Author + information
- Received July 31, 2017
- Revision received August 30, 2017
- Accepted September 7, 2017
- Published online February 19, 2018.
- Roopinder K. Sandhu, MD, MPHa,∗ (, )
- Dat T. Tran, MPHb,c,
- Robert S. Sheldon, MD, PhDd and
- Padma Kaul, PhDb,c
- aDivision of Cardiology, University of Alberta, Edmonton, Alberta, Canada
- bSchool of Public Health, University of Alberta, Edmonton, Alberta, Canada
- cCanadian VIGOUR Centre, University of Alberta, Edmonton, Alberta, Canada
- dDivision of Cardiology, University of Calgary, Calgary, Alberta, Canada
- ↵∗Address for correspondence:
Dr. Roopinder K. Sandhu, Division of Cardiology, University of Alberta, 8440-112 Street, 2C2 WMC, Edmonton, Alberta T6G 2B7, Canada.
Objectives This study sought to examine outcomes and costs of patients with syncope admitted and discharged from the emergency department (ED).
Background ED visits for syncope are common, yet the impact on health care utilization is relatively unknown.
Methods A total of 51,831 consecutive patients presented to the ED with a primary diagnosis of syncope (International Classification of Diseases-9 code 780.2 and International Classification of Diseases-10 code R55) in Alberta, Canada from 2006 to 2014. Outcomes included 30-day syncope ED and hospital readmissions; 30-day and 1-year mortality; and annual inpatient, outpatient, physician, and drug costs, cumulative.
Results Of adults presenting to the ED, 6.6% were hospitalized and discharged with a primary diagnosis of syncope (Cohort 1), 8.7% were hospitalized and discharged with a primary diagnosis other than syncope (Cohort 2), and 84.7% were discharged home with a syncope diagnosis (Cohort 3). The 30-day ED revisits for syncope varied from 1.2% (Cohort 2) to 2.4% (Cohort 1) (p < 0.001), and readmission rates were <1% among cohorts. Short- and long-term mortality rates were highest for Cohort 2 and lowest for Cohort 3 (30-day mortality: Cohort 1 of 1.2%, Cohort 2 of 5.2%, Cohort 3 of 0.4%; p < 0.001) (1-year mortality: Cohort 1 of 9.2%, Cohort 2 of 17.7%, Cohort 3 of 3.0%; p < 0.001). Total cost of syncope presentations was $530.6 million (Cohort 1: $75.3 million; $29,519/patient, Cohort 2: $138.1 million; $42,042/patient, Cohort 3: $317.3 million; $9,963/patient; p<0.001).
Conclusions Most patients with syncope presenting to the ED were discharged and had a favorable prognosis but overall costs were high compared with patients hospitalized. Further research is needed for cost-saving strategies across all cohorts.
Syncope is common, representing 1% to 3% of all emergency department (ED) visits (1,2). Diagnostic uncertainty and unclear prognosis mainly drive the high proportion of patients, ranging from 12% to 83%, who are admitted (3,4). Evaluating an underlying etiology for syncope is challenging and often results in numerous and expensive testing (5,6) yet these efforts have limited clinical utility and anywhere from a third to a half of patients with syncope are discharged without a clear diagnosis (7,8). Despite efforts to improve ED diagnostic evaluation and risk stratification (9–11), a 10-year analysis from the United States found no significant downward trend in the rates of syncope ED visits, admissions, or hospital discharge diagnosis of syncope while the use of advanced imaging rose significantly (12).
Single-center studies from multiple countries consistently report high costs of a hospital evaluation for syncope (13–15). The 2000 Cost and Utilization Project National Inpatient Sample provides the only national cost estimate of syncope-related hospitalizations based on discharge diagnosis of $2.4 billion U.S. dollars per annum (16). There are currently no data on contemporary costs of syncope and none that incorporate costs beyond the hospital setting.
Data evaluating the effect of acute health care visits on outcomes also remain sparse. A meta-analysis (17) of patients presenting to the ED with syncope found that 1.6% died within 30 days (n = 3,214) and 8.4% died at 1 year (n = 4,879). Among patients hospitalized for syncope, 1-year mortality rates have been shown to vary between 1.9% and 14.7% depending on comorbidity burden (18–20). No large study has evaluated patient characteristics and outcomes of unselected patients with syncope in various clinical settings (emergency vs. nonemergency) in 1 health care system. A better understanding of how outcomes and costs differ among patients with syncope based on ED triage decisions and hospital course could help to develop care pathways and provide baseline data against which the effectiveness of these interventions can be benchmarked.
Accordingly, we performed a population-based study of consecutive patients presenting to the ED with a primary diagnosis of syncope and evaluated comorbidities and short- and long-term outcomes for 3 syncope cohorts: 1) those admitted to hospital and discharged with a diagnosis of syncope; 2) those admitted to hospital and discharged with a diagnosis other than syncope; and 3) those discharged from the ED. We also compared resource utilization (inpatient, outpatient, physician, drugs) and costs among the 3 cohorts.
Data were obtained by linking the following 5 databases maintained by Ministry of Health and Wellness, in Alberta, Canada (20): 1) the Ambulatory Care Classification System database tracks all visits to the 101 EDs in the province and was used to identify the study cohort; 2) the Discharge Abstract Database records all admissions to acute care hospitals; 3) the Practitioners Claims database tracks all physician claims for outpatient services; 4) the Alberta Health Care Insurance Registry includes demographic and vital statistics data for all residents of the province; and 5) the Pharmaceutical Information Network database records information on drugs dispensed for prescribed medications (21).
Study population and variable elements
The study population included all patients 20 years or older who presented to the ED from April 2006 to March 2014 with a primary diagnosis of syncope (International Classification of Diseases-10th revision: R55 ). Patients were divided into 3 cohorts based on their ED discharge status: Cohort 1 consisted of patients who were hospitalized within 24 h of ED visit and discharged with a primary diagnosis of syncope; Cohort 2 consisted of patients who were hospitalized within 24 h of ED visit and discharged with a primary diagnosis other than syncope; and Cohort 3 consisted of patients who were discharged home from the ED. Patients were followed from the index ED presentation until death or being censored by March 31, 2014.
Comorbidities based on previously validated International Classification of Diseases codes were considered to be present if they were recorded for the index visit or hospitalization, within 24 h of ED discharge (if it occurred), and for all other contacts with the health care system in the 4 years preceding the index visit using the ambulatory care and hospitalization databases (23). A Charlson Comorbidity Index was calculated for each patient and categorized into as 0, 1 to 2, 3 to 4, or ≥5 (24). Canadian Classification of Health Interventions codes were used to track pacemakers (1.HZ.53 ending in NM, NK, NL, NN, FR), implantable cardioverter defibrillator (1.HZ.53 ending in FS) implants, and advanced imaging (computer tomography [CT] of the head [3.ER.20, 3.AN.20] and brain magnetic resonance imaging [3.AN.40]). The imaging codes search was restricted to ED index data before April 1, 2013, for 1-year of follow-up (n = 47,886) and all ED visits within a year was queried where syncope was the primary diagnosis. Invasive electrophysiology (EP) testing (2.HZ.24.GP-XJ, KJ, KL) was also queried after index ED visit performed in all ambulatory care visits and hospitalizations within 1 year.
Syncope costs were obtained from the Alberta Interactive Health Data Application (25) and calculated using Comprehensive Ambulatory Classification System, which provides average costs for an ED or outpatient visit; Case Mix Groups, similar in function to Diagnostic Related Groups, which provide average costs incurred in the direct care of patients admitted to hospital; and physician billing claims, which provide physician costs. Drug prices were extracted from the Alberta Drug Benefit List (26) from 2008 only and if a drug was not listed, we used the market price at Canada Drugs (27). Drug costs were derived by multiplying the price of a dispensed unit by the number of units dispensed. We calculated total costs and per patient costs of 1-year follow-up from the time of the index ED visit. Cost analysis was restricted to those with ED visits from 2006 to 2011 because Comprehensive Ambulatory Classification System and Case Mix Groups codes were only available during fiscal years of 2006 to 2012. All costs were converted to 2016 Canadian dollars using Bank of Canada inflation calculator (28).
The primary outcomes were 30-day revisits to the ED and rehospitalizations for a primary diagnosis of syncope and 30-day and 1-year mortality. Secondary outcomes included 1-year revisits to the ED and rehospitalizations for a primary diagnosis of syncope; 30-day and 1-year readmissions to the ED and hospital for any cause; in-hospital mortality; and annual costs including the distribution of inpatient, outpatient, physician, and drug costs for the 3 syncope cohorts.
We summarized patient characteristics using mean ± SD, medians (interquartile ranges), counts, and percentages, as appropriate. Values were compared across cohorts using chi-square test for categorical variables and the Kruskal-Wallis test for continuous variables. Cumulative mortality and readmission to ED and hospital were calculated using Kaplan-Meier curve. Patients who died during the index ED visit (for Cohort 3) and who died during the index hospital episode (for Cohorts 1 and 2) were excluded from the readmission analyses. All analyses were performed using Stata version 14 (Stata Corporation, College Station, Texas). Two-sided p values <0.05 were considered statistically significant. The study was reviewed and approved by the ethics board at the University of Alberta.
There were 51,831 patients who had an ED visit for a primary diagnosis of syncope during the 8-year study period. Of these, 6.6% were hospitalized and discharged with a primary diagnosis of syncope (Cohort 1), 8.7% were hospitalized and discharged with a primary diagnosis other than syncope (Cohort 2), and 84.7% were discharged home with a syncope diagnosis (Cohort 3). The number of syncope presentations remained stable from 2006 to 2010 (6,170 to 6,199) but increased subsequently (p < 0.001) (Figure 1), primarily as a result of an increase in patients in Cohort 3 (14.7% increase).
Patients discharged from the ED (Cohort 3) were younger, more likely to be female, and live in an urban residence with a higher median household income compared with patients with syncope who were hospitalized (Cohorts 1 and 2) (Table 1). Hospitalized patients had a higher comorbidity burden with more than a third having a Charlson comorbidity score of ≥3, whereas almost two-thirds of patients discharged from the ED had a score of 0.
A 30-day ED revisit for a primary diagnosis of syncope occurred in 2.4% of patients in Cohort 1, in 1.2% of patients in Cohort 2, and in 1.7% of patients in Cohort 3 (Figure 2) (p < 0.001). The rate of hospitalization at 30 days was low, ranging from 0.3% (Cohort 3) to 0.9% (Cohort 1; p < 0.001). At 1 year, 8.6% in Cohort 1, a total of 5.2% in Cohort 2, and 4.8% in Cohort 3 had repeat visit to the ED for a primary diagnosis of syncope (Figure 2) (p < 0.001). Compared with Cohort 3, the rate of a rehospitalization was 2-fold higher in Cohort 2 and 4-fold higher in Cohort 1 (p < 0.001).
After discharge, 19.8% of patients in Cohort 1, a total of 24% in Cohort 2, and 15.6% in Cohort 3 had a 30-day ED revisit for any cause (Online Figure 1) (p < 0.001). The rates of hospitalization for these patients differed significantly among the cohorts, ranging from 4.0% (Cohort 3) to 13.1% (Cohort 2) (p < 0.001). Syncope was the most common primary reason for rehospitalization for Cohort 1 (9.7%) and Cohort 3 (6.9%), whereas heart failure (6.8%) was the most common reason for Cohort 2. At 1 year, the rates of ED revisits for any cause was high for all cohorts (Cohort 1, 59.9%; Cohort 2, 64.9%; Cohort 3, 48.9%; p < 0.001) (Online Figure 1) and subsequent hospitalizations ranged from 20.2% in Cohort 3 to 43.7% in Cohort 2 (p < 0.001). The most common primary discharge diagnosis from hospital continued to be syncope for Cohort 1 (9.0%) and heart failure for Cohort 2 (8.0%) and care involving use of rehabilitation procedures for Cohort 3 (4.9%).
In-hospital mortality was 9 times higher for Cohort 2 (6.1%) compared with Cohort 1 (0.7%). Figure 3 shows the cumulative mortality rates according to discharge status. Short- and long-term mortality rates were highest for Cohort 2 and lowest for Cohort 3 (30-day mortality: Cohort 1, 1.2%; Cohort 2, 5.2%; Cohort 3, 0.4%; p < 0.001) (1-year mortality: Cohort 1, 9.2%; Cohort 2, 17.7%; Cohort 3, 3.0%; p < 0.001).
Among those hospitalized after the index ED visit, 3.4% and 0.4% of Cohort 1 and 10% and 0.6% of Cohort 2 underwent pacemaker and implantable cardioverter defibrillator implantation, respectively. Within 1 year, 6.6% and 0.9% of Cohort 1, 12% and 1.3% of Cohort 2, and 3.5% and 0.3% of Cohort 3 were implanted with pacemaker and implantable cardioverter defibrillators, respectively.
A total of 7,393 (15.4%) CT head scans were ordered for patients seen in the ED for a primary diagnosis of syncope. Of these, 25.1% occurred in Cohort 1, 23.9% for Cohort 2, and 13.7% for Cohort 3 (p < 0.001). Few brain magnetic resonance images were ordered for the entire population (n = 47). The rate of invasive EP procedures was 0.2% (78 of 44,855) and occurred in 20 patients in Cohort 1, in 14 patients in Cohort 2, and in 44 patients in Cohort 3 (p < 0.001).
The total cost of syncope presentations to the ED from 2006 to 2011 was $530.6 million (2016 Canadian Dollars) (Table 2). Of this, Cohort 1 accounted for $75.3 million ($29,519/patient), Cohort 2 accounted for $138.1 million ($42,042/patient), and Cohort 3 accounted for $317.3 million ($9,963/patient; p < 0.001). Across the 3 cohorts, the distribution of costs (inpatient, outpatient, physician, drugs) remained similar for each year. The highest proportion of annual costs was because of hospitalizations (Cohort 1, 69%; Cohort 2, 74%; Cohort 3, 48%) for all cohorts (Figure 4). Compared with the other cohorts, the proportion of outpatient visits and physician claims was highest for Cohort 3.
In this population-based study evaluating post-discharge outcomes and costs of 51,831 patients presenting to the ED with a primary diagnosis of syncope, we report the following findings: 1) most patients were considered low risk and discharged home; 2) short- and long-term outcomes for hospitalized and discharged patients with syncope are related to comorbidity burden; and 3) the most health care costs are incurred by patients with syncope discharged from the ED, whereas per patient cost is highest among patients hospitalized with syncope but discharged with a primary diagnosis other than syncope.
Impact of health care visits on outcomes
Although syncope is a frequent presentation, limited data exist evaluating whether ED visits and, in particular, hospitalization influence outcomes of syncope (19). In a meta-analysis of consecutive patients presenting to the ED, mortality rates were 1.6% and 8.4% at 30 days and 1 year, respectively (17). However, many of these observational studies included patients both discharged or admitted from the ED. An Italian study of 675 patients with syncope found 1-year mortality rates based on ED disposition and found 1.8% of the patients who were discharged from the ED died compared with 14.7% who were admitted (19). Almost one-half of admitted patients were 65 years or older and had significantly higher burden of cardiovascular comorbidity compared with those patients discharged from the ED. We found similar differences in mortality among patients discharged versus admitted from the ED. Short- and long-term mortality rates among patients with syncope discharged from the ED were very low (30 day, 0.4%; 1 year, 3.0%). Among admitted patients, mortality rates were 4 times higher at 30 day and almost 2 times higher at 1 year among patients in whom an underlying etiology was established (Cohort 2) than among those who were discharged with a diagnosis of syncope. Significant differences in comorbidity burden are likely to explain a large portion of the differences in mortality outcomes among the 3 cohorts of patients. Patients with syncope discharged home from the ED were younger and had significantly lower rates of comorbidities and mortality than admitted patients, suggesting appropriate triage was occurring in the ED. Both the short- and long-term mortality was highest among Cohort 2 who were sicker than the other cohorts. These data suggest ED visits or hospitalization does not seem to modify long-term prognosis.
Strategies to reduce hospitalizations
Although our study demonstrated a low rate of hospitalizations (15.3%) compared with studies from the United States and Europe (35% to 83%) (1,4,12), further reductions in the admission rates are needed, particularly because hospitalizations accounted for approximately 70% of costs of patients hospitalized (Cohorts 1 and 2) and 50% of costs for patients discharged from the ED (Cohort 3). A promising strategy to reduce hospitalizations for any country may be ED-based syncope units, which consist of time-limited observation with telemetry, a multidisciplinary team, and expedited access to cardiac testing (29). Two small, randomized clinical trials (30,31) in the United States demonstrated a protocol-driven ED syncope unit improved diagnostic yield and reduced hospitalizations and costs without adversely impacting outcomes compared with admission to an inpatient service in patients with intermediate risk factors (≥50 years, cardiac history, concerning electrocardiogram findings, family history of sudden cardiac disease, symptoms not suggestive of reflex syncope). The Falls and Syncope Service in England had also shown an initial detailed assessment by a physician syncope expert using evidence-based algorithms to advise admission, investigations, and referrals resulted in diagnosis reduced admission and implementation of good clinical practices (32). Before widespread use of these approaches consensus on factors for classifying low, intermediate, and high risk is needed, and additional large, clinical studies evaluating outcomes and cost-effectiveness in different countries.
Cost-saving ED measures
Multiple and expensive testing is often used in the ED to determine an underlying cause for syncope (5,6). A prospective, multicenter, observational study found a median number of 13 tests were performed per patient with unexplained syncope and almost half had advanced imaging (33). A similar finding was reported from a 10-year analysis of ED visits for syncope where CT/magnetic resonance imaging scan rates increased from 21% to 45% in recent years, and we also found CT head scans frequently ordered (12). Studies (1,34,35) have shown that advanced imaging in the absence of clinical suspicion affects diagnosis and management in very few patients (2%) yet is costly ($24,881 U.S. dollars) (6). The Choosing Wisely campaign is a program that encourages clinicians and patients to engage in conversation regarding unnecessary tests, treatments, and procedures. In Canada, the Choosing Wisely campaign in partnership with the Canadian Medical Association has targeted ED physicians in reducing unnecessary CT head scans in adult patients with syncope in the absence of defined high-risk predictors. This program exists in 20 countries and 5 continents and provides a unique platform to initiate similar campaigns globally. Evaluating practice changes with the Choosing Wisely campaign is needed. Similarly, appropriate use of expensive testing during a syncope work-up, such as invasive EP testing, is needed. Although our data demonstrate a low rate, an observational study in 10 European countries and Israel found invasive EP testing occurred in 25% of patients with recurrent unexplained syncope before implantable loop recorder insertion and was the highest cost per test (33,36). An implantable loop recorder–guided strategy has been shown to have high diagnostic yield (36) and is cost-effective compared with a conventional testing strategy including invasive EP testing (37).
A key quality indicator for syncope management is 30-day revisits to the ED. We found short-term revisits for syncope, ranging from 1.2% (Cohort 2) to 2.4% (Cohort 1) and 1-year revisits for syncope, ranging from 4.8% (Cohort 3) and 8.6% (Cohort 1). The rate of 30-day revisits was much higher in our study without a pre-defined syncope management algorithm than reported in the Evaluation of Guidelines in SYncope Study 2 (EGSYS 2) study (n = 380) (38). Italian patients managed according to the European Society of Cardiology syncope guidelines had a syncope relapse at 30 days of 0.3% after a first ED presentation. Application of novel evidence-based algorithms has also been shown to reduce 1-year ED visits. Readmission rates decreased from 12% to 0% after implementation of algorithms by the Falls and Syncope Service (32). The recently published 2017 American College of Cardiology/American Heart Association/Heart Rhythm Society guidelines on syncope evaluation and management provides simple algorithms that may be incorporated into clinical pathways to improve diagnosis, use of diagnostic testing, and assist with triage decisions but needs further research (29).
Improved discharge planning from the ED to an outpatient syncope clinic may also reduce readmission rates. A multidisciplinary, specialist nurse-led Rapid Access Blackouts Triage Clinic in Manchester provided risk stratification, comprehensive assessment using evidence-based algorithms, direct treatment, or specialist referral where appropriate (39). Readmission rates significantly decreased after evaluation in Rapid Access Blackouts Triage Clinic from 46.2% to 6.8% with no difference across risk strata. An outpatient syncope clinic may not be feasible in all health care systems; however, identifying physicians who have an interest in syncope evaluation and management in a geographic area may allow for scheduled follow-up appointments to be integrated into discharge planning (40). Patients stop fainting after assessment in the absence of specific therapy, although reasons for this are unclear (41). This may, in part, be caused by getting a diagnosis, education, and reassurance.
Our study provides population-based estimates of admission/discharge rates, outcomes, and costs, in a contemporary cohort of patients presenting with syncope to the ED in a geographically defined province with universal health care access. However, these data may not be generalizable to other provinces or other countries. We were unable to determine the type and frequency of tests and specialist consultations performed in the evaluation of syncope in inpatient and outpatient settings. These data may further aid in cost-saving measures. The study relies on administrative codes for the diagnosis of syncope. Although these codes have been validated in the ED and hospital settings (22), their accuracy in the outpatient clinic setting has not been established, and may have led to misclassifications and underestimation of costs. However, we were consistent in our use of the administrative code and focused on a primary diagnosis of syncope.
In this population-based study, we found most patients presenting to the ED with a primary diagnosis of syncope were discharged home and had favorable prognosis relative to patients admitted to the hospital. Regardless of ED discharge status, the high health care utilization and costs suggests further research is needed on syncope models of care in different health care settings that evaluate outcomes and cost-effectiveness of multidisciplinary syncope teams, evidence-based care pathways, discharge planning, and patient education.
COMPETENCY IN MEDICAL KNOWLEDGE: Decisions made at the first point of care, (i.e., the ED) have major implications on outcomes and costs of patients presenting with syncope. In this population-based study, most patients with syncope are discharged home. Short- and long-term mortality is low; however, overall health care costs are high compared with patients hospitalized.
TRANSLATIONAL OUTLOOK 1: Improved risk stratification is needed for correct classification of patients with high-risk features requiring urgent assessment and low-risk patients with a benign condition requiring minimal evaluation and reassurance.
TRANSLATIONAL OUTLOOK 2: Further research is needed on syncope care models evaluating outcomes and cost-effectiveness of a multidisciplinary syncope team, use of evidence-based algorithms, and discharge planning strategies in different countries.
Funded by the University Hospital Foundation of the University of Alberta and the Cardiac Arrhythmia Network of Canada. All authors have reported that they have no relationships relevant to the contents of this paper to disclose.
All authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the JACC: Clinical Electrophysiology author instructions page.
- Abbreviations and Acronyms
- computed tomography
- emergency department
- Received July 31, 2017.
- Revision received August 30, 2017.
- Accepted September 7, 2017.
- 2018 American College of Cardiology Foundation
- Thiruganasambandamoorthy V.,
- Stiell I.,
- Wells G.
- Linzer M.,
- Yang E.H.,
- Estes N.A. 3rd.,
- Wang P.,
- Vorperian V.R.,
- Kapoor W.N.
- Reed M.J.,
- Newby D.E.,
- Coull A.J.,
- Prescott R.J.,
- Jacques K.G.,
- Gray A.J.
- Probst M.A.,
- Kanzaria H.K.,
- Gbedemah M.,
- Richardson L.D.,
- Sun B.C.
- Farwell D.J.,
- Sulke A.N.
- Ruwald M.H.,
- Hansen M.L.,
- Lamberts M.,
- et al.
- Costantino G.,
- Perego F.,
- Dipaola F.,
- et al.
- Sandhu R.K.,
- Sheldon R.S.,
- Savu A.,
- Kaul P.
- ↵Alberta Health. Available at: http://www.health.alberta.ca/documents/Research-Health-Datasets.pdf. Accessed December 1, 2016.
- ↵Alberta Interactive Health Data Application. Available at: http://www.ahw.gov.ab.ca/IHDA_Retrieval/selectCategory.do. Accessed July 16, 2015.
- ↵Alberta Health 2016. Available at: https://www.ab.bluecross.ca/dbl/publications.html. Accessed September 10, 2016.
- ↵Canada Drugs 2016. Available at: https://www.canadadrugs.com/. Accessed September 10, 2016.
- ↵Bank of Canada. Available at: http://www.bankofcanada.ca/rates/related/inflation-calculator/?page_moved=1. Accessed April 14, 2016.
- Shen W.K.,
- Sheldon R.S.,
- Benditt D.G.,
- et al.
- Shen W.K.,
- Decker W.W.,
- Smars P.A.,
- et al.
- Parry S.W.,
- Frearson R.,
- Steen N.,
- Newton J.L.,
- Tryambake P.,
- Kenny R.A.
- Krahn A.D.,
- Klein G.J.,
- Yee R.,
- Hoch J.S.,
- Skanes A.C.
- Petkar S.,
- Bell W.,
- Rice N.,
- et al.