Article, Geriatrics

A novel multidimensional geriatric screening tool in the ED: evaluation of feasibility and clinical relevance

a b s t r a c t

Purposes: Geriatric problems frequently go undetected in older patients in emergency departments (EDs), thus increasing their risk of adverse outcomes. We evaluated a novel emergency geriatric screening (EGS) tool designed to detect geriatric problems.

Basic procedures: The EGS tool consisted of short validated instruments used to screen 4 domains (cognition, falls, mobility, and activities of daily living). Emergency geriatric screening was introduced for ED patients 75 years or older throughout a 4-month period. We analyzed the prevalence of abnormal EGS and whether EGS increased the number of EGS-related diagnoses in the ED during the screening, as compared with a preceding control period.

Main findings: Emergency geriatric screening was performed on 338 (42.5%) of 795 patients presenting during screening. Emergency geriatric screening was unfeasible in 175 patients (22.0%) because of life-threatening conditions and was not performed in 282 (35.5%) for logistical reasons. Emergency geriatric screening took less than 5 minutes to perform in most (85.8%) cases. Among screened patients, 285 (84.3%) had at least 1 abnormal EGS finding. In 270 of these patients, at least 1 abnormal EGS finding did not result in a diagnosis in the ED and was reported for further workup to subsequent care. During screening, 142 patients (42.0%) had at least 1 diagnosis listed within the 4 EGS domains, significantly more than the 29.3% in the control period (odds ratio 1.75; 95% confidence interval, 1.34-2.29; P b .001). Emergency geriatric screening predicted nursing home admission after the In-hospital stay (odds ratio for >=3 vs b 3 abnormal domains 12.13; 95% confidence interval, 2.79-52.72; P = .001).

Principal conclusions: The novel EGS is feasible, identifies previously undetected geriatric problems, and predicts determinants of subsequent care.

Introduction

Cognitive and functional problems are often overlooked in older patients presenting to emergency departments (EDs). When prob- lems go undetected, older patients have a high risk of adverse outcomes (eg, death, delirium, functional decline, ED readmission) [1-6]. Comprehensive geriatric assessment (CGA) is an established procedure that detects overlooked geriatric problems and improves the quality of care for older patients, but it is too time-consuming to apply in the ED [7-9]. Convenient tools are lacking for multidimen- sional geriatric screening in the ED. Although some CGA instruments

? Funding sources: This study was supported by the Geriatric Research Fund (Spital Netz Bern, University Hospital Bern, Bern, Switzerland).

* Corresponding author. Department of Geriatrics, Inselspital, Bern University Hospital, CH-3010 Bern, Switzerland. Tel.: +41 31 632 21 11.

E-mail address: [email protected] (A.W. Schoenenberger).

have been shortened and validated for use in an ED, they are organ- specific rather than multidimensional [10-14]. The “identification of seniors at risk” (ISAR) and the “interRAI ED screener” are frequently used, but these are not based on objective tests of cognition or mobility [15-18]. Therefore, these Screening tools only identify older at-risk patients who should receive CGA, but they do not detect undiagnosed geriatric problems.

Older patients account for 12% to 24% of all ED visits [19,20]. Visits for emergencies have increased in recent years, and the greatest increase is among older patients [21]. Use of EDs by older patients will continue to increase along with Life expectancy, and older people are a growing demographic. Detecting modifiable geriatric problems early is likely to improve patient outcomes [7-9]. A multidimensional instrument that briefly but effectively screens for previously overlooked geriatric problems would increase ED efficiency and improve patient prospects. This study evaluated the feasibility of a novel multidimensional emergency geriatric screening (EGS) tool specifically designed to detect

http://dx.doi.org/10.1016/j.ajem.2014.03.024

Table 1

Emergency geriatric screening tool, consisting of short validated instruments for screening 4 domains (cognition, falls, mobility, and ADL) [10,25-27]

Cognition
Instruction: Ask the patient the following questions. If the patient does not respond, the question is rated incorrect.
What day is today? What is the date today? (+-1 day is correct) What year is this?	Incorrect? Incorrect? Incorrect?	Correct Correct Correct
Spell “radio” backward.	Incorrect	Correct
Evaluation consistent with impairment of cognition (if one single response was incorrect):	Yes	No
Falls
Instruction: Rate the following questions considering all available sources (patient, proxy, observation, reports).
Did the patient present to the ED because of a fall?	Yes?	No
Did the patient have one or more falls during the last 12 months?	Yes	No
Evaluation consistent with patient history of falls (if one single response was yes):	Yes	No
Mobility
Instruction: Rate the following question considering all available sources (patient, proxy, observation, reports).
Did the patient require walking aids (cane, wheeled walker, or helping person) indoors or outdoors before presenting to the ED?	Yes	No
Instruction: Rate the following questions according to the current situation in the ED.
Is the patient currently confined to bed? Does the patient currently need help (walking aids or helping person) to get out of bed?	Yes? Yes?	No No
Does the patient need >=20 seconds for the Timed Up and Go Test?	Yes	No
Evaluation consistent with impairment of mobility (if one single response was yes):	Yes	No
ADLs
Instruction: Rate the following question considering all available sources (patient, proxy, observation, reports).
Did the patient require assistance for personal hygiene (sponge bath, tub bath, or shower) before presenting to the ED?	Yes?	No
Instruction: Rate the following questions according to the current situation in the ED.
Is the patient currently confined to bed or does he need help (walking aid or helping person) to get out of bed? Does the patient require assistance (for direct help or instruction) for dressing (clothes or shoes)? Does the patient require assistance (for direct help or instruction) for toileting?	Yes? Yes? Yes?	No No No
Does the patient require assistance (for direct help or instruction) for feeding?	Yes	No
Evaluation consistent with impairment in ADL (if one single response was yes):	Yes	No
* If one of the responses marked with an asterisk applies, the rater may directly proceed to evaluating the domain (hierarchical structure).

geriatric problems in an ED setting. Our goals were to determine the prevalence of abnormal EGS findings and to establish whether EGS increased the number of EGS-related diagnoses on ED discharge reports.

Methods

Study design and setting

This prospective controlled study included all patients 75 years or older consecutively presenting to the ED of Bern University Hospital between June 1, 2012, and February 21, 2013. The study period had 2 segments (pre- post design). The control period ran from June 1, 2012, to October 21, 2012; for this segment, patients received usual care (control group). The control period was followed by a 4-month screening period, from October 22, 2012, to February 21, 2013; for this segment, we recommended that the EGS tool be used to supplement usual care (screening group).

The ED is integrated into a tertiary care university hospital, manages acute problems, and refers patients to subsequent care as quickly as possible. Usual care does not include geriatric screening or the use of other geriatric risk Prediction tools. All patients visiting the ED (ie, inpatients and outpatients) receive an ED discharge report as part of usual care when leaving the ED. The Bern University Hospital ED is an academic unit that receives approximately 30000 visits annually.

Selection of participants

All patients 75 years or older presenting to the ED of Bern University Hospital during the study period were enrolled. There were no further inclusion or exclusion criteria for the overall cohort. This

study was approved by the Bern University Hospital Institutional Review Board (application no. 12-028).

Intervention: the EGS tool

We developed the EGS tool (Table 1) on the basis of a literature review. The tool met the following prerequisites: (1) EGS is multidi- mensional and covers relevant domains of geriatric problems; (2) EGS uses validated instruments; and (3) EGS must be feasible in an ED. We considered 4 domains relevant for older ED patients: cognition, falls, mobility, and activities of daily living (ADLs) [22-24]. For each, we selected short validated instruments. We used the Ottawa 3DY test, which assesses orientation and the ability to spell a word backward, to evaluate cognition [10]. To evaluate falls, we chose 2 self-report questions that predict future falls [25]. One self-report question screened for mobility prior to the ED visit. Current mobility in the ED was checked with the Timed Up and Go Test [26]. Activities of daily living were screened with a standard instrument [27]. All instruments were adapted and arranged hierarchically for quick use in the ED (Table 1).

Shortly before the start of the screening period, ED physicians were instructed in using the EGS form during a 30-minute seminar. Paper copies of the EGS tool were available in ED offices during the screening period, and screening was performed by ED physicians, at the patient’s bedside. If EGS identified 1 or more abnormal results on the screen, ED physicians were advised to check whether the abnormal EGS finding had to be mentioned as a diagnosis on the ED discharge report. Irrespective of the EGS result, ED physicians added a copy of the EGS findings as a supplemental sheet to the ED discharge report, in order to inform the physician in charge for subsequent care.

Fig. Flowchart.

Outcomes

Our objectives were to evaluate the clinical relevance and feasibility of the EGS. To evaluate clinical relevance, we analyzed the numbers of abnormal EGS findings. We also determined the number of EGS-related diagnoses on the ED discharge reports during screening, in comparison with the preceding control period. As indirect evidence of clinical

Table 2

Baseline characteristics

relevance, we analyzed whether EGS findings predicted determinants of subsequent care (ie, admission from the ED to an Inpatient unit, in- hospital length of stay [LOS], and nursing home admission). To evaluate feasibility, we analyzed time of administration and feedback from involved ED physicians.

Data collection

The original EGS forms were collected in a box located in the ED. A study nurse later checked the form for completeness and correctness.

Characteristic Control group

(n = 752)

Age (y)

Screening group (n = 795)

If the EGS form was not completed, the responsible physician had to report the primary reason why the EGS was not performed. Data from EGS were then entered into a database using 2-pass verification.

Emergency department discharge reports of all participating

Mean +- SD 82.6 +- 5.1 82.7 +- 5.0

IQR 78.3-85.9 78.7-86.1

maximum value 102.5 99.8

Female sex, n (%) 371 (49.3) 384 (48.3)

Main condition leading to ED visit^a

Cardiovascular, n (%) 179 (23.8) 188 (23.6)

ACS	43 (5.7)	40 (5.0)
Heart failure	44 (5.8)	38 (4.8)
Heart Rhythm disorder	16 (2.1)	17 (2.1)
Stroke	25 (3.3)	28 (3.5)
PAD	8 (1.1)	10 (1.3)
Syncope	30 (4.0)	44 (5.5)
Resuscitation	13 (1.7)	11 (1.4)
Infectious disease, n (%)	136 (18.1)	142 (17.9)
Other conditions, n (%)	437 (58.1)	465 (58.5)
Fall	106 (14.1)	136 (17.1)
Nonspecific pain	100 (13.3)	105 (13.2)
Gastrointestinal ^b	58 (7.7)	52 (6.5)
Unspecified condition	173 (23.0)	172 (21.6)

Disposition from ED, n (%)

Admitted from the ED to an in-house 551 (73.3) 558 (70.2) inpatient unit

Discharged home 176 (23.4) 196 (24.6)

Referral to other institution 20 (2.7) 33 (4.2)

Died in the ED 5 (0.7) 8 (1.0) LOS (d), mean +- SD

patients were electronically exported from the hospital informa- tion system. There were no missing ED discharge reports. Two authors (A.W.S. and C.B.) reviewed the ED discharge reports and used a standard protocol to abstract diagnoses related to EGS from them, for later analysis. The 2 reviewers determined if there were 1 or more cognition-, falls-, mobility-, and/or ADL-related diagnoses listed on the ED discharge report. A.W.S. and C.B. were blinded to the results of EGS when they reviewed the ED discharge reports. A.W.S. and C.B. rated the ED discharge reports independently, and their ratings were verified for agreement. In case of disagreement, a third author was involved and a consensus decision was made.

During both the screening and control periods, the following data were recorded for all study participants: birth date, sex, triage scale, date and time of presentation to the ED, date and time of leaving the ED, and disposition from ED (ie, admission from the ED to an in-house inpatient unit, Discharge home, referral to other institution, or death). For patients admitted from the ED to an in-house inpatient unit, date and time of discharge, and disposition (ie, discharge home, referral to other institution, nursing home admission, or death) were recorded. All these data were directly extracted from the hospital information system. No data were missing.

LOS in ED

In-hospital LOS^c

0.22 +- 0.28 0.22 +- 0.20

8.69 +- 7.32 8.07 +- 6.64

Emergency geriatric screening feasibility was determined by the following: (1) ED physicians noted how long each EGS took to

Abbreviations: ACS, acute coronary syndrome; IQR, interquartile range; PAD, peripheral

arterial disease.

^a According to ED discharge report.

^b Gastrointestinal, other than infectious diseases.

^c In the subgroup of inpatients at Bern University Hospital.

administer, rating speed on a 4-point scale (b 3, 3-4, 4-5, or N 5 minutes),

and (2) after the screening period, all ED physicians who performed EGS were asked to complete a standardized questionnaire of 10 questions about their experience with the EGS.

Statistical analysis

We descriptively analyzed baseline characteristics and EGS findings by counts, percentages, mean values, SDs, and interquartile ranges. Characteristics between groups were compared by a ?² test for categorical outcomes and by a Kruskal-Wallis test for continuous outcomes. Then we analyzed the numbers of geriatric diagnoses that were reported on ED discharge reports, using per-protocol analyses. The per-protocol analysis compared patients evaluated with EGS in the screening period with all patients presenting during the control period. We used logistic regression to give odds ratios (OR) and 95% confidence intervals (CI), with corresponding P values to measure the effect. All analyses were performed unadjusted and adjusted for age, sex, and main condition leading to the ED visit (ie, cardiovascular problems, infectious diseases, and other conditions). We then performed a predictive analysis in patients evaluated with the EGS. Emergency geriatric screening was dichotomized for use as an independent variable based on the number of abnormal domains: EGS with at least 3 abnormal single domains or EGS with less than 3 abnormal single domains. This was based on the a priori analytic plan in such a way that the cutoff had to identify 2 approximately equal-size groups. The associations of EGS findings with the following determinants of subsequent care were evaluated: (1) with admission from the ED to an inpatient unit (Bern University Hospital or in an external institution),

(2) with in-hospital LOS in the subgroup of inpatients at Bern University

Hospital, and (3) with nursing home admission of in-house inpatients who had not previously lived in a nursing home. For LOS as a dependent variable, we calculated time ratios with 95% CIs using accelerated failure time models. For binary outcomes (ie, admission from the ED to an inpatient unit and nursing home admission), ORs with 95% CIs were calculated using logistic regression. We then analyzed feasibility with descriptive statistics. Finally, we performed sensitivity analyses. We reanalyzed the numbers of geriatric diagnoses reported on ED discharge reports using intention-to-screen analysis. The intention-to-screen analysis compared all patients presenting to the ED during the screening period (ie, including both patients with and without EGS) with all patients presenting during the control period. We also repeated primary analyses after exclusion of readmissions in the control group and readmissions in the screening group, in order to avoid bias caused by readmission of the same patient. Data were analyzed with Stata 12.1 (StataCorp LP, College Station, TX) and R version 3.0.1 (2013, The R Foundation for Statistical Computing, Vienna, Austria).

Results

Study participants

During the study period, 1547 patients 75 years or older presented to the ED at Bern University Hospital. Of these, 752 patients (48.6%) presented during the control period and 795 patients (51.4%) during the

screening period (Fig.). Emergency geriatric screening was not per- formed for 457 patients (57.5%) who presented in the screening period: for 175 patients (22.0%), EGS was not feasible because of their clinical situation in the ED (ie, life-threatening condition, altered mental status such as coma); 282 instances of nonperformance (35.5%) were due to logistical reasons (ie, limited personnel availability, ED overcrowding, patient speaking foreign language, and patient’s refusal of diagnostic workup). Emergency geriatric screening was performed on 338 patients (42.5%) presenting during the screening period.

Baseline characteristics of study participants are shown in Table 2. According to ED discharge reports, cardiovascular or infectious diseases, falls, and nonspecific pain were the most common reasons for the ED visit. There were no statistically significant differences in baseline characteristics between the screening and the control group.

Emergency geriatric screening findings and related diagnoses on ED discharge reports

Table 3 shows findings from EGS and the numbers of related diagnoses on ED discharge reports among screened patients. In most screened patients, abnormal EGS findings were found. The numbers of related diagnoses on ED discharge reports were lower. A completely normal EGS was found in only 53 (15.7%) of 338 screened patients. Findings from 1 or 2 EGS domains were abnormal for 120 study participants (35.5%), and findings from 3 or 4 EGS domains were abnormal for 165 study participants (48.8%). Among the 285 patients with 1 or more abnormal EGS finding, in only 15 patients (5.3%), all abnormal EGS findings had a related diagnosis on the ED discharge report; in the remaining 270 patients (94.7%), there was at least 1 finding without a related diagnosis, which was reported for further diagnostic workup to subsequent care.

Table 4 shows the numbers of EGS-related diagnoses on ED discharge reports during screening as compared with the control period in the per-protocol analysis. There were statistically significant increases in the number of patients with EGS-related diagnoses on ED discharge reports during screening. During the screening period, 142 (42.0%) of the 338 screened patients had at least 1 diagnosis listed within the 4 EGS domains, significantly more than the 29.3% of the patients presenting during the control period. This was due to a marked increase of diagnoses in cognition-related and falls-related domains (Table 4).

Predictive analysis

Predictive analysis showed that patients with 3 or 4 abnormal EGS findings were more frequently admitted from the ED to an inpatient unit as compared with patients with 2 or less abnormal EGS findings (OR, 2.68; 95% CI, 1.65-4.35; P b .001). For inpatients, the presence of 3 or 4 abnormal EGS findings significantly predicted in-hospital LOS (time ratio, 1.26; 95% CI, 1.05-1.51; P = .01) and whether patients

Table 3

Findings from EGS and numbers of related diagnoses on ED discharge reports among screened patients

Domain Prevalence comparison of abnormal EGS findings and related diagnoses on ED discharge reports in all screened patients

Analysis of concordance between EGS and ED discharge report in the subgroup of patients with abnormal EGS findings

	Had abnormal EGS finding	Had domain-related diagnosis on ED discharge report	Proportion with concordance between EGS and ED discharge report (ie, had domain-related diagnosis on ED discharge report)	Proportion without concordance between EGS and ED discharge report (ie, had no domain-related diagnosis on ED discharge report)
Cognition, n (%)	170/338 (50.3)	69/338 (20.4)	58/170 (34.1)	112/170 (65.9)
Falls, n (%)	170/338 (50.3)	91/338 (26.9)	85/170 (50.0)	85/170 (50.0)
Mobility, n (%)	227/338 (67.2)	33/338 (9.8)	28/227 (12.3)	199/227 (87.7)
ADLs, n (%)	195/338 (57.7)	6/338 (1.8)	6/195 (3.1)	189/195 (96.9)

Table 4

Numbers of EGS-related diagnoses on ED discharge reports during screening as compared with the control period in the per-protocol analysis

Control group (n = 752), n (%)

Screening group (n = 338), n (%)

Per-protocol analysis

Unadjusted Adjusted^a

OR (95% CI) ^b P^b OR (95% CI) ^b P^b

Diagnoses related to single EGS domains
Cognition-related	92 (12.2)	69 (20.4)	1.84 (1.31-2.59)	b.001	1.77 (1.25-2.50)	.001
Falls-related	128 (17.0)	91 (26.9)	1.80 (1.32-2.44)	b.001	1.67 (1.21-2.29)	.002
Mobility-related	55 (7.3)	33 (9.8)	1.37 (0.87-2.16)	.17	1.32 (0.84-2.09)	.23
ADL-related	0 (0.0)	6 (1.8)	NA	NA	NA	NA
Diagnoses related to any of the 4 EGS domains
Any domain	220 (29.3)	142 (42.0)	1.75 (1.34-2.29)	b.001	1.64 (1.25-2.16)	b.001

^a Adjustment variables include age, sex, and main conditions leading to ED visit, according to Table 2 (ie, cardiovascular problems, infectious diseases, and other conditions).

^b OR with 95% CI and P value from logistic regression for the comparison of screening vs control group.

were institutionalized in a nursing home after their in-hospital stay (OR, 12.13; 95% CI, 2.79-52.72; P = .001).

Feasibility

The amount of time it took to administer EGS was less than 3 minutes for each of 114 (33.7%) of 338 screened patients; it took 3 to 4 minutes each for 107 patients (31.7%), 4 to 5 minutes each for 69

patients (20.4%), and 5 to 10 minutes each for 48 patients (14.2%). Thus, EGS took less than 5 minutes to perform in most (85.8%) cases. Of the 70 invited ED physicians, 41 (64.1%) returned the questionnaire that asked about their experience with the EGS. Most responders agreed or partially agreed that EGS domains are suited to detect geriatric problems: 73.0% agreed or partially agreed for cognition; 77.8%, for falls; 75.0%, for mobility; and, 72.2%, for ADLs.

Sensitivity analyses

We performed an intention-to-screen analysis including all patients presenting to the ED in both segments of the study. During the screening period, 274 (34.5%) of all 795 patients admitted during screening had at least 1 diagnosis listed within the 4 EGS domains (142 patients who received EGS and 132 patients who did not receive EGS during screening). Compared with the 220 patients (29.3%) in the control group, this corresponded to a significant increase (unadjusted OR, 1.27 [95% CI, 1.03-1.58; P = .03]; adjusted OR, 1.28 [95% CI, 1.03-

1.60; P = .03]).

We repeated primary per-protocol analyses by excluding read- mission of the same patient. During the control period, 78 (10.4%) of 752 admissions were excluded, and during the screening period, 87 (10.9%) of 795 admissions were excluded. For 130 (42.9%) of 303 patients for whom EGS was performed, the repeated analysis found that at least 1 diagnosis within the 4 EGS domains was listed on the ED discharge report (unadjusted OR, 1.79 [95% CI, 1.35-2.38; P b .001];

adjusted OR, 1.73 [95% CI, 1.29-2.31; P b .001], as compared with the control group).

Discussion

The EGS is a novel and feasible ED-based tool. It significantly improves the identification of previously undetected geriatric condi- tions in the ED and predicts hospital admission, hospital LOS, and admission to a long-term nursing facility. The EGS tool had a high yield, visible in the rate of abnormal cognitive and functional screening results it returned. Use of the EGS tool increased the number of cognition- and falls-related diagnoses listed on ED discharge reports and thus confirmed that it uncovered previously undetected problems. In addition, in a substantial proportion of patients with abnormal EGS finding, the detected problem was assigned for consideration in subsequent care, because no diagnosis

was possible during the patients’ ED stay. Because of its brevity, we found use of the novel EGS tool feasible in an ED setting.

The proportion of patients with positive EGS finding was high. Studies that used risk screening tools like the ISAR or the interRAI ED screener found cognition- or mobility-related problems to be less prevalent, whereas similar prevalence rates were found for falls- and ADL-related problems [15,18]. It is important to note that the ISAR or the interRAI ED screener relies on patient self-report or the subjective impression of the investigator, whereas the EGS tool uses perfor- mance tests of cognition and mobility. Therefore, it is likely that risk screening tools such as ISAR or interRAI ED screener fail to catch cognitive or mobility problems in many patients. This statement is supported by a previous study that was based on extensive investigation among older ED patients and that found similar prevalence rates of cognitive problems to our study [22]. Of course, the EGS tool may have been false positive in some patients, but this is not a serious disadvantage in an ED setting: it is far better to detect a potential problem than to overlook one that must be confirmed and eventually addressed in subsequent care.

Patient and logistical factors prevented ED physicians from perform- ing the EGS on 57.5% of the patients in our screening study. Although nonperformance in patients with life-threatening conditions is justified, nonperformance for logistical reasons is a cause for concern. Although the speed of evaluation and its accuracy make EGS a useful tool for clinicians, future work is needed to improve its performance rate. For example, integration in a software system could reduce time resources required for EGS and concomitantly increase the performance rate of EGS via automated control mechanisms [28-31].

This study has limitations. The nonrandomized pre-post design limited the comparability of screening and control group. However, we do not think that this introduced relevant bias because baseline characteristics were similar between the 2 groups and primary analyses were adjusted for baseline factors. Because the study was conducted in one academic center, generalizability is limited. We also did not address intratester and intertester reliability. Finally, our study does not demonstrate that geriatric screening in the ED ultimately improves patient outcomes.

This study has important clinical implications. The EGS tool can detect geriatric problems amenable to intervention, which might otherwise be overlooked. Thus, multidimensional geriatric screening based on performance tests should be performed in older ED patients. Comprehensive geriatric assessment is multidimensional and is based on performance tests, but it requires too much time for use in most ED settings. risk prediction tools such as ISAR or interRAI ED screener can quickly identify older ED patients at increased risk for adverse outcomes, but performance tests detect previously unrecognized problems more reliably. Overlooked geriatric problems increase the risk of adverse outcomes in older ED patients [1-9]. Therefore, multidimensional EGS based on performance tests has substantial potential to improve patient outcomes, if newly detected problems

are addressed in subsequent care. To achieve additional improvement in outcomes, methods are needed to ensure that EGS is conducted in all elderly patients who present to an ED and have no life-threatening condition. We recommend that EGS be integrated into clinical routine at EDs.

Acknowledgment

The authors wish to express their gratitude to Stephan Born and Romana Businger for their support. They also thank Kali Tal for her editorial contribution.

References

AJEM