Serum digoxin concentrations (SDCs) are frequently sampled before completion of drug distribution. If elevated, these concentrations may be misinterpreted, potentially leading to a misdiagnosis of digoxin toxicity.
To determine the frequency of elevated SDCs (>2.6 nmol/L [>2.0 ng/mL]) obtained at appropriate postdosing intervals and to evaluate the frequency of clinically defined digoxin toxicity in patients with elevated SDCs.
The medical records of adult patients with SDCs assayed at 5 general hospitals in North Carolina during a 3-month period (May 1 through July 31, 1996) were prospectively evaluated. Data on SDC, inpatient or outpatient status, and medical or surgical service were collected for all patients. Data on patient demographics, serum chemistry values, indication for digoxin treatment, clinical evidence of digoxin toxicity, and timing of the blood sample relative to administration of the last dose of digoxin were collected for patients with SDCs higher than 2.6 nmol/L (>2.0 ng/mL).
Of 3434 SDCs assayed in 2009 patients, 320 (9.3%) were higher than 2.6 nmol/L (>2.0 ng/mL). Fifty-one (15.9%) of the 320 SDCs were drawn at 6 hours or less following a digoxin dose. Sampling time relative to the digoxin dose could not be determined in 70 (21.9%) of the 320 elevated SDCs, leaving 199 (62.2%) of 320 SDCs in 138 patients evaluable for digoxin toxicity. Eighty-three of the 138 patients had clinical evidence of digoxin toxicity for an overall incidence of 4.1%.
Digoxin toxicity occurs less frequently than historically reported. Continued emphasis needs to be placed on obtaining appropriately timed SDCs.
DIGOXIN IS one of the most frequently prescribed medications1 and has historically been implicated as one of the most common causes of adverse drug reactions.2 In 1971, Beller et al3 published results of a prospective study of patients admitted to a medical service, in which 23% of patients receiving digitalis preparations were considered to have definite digitalis toxicity, and another 6% were judged as having possible toxicity.3 Mortality in the group with definite toxicity was 41%.3 In current clinical practice, however, there is a better understanding of the pharmacokinetics of digoxin,4 more astute monitoring of serum digoxin concentrations (SDCs),5 a prescribing trend toward lower doses, and greater awareness of drug interactions.6 Therefore, it is believed that digoxin toxicity occurs less frequently than previously reported. Furthermore, with the availability of digoxin-specific Fab fragments, mortality is also likely reduced.7
More recently, a retrospective chart review of all patients with heart failure admitted to an urban hospital during 1987 showed that 27 (4.8%) of 563 patients receiving digoxin had a discharge diagnosis of digoxin intoxication.8 These same investigators also reviewed the medical records of all patients with a discharge diagnosis of digoxin intoxication from 1980 through 1988. Of the 176 patients with definite or possible intoxication, only 2 (1.1%) died as a result.8
Despite the fact that digitalis preparations have been used therapeutically for more than 200 years, diagnosis of digoxin toxicity remains difficult. Signs and symptoms associated with toxicity are nonspecific, as are electrocardiographic changes. "Therapeutic" and "toxic" concentrations overlap; at a given SDC, 1 patient may have control of ventricular response with no adverse effects, whereas another may have evidence of toxicity. Therapeutic drug monitoring improves patient care and likely contributes to the suspected decrease in digoxin toxicity9; however, elevated concentrations alone do not constitute toxicity.
Blood samples for therapeutic drug monitoring of digoxin should be drawn at least 6 hours after administration of the last dose of digoxin to ensure adequate distribution between serum and myocardial digoxin concentrations.10,11 "Predistributional" serum samples assayed for digoxin are uninterpretable, and likely represent an unnecessary cost to the health care system. In addition, they may lead to inappropriate diagnoses of digoxin toxicity and unnecessary treatment, further increasing expenditures.
The primary objective of this study was to determine the frequency of elevated SDCs that were obtained at appropriate postdosing intervals. Secondary objectives were to evaluate the frequency of clinically defined digoxin toxicity in patients with elevated SDCs and to characterize outcome in these patients. Finally, we sought to estimate the financial impact of SDCs that were sampled inappropriately, ie, 6 hours or less after administration of the last digoxin dose.
The study sites included 5 general hospitals in North Carolina with a range of 222 to 950 beds. Of these, 1 was proprietary and the other 4 were nonproprietary. Three of the 4 nonproprietary hospitals were in academic health centers.
Using a standardized data collection instrument, SDCs assayed for adult patients from May 1 through July 31, 1996, were recorded along with a patient code, inpatient or outpatient status, and medical or surgical service. For patients with an SDC higher than 2.6 nmol/L (2.0 ng/mL), investigators recorded additional patient information including age, sex, serum chemistry values obtained at the time of the elevated SDC, and indication for digoxin treatment. The presence of symptoms and electrocardiographic changes consistent with digoxin toxicity were recorded. Symptoms included nausea, vomiting, anorexia, fatigue, abdominal pain, diarrhea, headache, dizziness, visual disturbances, confusion, or acute psychoses. Electrocardiographic changes included heart block, sustained or nonsustained ventricular tachycardia, sinus arrest, atrial fibrillation with a ventricular response of less than 50 beats per minute, junctional tachycardia, junctional escape, or sinus bradycardia at a rate of less than 40 beats per minute. If a diagnosis of digoxin toxicity was documented in the medical record by the patient's physician, this was also recorded.
Digoxin dosing data, including dose, route of administration, whether a loading dose was administered, and the timing of the blood sample relative to the last dose of digoxin (>6 or ≤6 hours after dosing), were also recorded on the standardized form. To minimize variability in data collection, medical records (patient chart, medication administration record, outside hospital records) were the primary source of information. Patients were followed up for outcome throughout their hospitalization or until the end of the study period, whichever occurred first. The likelihood of digoxin toxicity was assessed independently by 2 of us (K.M.W. and J.H.P.) based on the presence or absence of signs and symptoms adapted from those previously reported by other investigators.8 Each event was classified according to the following criteria: definite toxicity—events associated with symptoms and/or arrhythmias suggestive of digoxin intoxication that resolved after discontinuation of digoxin, treatment with digoxin immune Fab fragments, or a decrease in digoxin dose; possible toxicity—events associated with symptoms and/or arrhythmias suggestive of digoxin intoxication in the absence of documented resolution after discontinuation of digoxin; no toxicity—events associated with no signs or symptoms of digoxin intoxication; and uninterpretable—events associated with patient discharge prior to follow-up and/or medical record unattainable or blood samples obtained 6 hours or less after the previous dose of digoxin.
Serum digoxin concentrations were measured using immunoassay technology. Three centers used the Abbott TDx system (Abbott Laboratories, Abbott Park, Ill; assay sensitivity range, 0.3-6.4 nmol/L [0.2-5.0 ng/mL]), 1 center used the Technicon Immuno-1 system (Bayer, Tarrytown, NY; assay sensitivity range, 0.05-7.70 nmol/L [0.04-6.00 ng/mL]), and 1 used the ACMIA DGN A (DuPont Co, Wilmington, Del; assay sensitivity range, 0.26-5.10 nmol/L [0.2-4.0 ng/mL]).
Nominal data were analyzed by using χ2 or Fisher exact tests as appropriate. Continuous variables were analyzed by analysis of variance with multiple t tests and Bonferroni correction for all significant findings. All tests were 2-tailed. Computerized support for data analysis included the Statistical Analysis System (SAS), version 6.11 (SAS Institute, Cary, NC). An a priori level of significance was set at α=.05.
As a secondary objective of this study, we planned to determine the financial impact of blood samples drawn too soon after administration of the dose of digoxin. The laboratory at each study site was contacted. Actual hospital cost and patient charge for determination of an SDC were requested. Because hospital cost was calculated using different parameters, wide variability existed among the study sites. Thus, we report the average hospital charge to the patient for SDC assays.
A total of 3434 SDCs were assayed at the 5 participating hospitals in 2009 adult patients during the 3-month study period (Table 1). More than 80% of concentrations were obtained from inpatients. Elevated SDCs (>2.6 nmol/L [>2.0 ng/mL]) occurred in 320 (9.3%) of the 3434 samples analyzed. Fifty-one (15.9%) of the 320 elevated SDCs were drawn 6 hours or less after administration of the most recent dose of digoxin (Table 2). Sampling time relative to the last dose of digoxin could not be determined from the medical record for an additional 70 (21.9%) of the 320 elevated SDCs. Thus, 199 (62.2%) of 320 elevated SDCs were documented to have been sampled at an appropriate postdosing interval and were evaluable for presence or absence of digoxin toxicity. These 199 appropriately timed elevated SDCs were obtained from 138 patients.
Eighty-three of the 138 patients had at least 1 sign, symptom, or electrocardiographic change suggestive of digoxin toxicity (definite or possible), for an overall incidence of 4.1% (83/2009). The remaining 55 patients were asymptomatic and classified as "no toxicity." Two of us (K.M.W. and J.H.P.) independently assessed the likelihood of digoxin toxicity (definite or possible) in each of the 83 patients with events suggestive of digoxin toxicity and agreed on 78 after the initial review. The remaining 5 events were independently reviewed a second time, and agreement was reached in each case.
Of the 83 symptomatic patients, 33 were classified as having definite and 50 as having possible digoxin toxicity. Age, sex, and indication for digoxin treatment were not significantly different among the 3 groups (Table 3). Cases of definite digoxin toxicity were associated with a significantly higher mean SDC than those with possible or no toxicity (Table 4).
Symptoms of digoxin toxicity and characteristic electrocardiographic changes are presented in Table 5. The 55 asymptomatic patients were excluded from this table. Nausea was the most common symptom associated with digoxin toxicity. Definite toxicity was associated with significantly more episodes of anorexia and diarrhea (P<.05) than possible toxicity. Electrocardiographic changes occurred significantly more often in cases of definite toxicity. Heart block, bradycardia, and junctional tachycardia were significantly more frequent in cases of definite toxicity compared with cases of possible toxicity (P<.05).
A physician diagnosis of digoxin toxicity had been documented in the medical record of 61% (20/33) of the patients with definite toxicity, occurring significantly more frequently than patients with possible (32% [16/50]; P<.01) or no toxicity (0/55; P<.01) groups (Table 6). Death due to any cause occurred in 20 (14.5%) of 138 patients with appropriately sampled elevated SDCs; however, death occurred with equal frequency in each patient group, regardless of the presence or absence of digoxin toxicity. Seven patients with appropriately timed elevated SDCs received digoxin-specific Fab fragments, 3 of whom died despite this administration.
Based on an average patient charge of $51.50 (range, $35-$63) per SDC assay, the annualized estimate of total patient charge for uninterpretable SDCs ranged from $10,506 to $24,926 (Figure 1). The annualized estimated charges at individual institutions ranged from $412 to $9064.
Our results confirm the popularity of therapeutic drug monitoring in managing patients receiving digoxin and underscore the need for better education of health care professionals with respect to obtaining an appropriately timed SDC. Digoxin will likely continue to be one of the most commonly prescribed drugs as recent, well-designed clinical trials have further substantiated its efficacy in the treatment of heart failure.12-14 All new drugs for treatment of heart failure are studied with background therapy that usually includes digoxin, angiotensin-converting enzyme inhibitors, and diuretics. Moreover, digoxin continues to be the most widely prescribed medication to control ventricular response in atrial fibrillation.15 Therefore, it appears that digoxin will continue to be an integral part of treatment for heart failure and atrial fibrillation for the foreseeable future.
Although the necessity of drawing an appropriately timed blood sample after the completion of drug distribution (at least 6 hours after the dose) is well established, we found that a substantial number of SDCs were sampled inappropriately. At least 51 (15.9%) and as many as 121 (37.8%) of the 320 elevated SDCs were measured 6 hours or less following a dose of digoxin. These findings are consistent with those of Kumana et al,16 who monitored all SDCs during a 6-month period and found that 54 (31.4%) of 172 samples were drawn less than 6 hours after the dose. They observed that laboratory tests are frequently ordered merely because they are available, with little consideration for interpretation of the results. Concentrations drawn too soon after a digoxin dose represent an unnecessary financial burden to the health care system. In our study, the estimated annual expenditure for uninterpretable SDCs, based on charge to the patient, ranged from $10,506 to $24,926. Perhaps laboratories should mandate that all blood samples drawn for concentration monitoring indicate not only the time that the sample was drawn but also the time of the most recent digoxin dose. Samples obtained 6 hours or less after the dose should not be assayed until the clinical situation is verified with the requesting clinician.
In addition to the financial impact of inappropriately timed samples, predistributional SDCs that are elevated may contribute to an overestimation of the frequency of digoxin toxicity. If the concentration were the only criterion to establish toxicity, falsely inflated estimates of digoxin toxicity would arise and patients could be misdiagnosed as having digoxin intoxication. This may lead to inappropriate admissions and treatment and further misuse of health care resources.
Importantly, our study also supports the declining incidence of digoxin toxicity. Three hundred twenty of the 3434 concentrations were higher than 2.6 nmol/L (>2.0 ng/mL), 199 (5.8%) of which were sampled appropriately and evaluable for the presence or absence of digitalis toxicity. Signs or symptoms suggestive of digoxin intoxication were present in 33 patients classified with definite toxicity (1.6%) and in 50 patients with possible toxicity (2.5%), corresponding to an overall incidence of 4.1% (83/2009). This is lower than reported in the 1970s, and more similar to reports from the late 1980s.3,5,8,17-19
Our overall incidence of digoxin toxicity (4.1%) is similar to that found by Mahdyoon et al8 in a retrospective review. Of 994 patients admitted to their hospital in 1987 with a diagnosis of heart failure, 563 (56%) were receiving digoxin, and in 27 (5%), the diagnosis of digoxin intoxication was made by their clinicians. A diagnosis of definite digoxin toxicity could only be made in 4 (0.7%) of 563 patients and in another 16 (2.8%) the diagnosis could not be excluded, suggesting an overall incidence of 3.5%. In the recently published Digitalis Investigation Group study,14 the incidence of hospitalization for presumed digoxin toxicity was 0.9% in the placebo group and only 2% in the digoxin group. However, these figures do not account for patients who may have had clinical evidence of digoxin intoxication but did not require hospitalization.
It has been widely accepted that deteriorating renal function and electrolyte abnormalities (hypokalemia, hypomagnesemia, hypercalcemia) predispose patients to digoxin toxicity.20,21 In our study, none of these factors differed significantly (P>.05) among the 3 groups, although serum urea nitrogen and serum creatinine levels tended to be higher in patients with definite or possible digoxin intoxication than those without intoxication. However, serum urea nitrogen and serum creatinine are not the best predictors of renal function, and creatinine clearance would have likely been more indicative.22
Certain limitations of this study deserve comment. Our overall estimate of digoxin toxicity of 4.1% was potentially influenced by several factors that may have led to either an underestimation or an overestimation of this figure. Identification of subjects by elevated SDC may have underestimated the true incidence of digoxin toxicity, since patients may have had signs or symptoms of toxicity despite concentrations that were within the "therapeutic" range.19 Also, since we required documentation of all data in the patient's medical record, the sampling time relative to administration of the last dose of digoxin was unknown for 70 of 320 concentrations. These concentrations were eliminated from the toxicity analysis, potentially underestimating true incidence as well. In our analysis, we did not identify the actual number of patients who were taking digoxin during the study period at each hospital. Therefore, our toxicity estimate was based on the number of patients who had SDCs measured rather than the total number of patients receiving digoxin. Last, financial impact of inappropriately sampled SDCs was estimated using charge rather than cost data.
In summary, digoxin toxicity occurs less frequently than historically reported. Despite the fact that digoxin has been in clinical use for more than 200 years, improvement in therapeutic drug monitoring is still needed. Serum concentrations continue to be obtained without regard to digoxin pharmacokinetics, leading to uninterpretable SDCs that are costly. The time of the blood sampling as well as the most recent digoxin dose should be documented on all patient requisitions for analysis of SDCs and laboratories should not process samples drawn within 6 hours of administration of the previous digoxin dose unless exceptional clinical circumstances (ie, strong suggestion of digoxin toxicity) exist.
In addition to being cognizant as to the appropriate timing of the sample, we suggest that careful consideration be given to the need for obtaining any SDC. Although the majority of SDCs surveyed were drawn appropriately, we question the necessity of obtaining this number of SDCs and believe that the practice of routinely monitoring SDCs in otherwise stable patients should be closely reexamined. Clinical scenarios where an appropriately timed SDC may be necessary include (1) once a patient has reached steady state with a new digoxin dose, (2) significant change in renal function, (3) addition or discontinuation of a potentially interacting drug, or (4) confirmation of suspected digoxin toxicity in a patient with signs or symptoms and/or electrocardiographic changes consistent with this diagnosis. Following these simple guidelines would result in a considerable reduction in the number of samples assayed and a concomitant reduction in costs, and would provide more meaningful drug concentration data to be interpreted along with the patient's clinical information.
Accepted for publication April 13, 1998.
This research was supported in part by an unrestricted grant from Glaxo Wellcome Inc, Research Triangle Park, NC.
Reprints: J. Herbert Patterson, PharmD, FASHP, FCCP, BCPS, University of North Carolina School of Pharmacy, Beard Hall CB 7360, Chapel Hill, NC 27599-7360 (e-mail: firstname.lastname@example.org).
J Adverse drug reactions during hospitalization. CMAJ.
1967;971450- 1457Google Scholar
WB Digitalis intoxication: a prospective clinical study with serum level correlations. N Engl J Med.
1971;284989- 997Google ScholarCrossref
AD Digitalis: an update of clinical pharmacokinetics, therapeutic monitoring techniques and treatment recommendations. Clin Pharmacokinet.
1988;15165- 179Google ScholarCrossref
J Reduction of digoxin toxicity associated with measurement of serum levels. Ann Intern Med.
1974;80516- 519Google ScholarCrossref
BF Pharmacokinetic interactions with digoxin. Clin Pharmacokinet.
1988;15227- 244Google ScholarCrossref
TW Treatment of 150 cases of life-threatening digitalis intoxication with digoxin-specific Fab antibody fragments: final report of a multicenter study. Circulation.
1990;811744- 1752Google ScholarCrossref
M The evolving pattern of digoxin intoxication: observations at a large urban hospital from 1980 to 1988. Am Heart J.
1990;1201189- 1194Google ScholarCrossref
DJ Influence of serum digoxin concentration measurements on frequency of digitoxicity. Clin Pharmacol Ther.
1974;16284- 287Google Scholar
PH Digoxin. Evans
WJeds Applied Pharmacokinetics Principles of Therapeutic Drug Monitoring.
3rd ed. Vancouver, Wash Applied Therapeutics Inc1992;chap 20Google Scholar
AH Digoxin. Schumacher
GEed Therapeutic Drug Monitoring.
Norwalk, Conn Appleton & Lange1995;chap 15.Google Scholar
MK Randomized study assessing the effect of digoxin withdrawal in patients with mild to moderate chronic congestive heart failure: results of the PROVED trial. J Am Coll Cardiol.
1993;22955- 962Google ScholarCrossref
et al. Withdrawal of digoxin from patients with chronic heart failure treated with angiotensin-converting-enzyme inhibitors. N Engl J Med.
1993;3291- 7Google ScholarCrossref
SDigitalis Investigation Group, The effect of digoxin on mortality and morbidity in patients with heart failure. N Engl J Med.
1997;336525- 533Google ScholarCrossref
JS Atrial fibrillation: a review of mechanism, etiology and therapy. Clin Cardiol.
1997;20640- 650Google ScholarCrossref
M Audit exposes flawed blood sampling for "digoxin levels." Ther Drug Monit.
1992;14155- 158Google ScholarCrossref
L Studies on digitalis, XIII: a prospective study of 649 patients on maintenance treatment with digitoxin. Am Heart J.
1977;93434- 443Google ScholarCrossref
et al. A prospective survey of the incidence of cardiac intoxication with digitalis in patients being admitted to hospital and correlation with serum digoxin levels. Aust N Z J Med.
1973;3279- 284Google ScholarCrossref
et al. Digoxin serum concentration measurement in patients with suspected digitalis-induced arrhythmias. J Cardiovasc Pharmacol.
1980;2319- 329Google ScholarCrossref
U Correlation between manifestations of digoxin toxicity and serum digoxin, calcium, potassium, and magnesium concentrations and arterial pH. Br Med J (Clin Res Ed).
1983;2861089- 1091Google ScholarCrossref
SK Digoxin toxicity and electrolytes: a correlative study. Acta Cardiol.
1983;38115- 123Google Scholar
AJ A concurrent audit of high digoxin plasma levels. Clin Pharmacol Ther.
1994;55353- 358Google ScholarCrossref