thiazide cohort yielded 628,994 thiazide prescriptions within the VA nationwide. Most patients were male, with female patients representing 8.4%, 6.6%, and 5.6% of the HCTZ, chlorthalidone, and indapamide arms, respectively (Table 1). Utilization rates were significantly different between thiazide groups (P < .001). HCTZ was the most prescribed thiazide diuretic (84.6%) followed by chlorthalidone (14.9%) and indapamide (0.5%) (Table 2).

BP values documented after prescription initiation date were available for few individuals in the HCTZ, chlorthalidone, and indapamide groups (0.3%, 0.2%, and 0.5%, respectively). Overall, the mean BP values were similar among thiazide groups: 135/79 mm Hg for HCTZ, 137/78 mm Hg for chlorthalidone, and 133/79 mm Hg for indapamide (P = .32). BP control was also similar with control rates of 26.0%, 27.1%, and 33.3% for those on HCTZ, chlorthalidone, and indapamide, respectively (P = .75). The use of concomitant potassium or magnesium supplementation was significantly different between thiazide groups with rates of 12.4%, 22.6%, and 27.1% for HCTZ, chlorthalidone, and indapamide, respectively (P < .001). When comparing chlorthalidone to HCTZ, there was a significantly higher rate of concomitant supplementation with chlorthalidone (P < .001) (Table 3).

In the historic cohort, HCTZ utilization decreased from 90.2% to 83.5% (P < .001) and chlorthalidone utilization increased significantly from 9.3% to 16.0% (P < .001) (Figure). There was no significant change in the use of indapamide during this period (P = .73). Yearly trends from 2016 to 2021 are listed in Table 4.

Discussion

The findings of our evaluation demonstrate that despite the 2017 ACC/AHA BP guideline recommendations for using chlorthalidone, HCTZ predominates as the most prescribed thiazide diuretic within the VA. However, since the publication of this guideline, there has been an increase in chlorthalidone prescribing and a decrease in HCTZ prescribing within the VA.

A 2010 study by Ernst and colleagues revealed a similar trend to what was seen in our study. At that time, HCTZ was the most prescribed thiazide encompassing 95% of total thiazide utilization; however, chlorthalidone utilization increased from 1.1% in 2003 to 2.4% in 2008.⁸ In comparing our chlorthalidone utilization rates with these results, 9.3% in 2016 and 16.0% in 2021, the change in chlorthalidone prescribing from 2003 to 2016 represents a more than linear increase. This trend continued in our study from 2016 to 2021; the expected chlorthalidone utilization would be 21.2% in 2021 if it followed the 2003 to 2016 rate of change. Thus the trend in increasing chlorthalidone use predated the 2017 guideline recommendation. Nonetheless, this change in the thiazide prescribing pattern represents a positive shift in practice.

Our evaluation found a significantly higher rate of concomitant potassium or magnesium supplementation with chlorthalidone and indapamide compared with HCTZ in the active cohort. Electrolyte abnormalities are well documented adverse effects associated with thiazide diuretic use.⁹ A cross-sectional analysis by Ravioli and colleagues revealed thiazide diuretic use was an independent predictor of both hyponatremia (22.1% incidence) and hypokalemia (19% incidence) and that chlorthalidone was associated with the highest risk of electrolyte abnormalities whereas HCTZ was associated with the lowest risk. Their study also found these electrolyte abnormalities to have a dose-dependent relationship with the thiazide diuretic prescribed.¹⁰

While Ravioli and colleagues did not address the incidence of hypomagnesemia with thiazide diuretic use, a cross-sectional analysis by Kieboom and colleagues reported a significant increase in hypomagnesemia in patients prescribed thiazide diuretics.¹¹ Although rates of electrolyte abnormalities are reported in the literature, the rates of concomitant supplementation are unclear, especially when compared across thiazide agents. Our study provides insight into the use of concomitant potassium and magnesium supplementation compared between HCTZ, chlorthalidone, and indapamide. In our active cohort, potassium was more commonly prescribed than magnesium. Interestingly, magnesium supplementation accounted for 25.9% of the total supplement use for HCTZ compared with rates of 22.4% and 21.0% for chlorthalidone and indapamide, respectively. It is unclear if this trend highlights a greater incidence of hypomagnesemia with HCTZ or greater clinician awareness to monitor this agent, but this finding may warrant further investigation. In addition, when considering the overall lower rate of supplementation seen with HCTZ in our study, the use of potassium-sparing diuretics should be considered. These agents, including triamterene, amiloride, eplerenone, and spironolactone, can be supplement-sparing and are available in combination products only with HCTZ.

Low chlorthalidone utilization rates are concerning especially given the literature demonstrating CVD benefit with chlorthalidone and the lack of compelling outcomes data to support HCTZ as the preferred agent.^3,4 There are several reasons why HCTZ use may be higher in practice. First is clinical inertia, which is defined as a lack of treatment intensification or lack of changing practice patterns, despite evidence-based goals of care.¹² HCTZ has been the most widely prescribed thiazide diuretic for years.⁷ As a result, converting HCTZ to chlorthalidone for a patient with suboptimal BP control may not be considered and instead clinicians may add on another antihypertensive or titrate doses of current antihypertensives.

There is also a consideration for patient adherence. HCTZ has many more combination products available than chlorthalidone and indapamide. If switching a patient from an HCTZ-containing combination product to chlorthalidone, adherence and patient willingness to take another capsule or tablet must be considered. Finally, there may be clinical controversy and questions around switching patients from HCTZ to chlorthalidone. Although the guidelines do not explicitly recommend switching to chlorthalidone, it may be reasonable in most patients unless they have or are at significant risk of electrolyte or metabolic disturbances that may be exacerbated or triggered with conversion.

When converting from HCTZ to chlorthalidone, it is important to consider dosing. Previous studies have demonstrated that chlorthalidone is 1.5 to 2 times more potent than HCTZ.^13,14 Therefore, the conversion from HCTZ to chlorthalidone is not 1:1, but instead 50 mg of HCTZ is approximately equal to25 to 37.5 mg of chlorthalidone.¹⁴

Limitations

This study was limited by its retrospective design, gaps in data, duplicate active prescription data, and the assessment of concomitant electrolyte supplementation. As with any retrospective study, there is a potential for confounding and a concern for information bias with missing information. This study relied on proper documentation of prescription and demographic information in the Veterans Health Information Systems and Technology Architecture (VistA), as the CDW compiles information from this electronic health record. Strengths of the VistA include ease in clinical functions, documentation, and the ability for records to be updated from any VA facility nationally. However, there is always the possibility of user error and information to be omitted.

In our study, the documentation of BP values and subsequent analysis of overall BP control were limited. For BP values to be included in this study, they had to be recorded after the active thiazide prescription was written and from an in-person encounter documented in VistA. The COVID-19 pandemic shifted the clinical landscape and many primary care appointments during the active cohort evaluation period were conducted virtually. Therefore, patients may not have had formal vitals recorded. There may also be an aspect of selection bias regarding the chlorthalidone group. Although rates of thiazide switching were not assessed, some patients may have been switched from HCTZ or indapamide to chlorthalidone to achieve additional BP control. Thus, patients receiving chlorthalidone may represent a more difficult-to-control hypertensive population, making a finding of similar BP control rates between HCTZ and chlorthalidone an actual positive finding regarding chlorthalidone. Finally, this study did not assess adherence to medications. As the intent of the study was to analyze prescribing patterns, it is impossible to know if the patient was actively taking the medication at the time of assessment. When considering the rates of BP control, there were limited BP values, a potential for selection bias, and neither adherence nor patient self-reported home BP values were assessed. Therefore, the interpretation of overall BP control must be done with caution.

Additionally, duplicate prescriptions were noted in the active cohort. Rates of duplication were 0.2%, 0.08%, and 0.09% for HCTZ, chlorthalidone, and indapamide, respectively. With these small percentages, we felt this would not have a significant impact on the overall thiazide use trends seen in our study. Patients can receive prescriptions from multiple VA facilities and may have > 1 active prescriptions. This has been mitigated in recent years with the introduction of the OneVA program, allowing pharmacists to access any prescription on file from any VA facility and refill if needed (except controlled substance prescriptions). However, there are certain instances in which duplicate prescriptions may be necessary. These include patients enrolled and receiving care at another VA facility (eg, traveling for part of a year) and patients hospitalized at a different facility and given medications on discharge.

With the overall low rate of duplication prescriptions seen in each thiazide group, we determined that this was not large enough to cause substantial variation in the results of this evaluation and was unlikely to alter the results. This study also does not inform on the incidence of switching between thiazide diuretics. If a patient was switched from HCTZ to chlorthalidone in 2017, for example, a prescription for HCTZ and chlorthalidone would have been reported in this study. We felt that the change in chlorthalidone prescribing from January 1, 2016, to December 31, 2021, would reflect overall utilization rates, which may include switching from HCTZ or indapamide to chlorthalidone in addition to new chlorthalidone prescriptions.

Finally, there are confounders and trends in concomitant potassium or magnesium supplementation that were not accounted for in our study. These include concomitant loop diuretics or other medications that may cause electrolyte abnormalities and the dose-dependent relationship between thiazide diuretics and electrolyte abnormalities.¹⁰ Actual laboratory values were not included in this analysis and thus we cannot assess whether supplementation or management of electrolyte disturbances was clinically appropriate.

Conclusions

Thiazide utilization patterns have shifted possibly due to the 2017 ACC/AHA BP guideline recommendations. However, HCTZ continues to be the most widely prescribed thiazide diuretic within the VA. There is a need for future projects and clinician education to increase the implementation of guideline-recommended therapy within the VA, particularly regarding chlorthalidone use.

Hypertension is one of the most common cardiovascular disease (CVD) states, affecting nearly half of all adults in the United States.¹ Numerous classes of antihypertensives are available for blood pressure (BP) management, including thiazide diuretics, which contain both thiazide and thiazide-like agents. Thiazide diuretics available in the US include hydrochlorothiazide (HCTZ), chlorthalidone, metolazone, and indapamide. These agents are commonly used and recommended as first-line treatment in the current 2017 American College of Cardiology/American Heart Association (ACC/AHA) guideline for the prevention, detection, evaluation, and management of high BP in adults.²

The ACC/AHA guideline recommends chlorthalidone as the preferred thiazide diuretic.² This recommendation is based on its prolonged half-life compared with other thiazide agents, as well as the reduction of CVD seen with chlorthalidone in previous trials. The main evidence supporting chlorthalidone use comes from the ALLHAT trial, which compared chlorthalidone, amlodipine, and lisinopril in patients with hypertension. The primary composite outcome of fatal coronary artery disease or nonfatal myocardial infarction was not significantly different between groups. However, when looking at the incidence of heart failure, chlorthalidone was superior to both amlodipine and lisinopril.³ In the TOMHS trial, chlorthalidone was more effective in reducing left ventricular hypertrophy than amlodipine, enalapril, doxazosin, or acebutolol.⁴ Furthermore, both a systematic review and a retrospective cohort analysis suggested that chlorthalidone may be associated with improved CVD outcomes compared with HCTZ.^5,6 However, prospective randomized trial data is needed to confirm the superiority of chlorthalidone over other thiazide diuretics.

HCTZ has historically been the most common thiazide diuretic.⁷ However, with the available evidence and 2017 ACC/AHA BP guideline recommendations, it is unclear whether this trend continues and what impact it may have on CVD outcomes. It is unclear which thiazide diuretic is most commonly used in the US Department of Veterans Affairs (VA) health care system. The purpose of this project was to evaluate current thiazide diuretic utilization within the VA.

Methods

This retrospective, observational study evaluated the prescribing pattern of thiazide diuretics from all VA health care systems from January 1, 2016, to January 21, 2022. Thiazide diuretic agents included in this study were HCTZ, chlorthalidone, indapamide, and any combination antihypertensive products that included these 3 thiazide diuretics. Metolazone was excluded as it is commonly used in the setting of diuretic resistance with heart failure. Data was obtained from the VA Corporate Data Warehouse (CDW) and divided into 2 cohorts: the active and historic cohorts. The active cohort was of primary interest and included any active VA thiazide diuretic prescriptions on January 21, 2022. The historic cohort included thiazide prescriptions assessed at yearly intervals from January 1, 2016, to December 31, 2021. This date range was selected to assess what impact the 2017 ACC/AHA BP guideline had on clinician preferences and thiazide diuretic prescribing rates.

Within the active cohort, demographic data, vital information, and concomitant potassium or magnesium supplementation were collected. Baseline characteristics included were age, sex, race and ethnicity, and BP. Patients with > 1 race or ethnicity reported were categorized as other. The first BP reading documented after the active thiazide diuretic initiation date was included for analysis to capture on-therapy BPs while limiting confounding factors due to other potential antihypertensive changes. This project was ruled exempt from institutional review board review by the West Palm Beach VA Healthcare System Research and Development Committee.

The primary outcome was the evaluation of utilization rates of each thiazide in the active cohort, reported as a proportion of overall thiazide class utilization within the VA. Secondary outcomes in the active thiazide cohort included concomitant potassium or magnesium supplement utilization rates in each of the thiazide groups, BP values, and BP control rates. BP control was defined as a systolic BP < 130 mm Hg and a diastolic BP < 80 mm Hg. Finally, the change in thiazide diuretic utilization patterns from January 1, 2016, to December 31, 2021, was evaluated in the historic cohort.

Statistical Analysis

Data collection and analysis were completed using the CDW analyzed with Microsoft SQL Server Management Studio 18 and Microsoft Excel. All exported data to Microsoft Excel was kept in a secure network drive that was only accessible to the authors. Protected health information remained confidential per VA policy and the Health Insurance Portability and Accountability Act.

Baseline demographics were evaluated across thiazide arms using descriptive statistics. The primary outcome was assessed and a χ² test with a single comparison α level of 0.05 with Bonferroni correction to adjust for multiple comparisons when appropriate. For the secondary outcomes, analysis of continuous data was assessed using analysis of variance (ANOVA), and nominal data were assessed with a χ² test with a single comparison α level of 0.05 and Bonferroni correction to adjust for multiple comparisons where appropriate. When comparing all 3 thiazide groups, after the Bonferroni correction, P < .01667 was considered statistically significant to avoid a type 1 error in a family of statistical tests.

Results

As of January 21, 2022, the active thiazide cohort yielded 628,994 thiazide prescriptions within the VA nationwide. Most patients were male, with female patients representing 8.4%, 6.6%, and 5.6% of the HCTZ, chlorthalidone, and indapamide arms, respectively (Table 1). Utilization rates were significantly different between thiazide groups (P < .001). HCTZ was the most prescribed thiazide diuretic (84.6%) followed by chlorthalidone (14.9%) and indapamide (0.5%) (Table 2).

BP values documented after prescription initiation date were available for few individuals in the HCTZ, chlorthalidone, and indapamide groups (0.3%, 0.2%, and 0.5%, respectively). Overall, the mean BP values were similar among thiazide groups: 135/79 mm Hg for HCTZ, 137/78 mm Hg for chlorthalidone, and 133/79 mm Hg for indapamide (P = .32). BP control was also similar with control rates of 26.0%, 27.1%, and 33.3% for those on HCTZ, chlorthalidone, and indapamide, respectively (P = .75). The use of concomitant potassium or magnesium supplementation was significantly different between thiazide groups with rates of 12.4%, 22.6%, and 27.1% for HCTZ, chlorthalidone, and indapamide, respectively (P < .001). When comparing chlorthalidone to HCTZ, there was a significantly higher rate of concomitant supplementation with chlorthalidone (P < .001) (Table 3).

In the historic cohort, HCTZ utilization decreased from 90.2% to 83.5% (P < .001) and chlorthalidone utilization increased significantly from 9.3% to 16.0% (P < .001) (Figure). There was no significant change in the use of indapamide during this period (P = .73). Yearly trends from 2016 to 2021 are listed in Table 4.

Discussion

The findings of our evaluation demonstrate that despite the 2017 ACC/AHA BP guideline recommendations for using chlorthalidone, HCTZ predominates as the most prescribed thiazide diuretic within the VA. However, since the publication of this guideline, there has been an increase in chlorthalidone prescribing and a decrease in HCTZ prescribing within the VA.

A 2010 study by Ernst and colleagues revealed a similar trend to what was seen in our study. At that time, HCTZ was the most prescribed thiazide encompassing 95% of total thiazide utilization; however, chlorthalidone utilization increased from 1.1% in 2003 to 2.4% in 2008.⁸ In comparing our chlorthalidone utilization rates with these results, 9.3% in 2016 and 16.0% in 2021, the change in chlorthalidone prescribing from 2003 to 2016 represents a more than linear increase. This trend continued in our study from 2016 to 2021; the expected chlorthalidone utilization would be 21.2% in 2021 if it followed the 2003 to 2016 rate of change. Thus the trend in increasing chlorthalidone use predated the 2017 guideline recommendation. Nonetheless, this change in the thiazide prescribing pattern represents a positive shift in practice.

Our evaluation found a significantly higher rate of concomitant potassium or magnesium supplementation with chlorthalidone and indapamide compared with HCTZ in the active cohort. Electrolyte abnormalities are well documented adverse effects associated with thiazide diuretic use.⁹ A cross-sectional analysis by Ravioli and colleagues revealed thiazide diuretic use was an independent predictor of both hyponatremia (22.1% incidence) and hypokalemia (19% incidence) and that chlorthalidone was associated with the highest risk of electrolyte abnormalities whereas HCTZ was associated with the lowest risk. Their study also found these electrolyte abnormalities to have a dose-dependent relationship with the thiazide diuretic prescribed.¹⁰

While Ravioli and colleagues did not address the incidence of hypomagnesemia with thiazide diuretic use, a cross-sectional analysis by Kieboom and colleagues reported a significant increase in hypomagnesemia in patients prescribed thiazide diuretics.¹¹ Although rates of electrolyte abnormalities are reported in the literature, the rates of concomitant supplementation are unclear, especially when compared across thiazide agents. Our study provides insight into the use of concomitant potassium and magnesium supplementation compared between HCTZ, chlorthalidone, and indapamide. In our active cohort, potassium was more commonly prescribed than magnesium. Interestingly, magnesium supplementation accounted for 25.9% of the total supplement use for HCTZ compared with rates of 22.4% and 21.0% for chlorthalidone and indapamide, respectively. It is unclear if this trend highlights a greater incidence of hypomagnesemia with HCTZ or greater clinician awareness to monitor this agent, but this finding may warrant further investigation. In addition, when considering the overall lower rate of supplementation seen with HCTZ in our study, the use of potassium-sparing diuretics should be considered. These agents, including triamterene, amiloride, eplerenone, and spironolactone, can be supplement-sparing and are available in combination products only with HCTZ.

Low chlorthalidone utilization rates are concerning especially given the literature demonstrating CVD benefit with chlorthalidone and the lack of compelling outcomes data to support HCTZ as the preferred agent.^3,4 There are several reasons why HCTZ use may be higher in practice. First is clinical inertia, which is defined as a lack of treatment intensification or lack of changing practice patterns, despite evidence-based goals of care.¹² HCTZ has been the most widely prescribed thiazide diuretic for years.⁷ As a result, converting HCTZ to chlorthalidone for a patient with suboptimal BP control may not be considered and instead clinicians may add on another antihypertensive or titrate doses of current antihypertensives.

There is also a consideration for patient adherence. HCTZ has many more combination products available than chlorthalidone and indapamide. If switching a patient from an HCTZ-containing combination product to chlorthalidone, adherence and patient willingness to take another capsule or tablet must be considered. Finally, there may be clinical controversy and questions around switching patients from HCTZ to chlorthalidone. Although the guidelines do not explicitly recommend switching to chlorthalidone, it may be reasonable in most patients unless they have or are at significant risk of electrolyte or metabolic disturbances that may be exacerbated or triggered with conversion.

When converting from HCTZ to chlorthalidone, it is important to consider dosing. Previous studies have demonstrated that chlorthalidone is 1.5 to 2 times more potent than HCTZ.^13,14 Therefore, the conversion from HCTZ to chlorthalidone is not 1:1, but instead 50 mg of HCTZ is approximately equal to25 to 37.5 mg of chlorthalidone.¹⁴

Limitations

This study was limited by its retrospective design, gaps in data, duplicate active prescription data, and the assessment of concomitant electrolyte supplementation. As with any retrospective study, there is a potential for confounding and a concern for information bias with missing information. This study relied on proper documentation of prescription and demographic information in the Veterans Health Information Systems and Technology Architecture (VistA), as the CDW compiles information from this electronic health record. Strengths of the VistA include ease in clinical functions, documentation, and the ability for records to be updated from any VA facility nationally. However, there is always the possibility of user error and information to be omitted.

In our study, the documentation of BP values and subsequent analysis of overall BP control were limited. For BP values to be included in this study, they had to be recorded after the active thiazide prescription was written and from an in-person encounter documented in VistA. The COVID-19 pandemic shifted the clinical landscape and many primary care appointments during the active cohort evaluation period were conducted virtually. Therefore, patients may not have had formal vitals recorded. There may also be an aspect of selection bias regarding the chlorthalidone group. Although rates of thiazide switching were not assessed, some patients may have been switched from HCTZ or indapamide to chlorthalidone to achieve additional BP control. Thus, patients receiving chlorthalidone may represent a more difficult-to-control hypertensive population, making a finding of similar BP control rates between HCTZ and chlorthalidone an actual positive finding regarding chlorthalidone. Finally, this study did not assess adherence to medications. As the intent of the study was to analyze prescribing patterns, it is impossible to know if the patient was actively taking the medication at the time of assessment. When considering the rates of BP control, there were limited BP values, a potential for selection bias, and neither adherence nor patient self-reported home BP values were assessed. Therefore, the interpretation of overall BP control must be done with caution.

Additionally, duplicate prescriptions were noted in the active cohort. Rates of duplication were 0.2%, 0.08%, and 0.09% for HCTZ, chlorthalidone, and indapamide, respectively. With these small percentages, we felt this would not have a significant impact on the overall thiazide use trends seen in our study. Patients can receive prescriptions from multiple VA facilities and may have > 1 active prescriptions. This has been mitigated in recent years with the introduction of the OneVA program, allowing pharmacists to access any prescription on file from any VA facility and refill if needed (except controlled substance prescriptions). However, there are certain instances in which duplicate prescriptions may be necessary. These include patients enrolled and receiving care at another VA facility (eg, traveling for part of a year) and patients hospitalized at a different facility and given medications on discharge.

With the overall low rate of duplication prescriptions seen in each thiazide group, we determined that this was not large enough to cause substantial variation in the results of this evaluation and was unlikely to alter the results. This study also does not inform on the incidence of switching between thiazide diuretics. If a patient was switched from HCTZ to chlorthalidone in 2017, for example, a prescription for HCTZ and chlorthalidone would have been reported in this study. We felt that the change in chlorthalidone prescribing from January 1, 2016, to December 31, 2021, would reflect overall utilization rates, which may include switching from HCTZ or indapamide to chlorthalidone in addition to new chlorthalidone prescriptions.

Finally, there are confounders and trends in concomitant potassium or magnesium supplementation that were not accounted for in our study. These include concomitant loop diuretics or other medications that may cause electrolyte abnormalities and the dose-dependent relationship between thiazide diuretics and electrolyte abnormalities.¹⁰ Actual laboratory values were not included in this analysis and thus we cannot assess whether supplementation or management of electrolyte disturbances was clinically appropriate.

Conclusions

Thiazide utilization patterns have shifted possibly due to the 2017 ACC/AHA BP guideline recommendations. However, HCTZ continues to be the most widely prescribed thiazide diuretic within the VA. There is a need for future projects and clinician education to increase the implementation of guideline-recommended therapy within the VA, particularly regarding chlorthalidone use.

Heart failure (HF) is a chronic, progressive condition that is characterized by the heart’s inability to effectively pump blood throughout the body. In 2018, approximately 6.2 million US adults had HF, and 13.4% of all death certificates noted HF as a precipitating factor.¹ Patients not receiving appropriate guideline-directed medical therapy (GDMT) face a 29% excess mortality risk over a 2-year period.² Each additional GDMT included in a patient’s regimen significantly reduces all-cause mortality.³

The Change the Management of Patients with Heart Failure (CHAMP) registry reports that only about 1% of patients with HF are prescribed 3 agents from contemporary GDMT at target doses, highlighting the need for optimizing clinicians’ approaches to GDMT.⁴ Similarly, The Get With The Guidelines Heart Failure Registry has noted that only 20.2% of patients with HF with reduced ejection fraction (HFrEF) are prescribed a sodium-glucose cotransporter 2 inhibitor (SGLT2i) following hospital discharge for HFrEF exacerbation.⁵ Overall, treatment rates with GDMT saw limited improvement between 2013 and 2019, with no significant difference between groups in mortality, indicating the need for optimized methods to encourage the initiation of GDMT.⁶

Remote monitoring and telecare are novel ways to improve GDMT rates in those with HFrEF. However, data are inconsistent regarding the impact of remote HF monitoring and improvements in GDMT or HF-related outcomes.^6-10 The modalities of remote monitoring for GDMT vary among studies, but the potential for telehealth monitoring to improve GDMT, thereby potentially reducing HF-related hospitalizations, is clear.

Telemonitoring has demonstrated improved participant adherence with weight monitoring, although the withdrawal rate was high, and has the potential to reduce all-cause mortality and HF-related hospitalizations.^11,12 Telemonitoring for GDMT optimization led to an increase in the proportion of patients who achieved optimal GDMT doses, a decrease in the time to dose optimization, and a reduction in the number of clinic visits.¹³ Remote GDMT titration was accomplished in the general patient population with HFrEF; however, in populations already followed by cardiologists or HF specialists, remote optimization strategies did not yield different proportions of GDMT use.¹⁴ The aim of this study was to assess the impact of the home telehealth (HT) monitoring program on the initiation and optimization of HF GDMT among veterans with HFrEF at the Veterans Affairs Ann Arbor Healthcare System (VAAAHS) in Michigan.

Methods

This was a single-center retrospective study of Computerized Patient Record System (CPRS)data. Patients at the VAAAHS were evaluated if they were diagnosed with HFrEF and were eligible for enrollment in the HT monitoring program. Eligibility criteria included a diagnosis of stage C HF, irrespective of EF, and a history of any HF-related hospitalization. We focused on patients with HFrEF due to stronger guideline-based recommendations for certain pharmacotherapies as compared with HF with mildly reduced ejection fraction (HFmrEF) and preserved ejection fraction (HFpEF). Initial patient data for HT enrolling were accessed using the Heart Failure Dashboard via the US Department of Veterans Affairs (VA) Academic Detailing Service. The target daily doses of typical agents used in HFrEF GDMT are listed in the Appendix.

The HT program is an embedded model in which HT nurses receive remote data from the patient and triage that with the VAAAHS cardiology team. Patients’ questions, concerns, and/or vital signs are recovered remotely. In this model, nurses are embedded in the cardiology team, working with the cardiologists, cardiology clinical pharmacist, and/or cardiology nurse practitioners to make medication interventions. Data are recorded with an HT device, including weight, blood pressure (BP), heart rate, and pulse oximetry. HT nurses are also available to the patient via phone or video. The program uses a 180-day disease management protocol for HF via remote device, enabling the patient to answer questions and receive education on their disease daily. Responses to questions and data are then reviewed by an HT nurse remotely during business hours and triaged as appropriate with the cardiology team. Data can be communicated to the cardiology team via the patient record, eliminating the need for the cardiology team to use the proprietary portal affiliated with the HT device.

Study Sample

Patient information was obtained from a list of 417 patients eligible for enrollment in the HT program; the list was sent to the HT program for review and enrollment. Patient data were extracted from the VAAAHS HF Dashboard and included all patients with HFrEF and available data on the platform. The sample for the retrospective chart review included 40 adults who had HFrEF, defined as a left ventricular EF (LVEF) of ≤ 40% as evidenced by a transthoracic echocardiogram or cardiac magnetic resonance imaging. These patients were contacted and agreed to enroll in the HT monitoring program. The HT program population was compared against a control group of 33 patients who were ineligible for the HT program. Patients were deemed ineligible for HT if they resided in a nursing home, lacked a VAAAHS primary care clinician, or declined participation in the HT program.

Procedures

Patients enrolled in the HT program were assigned a nurse who monitored and responded to changes in vital signs, such as an increase in weight of ≥ 3 lb in 24 hours or ≥ 5 lb in 7 days. Each participant in HT received a home BP machine to check BP and heart rates and a scale for daily weights. The patient also responded to questions on a tablet device with the BP machine and scale. The vital signs were put into the tablet device, and the patient answered questions about their symptoms for that day. If the patient endorsed any symptoms of HF, such as pedal edema, abdominal distention, orthopnea, paroxysmal nocturnal dyspnea, dyspnea on exertion, or orthopnea, they received a call from the nurse to discuss symptoms and make appropriate adjustments in the diuretic or HF regimen. The HF-trained pharmacist served as a resource for drug therapy management and provided recommendations for diuretic adjustments on an as-needed basis.

Patients who declined participation in the HT program followed the standard of care, which was limited to visits with primary care clinicians and/or cardiologists as per the follow-up plan. Patient data were collected over 12 months. The study was approved by the VAAAHS Institutional Review Board (reference number, 1703034), Research and Development Committee, and Research Administration.

The primary goal of the study was to assess the impact of the HT program on drug interventions, specifically initiating and titrating HFrEF pharmacotherapies. Interventions were based on GDMT with known mortality- and morbidity-reducing properties when used at their maximum tolerated doses, including angiotensin-converting enzyme inhibitors (ACEi), angiotensin receptor-neprilysin inhibitor (ARNi), or angiotensin receptor blockers (ARB), with a preference for ARNi, β-blockers for HFrEF (metoprolol succinate, bisoprolol, or carvedilol), aldosterone antagonists, and SGLT2is.

Secondary goals included HF-related hospitalizations, medication adherence, time to enrollment in HT, time to laboratory analysis after the initiation or titration of an ACEi/ARB/ARNi or aldosterone antagonist, and time enrolled in the HT program. Patients were considered adherent if their drug refill history showed consistent fills of their medications. The χ² test was used for total interventions made during the study period and Fisher exact test for all others.

Results

Patient data were collected between July 2022 and June 2023. All 73 patients were male, and the mean age in the HT group (n = 40) was 72.6 years and 75.2 years for the control group (n = 33). Overall, the baseline demographics were similar between the groups (Table 1). Of 40 patients screened for enrollment in the HT program, 33 were included in the analysis (Figure 1).

At baseline, the HT group included more individuals than the control group on ACEi/ARB/ARNi (24 vs 19, respectively), β-blocker (28 vs 24, respectively), SGLT2i (14 vs 11, respectively), and aldosterone antagonist (15 vs 9, respectively) (Figure 2). There were 20 interventions made in the HT group compared with 11 therapy changes in the control arm during the study (odds ratio, 1.43; P = .23) (Table 2). In the HT group, 1 patient achieved an ACEi target dose, 3 patients achieved a β-blocker target dose, and 7 achieved a target dose of spironolactone (titration is not required for SGLT2i therapy and is counted as target dose). In the HT group, 17 patients were on ≥ 3 recommended agents, while 9 patients were taking 4 agents. Seven of 20 HT group interventions resulted in titration to the target dose. In the control group, no patients achieved an ARNi target dose, 3 patients achieved a β-blocker target dose, and 2 patients achieved a spironolactone antagonist target dose. In the control arm, 7 patients were on ≥ 3 GDMTs, and 2 were taking 4 agents. No patient in either group achieved a target dose of 4 agents. Five of 11 control group interventions resulted in initiation or titration of GDMT to the target dose.

We analyzed the overall percentage of VAAAHS patients with HFrEF who were on appropriate GDMT. Given the ongoing drive to improve HF-related outcomes, HT interventions could not be compared to a static population, so the HT and control patients were compared with the rates of GDMT at VAAAHS, which was available in the Academic Detailing Service Heart Failure Dashboard (Figure 3). Titration and optimization rates were also compared (Figure 4). From July 2022 to June 2023, ARNi use increased by 16.6%, aldosterone antagonist by 6.8%, and β-blockers by 2.4%. Target doses of GDMTs were more difficult to achieve in the hospital system. There was an increase in aldosterone antagonist target dose achievement by 4.7%, but overall there were decreases in target doses in other GDMTs: ACEi/ARB/ARNi target dose use decreased by 3.2%, ARNi target dose use decreased by 2.7% and target β-blocker use decreased by 0.9%.

Discussion

Telehealth yielded clinically important interventions, with some titrations bringing patients to their target doses of medications for HFrEF. The 20 interventions made in the HT group can be largely attributed to the nurses’ efforts to alert clinicians to drug titrations or ACEi/ARB to ARNi transitions. Although the findings were not statistically significant, the difference in the number of drug therapy changes supports the use of the HT program for a GDMT optimization strategy. Patients may be difficult to titrate secondary to adverse effects that make medication initiation or titration inappropriate, such as hypotension and hyperkalemia, although this was not observed in this small sample size. Considering a mean HT enrollment of 5.3 months, many patients had adequate disease assessment and medication titration. Given that patients are discharged from the service once deemed appropriate, this decreases the burden on the patient and increases the utility and implementation of the HT program for other patients.

A surprising finding of this study was the lower rate of HF-related hospitalizations in the HT group. Although not the primary subject of interest in the study, it reinforced the importance of close health care professional follow-up for patients living with HF. Telehealth may improve communication and shared decision making over medication use. Because the finding was unanticipated, the rate of diuretic adjustments was not tracked.

Patients were reevaluated every 6 months for willingness to continue the program, adherence, and clinical needs. These results are similar to those of other trials that demonstrated improved rates of GDMT in the setting of pharmacist- or nurse-led HF treatment optimization.^15,16 These positive results differ from other trials incorporating remote monitoring regarding patient continuation in HT programs. For example, in a study by Ding and colleagues, the withdrawal rate from their monitoring service was about 22%, while in our study only 1 patient withdrew from the HT program.¹¹

The HT program resulted in fewer hospitalizations than the control arm. There were 6 HF-related hospitalizations in the control group, although 5 involved a single patient. Typically, such a patient would be encouraged to follow HT monitoring after just 1 HF-related hospitalization; however, the patient declined to participate.

Early optimization of GDMT in patients who were recently discharged from the hospital for an HF-related hospitalization yields a reduction in hospital rehospitalization.¹⁷ GDMT optimization has unequivocal benefits in HF outcomes. Unfortunately, the issues surrounding methodologies on how to best optimize GDMT are lacking. While HT has been found to be feasible in the aid of optimizing medical therapy, the TIM-HF trial concluded that remote monitoring services had no significant benefit in reducing mortality.^7,8 On the other hand, the OptiLink HF Study showed that when clinicians respond to remote monitoring prompts from fluid index threshold crossing alerts, these interventions are associated with significantly improved clinical outcomes in patients with implantable cardioverter-defibrillators and advanced HF.⁹ In contrast to previous trials, the AMULET trial showed that remote monitoring compared with standard care significantly reduced the risk of HF hospitalization or cardiovascular death during the 12-month follow-up among patients with HF and LVEF ≤ 49% after an episode of acute exacerbation.¹⁰ Additionally, patients who received skilled home health services and participated in remote monitoring for their chronic HF experienced a reduction in all-cause 30-day readmission.¹⁸

Given the contrasting evidence regarding remote monitoring and variable modalities of implementing interventions, we investigated whether HT monitoring yields improvements in GDMT optimization. We found that HT nurses were able to nearly double the rate of interventions for patients with HFrEF. The HT program in providing expanded services will require adequate staffing responsibilities and support. The HT program is geared toward following a large, diverse patient population, such as those with chronic obstructive pulmonary disease, hypertension, and HF. We only evaluated services for patients with HFrEF, but the program also follows patients with HfmrEF and HfpEF. These patients were not included as GDMT optimization differs for patients with an LVEF > 40%.^19,20

The lower rates of achieving target doses of GDMTs were likely obstructed by continuous use of initial drug doses and further limited by lack of follow-up. When compared with the rest of the VAAAHS, there was a greater effort to increase ARNi use in the HT group as 7 of 33 patients (21%) were started on ARNi compared with a background increase of ARNi use of 17%. There was a lower mortality rate observed in the HT group compared with the control group. One patient in each group died of unrelated causes, while 2 deaths in the control group were due to worsening HF. The difference in mortality is likely multifactorial, possibly related to the control group’s greater disease burden or higher mean age (75.2 years vs 72.6 years).

Limitations

This was an observational cohort design, which is subject to bias. Thus, the findings of this study are entirely hypothesis-generating and a randomized controlled trial would be necessary for clearer results. Second, low numbers of participants may have skewed the data set. Given the observational nature of the study, this nonetheless is a positive finding to support the HT program for assisting with HF monitoring and prompting drug interventions. Due to the low number of participants, a single patient may have skewed the results with 5 hospitalizations.

Conclusions

This pilot study demonstrates the applicability of HT monitoring to optimize veteran HFrEF GDMT. The HT program yielded numerically relevant interventions and fewer HF-related hospitalizations compared with the control arm. The study shows the feasibility of the program to safely optimize GDMT toward their target doses and may serve as an additional catalyst to further develop HT programs specifically targeted toward HF monitoring and management. Cost-savings analyses would likely need to demonstrate the cost utility of such a service.

Acknowledgments

We thank the home telehealth nursing staff for their assistance in data collection and enrollment of patients into the monitoring program.

Heart failure (HF) is a chronic, progressive condition that is characterized by the heart’s inability to effectively pump blood throughout the body. In 2018, approximately 6.2 million US adults had HF, and 13.4% of all death certificates noted HF as a precipitating factor.¹ Patients not receiving appropriate guideline-directed medical therapy (GDMT) face a 29% excess mortality risk over a 2-year period.² Each additional GDMT included in a patient’s regimen significantly reduces all-cause mortality.³

The Change the Management of Patients with Heart Failure (CHAMP) registry reports that only about 1% of patients with HF are prescribed 3 agents from contemporary GDMT at target doses, highlighting the need for optimizing clinicians’ approaches to GDMT.⁴ Similarly, The Get With The Guidelines Heart Failure Registry has noted that only 20.2% of patients with HF with reduced ejection fraction (HFrEF) are prescribed a sodium-glucose cotransporter 2 inhibitor (SGLT2i) following hospital discharge for HFrEF exacerbation.⁵ Overall, treatment rates with GDMT saw limited improvement between 2013 and 2019, with no significant difference between groups in mortality, indicating the need for optimized methods to encourage the initiation of GDMT.⁶

Remote monitoring and telecare are novel ways to improve GDMT rates in those with HFrEF. However, data are inconsistent regarding the impact of remote HF monitoring and improvements in GDMT or HF-related outcomes.^6-10 The modalities of remote monitoring for GDMT vary among studies, but the potential for telehealth monitoring to improve GDMT, thereby potentially reducing HF-related hospitalizations, is clear.

Telemonitoring has demonstrated improved participant adherence with weight monitoring, although the withdrawal rate was high, and has the potential to reduce all-cause mortality and HF-related hospitalizations.^11,12 Telemonitoring for GDMT optimization led to an increase in the proportion of patients who achieved optimal GDMT doses, a decrease in the time to dose optimization, and a reduction in the number of clinic visits.¹³ Remote GDMT titration was accomplished in the general patient population with HFrEF; however, in populations already followed by cardiologists or HF specialists, remote optimization strategies did not yield different proportions of GDMT use.¹⁴ The aim of this study was to assess the impact of the home telehealth (HT) monitoring program on the initiation and optimization of HF GDMT among veterans with HFrEF at the Veterans Affairs Ann Arbor Healthcare System (VAAAHS) in Michigan.

Methods

This was a single-center retrospective study of Computerized Patient Record System (CPRS)data. Patients at the VAAAHS were evaluated if they were diagnosed with HFrEF and were eligible for enrollment in the HT monitoring program. Eligibility criteria included a diagnosis of stage C HF, irrespective of EF, and a history of any HF-related hospitalization. We focused on patients with HFrEF due to stronger guideline-based recommendations for certain pharmacotherapies as compared with HF with mildly reduced ejection fraction (HFmrEF) and preserved ejection fraction (HFpEF). Initial patient data for HT enrolling were accessed using the Heart Failure Dashboard via the US Department of Veterans Affairs (VA) Academic Detailing Service. The target daily doses of typical agents used in HFrEF GDMT are listed in the Appendix.

The HT program is an embedded model in which HT nurses receive remote data from the patient and triage that with the VAAAHS cardiology team. Patients’ questions, concerns, and/or vital signs are recovered remotely. In this model, nurses are embedded in the cardiology team, working with the cardiologists, cardiology clinical pharmacist, and/or cardiology nurse practitioners to make medication interventions. Data are recorded with an HT device, including weight, blood pressure (BP), heart rate, and pulse oximetry. HT nurses are also available to the patient via phone or video. The program uses a 180-day disease management protocol for HF via remote device, enabling the patient to answer questions and receive education on their disease daily. Responses to questions and data are then reviewed by an HT nurse remotely during business hours and triaged as appropriate with the cardiology team. Data can be communicated to the cardiology team via the patient record, eliminating the need for the cardiology team to use the proprietary portal affiliated with the HT device.

Study Sample

Patient information was obtained from a list of 417 patients eligible for enrollment in the HT program; the list was sent to the HT program for review and enrollment. Patient data were extracted from the VAAAHS HF Dashboard and included all patients with HFrEF and available data on the platform. The sample for the retrospective chart review included 40 adults who had HFrEF, defined as a left ventricular EF (LVEF) of ≤ 40% as evidenced by a transthoracic echocardiogram or cardiac magnetic resonance imaging. These patients were contacted and agreed to enroll in the HT monitoring program. The HT program population was compared against a control group of 33 patients who were ineligible for the HT program. Patients were deemed ineligible for HT if they resided in a nursing home, lacked a VAAAHS primary care clinician, or declined participation in the HT program.

Procedures

Patients enrolled in the HT program were assigned a nurse who monitored and responded to changes in vital signs, such as an increase in weight of ≥ 3 lb in 24 hours or ≥ 5 lb in 7 days. Each participant in HT received a home BP machine to check BP and heart rates and a scale for daily weights. The patient also responded to questions on a tablet device with the BP machine and scale. The vital signs were put into the tablet device, and the patient answered questions about their symptoms for that day. If the patient endorsed any symptoms of HF, such as pedal edema, abdominal distention, orthopnea, paroxysmal nocturnal dyspnea, dyspnea on exertion, or orthopnea, they received a call from the nurse to discuss symptoms and make appropriate adjustments in the diuretic or HF regimen. The HF-trained pharmacist served as a resource for drug therapy management and provided recommendations for diuretic adjustments on an as-needed basis.

Patients who declined participation in the HT program followed the standard of care, which was limited to visits with primary care clinicians and/or cardiologists as per the follow-up plan. Patient data were collected over 12 months. The study was approved by the VAAAHS Institutional Review Board (reference number, 1703034), Research and Development Committee, and Research Administration.

The primary goal of the study was to assess the impact of the HT program on drug interventions, specifically initiating and titrating HFrEF pharmacotherapies. Interventions were based on GDMT with known mortality- and morbidity-reducing properties when used at their maximum tolerated doses, including angiotensin-converting enzyme inhibitors (ACEi), angiotensin receptor-neprilysin inhibitor (ARNi), or angiotensin receptor blockers (ARB), with a preference for ARNi, β-blockers for HFrEF (metoprolol succinate, bisoprolol, or carvedilol), aldosterone antagonists, and SGLT2is.

Secondary goals included HF-related hospitalizations, medication adherence, time to enrollment in HT, time to laboratory analysis after the initiation or titration of an ACEi/ARB/ARNi or aldosterone antagonist, and time enrolled in the HT program. Patients were considered adherent if their drug refill history showed consistent fills of their medications. The χ² test was used for total interventions made during the study period and Fisher exact test for all others.

Results

Patient data were collected between July 2022 and June 2023. All 73 patients were male, and the mean age in the HT group (n = 40) was 72.6 years and 75.2 years for the control group (n = 33). Overall, the baseline demographics were similar between the groups (Table 1). Of 40 patients screened for enrollment in the HT program, 33 were included in the analysis (Figure 1).

At baseline, the HT group included more individuals than the control group on ACEi/ARB/ARNi (24 vs 19, respectively), β-blocker (28 vs 24, respectively), SGLT2i (14 vs 11, respectively), and aldosterone antagonist (15 vs 9, respectively) (Figure 2). There were 20 interventions made in the HT group compared with 11 therapy changes in the control arm during the study (odds ratio, 1.43; P = .23) (Table 2). In the HT group, 1 patient achieved an ACEi target dose, 3 patients achieved a β-blocker target dose, and 7 achieved a target dose of spironolactone (titration is not required for SGLT2i therapy and is counted as target dose). In the HT group, 17 patients were on ≥ 3 recommended agents, while 9 patients were taking 4 agents. Seven of 20 HT group interventions resulted in titration to the target dose. In the control group, no patients achieved an ARNi target dose, 3 patients achieved a β-blocker target dose, and 2 patients achieved a spironolactone antagonist target dose. In the control arm, 7 patients were on ≥ 3 GDMTs, and 2 were taking 4 agents. No patient in either group achieved a target dose of 4 agents. Five of 11 control group interventions resulted in initiation or titration of GDMT to the target dose.

We analyzed the overall percentage of VAAAHS patients with HFrEF who were on appropriate GDMT. Given the ongoing drive to improve HF-related outcomes, HT interventions could not be compared to a static population, so the HT and control patients were compared with the rates of GDMT at VAAAHS, which was available in the Academic Detailing Service Heart Failure Dashboard (Figure 3). Titration and optimization rates were also compared (Figure 4). From July 2022 to June 2023, ARNi use increased by 16.6%, aldosterone antagonist by 6.8%, and β-blockers by 2.4%. Target doses of GDMTs were more difficult to achieve in the hospital system. There was an increase in aldosterone antagonist target dose achievement by 4.7%, but overall there were decreases in target doses in other GDMTs: ACEi/ARB/ARNi target dose use decreased by 3.2%, ARNi target dose use decreased by 2.7% and target β-blocker use decreased by 0.9%.

Discussion

Telehealth yielded clinically important interventions, with some titrations bringing patients to their target doses of medications for HFrEF. The 20 interventions made in the HT group can be largely attributed to the nurses’ efforts to alert clinicians to drug titrations or ACEi/ARB to ARNi transitions. Although the findings were not statistically significant, the difference in the number of drug therapy changes supports the use of the HT program for a GDMT optimization strategy. Patients may be difficult to titrate secondary to adverse effects that make medication initiation or titration inappropriate, such as hypotension and hyperkalemia, although this was not observed in this small sample size. Considering a mean HT enrollment of 5.3 months, many patients had adequate disease assessment and medication titration. Given that patients are discharged from the service once deemed appropriate, this decreases the burden on the patient and increases the utility and implementation of the HT program for other patients.

A surprising finding of this study was the lower rate of HF-related hospitalizations in the HT group. Although not the primary subject of interest in the study, it reinforced the importance of close health care professional follow-up for patients living with HF. Telehealth may improve communication and shared decision making over medication use. Because the finding was unanticipated, the rate of diuretic adjustments was not tracked.

Patients were reevaluated every 6 months for willingness to continue the program, adherence, and clinical needs. These results are similar to those of other trials that demonstrated improved rates of GDMT in the setting of pharmacist- or nurse-led HF treatment optimization.^15,16 These positive results differ from other trials incorporating remote monitoring regarding patient continuation in HT programs. For example, in a study by Ding and colleagues, the withdrawal rate from their monitoring service was about 22%, while in our study only 1 patient withdrew from the HT program.¹¹

The HT program resulted in fewer hospitalizations than the control arm. There were 6 HF-related hospitalizations in the control group, although 5 involved a single patient. Typically, such a patient would be encouraged to follow HT monitoring after just 1 HF-related hospitalization; however, the patient declined to participate.

Early optimization of GDMT in patients who were recently discharged from the hospital for an HF-related hospitalization yields a reduction in hospital rehospitalization.¹⁷ GDMT optimization has unequivocal benefits in HF outcomes. Unfortunately, the issues surrounding methodologies on how to best optimize GDMT are lacking. While HT has been found to be feasible in the aid of optimizing medical therapy, the TIM-HF trial concluded that remote monitoring services had no significant benefit in reducing mortality.^7,8 On the other hand, the OptiLink HF Study showed that when clinicians respond to remote monitoring prompts from fluid index threshold crossing alerts, these interventions are associated with significantly improved clinical outcomes in patients with implantable cardioverter-defibrillators and advanced HF.⁹ In contrast to previous trials, the AMULET trial showed that remote monitoring compared with standard care significantly reduced the risk of HF hospitalization or cardiovascular death during the 12-month follow-up among patients with HF and LVEF ≤ 49% after an episode of acute exacerbation.¹⁰ Additionally, patients who received skilled home health services and participated in remote monitoring for their chronic HF experienced a reduction in all-cause 30-day readmission.¹⁸

Given the contrasting evidence regarding remote monitoring and variable modalities of implementing interventions, we investigated whether HT monitoring yields improvements in GDMT optimization. We found that HT nurses were able to nearly double the rate of interventions for patients with HFrEF. The HT program in providing expanded services will require adequate staffing responsibilities and support. The HT program is geared toward following a large, diverse patient population, such as those with chronic obstructive pulmonary disease, hypertension, and HF. We only evaluated services for patients with HFrEF, but the program also follows patients with HfmrEF and HfpEF. These patients were not included as GDMT optimization differs for patients with an LVEF > 40%.^19,20

The lower rates of achieving target doses of GDMTs were likely obstructed by continuous use of initial drug doses and further limited by lack of follow-up. When compared with the rest of the VAAAHS, there was a greater effort to increase ARNi use in the HT group as 7 of 33 patients (21%) were started on ARNi compared with a background increase of ARNi use of 17%. There was a lower mortality rate observed in the HT group compared with the control group. One patient in each group died of unrelated causes, while 2 deaths in the control group were due to worsening HF. The difference in mortality is likely multifactorial, possibly related to the control group’s greater disease burden or higher mean age (75.2 years vs 72.6 years).

Limitations

This was an observational cohort design, which is subject to bias. Thus, the findings of this study are entirely hypothesis-generating and a randomized controlled trial would be necessary for clearer results. Second, low numbers of participants may have skewed the data set. Given the observational nature of the study, this nonetheless is a positive finding to support the HT program for assisting with HF monitoring and prompting drug interventions. Due to the low number of participants, a single patient may have skewed the results with 5 hospitalizations.

Conclusions

This pilot study demonstrates the applicability of HT monitoring to optimize veteran HFrEF GDMT. The HT program yielded numerically relevant interventions and fewer HF-related hospitalizations compared with the control arm. The study shows the feasibility of the program to safely optimize GDMT toward their target doses and may serve as an additional catalyst to further develop HT programs specifically targeted toward HF monitoring and management. Cost-savings analyses would likely need to demonstrate the cost utility of such a service.

Acknowledgments

We thank the home telehealth nursing staff for their assistance in data collection and enrollment of patients into the monitoring program.

Heart failure (HF) is a chronic, progressive condition that is characterized by the heart’s inability to effectively pump blood throughout the body. In 2018, approximately 6.2 million US adults had HF, and 13.4% of all death certificates noted HF as a precipitating factor.¹ Patients not receiving appropriate guideline-directed medical therapy (GDMT) face a 29% excess mortality risk over a 2-year period.² Each additional GDMT included in a patient’s regimen significantly reduces all-cause mortality.³

The Change the Management of Patients with Heart Failure (CHAMP) registry reports that only about 1% of patients with HF are prescribed 3 agents from contemporary GDMT at target doses, highlighting the need for optimizing clinicians’ approaches to GDMT.⁴ Similarly, The Get With The Guidelines Heart Failure Registry has noted that only 20.2% of patients with HF with reduced ejection fraction (HFrEF) are prescribed a sodium-glucose cotransporter 2 inhibitor (SGLT2i) following hospital discharge for HFrEF exacerbation.⁵ Overall, treatment rates with GDMT saw limited improvement between 2013 and 2019, with no significant difference between groups in mortality, indicating the need for optimized methods to encourage the initiation of GDMT.⁶

Remote monitoring and telecare are novel ways to improve GDMT rates in those with HFrEF. However, data are inconsistent regarding the impact of remote HF monitoring and improvements in GDMT or HF-related outcomes.^6-10 The modalities of remote monitoring for GDMT vary among studies, but the potential for telehealth monitoring to improve GDMT, thereby potentially reducing HF-related hospitalizations, is clear.

Telemonitoring has demonstrated improved participant adherence with weight monitoring, although the withdrawal rate was high, and has the potential to reduce all-cause mortality and HF-related hospitalizations.^11,12 Telemonitoring for GDMT optimization led to an increase in the proportion of patients who achieved optimal GDMT doses, a decrease in the time to dose optimization, and a reduction in the number of clinic visits.¹³ Remote GDMT titration was accomplished in the general patient population with HFrEF; however, in populations already followed by cardiologists or HF specialists, remote optimization strategies did not yield different proportions of GDMT use.¹⁴ The aim of this study was to assess the impact of the home telehealth (HT) monitoring program on the initiation and optimization of HF GDMT among veterans with HFrEF at the Veterans Affairs Ann Arbor Healthcare System (VAAAHS) in Michigan.

Methods

This was a single-center retrospective study of Computerized Patient Record System (CPRS)data. Patients at the VAAAHS were evaluated if they were diagnosed with HFrEF and were eligible for enrollment in the HT monitoring program. Eligibility criteria included a diagnosis of stage C HF, irrespective of EF, and a history of any HF-related hospitalization. We focused on patients with HFrEF due to stronger guideline-based recommendations for certain pharmacotherapies as compared with HF with mildly reduced ejection fraction (HFmrEF) and preserved ejection fraction (HFpEF). Initial patient data for HT enrolling were accessed using the Heart Failure Dashboard via the US Department of Veterans Affairs (VA) Academic Detailing Service. The target daily doses of typical agents used in HFrEF GDMT are listed in the Appendix.

The HT program is an embedded model in which HT nurses receive remote data from the patient and triage that with the VAAAHS cardiology team. Patients’ questions, concerns, and/or vital signs are recovered remotely. In this model, nurses are embedded in the cardiology team, working with the cardiologists, cardiology clinical pharmacist, and/or cardiology nurse practitioners to make medication interventions. Data are recorded with an HT device, including weight, blood pressure (BP), heart rate, and pulse oximetry. HT nurses are also available to the patient via phone or video. The program uses a 180-day disease management protocol for HF via remote device, enabling the patient to answer questions and receive education on their disease daily. Responses to questions and data are then reviewed by an HT nurse remotely during business hours and triaged as appropriate with the cardiology team. Data can be communicated to the cardiology team via the patient record, eliminating the need for the cardiology team to use the proprietary portal affiliated with the HT device.

Study Sample

Patient information was obtained from a list of 417 patients eligible for enrollment in the HT program; the list was sent to the HT program for review and enrollment. Patient data were extracted from the VAAAHS HF Dashboard and included all patients with HFrEF and available data on the platform. The sample for the retrospective chart review included 40 adults who had HFrEF, defined as a left ventricular EF (LVEF) of ≤ 40% as evidenced by a transthoracic echocardiogram or cardiac magnetic resonance imaging. These patients were contacted and agreed to enroll in the HT monitoring program. The HT program population was compared against a control group of 33 patients who were ineligible for the HT program. Patients were deemed ineligible for HT if they resided in a nursing home, lacked a VAAAHS primary care clinician, or declined participation in the HT program.

Procedures

Patients enrolled in the HT program were assigned a nurse who monitored and responded to changes in vital signs, such as an increase in weight of ≥ 3 lb in 24 hours or ≥ 5 lb in 7 days. Each participant in HT received a home BP machine to check BP and heart rates and a scale for daily weights. The patient also responded to questions on a tablet device with the BP machine and scale. The vital signs were put into the tablet device, and the patient answered questions about their symptoms for that day. If the patient endorsed any symptoms of HF, such as pedal edema, abdominal distention, orthopnea, paroxysmal nocturnal dyspnea, dyspnea on exertion, or orthopnea, they received a call from the nurse to discuss symptoms and make appropriate adjustments in the diuretic or HF regimen. The HF-trained pharmacist served as a resource for drug therapy management and provided recommendations for diuretic adjustments on an as-needed basis.

Patients who declined participation in the HT program followed the standard of care, which was limited to visits with primary care clinicians and/or cardiologists as per the follow-up plan. Patient data were collected over 12 months. The study was approved by the VAAAHS Institutional Review Board (reference number, 1703034), Research and Development Committee, and Research Administration.

The primary goal of the study was to assess the impact of the HT program on drug interventions, specifically initiating and titrating HFrEF pharmacotherapies. Interventions were based on GDMT with known mortality- and morbidity-reducing properties when used at their maximum tolerated doses, including angiotensin-converting enzyme inhibitors (ACEi), angiotensin receptor-neprilysin inhibitor (ARNi), or angiotensin receptor blockers (ARB), with a preference for ARNi, β-blockers for HFrEF (metoprolol succinate, bisoprolol, or carvedilol), aldosterone antagonists, and SGLT2is.

Secondary goals included HF-related hospitalizations, medication adherence, time to enrollment in HT, time to laboratory analysis after the initiation or titration of an ACEi/ARB/ARNi or aldosterone antagonist, and time enrolled in the HT program. Patients were considered adherent if their drug refill history showed consistent fills of their medications. The χ² test was used for total interventions made during the study period and Fisher exact test for all others.

Results

Patient data were collected between July 2022 and June 2023. All 73 patients were male, and the mean age in the HT group (n = 40) was 72.6 years and 75.2 years for the control group (n = 33). Overall, the baseline demographics were similar between the groups (Table 1). Of 40 patients screened for enrollment in the HT program, 33 were included in the analysis (Figure 1).

At baseline, the HT group included more individuals than the control group on ACEi/ARB/ARNi (24 vs 19, respectively), β-blocker (28 vs 24, respectively), SGLT2i (14 vs 11, respectively), and aldosterone antagonist (15 vs 9, respectively) (Figure 2). There were 20 interventions made in the HT group compared with 11 therapy changes in the control arm during the study (odds ratio, 1.43; P = .23) (Table 2). In the HT group, 1 patient achieved an ACEi target dose, 3 patients achieved a β-blocker target dose, and 7 achieved a target dose of spironolactone (titration is not required for SGLT2i therapy and is counted as target dose). In the HT group, 17 patients were on ≥ 3 recommended agents, while 9 patients were taking 4 agents. Seven of 20 HT group interventions resulted in titration to the target dose. In the control group, no patients achieved an ARNi target dose, 3 patients achieved a β-blocker target dose, and 2 patients achieved a spironolactone antagonist target dose. In the control arm, 7 patients were on ≥ 3 GDMTs, and 2 were taking 4 agents. No patient in either group achieved a target dose of 4 agents. Five of 11 control group interventions resulted in initiation or titration of GDMT to the target dose.

We analyzed the overall percentage of VAAAHS patients with HFrEF who were on appropriate GDMT. Given the ongoing drive to improve HF-related outcomes, HT interventions could not be compared to a static population, so the HT and control patients were compared with the rates of GDMT at VAAAHS, which was available in the Academic Detailing Service Heart Failure Dashboard (Figure 3). Titration and optimization rates were also compared (Figure 4). From July 2022 to June 2023, ARNi use increased by 16.6%, aldosterone antagonist by 6.8%, and β-blockers by 2.4%. Target doses of GDMTs were more difficult to achieve in the hospital system. There was an increase in aldosterone antagonist target dose achievement by 4.7%, but overall there were decreases in target doses in other GDMTs: ACEi/ARB/ARNi target dose use decreased by 3.2%, ARNi target dose use decreased by 2.7% and target β-blocker use decreased by 0.9%.

Discussion

Telehealth yielded clinically important interventions, with some titrations bringing patients to their target doses of medications for HFrEF. The 20 interventions made in the HT group can be largely attributed to the nurses’ efforts to alert clinicians to drug titrations or ACEi/ARB to ARNi transitions. Although the findings were not statistically significant, the difference in the number of drug therapy changes supports the use of the HT program for a GDMT optimization strategy. Patients may be difficult to titrate secondary to adverse effects that make medication initiation or titration inappropriate, such as hypotension and hyperkalemia, although this was not observed in this small sample size. Considering a mean HT enrollment of 5.3 months, many patients had adequate disease assessment and medication titration. Given that patients are discharged from the service once deemed appropriate, this decreases the burden on the patient and increases the utility and implementation of the HT program for other patients.

A surprising finding of this study was the lower rate of HF-related hospitalizations in the HT group. Although not the primary subject of interest in the study, it reinforced the importance of close health care professional follow-up for patients living with HF. Telehealth may improve communication and shared decision making over medication use. Because the finding was unanticipated, the rate of diuretic adjustments was not tracked.

Patients were reevaluated every 6 months for willingness to continue the program, adherence, and clinical needs. These results are similar to those of other trials that demonstrated improved rates of GDMT in the setting of pharmacist- or nurse-led HF treatment optimization.^15,16 These positive results differ from other trials incorporating remote monitoring regarding patient continuation in HT programs. For example, in a study by Ding and colleagues, the withdrawal rate from their monitoring service was about 22%, while in our study only 1 patient withdrew from the HT program.¹¹

The HT program resulted in fewer hospitalizations than the control arm. There were 6 HF-related hospitalizations in the control group, although 5 involved a single patient. Typically, such a patient would be encouraged to follow HT monitoring after just 1 HF-related hospitalization; however, the patient declined to participate.

Early optimization of GDMT in patients who were recently discharged from the hospital for an HF-related hospitalization yields a reduction in hospital rehospitalization.¹⁷ GDMT optimization has unequivocal benefits in HF outcomes. Unfortunately, the issues surrounding methodologies on how to best optimize GDMT are lacking. While HT has been found to be feasible in the aid of optimizing medical therapy, the TIM-HF trial concluded that remote monitoring services had no significant benefit in reducing mortality.^7,8 On the other hand, the OptiLink HF Study showed that when clinicians respond to remote monitoring prompts from fluid index threshold crossing alerts, these interventions are associated with significantly improved clinical outcomes in patients with implantable cardioverter-defibrillators and advanced HF.⁹ In contrast to previous trials, the AMULET trial showed that remote monitoring compared with standard care significantly reduced the risk of HF hospitalization or cardiovascular death during the 12-month follow-up among patients with HF and LVEF ≤ 49% after an episode of acute exacerbation.¹⁰ Additionally, patients who received skilled home health services and participated in remote monitoring for their chronic HF experienced a reduction in all-cause 30-day readmission.¹⁸

Given the contrasting evidence regarding remote monitoring and variable modalities of implementing interventions, we investigated whether HT monitoring yields improvements in GDMT optimization. We found that HT nurses were able to nearly double the rate of interventions for patients with HFrEF. The HT program in providing expanded services will require adequate staffing responsibilities and support. The HT program is geared toward following a large, diverse patient population, such as those with chronic obstructive pulmonary disease, hypertension, and HF. We only evaluated services for patients with HFrEF, but the program also follows patients with HfmrEF and HfpEF. These patients were not included as GDMT optimization differs for patients with an LVEF > 40%.^19,20

The lower rates of achieving target doses of GDMTs were likely obstructed by continuous use of initial drug doses and further limited by lack of follow-up. When compared with the rest of the VAAAHS, there was a greater effort to increase ARNi use in the HT group as 7 of 33 patients (21%) were started on ARNi compared with a background increase of ARNi use of 17%. There was a lower mortality rate observed in the HT group compared with the control group. One patient in each group died of unrelated causes, while 2 deaths in the control group were due to worsening HF. The difference in mortality is likely multifactorial, possibly related to the control group’s greater disease burden or higher mean age (75.2 years vs 72.6 years).

Limitations

This was an observational cohort design, which is subject to bias. Thus, the findings of this study are entirely hypothesis-generating and a randomized controlled trial would be necessary for clearer results. Second, low numbers of participants may have skewed the data set. Given the observational nature of the study, this nonetheless is a positive finding to support the HT program for assisting with HF monitoring and prompting drug interventions. Due to the low number of participants, a single patient may have skewed the results with 5 hospitalizations.

Conclusions

This pilot study demonstrates the applicability of HT monitoring to optimize veteran HFrEF GDMT. The HT program yielded numerically relevant interventions and fewer HF-related hospitalizations compared with the control arm. The study shows the feasibility of the program to safely optimize GDMT toward their target doses and may serve as an additional catalyst to further develop HT programs specifically targeted toward HF monitoring and management. Cost-savings analyses would likely need to demonstrate the cost utility of such a service.

Acknowledgments

We thank the home telehealth nursing staff for their assistance in data collection and enrollment of patients into the monitoring program.