New–onset Atrial Fibrillation: When Is Admission Medically Justified?

114 ACADEMIC EMERGENCY MEDICINE FEB 1996 VOL 3/NO 2

New-onset Atrial Fibrillation: When Is Admission Medically Justified? Beth Mulcahy, MD, Wendy C. Coates, MD, Philip L. Henneman, MD, Roger J . Lewis, MD, PhD

I ABSTRACT

Objective: To determine 1) the percentage of patients with new-onset atrial fibrillation for whom admission is medically justified and 2) whether those patients for whom hospitalization is medically justified can be reliably identified in the ED.

Methods: A retrospective, descriptive cohort analysis was performed using consecutive adult patients with new-onset atrial fibrillation seen in an urban, county hospital ED from 1987 through 1992. Admissions were categorized as medically justified if patients were hypotensive (systolic blood pressure <90 mm Hg), had a diagnosis other than new-onset atrial fibrillation that warranted admission, or had a significant complication during the ED stay or during the subsequent hospitalization. The need for admission was considered to have been apparent during the ED evaluation if the patient fulfilled the above criteria for a medically justified admission while in the ED.

Results: Admission was medically justified for 143 of the 216 patients (66%; 95% CI 60-71%) admitted to our institution. For those patients whose admissions were medically justified, the most common concurrent conditions were congestive heart failure and chest pain suggestive of myocardial ischemia. The need for admission was apparent during the ED evaluation for 140 of the 143 patients (98%; 95% CI 94-100%) whose admissions were categorized as medically justified.

Conclusion: Approximately one third of patients with new-onset atrial fibrillation may not require admission to the hospital. Most patients (98%) for whom admission is medically justified can be reliably identified during the ED evaluation.

Key words: atrial fibrillation; hospital admission; emergency department; complications.

Acad. Emerg. Med. 1996; 3:114-119.

I Atrial fibrillation is one of the most common dys- rhythmias encountered in clinical practice.’ Patients presenting to an ED with atrial fibrillation of new onset are usually admitted to the hospital to exclude the diagnosis of acute myocardial infarction (AMI) and to monitor for serious cardiovascular complications.2 This practice originally stemmed from studies performed in the 1930s demonstrating an association of new-onset

From the Department of Emergency Medicine, Harbor- UCLA Med- ical Center, Torrance, CA, and the UCLA School of Medicine, Los Angeles, CA (BM, WCC, PLH, WL).

Received: May 2, 1995; revision received: August 14, 1995; accepted: Auglrst 16, 1995; updated: August 22, 1995.

Prior presentation: American College of Emergency Physicians Re- search Forum, San Diego, CA, February 1994.

Address for correspondence and reprints: Philip L. Henneman, MD, Department of Emergency Medicine, Box21, Harbor- UCLA Medical Center, lo00 West Carson Street, P. 0. Box2910, Torrance, CA 90509- 2910. Fax: 310-782-1763.

atrial fibrillation with rheumatic heart disease, athero- sclerotic heart disease, and thyroid d i s e a ~ e . ~ Yet, the incidence of AM1 and serious cardiovascular complications in patients with new-onset atrial fibrillation has been reported to be 20%,2-4 leading some to conclude that select pptients at “low risk” for serious complications may not require ho~pitalization.~

Many patients with new-onset atrial fibrillation have concurrent cardiac conditions that, in and of them- selves, necessitate inpatient management.2-4 Previous studies have identified congestive heart failure, myocardial infarction (MI), and typical ischemic chest pain as being associated with subsequent complications in patients with new-onset atrial f ibri l lati~n.~-~ The majority of such patients require hospitalization whether or not they have new-onset atrial fibrillation.

The important question for the practicing emergency physician is whether a patient with new-onset atrial fibrillation, who has no other indication for admission, requires hospitalization. We sought to determine the prevalence of conditions or presentations war-

Admission in New-onset Atrial Fibrillation, Mulcahy ef af. 115

ranting hospital admission in emergency patients with new-onset atrial fibrillation and the characteristics of patients with new-onset atrial fibrillation who presum- ably could be released home safely from the ED.

I METHODS

Study Design

A retrospective, descriptive, cohort analysis was performed using consecutive case series of emergency patients presenting with new-onset atrial fibrillation. The study was approved by the institutional review board at Harbor-UCLA Medical Center.

Population and Setting Harbor-UCLA Medical Center is an urban county

hospital with an annual adult E D census of >8O,OOO visits. The adult E D is staffed by emergency medicine (EM) and internal medicine residents who are super- vised by EM attending physicians. It is standard practice at Harbor-UCLA Medical Center to admit all patients with new-onset atrial fibrillation.

The ED logbooks from January 1, 1987, through December 31, 1992, were reviewed, and the names of all patients with a diagnosis of atrial fibrillation were collected. The logbooks contain the name, hospital identification number, chief complaint, E D diagnosis, and disposition for all patients seen in the ED. Medical records were reviewed to determine whether the patient had an ED diagnosis of new-onset atrial fibrillation (defined below). All such patients were entered in the study.

Measurements Medical records were reviewed retrospectively by a

single reviewer (BM). and a data sheet was completed containing demographic information, E D evaluation, hospital course, final diagnosis, and disposition of all adult patients (age >15 years) with a diagnosis of new- onset atrial fibrillation. Data sheets from patients admitted to Harbor-UCLA Medical Center were evaluated retrospectively by two of the authors (PLH, WCC) to determine whether the admissions were medically justified. Disagreements regarding the categorization of admissions were resolved by a third reviewer (RJL), and a majority decision prevailed.

Defmi t ions New-onset atrial fibrillation was defined as the pres-

ence of atrial fibrillation in a patient who had not been previously diagnosed as having atrial fibrillation.

Admissions were categorized as medically justified

if a patient was hypotensive in the ED (systolic blood pressure <90 mm Hg), had an additional diagnosis other than new-onset atrial fibrillation with a clinical presentation that warranted admission [e.g., chest pain consistent with AMI, congestive heart failure (CHF) necessitating admission, or noncardiac diseases necessitating admission], or had a significant complication while in the ED or during the subsequent hospitalization. Pa- tients with CHF were considered to require admission if they: 1) were in moderate to severe distress; 2) had new or abrupt onset of their symptoms; 3) developed CHF as a result of renal failure, hyperthyroidism, or uncontrolled hypertension; or 4) required prolonged treatment to correct fluid overload.

Often patients were found to fulfill multiple criteria that would justify their admission to the hospital. When categorizing the reason justifying such a patient’s admission to the hospital, the first criterion fulfilled was chosen. In most cases, the criterion justifying admission was present during the E D evaluation, in which case that criterion was chosen. In some cases, the patient’s only criterion justifying admission developed after admission to the hospital.

The medical justification for admission was considered to be apparent during the ED evaluation if the patient was hypotensive in the E D or if he or she had a diagnosis other than new-onset atrial fibrillation that warranted admission. Patients who were not hypotensive in the E D and were admitted only for their new- onset atrial fibrillation, and later had a significant complication during the hospitalization, were categorized as having a medically justified admission that was not apparent during the ED evaluation.

Laboratory and radiographic results in the E D were considered to be clinically significant if they assisted in diagnosing the etiology of the atrial fibrillation, resulted in admission, or met predefined criteria for critical values. These critical values were: hematocrit <0.30 (30%); white blood cell count >10 x 109/L in an afebrile patient with an infection; sodium e l 3 0 mmol/L; po- tassium <3 mmol/L or >6 mmoYL; bicarbonate <19 mmol/L; BUN >14 mmol/L (40mg/dL); creatinine >I77 pmol/L (2 mg/dL); arterial blood gas (ABG) with a pH c7.35, a PcoZ >45 torr, o r a PO, <60 tom, and creatine kinase values (CK and CK-MB) diagnostic of AM1 [CK >240 U/L, CK-MB ratio >0.04 (4%)]. The hematologic abnormalities were collectively referred to as abnormalities on the complete blood count (CBC). The elec- trolyte, BUN, and creatinine abnormalities were collectively referred to as CHEM-7 abnormalities.

Data Analysis Data were entered into a database (Paradox 3.5,

Borland, Scotts Valley, CA) and analyzed using a com-


I TABLE 1 Population Demographics and Presenting Charac- teristics (n = 229)

Gender Men Women

Age (years) Mean 2 SD Median Interquartile range Range

Race Hispanic White Black Asian Other

Presentation. Shortness of breath Palpitations Chest pain

116 (51%) 113 (49%)

61 -c 16 61

50-71 16- 101

78 (34%) 75 (33%) 37 (16%) 30 (13%) 9 (4%)

127 (55%) 94 (41%) 75 (33%)t

Duration of symptoms (days) Mean 2 SD 7 2 13 Median 2 Interquartile range 1-7 Range 0-90

'Patients may have had more than one presenting complaint. tChcst pain refers to any chest pain during patient presentation.

even i f the quality of the pain was not suggestive of acute myocardial infarction.

I TABLE 2 E D Diagnostic Testing, Treatment, and Disposition ( n = 229) .............................................................................................

Number Tcstcd (9'0,) Number Clinically Diagnostic testing' Signijicanrt (%)

CHEM-7 226 (99%) 42 (19%)

Arterial blood gas 95 (41%) 25 (26%) Chest radiography 218 (95%) 80 (37%)

CBC 225 (98%) 5 (2%)

CK or CK-MB 163 (71%) 2 (1%)

Treatment Digoxin Diuretic Calcium-channel blocker Beta blocker Electrical cardioversion

Disposition Admitted Transferred Released home$ Left against medical advice

Number Treated (8) 189 (83%) 48 (21%) 45 (20%) 39 ( 1 7 4 ) 13 (6%)

Number (%) 216 (94%)

7 (3%) 5 (2%) 1 ( < I % )

'Terms are explained in "Definitions" in the Methods section of

*Clinical significance is defined for each variable in the Methods

SAt the discretion of the treating physician.

the text.

section.

mercially available statistical package (SAS 6.07, SAS Institute, Cary, NC).5.6 Data are expressed as mean ? SD and median with interquartile ranges (IQRs). Dif- ferences in continuous data were evaluated using the Wilcoxon rank sum test; categorical data were analyzed using chi-square or Fisher's exact test, where appro- priate .7 Logistic regression analysis was used to assess the independent association of specific variables with medical justification for admission.8 A p-value cO.01 was considered to be significant because of the multiple statistical tests used in the analy~is .~

I RESULTS

During the six-year study period, 394 consecutive patients had atrial fibrillation as one of their admission diagnoses. Medical records were available for review for 379 (96%) of these patients (six had been destroyed and nine were incomplete or not available). Of these 379 patients, 229 had atrial fibrillation of new onset and were entered into the study. Patient demographics and clinical presentation are shown in Table 1.

The ED diagnostic evaluation, treatment, and outcome for the 229 patients are listed in Table 2. Thirteen of the patients were transferred, were released home, or left against medical advice. Eighty-eight of the remaining 216 patients (41%) admitted to Harbor-UCLA Medical Center were admitted to an intensive care unit. Patients were hospitalized at Harbor- UCLA Medical Center for a mean of 7 2 5 days (median 6 days; IQR 3-9 days). One hundred twelve of the admitted patients (52%) converted to normal sinus rhythm during their stays.

Thirty patients (14%) who were admitted to our hospital developed 36 complications after admission to the hospital: nine died; six developed infections; five developed gastrointestinal (GI) bleeding; four had re- currence of the atrial fibrillation with a fast ventricular response; three suffered cardiac arrest but survived; three developed symptomatic bradycardia; two developed acute renal failure; one developed a stroke; one developed symptomatic hypotension; one developed chest pain; and one developed a drug fever that resulted in a second hospital admission. Of the nine patients who died, five had infections with systemic manifestations, one had an AMI, one had cardiogenic shock, one patient arrested after developing chest pain, and one had metastatic cancer and was not to be resuscitated. Of the 30 patients who had in-hospital complications, 25 (83%) had CHF ( n = 15), chest pain or a recent history of chest pain (n = 9), evidence of infection (n = 5), or hypotension (n = 4) in the ED.

Complications for nine of the 30 patients (30%) were iatrogenic to some extent: two bradycardias were related to treatment (digoxin or propranolol); two cardiac

........................................................................................

Admission in New-onset Atrial Fibrillation, Mulcuhy et ul. 117

arrests were believed to be due to drug toxicity (quinidine-induced torsade de pointes and procainamide-induced ventricular tachycardia); one recurrent atrial fibrillation occurred during an exercise treadmill test; one stroke occurred after mitral valve replacement; one hypotension was due to excessive diuresis; one patient developed quinidine fever; and one patient developed renal failure following IV contrast administration.

One hundred forty-three of the 216 admissions (66%; 95% CI 60-71 %) were categorized retrospectively as medically justified (Table 3). Patients who had CHF, hypotension. or chest pain consistent with AM1 ac- counted for 69% (98/143) of these admissions. The 17 justifiably admitted patients in the “other” category in Table 3 had: ECG changes consistent with AM1 without chest pain (n = 3); GI bleeding (n = 3); altered mental status (n = 2); and acute renal failure, asthma, deep venous thrombosis, hip fracture, hyperkalemia, mural thrombus, pancreatitis, pericardial effusion, and un- controllable ventricular rate (n = 1 each). Neither the duration of symptoms nor the time until conversion to normal sinus rhythm after the patient left the E D was related to whether an admission was categorized as medically justified. Older patients were more likely to have their admissions categorized as medically justified

The medical justification for admission was apparent during the ED evaluation for 140 of the 143 (98%: 95% CI 94- 100%) admissions categorized as justified. There were three patients (2%) who did not appear to require admission while they were in the ED, but subsequently developed complications as inpatients: 1) a 72-year-old developed chest pain with ST-segment elevation on hospital day 6, but MI was ruled out and the patient had a normal coronary artery catheterization; 2) an 86-year- old died on hospital day 3 after developing GI bleeding (not related to anticoagulation) and recurrent bradycardia that required pacemaker placement and intu- bation; and 3) a 63-year-old with metastatic cancer, who was not to be resuscitated. died on hospital day 6.

(p = 0.Oool).

TABLE 4 Etiology of New-onset Atrial Fibrillation ( n = 216) ................................................................................................... . . . .

I TABLE 3 Reasons Admissions Were Retrospectively Cate- gorized as Medically Justified (n = 143) .............................................................................................................

Medically Justified Admission’ Number (%)

Congestive heart failure Chest pain-rule out AMlt Hypotension$ Infection5 Syncope Stroke Diabetes out of control7 Cancer Other

47 (33%) 36 (25%) 15 (10%) 10 (7%) 6 (4%) 4 (3%) 4 (3%) 4 (3%)

17 (12%)

‘For descriptive purposes, all patients were assigned a single justification for admission (see text).

tAcutc myocardial infarction. $Systolic blood pressure <90 mm Hg. IRefers to an infection with systemic manifestations. Whree patients had diabetic ketoacidosis and the fourth had a

glucose level of 37 mmoVL (669 mg/dL).

The etiologies of new-onset atrial fibrillation for the 143 patients whose admissions were categorized as medically justified and for the 73 patients whose admissions were not categorized as medically justified are listed in Table 4. The patients who had CHF (not due to valvular disease) as the cause of atrial fibrillation were more likely to require admission than were the other patients (p < 0.01), and the patients who had atrial fibrillation related to alcohol use were less likely than the other patients to have their admissions categorized as medically justified (p = 0.02). The patients whose admissions were categorized as medically justified stayed in the hospital a mean of 8 k 5 days (median 6 days; IQR 4-9 days); the patients whose admissions were not categorized as medically justified stayed a mean of 5 5 5 days (median 4 days; IQR 2-6 days; p = 0.OOOl).

I DISCUSSION

To our knowledge, this is the largest study to date ex- amining the need for admission of patients with new-

..............................................................................................

Admission Categorized as

Etiology (n = 143) (n = 73)

Valvular condition 32 (22%) 13 (18%) CHF‘ 27 (19%) 4 (6%) Ethanol abuse 8 (6%)

Admission Not Categorized as Medically Justified Medically Justified

Unknown 49 (34%) 32 (44%)

11 (15%) Hyperthyroidism I (5%) 6 (8%) Myocardial infarction 9 (6%) 0 (0%) CADt 4 (3%) 3 (446) Other 7 (5%) 4 (6%)

‘Congestive heart failure not due to valvular disease. tCoronary artery disease without acute myocardial infarction or congestive heart failure.


onset atrial fibrillation. One third of our patients who had new-onset atrial fibrillation were stable in the ED and, based on a retrospective review of their other medical conditions and their hospital courses, could have been safely treated as outpatients. Ninety-eight percent of the 143 remaining patients whose admissions were medically justified could have been identified in the ED.

Prior studies of patients with new-onset atrial fibrillation have found vastly different percentages of patients who required admission. In the study by Fried- man et al.,2 83% of the patients met their high-risk characteristics (jugular venous distention, pulmonary rales, peripheral edema, history of CHF, and failure to convert to sinus rhythm within six hours) and 18% had complications. However, in the study by Shlofmitz et al.,3 only 5% had significant morbidity. This variability probably results from differences in inclusion criteria for the study populations, differences in definitions of endpoints, and differences in study designs (retrospective vs prospective). Friedman et al. excluded patients who obviously required admission and those who had prolonged symptoms. Both we and Shlofmitz et al. have studied consecutive patients with new-onset atrial fi- b r i l l a t i~n .~ .~ Friedman et al. and Shlofmitz et al. both defined patient risk groups using observed complica- ti on^.,^.^ while we used a more conservative definition to determine which admissions were medically justified, based on the patient's presentation, admitting diagnoses, and complications occurring during the hospital course. The percentage of patients in the prior works whose admissions would have been considered medically justified by our current criteria is unknown. Like Friedman et al.,z we found CHF to be a risk factor for subsequent complications. The fraction of patients found by Friedman et al. to be high risk (83%) and the fraction we found to have medically justified admissions (66%) are qualitatively similar.

The incidence of AM1 in patients with new-onset atrial fibrillation ranges from 1% to l l%.2-4 Although chest pain suggestive of AM1 was the second most common reason for admissions to be categorized as medically justified, occurring in 25% (36/143), AM1 occurred in only 6% (91143) of the patients who had justified admissions and in only 4% (91216) of all the admitted patients. Each patient who had AM1 had a clinical pic- ture consistent with AMI, in addition to the atrial fibrillation, and therefore met our criteria for justified admission during the E D evaluation. Four had ongoing chest pain in the ED, two were hypotensive on presentation, two were in severe CHF, and one had ST-seg- mcnt elevation on the initial ECG.

Many patients with new-onset atrial fibrillation convert to normal sinus rhythm early in their treatment. In our study, 112 admitted patients (52%) converted to

sinus rhythm before hospital discharge, 95 of the 112 (85%) within the first two days. In the studies by Fried- man et al.* and Shlofmitz et 75% and 82% of the patients, respectively, converted to sinus rhythm during their hospitalizations, It is unclear why fewer of our patients converted to sinus rhythm, although our patients as a group may have been in atrial fibrillation for a longer interval prior to presentation. Although it has been recommended that patients who do not convert to sinus rhythm in a timely fashion receive anticoagu- lafion therapy for three weeks prior to elective cardioversion,l0 this can be instituted on an outpatient basis.

It is important to note that admission to the hospital was not completely benign for the patients in this study. Thirty percent (9130) of the patients who had in-hospital complications suffered complications that were at least partially iatrogenic.

Of greatest concern is the patient with new-onset atrial fibrillation who is released frqm the E D and later suffers serious complications. In ou; study, there were three patients (2% of the 143 patients whose admissions were categorized as medically justified) whose criteria for admission developed after leaving the ED, and therefore would not have been identified as high risk in the ED. Although two patients died; one death was expected from terminal cancer, and the second patient had a history of peptic ulcer disease and died of GI bleeding complicated by recurrent bradycardia. The third patient had chest pain on hospital day 6 and subsequent normal coronary angiography .

I LIMITATIONS AND FUTURE OUESTIONS - . , . . . . , . . . . . , . . . . . . . . ,

Our study has several limitations. The retrospective design of the study may have affected data acquisition. Patients with new-onset atrial fibrillation may not have been identified as such in the ED logbooks. Medical records were not available for 4% (151339) of potential study patients identified by the logbooks. Pertinent signs and symptoms may not have been recoided in the hospital chart, and medical treatment was not standard- ized. Although most patients underwent similar medical workups, not all patients received CK determinations, and 32% of the admitted patients did not have echo- cardiography. Outcome information was incomplete or unavailable for the 13 patients not admitted to our hospital (4% of total patients). The determination of a "medically justified" admission was retrospective and subjective, and in a small number of cases (n < 10) a third reviewer was needed. Although our values for critically important laboratory values were determined prospectively, they were subjective. Finally, our county hospital in southern California cares for a population that may be different from that seen in other areas.

Atrial fibrillation may not have been of recent onset

Admission in New-onset Atrial Fibrillation, Mulcahy et al. 119

for some patients, but rather "newly diagnosed." Eleven percent of the patients had symptoms for >14 days. The duration of symptoms may not be reliably related to the duration of atrial fibrillation in all patients, thus some patients in our study may have had atrial fibrillation for an extended period.

Our data suggest that adults who present to the E D with new-onset atrial fibrillation and no other indication for admission to the hospital can be managed safely as outpatients if they are hemodynamically stable and the ventricular rate can be controlled in the ED. Patients with concurrent medical conditions, especially CHF, chest pain consistent with AMI, infections with systemic manifestations, and hypotension, are at highest risk for in-hospital complications. Validation of a selected outpatient management approach for new-onset atrial fibrillation should incorporate a prospective trial that lon- gitudinally monitors released patients with new-onset atrial fibrillation who have no other indication for admission to the hospital.

I CONCLUSION

Our study suggests that approximately one third of patients with new-onset atrial fibrillation may be managed safely as outpatients. Most patients (98%) whose admission is medically justified can be reliably identified during the ED evaluation.

The authors thank Larry Stock, MD, for his help in data collection, and the staff at Harbor-UCLA who cared for these patients and

helped find their medical records. The authors also thank A. F. Parlow, MD. for the use of the VAX computer network at the Har- bor-UCLA Research and Education Institute. Dr. Lewis was sup- ported in part by a Career Development Award from the Emergency Medicine Foundation.

REFERENCES

1. Podrid PJ. Falk RH. Management of atrial fibrillation-an overview. In: Falk RH. Podrid PJ. Atrial Fibrillation. Mecha- nisms and Management. New York: Raven Press, I W , pp 389- 411.

2. Friedman HZ. Goldberg SF, Bonema JD. C r a g DR, Hauser AM. Acute complications associated with new-onset atrial fibrillation. Am J Cardiol. 1991; 67:437-9.

3. Shlofmitz RA. Hirxh BE, Meyer BR. New-onset atrial fibrillation: is there need for emergent hospitalization? J Gen Intern Med. 1986; 1:139-42.

4. Friedman HZ. Weber-Bornstein N , Deboe SF. Mancini GBJ. Cardiac care unit admission criteria for suspected acute myocardial infarction in new-onset atrial fibrillation. Am J Cardiol. 1987; 59: 866- 9.

5 . SAS Institute Inc. SAS/STAT User's Guide, Version 6, Fourth Edition, Volume 1. Cary, NC: SAS Institute Inc.. 1989.

6. SAS Institute Inc. SAS/STAT User's Guide, Version 6, Fourth Edition, Volume 2. Cary, NC: SAS Institute Inc., 1989.

7. Dixon WJ. Massey FJ Jr. Introduction to Statistical Analysis. Fourth Edition. New York: McGraw-Hill, 1983.

8. Hosmer DW. Lemeshow S. Applied Logistic Regression. New York: John Wiley & Sons. 1989.

9. Smith DG, Clemens J, Crede W, Harvey M. Gracely EJ. Impact of multiple comparisons in randomized clinical trials. Am J Med.

10. Laupacis A, Albers G, Dunn M, Feinberg W. Antithrombotic therapy in atrial fibrillation. Chest. 1992; 102(4 suppl):426S- 433s.

1987; 83:545-50.

Documents

New–onset Atrial Fibrillation: When Is Admission Medically Justified?