Working Toward a Gold Standard: The Severity of Ethanol Withdrawal Scale (SEWS) Versus the Clinical Institute Withdrawal Assessment Alcohol Scale (CIWA-Ar)

Abstract

Aim

Proving the Severity of Ethanol Withdrawal Scale (SEWS) significantly reduces Alcohol Withdrawal Syndrome (AWS) treatment Time on Medication Protocol (TOMP).

Method

Head-to-head Quality Assurance outcome compared separate cohorts of SEWS or Clinical Institute Withdrawal Assessment Alcohol Scale, Revised (CIWA-Ar) data using Student’s t and Wilcoxon tests.

Results

SEWS-driven treatment (n = 244) reduced TOMP to 2.2 days versus 3.4 days for CIWA-Ar (n = 137); P < 0.0001.

Conclusion

The SEWS is the superior measure of AWS symptoms.

INTRODUCTION

The Alcohol Withdrawal Syndrome (AWS), along with Tolerance to ethanol, indicates physical dependence, a primary feature (Beresford and Lucey, 2018) of ICD-10 Alcohol Dependence, or AlcD (alternatively Alcohol Use Disorder, Severe, DSM-5). With Alcohol Dependence/Abuse affecting 7–10% of the general population (Grant et al., 1994; Grant et al., 2004), we conservatively estimate AlcD itself at half that frequency, or 3.5%, suggesting that AlcD affects an estimated 9 million persons in the USA. The great majority of these will experience AWS. As many as 20% of AWS cases, or about 1.8 million persons annually, may suffer life-threatening AWS conditions (Grover and Ghosh, 2018). The estimated death rate from severe AWS varies from 5% in treated cases (Cushman Jr., 1987) to mention of as high as 37% in untreated cases (Rahman and Paul, 2022), including deaths from Delirium Tremens (DTs), seizure, stroke, heart disease and other causes. From these figures, we estimate about 270,000 US deaths annually. One study notes a 37% death rate among DT cases within 8 years of the initial occurrence (Perala et al., 2010).

Among US Veterans the estimated rate of AlcD or abuse is 32%, about three times higher than the general population (Lan et al., 2016), using applied diagnostic criteria. The actual figures for AWS related life-threatening illness may be much higher owing to inaccurate estimates from medical records, possibly reaching as high as 45% or 810,000 persons in the USA.

Measuring AWS Severity: Early intervention can prevent or mitigate later, more severe presentations. This requires accurate, assessment-driven treatment of AWS symptoms (Victor and Adams, 1953; Lee et al., 2005).

Our detailed pilot comparison (Beresford et al., 2017) of the Severity of Ethanol Withdrawal Scale (SEWS) against the Clinical Institute Withdrawal Assessment Alcohol Scale, Revised (CIWA-Ar), (Reoux and Miller, 2000; Schuckit, 2014) demonstrated the SEWS’ superior effectiveness in guiding common clinical AWS medication treatment (Beresford et al., 2017). SEWS-driven medication treatment reduced the Time on Medication Protocol (TOMP) by 1 day, on average, over that seen with the CIWA-Ar. This reduces hospital lengths of stay from 3 days to 2 days per AWS episode. Here, we report the results of the SEWS/CIWA-Ar comparison in an extended sample. We hypothesized that the same significant differences in TOMP would characterize outcomes in this much larger size sample.

MATERIALS AND METHODS

Development

Following our literature review (Taub and Beresford, 2015), we developed the SEWS scale (Beresford et al., 2017) to advance the care of AWS cases. A panel of Internal Medicine specialists reviewed the SEWS and approved its local use.

SEWS item weightings followed a secondary analysis of data from a large clinical study of alcoholic persons (n = 1128) admitted to a general hospital (Beresford et al., 1990). Fig. 1 lists the several differences in SEWS scale items contrasted with those of the CIWA-Ar and enumerated as follows.

1. The clinical literature on ‘headache’ during AWS ‘hangover’ relates to brain fluid shifts rather than AWS processes mediated by the Sympathetic Nervous System. (Boness et al., 2016) The SEWS does not include this irrelevant item.

2. The SEWS recognizes that visual and tactile hallucinations are commonly found in delirium—including DTs—in which the brain suffers from a toxic or metabolic insult. Auditory hallucinations may occur in delirium but occur more characteristically in primary psychiatric illnesses. The SEWS therefore gives auditory hallucinations 1 point while weighting hallucinations in any of the other four modalities with 3 points. This condenses hallucinations into one scale item, rather than three CIWA-Ar items.

3. Designed for accuracy and ease of use, the SEWS requires only a dichotomous present/absent symptom determination, rather than the CIWA-Ar 7-point Likert-type scale. The SEWS’ yes/no variables can be done easily during busy nursing shifts to give better factual information.

4. ‘Nausea’ by itself is far too unspecific to indicate AWS. When vomiting is also present however, the AWS severity often appears worse (Beresford et al., 1990). The SEWS keeps the more specific item.

5. CIWA-Ar ‘Orientation’ presents an arithmetic calculation, a function referrable to the brain’s parietal lobe that is not a test of time-space orientation. The SEWS opts for a standard, scored orientation test.

6. Vital signs are integral to AWS pathophysiology and therefore to the SEWS assessment. The SEWS combines the most serious AWS vital sign alterations and weights them heavily. Early CIWA-Ar versions omit vital signs and later versions attach them without a score.

7. Three items find a relative agreement between the two scales—sweats, tremor and agitation defined as a physical restlessness. Again, the SEWS measures each dichotomously while the CIWA-Ar uses a poorly defined Likert scale.

8. Agitation in the CIWA-Ar assessment is confused with anxiety, reported as feeling ‘nervous.’ The SEWS uses a more common clinical definition of anxiety—a sense of impending disaster—differentiating it from the physical restlessness of an agitated state.

Fig. 1

Item comparisons, SEWS versus CIWA-Ar. SEWS item weights indicate absent (0) or present (1, 2 or 3) on that item, either X or Y, for example either 0 or 3. Each item is scored dichotomously. CIWA-Ar item scores, by contrast, are assigned by an observer based on their subjective ratings of each item separately on a 0–7 Likert-type scale continuum.

Use approvals

Our hospital’s Pharmacy and Therapeutics Committee approved a final version of the SEWS for clinical deployment with the condition of an ongoing Quality Assurance (QA) review of safety and effectiveness. Both the local, joint Denver VA Medical Center/University affiliate Institutional Review Board and the Denver VA Research and Development Committee approved the submission of the present QA outcome report.

Procedures

The SEWS and CIWA-Ar, respectively, were each inserted into separate, pre-approved, symptom-triggered treatment protocols for AWS used in our medical center. Following a physician’s order, nursing assessments provided symptom severity data on the respective scales. The item scores were then entered into an electronic record that calculated a total observed AWS symptom score on one or the other scale. The total score determined the administered dose of chlordiazepoxide (CDZ). Patients with either a SEWS score of ≥6 and <12, or the published CIWA-Ar score of ≥8 and <16 (Koch-Weser et al., 1976; Sullivan et al., 1989), received 25 mg of CDZ. Patients with a SEWS score of ≥12 or a CIWA-Ar score of ≥16 received 50 mg of CDZ.

In both protocols, the score was reassessed as frequently as hourly when the SEWS reached ≥6 or the CIWA-Ar ≥8. If the score dropped to <6 for SEWS or <8 for CIWA-Ar—not indicating medication—the scale was administered every 4 h and was discontinued 48 h after last dose of CDZ.

Data collection

De-identified data prospectively entered for consecutively admitted patients accrued during an 8-month period; these were extracted from our VA medical center’s electronic medical record.

For QA outcome analysis, the recorded time of the first scale score entry for either scale defined the starting point and the recorded time of the last CDZ dose served as the end of treatment. The difference between the two is TOMP measuring the primary outcome. Only medicated cases were included for analysis. Dosing data were not available.

Statistical analysis

Comparison of the TOMP means for the respective scales used Student’s t-test and the Wilcoxon test for score ranges. Analyses were conducted with SAS 11.0 (SAS Institute, Cary, NC) and Excel (Microsoft, Seattle, WA) software.

RESULTS

Sample

All the patients were eligible Veterans at the Rocky Mountain Regional VA Medical Center, a general hospital (Holleck et al., 2019) in Aurora, Colorado. All cases were male with mean age of 47.0 ± 13.2 years. The study prospectively harvested, electronic-only clinical score data from admitted AWS inpatients (SEWS n = 244, and CIWA-Ar n = 137, N = 381 total) who had received CDZ. Group demographics were 69.6% Caucasian, 16.5% African American, 12.7% Hispanic and 1.3% Native American. Neither age nor ethnic origin were associated with AWS outcome.

Outcome

This head-to-head, prospective comparison of the SEWS against the CIWA-Ar in driving medication treatment shortened the average AWS illness course from 3.4 days down to 2.2 days, P < 0.0001 (see Table 1).

Table 1

Time on Medication Protocol (TOMP), N = 381

	Mean (h)	SD	t-test	Median (h)	Range	Wilcoxon test
SEWS (n = 244)	53	48.4		37.3	367.9
			P < 0.0001			P < 0.0001
CIWA-AR (n = 137)	81.7	43.4		71	296

SEWS driven treatment significantly shortened the course of AWS as measured by Time on Medication Protocol (TOMP) as compared to the CIWA-Ar.

DISCUSSION

Our work to date strongly suggests that SEWS-driven treatment (a) shortens the AWS treatment course, (b) and drives benzodiazepine dosing for the treatment of AWS more effectively. While this report does not specifically address ease of use, qualitative follow up has been positive. In our institution, the nurses complimented the SEWS on systematizing AWS treatment. This has led directly to its implementation in all the general medical and surgical units of our facility. Further, our local community public hospital, the Denver Health Medical Center with its high AWS volume, adopted the SEWS in their clinical services and noted the SEWS’ ease of use in that setting (Sankoff et al., 2013). Qualitative ease of use comments have reached us from our University affiliate hospital as well.

Limitations

This report presents an exclusively male, middle-aged sample using a QA methodology. Going forward, we have proposed a replication of this study at multiple sites, using a diverse gender sample, and randomized, double-blind, point-of-care subject assignment.

Strengths

The QA method used here also presents a strength of the study in assessing the two scales in real-time, prospective, treatment comparisons of actual persons suffering AWS. The prospective collection of the clinical data reported carries far more precision than a retrospective method.

Our pilot study identified a potential mechanism to explain the more effective AWS resolution. The SEWS-driven treatment protocol administered nearly twice the medication dose in the first 24 h and over the total AWS episode. (Beresford et al., 2017) This is consistent with other clinical reports that symptom-triggered AWS pharmacotherapy must be vigorously applied early in the AWS course (Jaeger et al., 2001; Lange-Asschenfeldt et al., 2003; McGregor et al., 2003; McKay et al., 2004; Hardern and Page, 2005; Hecksel et al., 2008; Bonnet et al., 2010; Cassidy et al., 2011; Elholm et al., 2011; Maldonado et al., 2012).

Some have asked, ‘Why not merely increase the medication doses triggered in the CIWA-Ar driven protocol?’ Following equipoise, we did not know which scale was better before comparing them. We conclude from the evidence that the observed difference in treatment is due to the improved accuracy of a better symptom severity scale. Merely changing doses does not justify using an empirically inaccurate scale.

Last, the QA data reported here argue for better understanding of the neuro-pathophysiology that underlies AWS itself. For example, in an animal AWS model, Ulrichsen and colleagues reported that early delivery of substantial medication doses decreases the probable frequency, and possibly the severity, of further AWS episodes (Ulrichsen et al., 1995). From this, they hypothesized that AWS reflects a kindling phenomenon that proper pharmacotherapy likely interrupts. If this is true in humans, appropriate treatment may itself lessen the long-term AWS healthcare burden of Alcohol Dependence.

CONCLUSION

The relatively large cohort in this report has extended the results of our pilot study to demonstrate further the effectiveness of the SEWS over CIWA. The SEWS significantly improved treatment and lessened both treatment time and hospital stay. The development of the SEWS (Fig. 2) grew out of the perceived clinical necessity to provide optimal care through an improved and specifically targeted approach to AWS medication treatment. Subjective responses from outside our doors have been exclusively positive in favor of the SEWS over the CIWA-Ar in some 20+ other medical centers where it is in use. It is available for others to adopt and assess. For discussion of the strengths and limitations of this study in a VA population sample, please see our previous report. (Beresford et al., 2017) Questions or concerns about its use—such as the specific medical center protocol that integrates the SEWS—may be directed to the principal author.

Fig. 2

Severity of SEWS electronic medical record version.

An AWS standard measure

Despite its ubiquity—and despite its medical danger—AWS has never merited a ‘gold standard’ scale for measuring its occurrence and its severity. The CIWA-Ar, to many, was long assumed to be the ‘gold scale’ by default until our work showed that this is not the case (Beresford et al., 2017). In our view, the SEWS is empirically a far better clinical standard with which to characterize and treat AWS. It has become the ‘gold standard’ in our clinical work and in that of many others. The SEWS can be applied easily, both in clinical treatment and in research. Its use improves the standardization of clinical treatment for AWS, a widespread, dangerous syndrome. We recommend it on that basis.

DISCLAIMER

The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the US government.

FUNDING

T.P.B. and P.J.R. VA Merit Award I01 BX004712. P.J.R. Great Plains Veterans Research Foundation, NIMH R01 MH122954, NIH NIGMS U54GM128729, NIMH R15 MH125306-01A1, NIMH R01 MH122954

CONFLICT OF INTEREST

The authors certify that they have no conflicts of interest related to this work.

DATA AVAILABILITY

Raw data, either electronic or paper, will be shared on a collaborative basis with other researchers with similar interests in accordance with VA and local research review entities.