Evaluation of digoxin dosing and therapeutic drug monitoring in adult patients receiving continuous venovenous hemodiafiltration

Abstract

Background

Digoxin dosing guidance during continuous renal replacement therapy (CRRT) is scarce, relying on two case reports with pharmacokinetic modeling not subsequently validated by TDM and expert opinion.

Objective

To examine varying digoxin dosing regimen effects on serum digoxin concentrations (SDCs) in patients receiving continuous venovenous hemodiafiltration (CVVHDF).

Design

Retrospective, multicenter, single-health-system case series.

Methods

Adult patients admitted to an intensive care unit over a 3.5-year period who received digoxin, with concurrent SDC monitoring and CVVHDF were included. The primary outcome of this study was to describe digoxin dosing and associated SDCs in adult patients receiving CVVHDF. Secondary outcomes aimed to determine the safety and tolerability of digoxin for patients receiving CVVHDF, including the incidence of bradycardia, digoxin immune fab administration, and in-hospital mortality.

Results

Six patients met criteria for inclusion, yielding fourteen digoxin levels drawn during CVVHDF. The median CRRT effluent rate was 30.7 mL/kg/hr with a median digoxin level of 0.9 (IQR 0.7-1.1) ng/mL. Both levels ≥ 1.2 ng/mL (above the safety-oriented upper threshold) occurred in patients following a loading dose exceeding 8 mcg/kg ideal body weight. Ten total SDC were drawn in 4 patients receiving 125 mcg every 48-hours maintenance regimens and all resulted in pre-steady state levels < 1.2 ng/mL. Two patients (33.3%) experienced bradycardia, both of whom were receiving concomitant amiodarone. No patients received digoxin immune fab, and two patients (33.3%) experienced in-hospital mortality.

Conclusion

A maintenance regimen of 125 mcg every 48 hours during CVVHDF achieved levels below the safety SDCs, although notably all levels reflect the first several doses during CVVHDF and may not reflect steady state monitoring. These data help address the current gap in data describing digoxin dosing during CRRT. Although this adds limited insight for digoxin dosing during CRRT, standard dosing and monitoring recommendations remain unconfirmed.

Introduction

Digoxin is a cardiac glycoside used for heart failure (HF) treatment and acute rate control managing arrhythmias such as atrial fibrillation (AF) when first-line treatment options are contraindicated or ineffective. ¹ Digoxin has different mechanisms of action depending on the indication for use. When used for HF, digoxin acts as a positive inotrope by inhibiting the Na-K ATPase pump allowing the influx of calcium ions into the heart and increasing myocardial contractility. For AF, digoxin has negative chronotropic actions at the atrioventricular node slowing the conduction of the heart. ² Digoxin is 20-30% protein-bound with a large volume of distribution and 72-75% renal excretion. ³ Goal levels have historically been defined as 0.8 to 2.0 ng/mL for AF and 0.5 to 0.9 ng/mL for HF. ⁴ A post hoc analysis of the Digitalis Investigation Group (DIG) trial showed an 11.8% increase in mortality in patients with HF and a serum digoxin concentration (SDC) >1.2 ng/mL (95% CI, 5.7%-18.0%) receiving digoxin compared to placebo. ⁵ The association with SDC and mortality persisted after a multivariate analysis (SDC ≥1.2 ng/mL; HR 1.16, 95% CI, 0.96-1.39). Another retrospective analysis of data from the DIG trial showed SDC between 1.2 and 2.0 ng/mL were associated with a hazard ratio for death in women of 1.33. ⁶ The ARISTOTLE trial showed a 56% increased hazard of mortality (adjusted HR: 1.56; 95% CI: 1.20 to 2.04) in patients with AF and a SDC ≥1.2 ng/mL compared with those not on digoxin. ⁷ Notably, the 2023 American College of Cardiology/American Heart Association (ACC/AHA) Atrial Fibrillation guidelines support a goal of <1.2 ng/mL which provides benefit without increasing mortality risk.^1,8 The narrow therapeutic range of digoxin is further complicated by renal dysfunction, which may lead to toxicity, such as nausea, vomiting, visual symptoms, heart palpitations, bradycardia, and heart block. ² These symptoms are more likely to occur at SDC >2.0 ng/mL but may be experienced by patients with lower SDCs. ² Digitoxin, another cardiac glycoside with more limited availability worldwide (unavailable currently in the Unites States), exhibits different pharmacokinetic properties. ⁹ In addition to being more lipophilic, digitoxin has increased enteral absorption (complete vs. 60-80%) and protein binding (90-97% vs. 25%), with decreased volume of distribution (0.7 vs. 7 L/kg) relative to digoxin.^9–11 Given its primary route of enterohepatic excretion, digitoxin is markedly less affected by decreased renal function than digoxin.^9,11

Recent randomized controlled trials have further refined contemporary roles of cardiac glycosides, emphasizing potential clinical benefit in select patients. The DIGOBET-AF trial recently demonstrated that de novo therapy with digoxin compared with bisoprolol was associated with improved left atrial strain in patients with AF initiated on therapy using a rate control strategy who remained in AF at 30 days. This improvement was demonstrated despite similar ventricular rate control, suggesting possible hemodynamic advantage in this population. ¹²

More recently, the DIGIT-HF trial compared digitoxin to placebo in patients with HF with reduced ejection fraction receiving guideline-directed medical therapy (GDMT). The primary composite outcome of death from any cause or hospital admission for worsening HF showed benefit (39.5% versus 44.1%, HR 0.82; 95% CI 0.69 to 0.98; P = 0.03); although trending towards benefit, neither component of the primary composite reached statistical significance individually as secondary outcomes. ¹³ While benefit of reduced hospitalizations demonstrated in the DIG trial was not as clearly demonstrated, it is important to note that DIGIT-HF was completed in the era of more robust GDMT. Collectively, these trials underscore likely benefits of cardiac glycoside therapy, recognizing safe use requires conservative dosing, careful patient selection, and therapeutic drug monitoring (TDM), critical principles in patients with renal failure.

In critically ill patients with acute kidney injury (AKI), acute on chronic kidney disease (CKD), or end-stage renal disease (ESRD), continuous renal replacement therapy (CRRT) allows for solute and fluid removal to facilitate restoration of acid-base balance, electrolyte abnormalities, and/or toxin or fluid removal. ¹⁴ Due to digoxin’s narrow therapeutic index and potential for accumulation in the setting of renal impairment, it is often withheld in patients undergoing CRRT; however, certain clinical scenarios may still necessitate its use in this setting. For patients receiving digoxin and CRRT, dose adjustments and TDM are necessary due to the individualization of the CRRT prescription and pharmacokinetic variability. There are three CRRT modalities: continuous venovenous hemofiltration (CVVH), continuous venovenous hemodialysis (CVVHD), and continuous venovenous hemodiafiltration (CVVHDF). CVVH uses high ultrafiltration rates with a hydrostatic gradient across a semi-permeable membrane to move solutes via convection. CVVHD utilizes dialysate to move solutes out of the blood down their concentration gradient. Lastly, CVVHDF is a hybrid of these modalities and combines dialysate with ultrafiltration and replacement fluids^.14 While digoxin’s large volume of distribution decreases the amount of drug expected to be removed during CRRT, its molecular weight (780.95 g/mol) and relatively low percent protein binding are factors that would increase expected removal during CRRT.^11,15,16

Currently, there is minimal evidence guiding digoxin dosing in CRRT. Practices are guided by data from single case reports of digoxin utilization during CVVH and CVVHDF, respectively, and expert opinion. ¹⁷ Benken et al. measured SDCs and dialysate digoxin concentrations in a 46-year-old patient receiving CVVH to calculate the sieving coefficient, transmembrane clearance, digoxin concentration in the ultrafiltration fluid, and need for supplemental doses of digoxin. Investigators estimated that 38.4 mcg of digoxin was needed daily (approximately 250 mcg weekly) to replace losses from CVVH in addition to a maintenance regimen of 125 mcg every other day. ¹⁷ Srinivasan et al. subsequently reported detailed pharmacokinetic modeling of digoxin during CVVHDF. Utilizing sieving coefficient from the Benken et al. case report, authors modeled digoxin clearance incorporating physiologically based pharmacokinetic modeling utilizing a single post-load digoxin level. ¹⁸ Although the breadth of outcomes data linked to SDCs were not established in critically ill patients, a recent single-center retrospective study found patients who died had significantly higher mean digoxin concentrations compared to survivors (1.33 ± 0.59 ng/mL vs. 1.03 ± 0.43 ng/mL, p = 0.003). While extremely limited, this data in critically ill patients aligns with established thresholds portending worse outcomes. ¹⁹

Collectively, these reports underscore the complexity of digoxin pharmacokinetics during CRRT and the absence of standardized, clinically validated dosing strategies for patients receiving CVVHDF. Additional descriptive clinical data are needed to contextualize pharmacokinetic findings within real-world critical care practice, particularly with respect to dosing decisions, TDM, and safety signals. The present study conforms to the STROBE statement and checklist (see Supplemental Materials). ²⁰

Methods

Study design

This was a retrospective, multicenter, single-health-system case series.

Outcomes

The primary outcome of this study was to describe digoxin dosing and associated SDC in adult patients receiving CVVHDF. Secondary outcomes aimed to determine the safety and tolerability of digoxin for patients receiving CVVHDF and included the incidence of bradycardia, digoxin immune fab administration, and in-hospital mortality. Bradycardia was defined as documented heart rate less than 60 beats per minute. Weight-based loading dose calculations (mcg/kg) utilized ideal body weight (IBW) unless actual body weight was less than IBW, in which case actual body weight was utilized.

Study population

Patients were included if they were at least 18 years old, admitted to an intensive care unit at a Prisma Health hospital from March 1, 2021 to September 1, 2024, and received digoxin with TDM and CVVHDF. Patients were excluded if TDM was not completed concurrently while receiving CVVHDF. Levels below the safety-oriented upper threshold of digoxin were defined as <1.2 ng/mL; levels above this threshold were considered elevated.

Statistical analysis

The primary and secondary objectives were analyzed using descriptive statistics. Medians and interquartile range (IQR) were calculated for continuous variables and counts and percentages for categorical variables.

Results

Baseline characteristics

Forty-four patients received digoxin and CVVHDF during admission, with 27 patients excluded due to lack of TDM. All patients received CVVHDF via PRISMAFLEX system utilizing the HF 1000 hemofilter sets with PAES membrane. An additional 11 patients did not have digoxin levels drawn while on CRRT, leaving 6 patients in the analysis (Figure 1). The majority of the patients were male, Black or African American, and non-Hispanic (Table 1). The majority of patients had a HF diagnosis before admission (83.3%) compared to only 33.3% of patients that had an AF diagnosis before admission. An equal amount of patients received digoxin for HF and AF indications. Four patients (all patients with HF, three with digoxin initiated following durable left ventricular assist device insertion) were noted to have right ventricular dysfunction. Two-thirds of included patients were receiving amiodarone prior to the initiation of CRRT. The median CRRT effluent rate was 30.7 mL/kg/hr (Table 2).

Figure 1.

Patient screening.

Table 1.

Baseline characteristics.

Patient characteristic	Included patients (n =6)
Age (years), median (IQR)	61.5 (51-67)
Actual body weight (kg), median (IQR)	73.55 (71.1-97.5)
Ideal body weight (kg), median (IQR)	70.55 (57-73)
Height (cm), median (IQR)	176.55 (165.1-180)
Sex (male), n (%)	4 (66.7)
Race/ethnicity, n (%)
Black/African American	5 (83.3)
Non-Hispanic	6 (100)
Heart failure diagnosis PTA (yes), n (%)	5 (83.3)
Atrial fibrillation diagnosis PTA (yes), n (%)	2 (33.3)
Digoxin indication listed inpatient, n (%)
Heart failure	3 (50)
Atrial fibrillation	3 (50)
Digoxin PTA med (yes), n (%)	1 (16.7)
Amiodarone PTA med (yes), n (%)	0 (0)
Amiodarone inpatient before digoxin level (yes), n (%)	4 (66.7)
ICU subset, n (%)	​
Cardiovascular ICU	5 (83.3)
Cardiopulmonary ICU	1 (16.7)
LVAD, n (%)	3 (50)
RV dysfunction, n (%)	4 (66.7)

PTA = prior to admission; ICU = intensive care unit; LVAD = left ventricular assist device; RV = right ventriclular; IQR = interquartile range; kg = kilogram.

Table 2.

CRRT prescription.

CRRT variable	Included patients (n=6)
Length of CRRT (days), median (IQR)	7 (4-10.8)
CRRT effluent rate (mL/hr), median (IQR)	2400 (2200-3200)
Pre-filter rate	1100 (1000-1500)
Dialysate rate	1100 (1000-1500)
Post-filter rate	200 (200)
CRRT effluent rate (mL/kg/hr), median (IQR)	30.7 (30.3-37.4)
Max ultrafiltration rate (mL/hr), median (IQR)	150 (125-200)

CRRT = continuous renal replacement therapy; IQR = interquartile range; mL = milliliter; kg = kilogram; hr = hour.

Primary outcome

Fourteen levels from six patients were drawn during CRRT. The median digoxin level was 0.9 (IQR 0.7-1.1) ng/mL. Only three levels drawn occurred at least 72 hours into digoxin therapy with concomitant CRRT for entire duration. All three levels were in the same patient with residual urine output. The median total loading dose was 8 mcg/kg using ideal body weight (IBW). Loading doses greater than 8 mcg/kg IBW (n=2) resulted in levels above 1.2 ng/mL, whereas loading doses below 8 mcg/kg IBW (n=2) resulted in levels less than 1.2 ng/mL. One-third of patients received a digoxin load but were not initiated on maintenance regimens. Of the patients that received maintenance therapy, all received 125 mcg every 48-hour regimens. All maintenance doses on this regimen resulted in levels less than 1.2 ng/mL (n=10) (Table 3). Maintenance dosing was initiated in patients when clinicians determined that ongoing digoxin exposure was necessary to attain or sustain clinical response.

Table 3.

Patient-specific demographics and primary outcome.

Patient	ABW (kg)	IBW (kg)	LD (mcg)	MD (mcg)	Flow rates (mL/hr) Pre-filter Dialysate Post-filter	Net ultrafiltration (mL/hr)	CRRT effluent rate (mL/kg/hr)	Digoxin level (ng/mL)	Approximate timing of level	Urine output from CRRT initiation until level
1	97.5	79.9	500	125 q48h	1200	100	27.7	1.0	43h after 4th MD, 18h after CRRT	None
					1200
					200
					1500	0	27.7	1.1	18h after 5th MD, 40h after CRRT	None
					1000
					200
					1500	25	33.1	0.8	43h after 5th MD, 65h after CRRT	None
					1500
					200
2	70.4	75.4	625	None	1500	0	45.5	2.5	20h after LD, 13h after CRRT a	175 mL
					1500
					200
			500	None	1500	0	45.5	1.1	14.5h after LD, 163.5h after CRRT resumed b	None
					1500
					200
			None	125 q48h				1.1	17h after 1st MD, 184h after CRRT resumed b	None
			None	125 mcg q48h				1	10h after 2nd MD, 232h after CRRT resumed b	None
3	68.2	57	None	125 mcg q48h	1000	75	40.7	0.6	13h after 1st MD c , 213h after CRRT	50 mL
					1500
					200
			None	125 mcg q48h	1500	125	41.4	0.6	37h after 1st MD, 237h after CRRT	50 mL
					1000
					200
4	72.1	47.8	250	None	1000	0	30.5	0.7	17h after LD, 33h after CRRT	10 mL
					1000
					200
5	128	73	750	None	1500	0	25.0	1.6	11h after LD, 44h after CRRT	50 mL
					1500
					200
6	71.1	70.7	None	125 mcg q48h	1000	0	30.9	0.7	43h after 2nd MD, 78h after CRRT	1980 mL
					1000
					200
					1000	150	33.1	0.6	33h after 3rd MD, 126h after CRRT	3055 mL
					1000
					200
					1000	150	33.1	0.7	44h after 4th MD, 175h after CRRT	4045 mL
					1000
					200

^aReceived ten days of 62.5 mcg every other day MD prior to LD.

^bCRRT received for 9 days after initial level and held for three days prior to resuming; no further digoxin dosing between loading regimens.

^cHeld for 10 days after AKI onset; level was after 8 days on CRRT (<100 mL total UOP during time held).

ABW = actual body weight; IBW = ideal body weight; LD = loading dose; MD = maintenance dose; CRRT = continuous renal replacement therapy; kg = kilogram; mcg = microgram; ml = milliliter; hr = hour; ng = nanogram.

Secondary outcomes

One-third (33%) of patients experienced bradycardic events while receiving digoxin and CRRT. Half of patients with a digoxin level of 1.2 ng/mL or more experienced bradycardia, compared with one-quarter (n=1) that had levels below 1.2 ng/mL. No patients received digoxin immune fab (Table 4). The majority (83.3%) of patients were on amiodarone during CRRT. Only 33.3% of patients each were on dexmedetomidine, midodrine or propofol and no patients received a beta-blocker while on CRRT. Both patients that experienced bradycardia on digoxin and CRRT received amiodarone concurrently (Tables 4 and 5). In-hospital mortality occurred in two of the six patients (33.3%). Neither patient who experienced in-hospital mortality had elevated SDC, however, one experienced bradycardia.

Table 4.

Secondary outcomes.

Patient	Bradycardia	Receipt of digoxin immune fab	Concurrent medications associated with bradycardia					In-hospital mortality
			Amiodarone	Beta-blocker	Dexmedetomidine	Midodrine	Propofol
1	No	No	No	No	No	No	No	No
2	No	No	Yes	No	Yes	No	Yes	No
3	No	No	Yes	No	No	Yes	No	No
4	No	No	Yes	No	No	No	No	Yes
5	Yes	No	Yes	No	Yes	No	Yes	No
6	Yes	No	Yes	No	No	Yes	No	Yes

Table 5.

Incidence of bradycardia with associated concurrent medications.

Medication, n (%)	No bradycardia (n =4)	Patients who experienced bradycardia (n = 2)	All included patients (n = 6)
Amiodarone	3 (75)	2 (100)	5 (83.3)
Beta-blocker	0 (0)	0 (0)	0 (0)
Dexmedetomidine	1 (25)	1 (50)	2 (33.3)
Midodrine	1 (25)	1 (50)	2 (33.3)
Propofol	1 (25)	1 (50)	2 (33.3)

Discussion

Recent randomized trials further contextualize the role of cardiac glycosides in contemporary practice and inform interpretation of our findings. The DIGOBET-AF trial demonstrated that digoxin achieved comparable ventricular rate control to bisoprolol while improving left atrial strain in patients initiated on rate control strategy who remained in AF at one month, suggesting a potential hemodynamic advantage. ¹² This may be informative for patients unable to tolerate standard rate-control agents due to hypotension or myocardial dysfunction, both of which are expected to be seen at a higher frequency for patients in AF requiring CRRT. ¹² The DIGIT-HF trial demonstrated reductions in the composite primary endpoint of mortality and heart-failure–related hospitalizations among patients with HF with reduced ejection fraction receiving GDMT. ¹³ While neither component of the composite endpoint reached statistical significance, trends were towards benefit, and may be more difficult to attain in the era of more comprehensive GDMT than prior large-scale historical data.

Importantly, neither DIGOBET-AF nor the DIGIT-HF trial directly addressed patients with advanced kidney failure or those receiving renal replacement therapy. In fact, DIGIOBET-HF specifically did not have routine TDM available and therefore excluded patients with renal dysfunction. Regardless, both studies acknowledged the narrow therapeutic window of cardiac glycosides and clinical consequences of excessive exposure. This supports our rationale for prioritizing safety-oriented serum concentration thresholds in our study, rather than attempting to achieve higher historical targets that may increase mortality risk. Multiple factors expected in critically ill patients receiving digoxin during CVVHDF may amplify prioritizing safety-oriented targets: predictability of drug clearance is decreased due to extracorporeal removal, residual renal function is often dynamic, and drug–drug interactions may compound exposure and/or hemodynamic effects.

In patients with normal renal function, several validated tools are available to guide digoxin dosing for both AF and HF. For HF, the Bauman-DiDomenico nomogram provides dosing recommendations based on IBW and creatinine clearance. ²¹ For AF, calculators such as Jelliffe or Koup and Jusko are commonly used to estimate appropriate dosing.^22–24 However, no equivalent nomogram or standardized dosing algorithm exists for patients undergoing renal replacement therapies, including intermittent hemodialysis (iHD) or CRRT. In these populations, dosing is based on limited data described. Although the Benken et al. case report in a patient receiving CVVH measured serum and dialysate digoxin concentrations allowed calculation of a sieving coefficient, it had notable limitations. Digoxin dosing was held for the first five days of CVVH until levels fell below 1.0 ng/mL, when a maintenance regimen of 125 mcg every 48 hours was initiated. The estimated weekly supplemental dose required was divided into a 125 mcg twice weekly dosing strategy. Since these two weekly doses alternated with scheduled every 48-hour dosing, the patient actually received 125 mcg daily for the first five days after initiation of CRRT. Importantly, no TDM was conducted after the first five days of digoxin therapy during CVVH until levels were reassessed after the patient transitioned to iHD. Therefore, no TDM was available to confirm SDC in patients receiving greater than five days of maintenance dosing during CVVH. Additionally, the timing of digoxin levels relative to dose administration and presence of residual urine output was not reported. Given this only involved CVVH, Srinivasan et al. subsequently reported a case report with detailed physiologically based pharmacokinetic modeling of digoxin during CVVHDF.

Their analysis quantified extracorporeal clearance and demonstrated that CVVHDF can contribute non-negligibly to digoxin elimination, particularly when combined with pharmacokinetic interactions. ¹⁸ While this remains the only detailed pharmacokinetic evaluation of digoxin during CVVHDF to date, this report also had significant limitations. Notably, authors utilized variables (i.e., saturation coefficient) calculated in the Benken et al. CVVH case report to execute their pharmacokinetic calculations, introducing potential limitations in modeling. Additionally, modeling was based upon a single post-load level. While efforts to include residual renal function in modeling are applauded, utilization of a serum creatinine-based equation to determine is problematic. Creatinine values respond proportionally to the CRRT effluent prescription; therefore, the proposition this would represent intrinsic residual renal function is problematic. Residual urine output during CRRT support would be a more clinically appropriate variable to consider as a surrogate marker. Finally, neither case reported TDM to corroborate modeled equations.

In contrast, our study did not aim to perform formal pharmacokinetic modeling but instead sought to describe real-world dosing practices, SDCs, and safety outcomes in a multicenter sample of critically ill patients. These approaches are complementary rather than redundant. While Srinivasan et al. provide mechanistic insight into digoxin disposition during CVVHDF, the present case series adds clinical context by illustrating how digoxin is dosed and monitored in practice, how frequently serum concentrations fall within contemporary safety-oriented targets, and what adverse events may occur in the presence of common confounders such as amiodarone use and residual renal function. Together, these data emphasize that digoxin exposure during CVVHDF is highly individualized and influenced by multiple dynamic factors beyond extracorporeal clearance alone.

The novelty of the present study lies not in proposing a definitive dosing algorithm, but in providing multicenter descriptive data on digoxin TDM during CVVHDF, an area where available evidence remains extremely sparse. Additionally, our study helps address this gap by contributing novel data on digoxin dosing and monitoring specifically in patients receiving CVVHDF. Residual renal function was assessed by reporting documented urine output, providing ability for a more complete evaluation of medication exposure secondary to CRRT clearance alone versus in addition to residual intrinsic renal function.

To our knowledge, this represents the largest published series reporting SDCs obtained during CVVHDF and the first to systematically describe dosing decisions, concomitant medications, residual urine output, and safety outcomes across multiple patients receiving this CRRT modality.

In this study population, digoxin loading doses varied, with a median cumulative exposure of 8 mcg/kg based on IBW (unless actual body weight was less than IBW). Two patients developed elevated SDCs (1.6 ng/mL and 2.5 ng/mL) following cumulative loading doses exceeding 8 mcg/kg IBW. Notably, the highest observed concentration (2.5 ng/mL) was obtained shortly after CRRT initiation in a patient who had received maintenance digoxin therapy (62.5 mcg every 48 hours) for 10 days prior to loading, suggesting the contribution of pre-existing digoxin exposure. Collectively, these findings indicate that higher cumulative loading doses, particularly in the presence of prior digoxin therapy, were associated with higher measured serum concentrations during CVVHDF.

Bradycardia occurred in two patients, both of whom were receiving concomitant amiodarone, a known P-glycoprotein inhibitor that increases digoxin exposure. Findings highlight the importance of drug–drug interactions when interpreting safety outcomes and suggest that observed adverse events may reflect combined pharmacologic properties. Accordingly, safety conclusions based strictly upon reported TDM should be interpreted cautiously.

This study presents several important strengths that enhance its clinical relevance. First, its multicenter design allows for the inclusion of a diverse patient population across various institutions. Second, the study makes a meaningful contribution to the currently limited body of evidence surrounding digoxin dosing in patients undergoing CVVHDF. Given the pharmacokinetic complexities of digoxin and the unique challenges posed by renal replacement therapies, we provide much-needed data to guide clinicians in optimizing therapy for this patient population. Finally, the study accounted for confounding variables in the analysis of secondary outcomes by evaluating concomitant medications that can lead to bradycardia. Collectively, these strengths position the study as a valuable contribution to both clinical practice and scientific literature.

Limitations

Our retrospective study design and small sample size limits the ability to establish causal relationships. Although multicenter, it was conducted within a single health system, and no standardized institutional protocol exists for digoxin dosing in this specific population. As such, dosing decisions were based on the clinical judgment of the medical providers and pharmacists, introducing the potential for selection bias. Although 44 patients received concomitant digoxin therapy and CVVHDF during the study period, only six patients had SDCs obtained during CVVHDF and therefore eligible to be included in the analysis. The decision to perform TDM was part of routine clinical care rather than according to a predefined study protocol and may have preferentially identified patients with specific clinical characteristics, such as longer treatment duration, concern for digoxin toxicity, altered renal function, hemodynamic instability, or uncertainty regarding dosing adequacy. As a result, the included patients may not represent the broader population of patients receiving digoxin during CVVHDF, potentially limiting external validity of the findings.

Disease-related confounders also impact interpretation. Digoxin was often initiated for HF, but dose escalation—through loading and maintenance regimens—frequently occurred in response to acute AF episodes. Notably, 50% of patients received digoxin following durable left ventricular assist device implantation, accounting for three of the four patients who exhibited RV dysfunction. Although digoxin loading is not standard for chronic HF, 75% of patients with RV dysfunction received loading doses, likely reflecting provider concern over the delay in achieving therapeutic levels with maintenance dosing alone. The only patient with RV dysfunction who did not receive a loading dose had pre-existing RV dysfunction prior to their admission. Patients were not evaluated for whether digoxin achieved the intended clinical goal or response, as our objective was to describe digoxin TDM and safety in patients receiving CVVHDF support. We recognize SDC < 1.2 ng/mL is not a universal therapeutic range. Based on findings from the DIG and ARISTOTLE trials described in the introduction, SDC ≥ 1.2 ng/mL are associated with increased mortality in both HF and atrial fibrillation populations.^5–7 Consequently, we used <1.2 ng/mL as a clinically relevant therapeutic target consistent with current evidence-based practice in regard to a safety-oriented upper threshold rather than historical ranges. Although clinical efficacy outcomes were not assessed in this study, concentrations below 1.2 ng/mL were considered to align with currently accepted SDC goals associated with a more favorable benefit–risk profile.

A key limitation of this analysis is that the 14 SDC were derived from only six individual patients and therefore do not represent 14 independent observations. Instead, several SDCs were obtained repeatedly from the same patients over time, introducing within-subject correlation that may reduce the effective sample size and limit the generalizability of concentration-level findings. Consequently, results should be interpreted primarily as descriptive observations of drug exposure within this small patient cohort rather than as independent concentration-based data points. While the repeated measurements provide valuable insight into the pharmacokinetic variability and temporal trends observed in these patients, patient-level conclusions are inherently limited by the small number of individuals included.

In-hospital mortality occurring in one-third of patients likely reflects the severity of illness in this critically ill sample versus digoxin exposure, given neither patient experiencing elevated levels experienced in-hospital mortality. Mortality was attributed to underlying cardiac disease and multiorgan failure rather than digoxin-related toxicity. Observed mortality not correlating with elevated SDC suggests that higher post-load levels, perhaps transient, may not confer the same risk. It does underscore the severity of illness in this population, highlighting the need for cautious interpretation of safety and efficacy outcomes in the small study.

Interpretation of SDCs in this sample is complicated by substantial heterogeneity in residual renal function. Most notably, Patient 6 demonstrated daily urine output after CRRT initiation of roughly 555 mL per day (0.33 mL/kg/hr). Given digoxin is predominantly eliminated unchanged by the kidneys, a degree of preserved intrinsic renal function likely played a role in digoxin clearance for this patient. In such cases, CVVHDF should be viewed as an adjunctive or supportive elimination pathway rather than the sole, or primary determinant of SDCs. Consequently, the observation that SDCs < 1.2 ng/mL were maintained with a maintenance regimen of 125 mcg every 48 hours in Patient 6 a week into maintenance therapy on CVVHDF may largely reflect additive native renal clearance rather than the adequacy of CVVHDF clearance alone. Grouping patients with preserved residual renal function alongside anuric patients risks obscuring important pharmacokinetic differences. These differences underscore the need for continued cautious extrapolation of dosing recommendations and highlight the importance of stratifying digoxin dosing strategies considering residual renal function in patients receiving CVVHDF. All CVVHDF treatment interruptions less than 4 hours were not systematically captured, which may have influenced cumulative digoxin exposure in certain cases. Any longer breaks in therapy are described in Table 3 annotation. To maintain consistency across patients despite these limitations, we standardized reporting of SDC timing relative to the initiation of CRRT and receipt of digoxin doses rather than attempting to adjust for intermittent therapy interruptions. While this approach provides a uniform reference point for interpretation, it may not fully reflect dynamic changes in drug clearance associated with limited pauses and resumptions of therapy.

As this study was descriptive in nature, we did not evaluate how concomitant medications or clinical factors modifying digoxin toxicity risk were actively managed by the primary treatment teams. Documentation describing how drug–drug interactions or electrolyte and acid–base abnormalities were addressed was inconsistent and therefore could not be assessed. However, concomitant medications associated with bradycardia, including amiodarone, are reported in Tables 4 and 5, and their potential contribution to observed safety outcomes is discussed.

Finally, a major limitation of this case series is the use of pre–steady-state digoxin levels. Digoxin’s half-life ranges from 36–48 hours in patients with normal renal function but may extend to 3-5 days in anuric patients.^14,15 The median length of time on CRRT in our study was 7 days suggesting the inability to reach a true steady state for most patients. While the present dataset describes SDCs < 1.2 ng/mL resulting from 125 mcg every 48-hour regimens, caution must be taken interpreting these values. For example, SDCs for Patient 1 were drawn after the fourth and fifth maintenance doses. Although it may quickly be deduced this reflects at least 8 to 10 days of therapy, values were all collected less than three days after CRRT initiation, meaning a significant portion of the regimen was administered prior to the initiation of CRRT. No levels were drawn reflecting more than two doses of a maintenance regimen with concomitant CRRT in patients without residual urine output (Table 3). Furthermore, in real life practice, CRRT is not typically anticipated for a prolonged duration of time, thus inherently patients receiving digoxin and concomitant CRRT may not reach a true steady state.

Conclusion

Rather than establishing prescriptive dosing recommendations, this case series highlights the variability and complexity of digoxin exposure during CVVHDF and underscores the need for individualized dosing guided by conservative SDC targets and close TDM. Frequent TDM is generally advised, albeit with greater urgency as duration of therapy exceeds several days of therapy during CRRT support, where in vivo data is completely lacking without residual urine output. Pre-steady state levels in this study reflect initial safety and do not guarantee safety in prolonged CRRT courses, where tissue accumulation of digoxin could still occur. When interpreted alongside prior pharmacokinetic analyses, these findings reinforce that both mechanistic modeling and real-world clinical observation are necessary to inform safe digoxin use in patients receiving CRRT. Future studies with levels reflecting both loading and maintenance dosing are needed to validate current weight-based recommendations, which are based on expert opinion and/or pharmacokinetic modeling from case reports.

ORCID iDs

Madison P. Nordin https://orcid.org/0009-0007-5565-9821

Jenna F. Cox https://orcid.org/0000-0002-7516-4590

Breanne Mefford https://orcid.org/0000-0003-2144-2724

Ethical considerations

The Prisma Health Institutional Review Board (approval 2233384-1) approved an expedited review based on an appropriate risk/benefit ratio and project design wherein risks have been minimized. Based on regulatory requirements, this project was approved for a waiver of informed consent as regulated by 45CFR46.116 and 46.117 and a waiver for HIPAA authorization as required by 45CFR164.512(i) (1) (i).

Consent to participate

The requirement for informed consent was waived by the Prisma Health IRB.

Data Availability Statement

The de-identified data collected supporting this study’s findings have been described within the manuscript. Additional data are not publicly available due to privacy restrictions.*