a b s t r a c t

Introduction: The radiation risk posed by diagnostic computed tomography (CT) is a growing concern. The use of model-based iterative reconstruction (MBIR) technology reduces radiation exposure but requires additional processing time. The goal of this study was to compare MBIR and a standard CT reconstructive protocols in terms of emergency department (ED) Visit duration and reduction in radiation exposure.

Methods: A retrospective, matched, case-control design was used to compare patients who received MBIR and standard protocol abdomen and pelvis CTs. ED length of stay and radiation exposure were the 2 primary outcome variables.

Results: During the study period, 121 patients met inclusion criteria and were matched to controls for a total of 242 subjects. Although the low-dose group LOS was slightly longer, there was no significant difference in LOS. Mean differences were 18 minutes overall (520 vs 502 minutes; P = .497), 11 minutes for admitted patients (587 vs 576 minutes; P = .839), and 22 minutes for discharged patients (490 vs 468 minutes; P = .482). The mean volume CT dose index for the standard-dose CT was 11.6 +- 8.3 and 7.7 +- 4.6 mGy for the reduced-dose CT, a 34% decrease (P b .001).

Conclusion: Use of MBIR in the ED may provide decreased radiation exposure while minimally impacting ED LOS.

(C) 2015

1. Introduction

Over the last decade, there has been increasing concern about the radiation risk posed by diagnostic Computed tomography use in the emergency department (ED) [1]. The concerns stem from both increasing use of CT [2,3] and increasing attention paid to the associated risks in both the medical literature [1,4] and the lay press [5].

One promising way to significantly decrease radiation exposure is to use model-based iterative reconstruction (MBIR) technology, which requires a much smaller dose. Studies of such forms of iterative reconstruction have shown comparable diagnostic Image quality with up to 78% radiation dose reduction [6]. A notable limitation to this technology is a much more complex reconstruction algorithm requiring extensive additional computing hardware and processing time-up to

? Index Medicus sub-headings: Computed Tomography, Radiation, Length of Stay.

?? This work was presented at the 2015 American College of Emergency Physicians Research Forum in October 2014. There was no funding or financial support.

* Corresponding author at: Division of Emergency Medicine University of Washington Medical Center, Box 356123 1959 NE Pacific Street Seattle, WA 98195. Tel.: +1 206 598

0103; fax: +1 206 598 4569.

E-mail address: [email protected] (J. Strote).

an hour for a CT of the abdomen and pelvis. This has been cited as a likely obstacle to use in EDs [7,8].

Given the time sensitivity of ED evaluations, both in terms of diagno- sis and Disposition decision, the advantage of radiation reduction may be outweighed by the expected delays. Alternatively, the complex inter- play of myriad variables that affect ED flow may offset the processing delay, creating a negligible difference in the Time to discharge.

The goal of this study was to compare MBIR and a standard CT reconstructive protocol in terms of ED visit duration and reduction in radiation exposure.

1. Methods
   1. Study design and setting

The study used a retrospective, matched, case-control design.

An a priori analysis of the types of low-dose CTs performed revealed that only abdomen and pelvis studies and Kidney stone protocol studies had enough data points for meaningful analysis; other less frequently used protocols (eg, abdomen only, pelvis only, pancreas, liver,

http://dx.doi.org/10.1016/j.ajem.2015.01.010

0735-6757/(C) 2015

560 M.O. Gatewood et al. / American Journal of Emergency Medicine 33 (2015) 559-562

intravenous pyelogram, hernia, chest) were therefore not studied given the extremely small number performed.

Inclusion criteria were all patients who received a low-dose radia- tion protocol CT of the abdomen and pelvis using MBIR reconstruction and matched controls who underwent the standard radiation dose pro- tocol scans during the same period. Exclusion criteria were patients for whom a match could not be made.

The study was performed at an Urban academic ED from October 2012 to June 2013. Emergency physicians could order low-dose or standard-dose protocol CT at their discretion. This was a previously established option in the ED that had been ongoing since May 2012.

CT acquisition and post-processing

All patients were scanned on a 64-multidetector CT scanner (Dis- covery CT750 HD; GE Healthcare, Waukesha, Wisconsin). Standard- dose CT examinations of the abdomen and pelvis were reconstructed using a blend of filtered back projection and 40% adaptive statistical it- erative reconstruction (GE Healthcare, Milwaukee, Wisconsin). Reduced-dose CT exams were reconstructed using model-based itera- tive reconstruction (VEO, GE Healthcare). Processing times were ap- proximately 45 minutes for low-dose studies and less than 1 minute for standard-dose studies. All CT examinations and post-processing pa- rameters are outlined in Table 1.

Selection of participants

All patients who underwent a CT scan of the abdomen and pelvis using the reduced-dose radiation protocol while in the ED during the study period were identified retrospectively.

A matched group was selected from patients undergoing standard protocol CT scans while in the ED during the same period. Cases were matched by the following criteria: sex, age (b 30, 30-39, 40-49, 50-9, N 60), intravenous and/or oral contrast use, disposition (admitted or discharged), scan indications, and study ordered.

Scan indications were ascertained by reviewing the imaging requisi- tion and ED chart, and matching was performed on the basis of the pa- tient symptoms, clinical diagnostic concern, and abdominal region(s) of symptoms.

Cases were excluded if control matching was not possible due to lack of case-control similarity within the study’s time interval. All included cases were matched exactly for CT protocol type, sex, age, and disposi- tion. Scan indication matching involved interpretive analysis and by its nature was less exact; cases were excluded when a match for reason- ably similar scan indications could not be made. The control cohort was matched case-by-case, 1 control per case.

The study was approved by the lead author’s institution’s Human Subjects Division.

Table 1

CT exam and postprocessing parameters for standard-dose and reduced-dose CT

standard-dose CT Low-Dose CT

Noise index 41 50

Detector collimation, mm 0.625 0.625

Scan field of view, cm 50 50

Recon slice, mm 2.5 2.5

Recon slice interval, mm 2.5 2.5

Pitch 1.375:1 1.375:1

Gantry rotation time, s 0.4-0.8 0.4-0.8

Tube voltage, kVp 120 for BMI N 20 120 for BMI N 20

100 for BMI b 20 120 for BMI b 20

Tube current control ATCM ATCM

Iterative reconstruction ASIR 40% MBIR

Noise Index is based on 0.625-mm slice. Abbreviations: ATCM, automatic tube current modulation; Recon, reconstruction.

Data collection

Two emergency physicians abstracted and matched all the study cases. Variables extracted included the matching variables as well as ED arrival time, CT scan time, ED departure time, and radiation exposure.

Mean volume CT dose index (CTDI_vol) was used to calculate radia- tion exposure and was recorded for both the standard radiation dose and the reduced radiation dose examinations. The anterior-posterior and lateral dimensions of each patient were measured on a CT image from each examination.

Outcome measures and data analysis

ED length of stay and radiation exposure were the 2 primary outcome variables. A paired two-tailed t test was used to compare dif- ference in ED LOS between the case and control group. Mean differences in LOS were also calculated.

A paired two-tailed t test was also used to compare CTDI_vol for the case and control group. The percent average reduction in CTDI_vol was also calculated. Size-specific dose estimates (SSDE) were calculated using the sum of the anterior-posterior and lateral dimensions from each patient [9]. The admitted and discharged subgroups were then an- alyzed in the same fashion.

Chi-square analyses (not shown in tables) of proportions were per- formed comparing the demographics of the unmatched cases to the matched cases (which are proportionately equal to matched controls). Independent sample two-tailed t test analyses were also performed (not shown in tables) to compare the unmatched group’s LOS to the matched cases and controls.

Because there were not enough cases to match to time of day, a sep- arate subanalysis of LOS compared case and control groups for arrival time of day and scan time of day. Time of day was classified into 3 pe- riods (6:00 AM to 1:59 PM, 2:00 PM to 9:59 PM, and 10:00 PM to 5:59

AM). Chi-square analyses of proportions were performed comparing the distribution of subjects by time of day between the cases and con- trols. One-way analysis of variance analyses were performed to assess variance for LOS within the periods.

Microsoft Excel version 14.3.8 (Microsoft Corp, Redmond, Washingtom) was used to organize and maintain study data. Data anal- ysis was conducted with SPSS version 22.0 (SPSS Inc, Chicago, Illinois).

1. Results

During the study period, 144 low-dose MBIR protocol abdomen and pelvis CT studies were performed, of which 121 (84%) met inclusion criteria and were matched to 121 controls. The unmatched low-dose MBIR cases were younger than the matched group (P = .138) but were similar in sex (P = .964), disposition (P = .612), and CT scan pro- tocol performed (P = .806). A summary of the case group characteristics is shown in Table 2.

ED LOS was slightly longer but not significantly different (P b .05) when low-dose studies were used. This was true when looking at all pa- tients as well as the subgroups. Mean differences were 19 minutes over- all, 11 minutes for admitted patients, and 22 minutes for discharged patients (Table 3). In addition, the unmatched group’s LOS was not sig- nificantly different when compared to matched cases (P = .407) or matched controls (P = .637). The data distribution was analyzed and determined to fairly meet the assumption of normal distribution. Log transformation of the data did not demonstrate significance in mean LOS in any of the study groups (all patients P = .107, admitted patients P = .438, and discharged patients P = .157).

There was no significant difference in the sum of the anterior- posterior and lateral dimensions of patients between the standard- dose CT group and the reduced-dose CT group. The mean CTDI_vol for the standard-dose CT was 11.6 +- 8.3 and 7.7 +- 4.6 mGy for the reduced-dose CT, a 34% decrease (P b 0.001). The mean SSDE for the

M.O. Gatewood et al. / American Journal of Emergency Medicine 33 (2015) 559-562 561

standard-dose CT was 14.5 +- 7.7 and 10.5 +- 6.8 for the reduced-dose

CT, a 28% decrease (P b 0.001) (Table 4).

There were no statistically significant differences in distribution of subjects or in LOS when analyzed by time of day. Low-dose protocolED presentations LOS were shorter early in the day (mean, 11 minutes)

Table 3

Length of stay

LOS

Total

(N = 242)

Case

(n = 121)

Control

(n = 121)

Unmatched (n = 23)^a

and longer in the evening (mean, 65 minutes) and overnight (mean, 23 minutes) (Table 5).

Total nean (SD) 511 (226) 520 (206) 502 (245) 483 (156)

Admit mean (SD) 581 (225) 587 (171) 576 (271) 511 (175)

Disch mean (SD) 478 (219) 490 (214) 468 (386) 473 (154)

1. Discussion

Case-control total (N = 242)

Case-control admitted (n = 76)

Case-control discharged (n = 166)

The results of this study suggest that it may be possible to use reduced-dose protocol CT studies in the ED without significant increases	LOS difference Mean 19		11	22
in LOS. We found an average 19-minute increase in ED LOS for our MBIR	SD	304	342	287
studies, which took an additional 45 minutes to process, suggesting that a large portion of the increased processing time in our department was	95% CI P	36-4 .497	101-124 .839	41- 85 .482

offset by other ED flow processes.

Model-based iterative reconstruction involves post-processing methods that use successive steps of converting raw data to images and back again with different algorithms, comparing the results to remove noise. Numerous studies have demonstrated that using a fraction of stan- dard radiation doses, MBIR can increase subjective image quality and signal-to-noise ratio while decreasing mean image noise [7,8,10-12]. The final effect is that high-quality images can be obtained with much lower radiation exposure. The increase in processing time is considerable, which has raised questions about its usefulness in the ED [13].

With increased focus on the effect of ED throughput and efficiency on safety and quality as well as a growing appreciation of the large im- pact testing has on LOS [14], even relatively small delays may be unac- ceptable. ED overcrowding and increased LOS have been associated with worse care and outcomes, as well as lower patient and Physician satisfaction [15].

Emergency department LOS is a function of many variables, only some of which are in the emergency physician’s control [14,16,17]. De- lays related to radiology, and especially CT, have been noted to play a large role [18,19], and ordering CTs is very common. CTs were part of ED evaluations in more than 14% of visits in 2008, a percentage that has grown steadily over the last decade [2]. These real issues make the introduction of a CT technology with large additional processing time challenging.

The question then becomes whether the radiation reduction benefits of MBIR offset potential increases in LOS. Recently published estimates of the effects of CT use make the positive impact of decreasing radiation dose clear. Calculations of increased risk suggest that as many as 2% of future cancers will be attributable to CT; [1] 29,000 excess cancers with a 50% mortality are projected to occur from CTs performed in 2007, a time when the concern about radiation exposure as well as the technology to limit it was notably less [20]. Despite these effects,

Abbreviations: CI, confidence interval; Disch, discharge.

^a Unmatched data not included in total.

there remains an underestimation of the risk by both physicians and pa- tients [21-25], and CT use in EDs continues to rise [2].

Although one solution to this problem is to decrease CT use in the ED, some have argued that diagnostic accuracy and speed of CT can off- set associated risk [26]. This becomes an increasingly convincing argu- ment if the amount of radiation exposure from CT decreases considerably.

In addition, as radiation risk is discussed increasingly in the lay press

[5] and patients continue to feel that CT is an important part of emer- gent medical evaluations [27], there is an additional potential for real benefit from increased patient satisfaction.

Our institution’s standard prior to the introduction of MBIR was to use an older generation iterative reconstruction that leads to a smaller reduction of radiation but without notable processing times. Our expe- rience of decreasing radiation significantly further with a relatively low increase in LOS has led to an increase in MBIR use in our ED, championed by leaders in both the radiology and EDs. Whether, in other departments, such a limited increase in total LOS is reproducible and a tolerable tradeoff for radiation reduction would depend on many different variables, including patient preferences, and be institution-dependent.

The speed of iterative reconstruction will certainly decrease in com- ing years as computing power increases. In the mean time, this study suggests that the processing time of iterative reconstruction may not correspond to substantial LOS increases, making the tradeoff for de- creased radiation exposure more attractive.

1. Limitations

Table 2

Demographics

Study performed, no. (%)

Total

(N = 265)

Case

(n = 121)

Control

(n = 121)

Unmatched (n = 23)

There were many limitations to this study that affect its interpretation.

This study is underpowered to detect small differences in LOS: the range of ED visit length for our entire ED population is extremely large; therefore, a considerably larger sample size would be required to do so. However, we feel that the similarity in mean LOS in these

Ab pel 162 (67) 81 (67) 81 (67) 16 (70)

Kid stone prot 87 (33) 40 (33) 40 (33) 7 (30) Sex

% male 43 (n = 114) 43 (n = 52) 43 (n = 52) 44 (n = 10)

Age

% b 30	52	50	50	74
% 30-39	35	36	36	17

matched patients is noteworthy when considered with the substantial decrease in radiation exposure.

Table 4

Radiation values

% 40-49	12	12	12	4		Standard-dose CT	Low-dose CT	P
% 50-59	1	1	1	4	PA/LAT sum, cm	56 +- 11	56 +- 8	.67
% N 60	b 1	1	1	0	CTDIvol, mGy	11.6 +- 8.3	7.7 +- 4.6	b.0001
Disposition, no. (%)					SSDE, mGy	14.5 +- 7.7	10.5 +- 6.8	b.0001

Abbreviations: Ab pel, abdomen and pelvis protocol CT; Kid stone prot, kidney stone protocol CT.

Admit	82 (31)	38 (31)	38 (31)	6 (26)
Discharge	183 (69)	83 (69)	83 (69)	17 (74)

Data are means +- standard deviations. Abbreviations: CTDI, volume CT dose index; PA/Lat sum, sum of anterior-posterior and lat- eral patient dimensions.

562 M.O. Gatewood et al. / American Journal of Emergency Medicine 33 (2015) 559-562

Table 5

Time of day

Period	Case (n = 121)				Control (n = 121)	Case (n = 121)				Control (n = 121)
	Arrival time				Arrival time	Scan time				Scan time
6:00 AM to 1:59 PM	59 (49%)				46 (38%)	22 (18%)				32 (26%)
2:00 PM to 9:59 PM	42 (35%)				44 (36%)	60 (50%)				47 (39%)
10:00 PM to 5:59 AM	20 (17%)				31 (26%)	39 (32%)				42 (35%)
P	.12					.17
Mean LOS	Case		Control		Difference	Case		Control		Difference
	Arrival time		Arrival time			Scan time		Scan time
6:00 AM to 1:59 PM	487 (127)		498 (252)		-11	548 (207)		491 (290)		57
2:00 PM to 9:59 PM	553 (286)		488 (230)		65	530 (236)		522 (237)		8
10:00 PM to 5:59 AM	549 (187)		526 (259)		23	490 (148)		488 (217)		2
P	.178		.797			.514		.777

The upper portion of this table details the distribution of subjects by time of day defined by both arrival and scan times. The lower portion compares the LOS by time of day defined by both arrival and scan times.

This study was also performed in a single institution with unique ra- diology and EDs; the findings may not be applicable more broadly.

Given that low-dose scans were ordered per provider discretion, se- lection bias is likely present. The choice of which dosing protocol was ordered may have been influenced by age, sex, case complexity, prior medical history, and radiologist input when scan protocol decisions are discussed. We did not match for comorbidities or prior medical his- tory. It is unclear what effect these other differences may have made on the outcomes.

Although analyzing the presentation time of day did not yield statis- tically significant differences, the LOS increase of over an hour for pa- tients presenting in the afternoon raises the question of whether other unmeasured variables may have affected outcomes.

We were forced to exclude 23 cases due to an inability to match them closely to a standard-protocol patient. These cases may have dif- fered in a substantial way from the included cases, which may have bi- ased our results.

Importantly, we did not assess for impact on diagnosis by using MBIR. Although the comparable image quality achieved by MBIR should yield no significant impact on diagnosis, any such impact would be a critical factor in the decision to use MBIR or not.

Finally, this study only looked at abdomen/pelvis and kidney stone protocol studies. Length of stay differences might be different for ED presentations requiring chest CT or other specialized studies.

References

1. Bremner DJHE. Computed tomography: an increasing source of radiation exposure. NEJM 2007;357:2277-84.
2. Berdahl CT, Vermeulen MJ, Larson DB, Schull MJ. Emergency department computed tomography utilization in the United States and Canada. Ann Emerg Med 2013;62: 486-94 [e3].
3. Korley FKPJ, Kirsch TD. Use of advanced radiology during visits to US emergency de- partments for injury-related conditions 1998-2007. JAMA 2010;304:1465-71.
4. Smith-Bindman R, Lipson J, Marcus R, Kim KP, Mahesh M, Gould R, et al. Radiation dose associated with common computed tomography examinations and the associ- ated lifetime attributable risk of cancer. Arch Intern Med 2009;169:2078-86.
5. Redberg RF. We are giving ourselves cancer. New York Times; 2014.
6. Liu L. Model-based iterative reconstruction: a promising algorithm for today’s com- puted tomography imaging. J Med Imaging Radiat Sci 2014;45:131-6.
7. Pickhardt PJ, Lubner MG, Kim DH, Tang J, Ruma JA, del Rio AM, et al. Abdominal CT with model-based iterative reconstruction (MBIR): initial results of a prospective trial comparing ultralow-dose with standard-dose imaging. AJR Am J Roentgenol 2012;199:1266-74.
8. Singh S, Kalra MK, Do S, Thibault JB, Pien H, O’Connor OJ, et al. Comparison of hybrid and pure iterative reconstruction techniques with conventional filtered back projec-

tion: dose reduction potential in the abdomen. J Comput Assist Tomogr 2012;36: 347-53.

1. AAPM Task Group 204. Size Specific Dose Estimates (SSDE) in Pediatric and Adult CT Examinations; 2011[Accessed July, 2014, at http://aapm.org/publications.].
2. Fuchs TA, Stehli J, Bull S, Dougoud S, Clerc OF, Herzog BA, et al. Coronary computed tomography angiography with model-based iterative reconstruction using a radia- tion exposure similar to chest X-ray examination. Eur Heart J 2014;35:1131-6.
3. Ning P, Zhu S, Shi D, Guo Y, Sun M. X-ray dose reduction in abdominal computed tomog- raphy using advanced iterative reconstruction algorithms. PLoS One 2014;9:e92568.
4. Yanagawa M, Gyobu T, Leung AN, Kawai M, Kawata Y, Sumikawa H, et al. Ultra-low- dose CT of the lung: effect of iterative reconstruction techniques on image quality. Acad Radiol 2014;21:695-703.
5. Katsura M, Matsuda I, Akahane M, Sato J, Akai H, Yasaka K, et al. Model-based itera- tive reconstruction technique for radiation dose reduction in chest CT: comparison with the adaptive statistical iterative reconstruction technique. Eur Radiol 2012; 22:1613-23.
6. Kocher KE, Meurer WJ, Desmond JS, Nallamothu BK. Effect of testing and treatment on emergency department length of stay using a national database. Acad Emerg Med 2012;19:525-34.
7. Casalino E, Choquet C, Bernard J, Debit A, Doumenc B, Berthoumieu A, et al. Predic- tive variables of an emergency department quality and performance indicator: a 1- year prospective, observational, cohort study evaluating hospital and emergency census variables and emergency department time interval measurements. Emerg Med J 2013;30:638-45.
8. Capuano F, Lot AS, Sagnes-Raffy C, Ferrua M, Brun-Ney D, Leleu H, et al. Factors as- sociated with the length of stay of patients discharged from emergency department in France. Eur J Emerg Med 2014 [Epub ahead of print].
9. Perimal-Lewis L, Ben-Tovim DI, Li JY, Hakendorf PH, Thompson CH. Emergency depart- ment lengths of stay: characteristics favouring a delay to the admission decision as dis- tinct from a delay while awaiting an inpatient bed. Intern Med J 2014;44:384-9.
10. Kanzaria HK, Probst MA, Ponce NA, Hsia RY. The association between advanced diag- nostic imaging and ED length of stay. Am J Emerg Med 2014;32:1253-8.
11. Perimal-Lewis L, Ben-Tovim DI, Li JY, Hakendorf PH, Thompson CH. Emergency de- partment lengths of stay: characteristics favouring a delay to the admission decision as distinct from a delay while awaiting an inpatient bed. Inter Med J 2014;44:384-9.
12. Berrington de Gonzalez A, Mahesh M, Kim KP, Bhargavan M, Lewis R, Mettler F, et al. Projected cancer risks from Computed tomographic scans performed in the United States in 2007. Arch Intern Med 2009;169:2071-7.
13. Brown N, Jones L. Knowledge of medical imaging radiation dose and risk among doc- tors. J Med Imaging Radiat Oncol 2013;57:8-14.
14. Sodickson A. Strategies for reducing radiation exposure from multidetector comput- ed tomography in the acute care setting. Can Assoc Radiol J 2013;64:119-29.
15. Takakuwa KM, Estepa AT, Shofer FS. Knowledge and attitudes of emergency department patients regarding radiation risk of CT: effects of age, sex, race, education, insurance, body mass index, pain, and seriousness of illness. AJR Am J Roentgenol 2010;195:1151-8.
16. Weigner MB, Dewar KM, Basham HF, Rupp VA, Greenberg MR. Impact of education on physician attitudes toward computed tomography utilization and consent. J Emerg Med 2012;43:e349-53.
17. Youssef NA, Gordon AJ, Moon TH, Patel BD, Shah SJ, Casey EM, et al. Emergency de- partment patient knowledge, opinions, and risk tolerance regarding computed to- mography scan radiation. J Emerg Med 2014;46:208-14.
18. Schwartz DT. US emergency physicians order too many computed tomography scans-or do they? Ann Emerg Med 2013;62:495-7.
19. Baumann BM, Chen EH, Mills AM, Glaspey L, Thompson NM, Jones MK, et al. Patient perceptions of computed tomographic imaging and their understanding of radiation risk and exposure. Ann Emerg Med 2011;58:1-7 [e2].