FormalPara Key Points

Digoxin immune fab therapy may limit duration of digoxin toxicity and intensive care unit length of stay in digoxin toxicity.

Digoxin immune fab therapy was appropriately utilized in most of the study population, however there were many patients in which it was underutilized or not indicated.

Education to improve clinician knowledge on appropriate patient selection for digoxin immune fab therapy may improve recognition and management of digoxin toxicity.

1 Background

Digoxin remains a widely prescribed drug for congestive heart failure (CHF) and atrial fibrillation [1]. A recent analysis of the ENGAGE AF-TIMI 48 Trial reported that 30% of CHF patients were treated with digoxin at baseline and 20% of patients without CHF reported digoxin use at trial entry [2]. With such widespread use, digoxin toxicity occurs frequently, upwards of 1% incidence per year [3]. Having a narrow therapeutic index, digoxin is a common medication implicated in emergency department (ED) admissions and has a high proportion of ED visits resulting in medication-related hospitalization [4]. When stratifying for age (< 85 or > 85 years), ED admissions for digoxin toxicity were 3.5-fold greater in the elderly than in the younger cohort [4]. This relatively high index for drug toxicity requires careful therapeutic drug monitoring, particularly in the elderly where digoxin toxicity is associated with increased risk of death [5].

When digoxin toxicity occurs, it must be quickly recognized and managed. In one report, serum digoxin concentrations (SDC) were above the toxic threshold (≥ 2.0 ng/mL) for nearly 96% of estimated ED visits [4]. Toxicity can also be present with lower SDCs [5]. In either case, digoxin toxicity symptoms can be vague and difficult to recognize from other underlying diseases, particularly in chronic digoxin toxicity. Patients with chronic digoxin toxicity are a heterogeneous group of patients with multiple co-morbidities. The complexity of these patients can confound digoxin toxicity diagnosis, potentially delaying appropriate treatment, and lead to worse outcomes.

Fortunately, there is an effective antibody therapy for severe and life-threatening digoxin toxicity—digoxin immune fab antibodies (DIF) [1, 6]. However, studies demonstrating the impacts of DIF on clinical outcomes in real-world healthcare settings are limited [7,8,9,10,11]. There are also limited observational data characterizing DIF’s utilization (appropriate utilization, underutilization, or administration when not indicated) in clinical care. Inappropriate DIF utilization may impact healthcare costs, adverse drug effects and patient outcomes. Given the limited published data, there is a need to characterize utilization of DIF and its impact on patient outcomes. The objectives of this study were to (1) characterize DIF use in confirmed/suspected digoxin toxic patients; (2) identify demographic and clinical factors associated with DIF use; and (3) evaluate the outcomes of patients with confirmed/suspected digoxin toxicity receiving DIF compared with those not receiving DIF.

2 Methods

2.1 Study Design, Settings, and Patient Population

This retrospective chart review was approved by the University of Florida’s Institutional Review Board as an exempt study (IRB study approval number IRB202201650). The University of Florida (UF) Health Integrated Data Repository (IDR) was utilized to identify patients with confirmed or suspected digoxin toxicity. The IDR is a large-scale database that collects and organizes information from across UF Health’s clinical and research enterprises. This clinical data warehouse aggregates data from the university’s various clinical and administrative information systems, including the EpicCare electronic health record system.

The IDR was queried for adult patients receiving digoxin therapy between 2011–2020 and having a serum digoxin concentration > 1.8 ng/mL or having confirmed digoxin toxicity (ICD-10 T46.0X1, T46.0X2, T46.0X5, ICD-9 972.1 and E942.1). If an individual patient had more than one inpatient hospitalization fulfilling either criterion, the earliest hospitalization was selected. Exclusion criteria included use of DIF for reasons other than digoxin toxicity, including any diagnosis indicating possible severe pre-eclampsia or a related diagnosis (ICD-10-CM code 642.x). In cases where the serum digoxin concentration was > 1.8 ng/mL without an accompanying diagnostic code confirming toxicity, a manual review of the chart was performed by a physician, clinical toxicologist, or medical toxicologist to determine if symptoms consistent with digoxin toxicity were also present (such as mental status disturbances, cardiac arrhythmias, and hyperkalemia). Patients who were asymptomatic with supratherapeutic digoxin concentrations were excluded from the analysis. Figure 1 depicts the schematic for inclusion of patient encounters identified by the IDR.

Fig. 1
figure 1

Inclusion schematic of patient encounters identified by the IDR. *Encounters associated with serum digoxin concentrations performed as outpatient laboratory testing, documented in outpatient clinic visits or telephone calls. ^Encounters not related to digoxin (examples include mismatch between physician-selected diagnostic code [such as adverse effect of other antidysrhythmic drugs, initial encounter] and final billed diagnostic code [adverse effect of cardiac-stimulant glycosides and drugs of similar action, initial encounter]). DIF digoxin immune fab, EHR electronic health record, IDR University of Florida Integrated Data Repository

Full size image

Data for this query were derived from two EDs located within a similar geographic area. One ED (site 1) is located within an urban safety-net academic hospital serving a largely indigent socially vulnerable patient population. The other ED (site 2) is located within a large academic tertiary center. During the study period, site 1 had an annual ED volume of 73,308. The average ED patient was 44 years old, 50% were female, 45% self-reported as Black/African American and 46% as White, 33% were on Medicare/Medicaid, and 36% were self-pay/charity pay. Site 2 had an annual ED volume of 62,314. The average ED patient was 45 years old, 52% were female, 20% self-reported as Black/African American and 70% as White, 49% were on Medicare/Medicaid, and 23% were self-pay.

Five independent, non-blinded reviewers performed chart reviews. All chart abstractors were trained prior to data collection, using standardized definitions for each variable and standardized data collection forms. TM then performed a secondary review and SS was available to resolve any conflicting coding. Inter-rater reliability was 99%.

2.2 Variables Collected

Demographics and clinical data were collected by manual chart review. Clinical data included whether patient was admitted from emergency department (yes/no); acute digoxin toxicity as suggested by clinical history (yes/no); and severe digoxin toxicity (yes/no) (defined as having at least one of the following—mental status changes/coma; requiring heart rhythm stabilization medications, temporary pacemaker, or external cardioversion; chronic dehydration or acute renal failure; serum digoxin > 4 ng/mL or serum potassium > 5 mEq/mL if serum creatinine is < 2 mg/dL or serum potassium level > 5.5 mEq/mL if serum creatinine is ≥ 2 mg/dL). DIF utilization (yes/no) and DIF date/time of administration were also collected. DIF treatment indications included potentially life-threatening dysrhythmias, progressive bradycardia, second- or third-degree heart block not responsive to atropine, serum potassium > 5.5 mEq/L, acute ingestion of ≥ 10 mg of digoxin, or chronic ingestion leading to steady-state serum concentration of > 6 ng/mL. These criteria were used to categorize DIF utilization as ‘appropriately’ utilized (DIF treatment given only if DIF indications met), ‘not indicated’ (DIF administered but not indicated), or ‘underutilized’ (DIF indicated but not administered) as determined by a clinical/medical toxicologist.

Other variables collected included All Patient Refined Diagnosis-Related Group (APR-DRG) severity score (severity of illness and mortality risk determination that considers comorbidities and admission indication); Charlson Comorbidity Index and estimated 10-year survival; principal diagnosis (digoxin toxicity, ventricular arrhythmia, atrioventricular conduction disturbance, bradycardia, hyperkalemia, altered level consciousness [AMS], acute kidney injury or failure [AKI]); principle diagnosis not digoxin related (CHF, ischemic heart disease [CAD], chronic renal insufficiency [CKD], diabetes mellitus [DM], end-stage renal disease [ESRD], electrolyte disorder other than hyperkalemia); digoxin toxicity was a secondary diagnosis (yes/no); and time to toxicity resolution (defined as time interval when symptoms related to digoxin toxicity resolved) determined by the clinical/medical toxicologist reviewing the medical record.

Collected laboratory, procedure, medications, and vital sign data included serum digoxin concentration (ng/mL), serum potassium, and heart rate at time of toxicity; external cardioversion/pacing (yes/no); permanent pacemaker placement (yes/no); mechanical ventilation (yes/no); administration of magnesium, lidocaine, amiodarone, atropine, kayexalate, and catecholamines.

The primary outcome included time to toxicity resolution and secondary outcomes of hospital admission, intensive care unit (ICU) admission, hospital length of stay (LOS), ICU LOS, APR-DRG severity, all-cause 60-day readmissions, and all-cause mortality during admission. Hospital LOS captured all time spent in the hospital, including the ICU when applicable.

2.3 Data Analysis

Demographic and clinical data were summarized descriptively for the overall study group as well as by DIF indication categories (appropriate utilization, not indicated, underutilized). Patients were stratified by whether they received DIF treatment or not. Univariate regressions were performed to compare demographic and clinical outcome differences between treatment groups. Both logistic and linear regressions were used to test the effect of DIF treatment on time to toxicity resolution (linear regression, log transformed), ICU admission (logistic regression), ICU LOS (linear regression, log transformed), all-cause 60-day readmission (logistic regression), and all-cause mortality during admission (logistic regression); while controlling for gender, race, age, SDC, severe digoxin toxicity designation, Charlson comorbidity estimated 10-year survival and APR-DRG score. All statistical analyses on the data set were performed in R version 4.1. Sample size calculations were performed using the R pwr package version 1.3-0 [12].

3 Results

A total of 796 unique patient encounters were identified by the IDR. Of these, 700 encounters were excluded after chart review, leaving 96 that met study inclusion criteria (30% from site 1; 70% from site 2) (Fig. 1). To model time to resolution of toxicity (primary study outcome) using multivariable linear regression, our study cohort size of 96 allowed detection of a moderate effect size of 0.25 with at least 80% power assuming a 0.05 level of significance.

Figure 2 displays temporal trends in digoxin toxicity and administrations of DIF. Data from 96 patients (non-DIF treated group = 49; DIF treated group = 47) were analyzed (Table 1). The study cohort was predominantly elderly (71 years old, SD 16.4), White (70%), female (66%), with a mean potassium of 4.93 (SD 1.27) and mean SDC of 3.13 (SD 1.43), with 53% presenting in acute toxicity, and 67% meeting study definition of severe toxicity. The average number of vials given was 3.7 (range 2–10). There were 12 (12%) deaths during hospital admission.

Fig. 2
figure 2

Temporal trends in digoxin toxicity and digoxin immune fab antibody administrations (2011–2020)

Full size image
Table 1 Comparison of demographic, clinical factors and outcomes in digoxin toxic patients stratified by digoxin immune fab (DIF) treatment
Full size table

3.1 Regressions

Several clinical parameters differentiated the DIF from the non-DIF group (p < 0.05): higher mean SDC (3.41 ± 1.63 vs 2.87 ± 1.17), higher mean potassium (5.33 ± 1.48 vs 4.55 ± 0.87), more patients who were severely toxic (85% vs 49%), and more likely to require pacing (Table 1). The DIF group was more likely to be appropriately managed (i.e., received DIF treatment only if DIF study indications were met) versus the non-DIF group (85% vs 61%, p < 0.001). In patients admitted to the ICU (n = 49), the DIF group (n = 33) had shorter ICU LOS compared with the non-DIF (n = 16) group (12.4 ± 20.3 vs 24.4 ± 28.7 days, p = 0.018). Electronic supplementary material (ESM) Table 1 compares additional clinical parameters for patients admitted to the ICU by DIF treatment group. Additionally, digoxin toxicity resolved sooner in the DIF group (0.6 ± 1.1 log days) compared with the non-DIF group (1.1 ± 0.4 log days) (coefficient − 0.702, 95% CI − 1.137 to − 0.267) (p < 0.01).

There were 12 deaths during hospital admission. Of the seven deaths in the non-DIF group, 86% (n = 6) were admitted to the ICU and had a mean ICU LOS of 37.6 days (range 4–121). In the DIF group, all five patients who died were admitted to the ICU and had a mean ICU LOS of 26 days (range 1–86). Groups did not differ in demographic factors, toxicity type (acute versus chronic), hospital LOS, digoxin-related diagnoses, mortality rates during admission, adjunctive medication administrations, APR-DRG severity or Charlson comorbidity estimated 10-year survival.

DIF was not indicated in 15% (n = 7) of the DIF group. DIF was underutilized in 39% (n = 19) of the untreated group. Of the 64 patients who met study criteria for severe toxicity, 24 (38%) did not receive DIF treatment.

3.2 Comparison of DIF Utilization

Table 2 compares demographic and clinical factors by DIF utilization. Symptoms of heart block/arrhythmia (55, 79%) and heart rate ≤ 50 (25, 53%) were more common in patients where DIF was utilized appropriately. Altered mental status was present in a higher proportion of patients where DIF treatment was underutilized (6, 32%) versus patients in which DIF was appropriately utilized (12, 17%). Digoxin toxicity was the primary diagnosis in only 26% (n = 5) of patients in which DIF was underutilized compared with 60% (n = 42) of patients where DIF was appropriately utilized. Severe digoxin toxicity was present in all patients who did not receive DIF (underutilized group), while none of the not-indicated group had severe digoxin toxicity. SDC were highest in the underutilized group (3.38 ± 1.69 ng/mL), while the not-indicated group had the lowest SDC (2.56 ± 1.36 ng/mL). The all-cause mortality rate in patients appropriately managed with DIF therapy versus those patients where DIF was underutilized was 11% and 21%, respectively.

Table 2 Comparison of demographic and clinical factors by DIF utilization
Full size table

3.3 Comparison of Demographic and Clinical Factors with Study Outcomes

ESM Tables 2–6 characterize the association between demographic and certain clinical factors with time to toxicity resolution, ICU admission, ICU LOS, all-cause 60-day readmissions and all-cause fatalities during admission for digoxin toxicity. APR-DRG severity scores were inversely associated with ICU LOS (p = 0.012) (ESM Table 4). The likelihood of 60-day all-cause re-admissions differed by race and age; with non-White patients having a 4.8-fold increased likelihood compared with White patients (p = 0.035) and each additional year of age associated with a decrease in the likelihood of re-admission (p = 0.038) (ESM Table 5). All-cause fatalities during admission were associated with increased age (p = 0.027) (ESM Table 6).

3.4 Costs Analysis

In the US, the average cost for one day of ICU level care is around US$2000 [13]. Per communication with the manufacturer, the average wholesale price of DigFab in the US is reported to be US$4599 [14]. The cost of DigiFab administration in our DIF cohort was US$17,016.30 per patient (US$4599 × 3.7 vials, average number of vials per patient). The cost for the average ICU stay in the DIF group was US$24,800 (12.4 days × US$2000); whereas the average ICU cost was US$48,800 in the non-DIF group (24.4 days × US$2000).

4 Discussion

Despite the DIF group exhibiting greater toxicity (higher SDC, higher potassium levels, and higher proportions of patients requiring cardiac pacing and meeting study definition of severe toxicity), digoxin toxicity resolved sooner in this group and they experienced shorter ICU stays compared with the non-DIF group (nearly 50% reduction in both mean time to toxicity resolution and ICU LOS). These outcomes were not associated with any other measured demographic or clinical factor, including Charlson Comorbidity and APR-DRG severity score, suggesting DIF treatment alone may be responsible for these findings. Previous studies have shown mixed results in the association of DIF with reduction in mean hospital LOS [8,9,10, 15]. Study population differences (demographics, proportion of patients with acute, subacute, or chronic toxicity, comorbidity status) may potentially explain these discrepant findings. Our findings were derived from a cohort of all digoxin toxic patients and suggests DIF, when clinically indicated, may improve patient morbidity (shorten toxicity) and reduce ICU healthcare costs, but not overall hospital LOS.

Although DigiFab can be an expensive therapy, within our cohort, it was associated with a substantial reduction in ICU LOS (12.4 vs 24.4 days) and, subsequently, may actually reduce overall healthcare costs. However, this cost savings may not translate to other cohorts composed of a different mix of chronic and acute digoxin toxic patients, patients within a different hospital system (particularly within other geographic regions of the US), or those managed in different countries where DigiFab pricing and healthcare costs may differ [16]. Future studies employing a formal cost-effectiveness analysis are needed to fully address the potential for healthcare savings.

A recent report compared clinical features of patients with elevated digoxin serum concentrations treated with DIF therapy with those who were not administered DIF [11]. Patients receiving DIF were more likely to have a heart rate < 51 beats per minute, evidence of abnormal renal function, serum potassium > 5.1 mmol/L and require multiple medical therapies for the treatment of their atrial fibrillation and/or CHF. In our cohort of digoxin toxic patients, we also found patients receiving DIF to have higher potassium levels, but also higher SDC, more severe digoxin toxicity, and a primary diagnosis of digoxin toxicity. DIF-treated patients also had heart rates that were lower (≤ 50 beats/min), but this was not statistically significant. It is not surprising that hyperkalemia is a common factor in patients receiving DIF therapy, as it is a well-known indication of serious, acute digoxin toxicity and is a clinical finding that is easy to diagnose [17].

Proper utilization of DIF is unknown. This real-world evidence study indicates that DIF was appropriately utilized (i.e., DIF treatment was given only if DIF study indications were met) in the majority of patients within our cohort (73%). However, there was a substantial proportion of patients in which DIF was underutilized in our cohort (20%). We found the all-cause mortality during hospital admission was lower when DIF was appropriately utilized compared with when it was underutilized. A similar trend in nonsignificant, numerically lower in-hospital mortality rates has also previously been noted [10].

Altered mental status was present in a higher proportion of patients where DIF treatment was underutilized compared with patients in which DIF was appropriately utilized. Additionally, digoxin toxicity was the primary diagnosis in only 26% of patients in whom DIF was underutilized versus 60% of patients where DIF was appropriately utilized. Although there could be other possible reasons for this discrepancy, namely incomplete diagnostic coding by the clinical team despite their recognition of patients’ toxicity, these findings suggest recognizing mental status changes as a potential sign of digoxin toxicity and the need for DIF treatment may be difficult to recognize by clinical care teams. Deciphering the cause of a patient’s altered mental status can be challenging depending on the patient’s comorbid conditions and other concomitant acute medical conditions. This may be particularly difficult in the older adult patient population, some of whom may have pre-existing dementia, and when it is early in the clinical course and collateral information may be limited, it may not be apparent if a patient’s mental status is altered from their baseline, altered from digoxin toxicity, or altered from a co-existing acute medical condition (such as a urinary tract infection).

We also examined the relationship between demographic and clinical factors with the study outcomes. We identified that non-White patients had a 4.8-fold increased likelihood compared with White patients of having an all-cause 60-day readmission. Although we did not measure social determinants of health and other health disparity factors known to influence healthcare utilization, it is unlikely that DIF treatment would be related to racial differences in re-admission rates, particularly since we did not find the DIF and non-DIF treatment groups differed by race or ethnicity (i.e., one racial/ethnic group was not more likely to receive DIF treatment than another).

We found patients in our cohort had a decrease in the likelihood of all-cause 60-day re-admission with each additional year of age; a finding potentially explained by the life expectancy of our cohort given their older age and comorbidity status (estimated 10-year survival of only 33%). Additionally, APR-DRG severity scores were inversely associated with ICU LOS, which may seem counterintuitive, but perhaps suggests that these patients may have died early in the course of their ICU admission. We also found that all-cause fatalities during admission were associated with increasing age.

There are several limitations to note. First, all patients and patient data were identified and collected retrospectively via review of the electronic health record. Thus, data is limited by the inherent flaws associated with chart reviews and in that it was not collected prospectively, such as incomplete or missing data (although our dataset had very few encounters with missing data—APR-DRG scores were missing in 13 patients), bias in electronic health record documentation by clinical teams and potentially by data collectors as they were not blinded to treatment groups, and difficulty assessing temporal relationships. Second, due the study design, there likely was bias in the selection of patients for DIF treatment by clinical care teams which may have impacted study findings. However, our findings reflect real-world clinical practice for the management of digoxin toxicity and utilization of DIF therapy. Third, our cohort is composed of a relatively small number of patients presenting to only two clinical sites within the same geographic region, thus our findings may not be generalizable to other clinical settings and populations, particularly those residing outside the United States. Fourth, it is possible that our inclusion and exclusion criteria did not sufficiently include/exclude relevant cases potentially biasing the study sample and results. However, as we performed manual chart reviews of each included case, we are certain that we only included cases fitting the study criteria.

5 Conclusions

Based on our study population, DIF therapy appears to be beneficial in limiting duration of digoxin toxicity and ICU LOS in patients with digoxin toxicity. Although DIF was appropriately utilized in most cases of digoxin toxicity, there was a relatively high proportion of cases in which DIF treatment was either underutilized or not indicated. Patient selection/treatment bias on the part of the clinical care teams may have impacted study outcomes, particularly ICU LOS. Additionally, digoxin toxicity-related altered mental status, whether stemming from or exacerbated by digoxin toxicity, may be challenging to diagnose, particularly during the ED clinical course when details regarding a patient’s illness, history, and other collateral information may be limited. Education to improve and/or reinforce clinician knowledge on appropriate patient selection for DIF therapy and recognition of digoxin toxicity-related mental status changes may help address these issues.