Evaluation for occult sepsis incorporating NIRS and emergency sonography

a b s t r a c t

Purpose: We aim to determine whether the combination of regional tissue oxygen saturation (StO₂) measure- ment using Near-infrared spectroscopy (NIRS), Inferior vena cava collapsibility and ejection fraction (EF) is able to detect occult sepsis.

Methods: We included adult patients in the emergency department with at least one of the following: fever; any one component of the quick sepsis-related Organ function assessment (SOFA) score; heart rate >= 100 beats per minute; or white cell count b4.0 x 10⁹/L or N12.0 x 10⁹/L. StO₂ parameters, IVC collapsibility and EF were assessed.

Primary outcome was composite of admission to intensive care unit, hypotension requiring fluid resuscitation or vasopressor use, and antibiotic escalation.

Results: We included 184 patients with mean age of 55.4 years and slight male predominance (51.6%). Increase in temperature (adjusted odds ratio [aOR] 3.05; 95% confidence interval [CI] 1.16 to 8.02), higher white cell counts (aOR 1.10; 95% CI 1.03 to 1.19), increase in time taken to new StO₂ baseline (aOR 1.03; 95% CI 1.01 to 1.06) and reduced EF (aOR 33.9; 95% CI 2.19 to 523.64) had higher odds of achieving the primary outcome.

Conclusion: Change in StO₂ and time taken to reach new StO₂ baseline, combined with EF could potentially pre- dict sepsis among patients with infection.

Introduction

Early recognition and intervention is imperative for patients pre- senting with sepsis. The most recent international consensus definitions characterized sepsis as life-threatening organ dysfunction caused by a dysregulated host response to infection [1]. Key pathophysiological de- rangements in sepsis include tissue ischemia from microcirculatory and endothelial lesions, and vasodilatation as a consequence of vasoactive mediator release [2-4]. The mainstay of hemodynamic management in

* Corresponding author at: Emergency Medicine Department, National University Hospital, 5 Lower Kent Ridge Road, 119074, Singapore.

E-mail addresses: [email protected] (H.H.E. Ang), [email protected] (J.C.L. Tan), [email protected] (W.H. Ho), [email protected] (W.S. Kuan), [email protected] (M.T. Chua).

sepsis is to overcome these deficits through improvement of perfusion and simultaneously oxygenation to the tissues and organs [5,6].

Methods that have been used to ascertain the balance between oxy- gen delivery and consumption in patients with sepsis include Central venous oxygen saturation and serial lactate measurements [7,8]. However, these processes are invasive and require time to obtain the relevant parameters. Near-infrared spectroscopy (NIRS) has been inves- tigated as a potential surrogate tool to invasive methods for measure- ment of the adequacy of organ perfusion, especially in the setting of trauma [9]. NIRS utilizes infrared light waves (wavelengths between 680 and 800 nm) emitted from a sensor placed above the skin to mea- sure the ratio of oxygenated to total hemoglobin in peripheral tissue in vivo resulting in a parameter called tissue oxygen saturation (StO₂) [10]. This technology has been shown to provide reproducible, continu- ous, noninvasive, real-time monitoring of regional Tissue oxygenation at

https://doi.org/10.1016/j.ajem.2018.02.020

the bedside and may be a useful adjunct in the emergency department (ED) to identify patients with sepsis before its florid clinical manifesta- tion [11,12].

In addition to measuring absolute StO₂ values, the inclusion of the vascular occlusion test (VOT) could improve the discriminatory func- tion between normal and abnormal microcirculatory states [13-15]. VOT involves transient occlusion of the blood vessels supplying a pe- ripheral site whilst observing the steady decline of the StO₂ and its sub- sequent rise during the reperfusion phase. Dynamic parameters that can be derived from this ischemic test are the rate of deoxygenation (DeStO₂), reflecting local oxygen consumption and cardiovascular re- serve, and the rate of reoxygenation (ReStO₂), indicating the post- ischemic reactivity of the microcirculation [16,17].

Point-of-care ultrasonography is ubiquitous in the ED. Real-time vi- sualization of the cardiovascular system through bedside echocardiog- raphy and measurement of inferior vena cava collapsibility could provide a real-time gauge of the cardiac pump function, volume status and fluid responsiveness of the patient with sepsis [18,19].

This study aims to determine if the combination of StO₂ measure-

ment, IVC collapsibility assessment and Bedside echocardiography is able to detect the presence of occult sepsis in the ED before its overt clin- ical and laboratory manifestation.

Methods

Study design and setting

This was a prospective observational study of patients with suspected sepsis at the ED of National University Hospital, a 1225-bed tertiary academic medical center in Singapore from February 2016 to April 2016. The ED receives over 110,000 visits annually, of which about 47% of the cases require urgent (42.5%) or immediate (4.5%) care. Recruitment of subjects was carried out on weekdays from 0700 h to 1900 h and occasionally on weekends. Patients or their legally acceptable representatives provided written informed consent, and all protocols were approved by the Domain Specific Review Board, Na- tional Healthcare Group, Singapore (DSRB 2015/01266).

Selection of participants

Eligible patients were adults 21 years of age and above who met at least one of following inclusion criteria: (1) presence of a documented temperature of N38.0 ?C in an outpatient setting; or (2) any one compo- nent of the quick SOFA score [1] (altered mentation with Glasgow Coma Scale <=13, respiratory rate >= 22 breaths per minute, systolic blood pressure <= 100 mm Hg); or (3) heart rate N100 beats per minute; or

(4) white cell count b4.0 x 10⁹/L or N12.0 x 10⁹/L.

Exclusion criteria were those younger than 21 years, known preg- nancy, prisoners, do-not-resuscitate status, requirement for immediate surgery, oncological patients on active chemotherapy or hematological malignancy, and treating physician deems aggressive care unsuitable. Patients who had obvious sepsis or septic shock, such as requiring intu- bation, vasopressors and hypotension not responding to fluid challenge were also excluded.

Patients were followed-up for 28 days electronically for primary and secondary outcomes of interest.

Study protocol

StO₂ measurement

StO₂ was measured using the InSpectra(TM) Tissue Oxygen Monitor (Model 650) (Hutchinson Technology Inc., Hutchinson, MN). The StO₂ Sensor was placed noninvasively onto the patient’s thenar eminence and secured using an adhesive tape. The patient’s initial StO₂ was mea- sured and recorded as soon as they arrived in the ED and written in- formed consent was obtained. Baseline StO₂ was recorded only after a

tissue hemoglobin index (THI) value above 10.0 was obtained and a sta- ble StO₂ value was observed on the Tissue Oxygen Monitor after 1 min of stabilization. The manufacturer recommends a minimum THI of 5 to obtain a sufficient signal in most circumstances. However, we opted to employ a minimum THI of 10 in our study, which is twice the signal strength to increase the reliability of the StO₂ readings that were obtained.

Vascular occlusion test

Subsequently, the VOT was conducted through placement of a sphygmomanometer cuff over the ipsilateral arm and inflated 50 mm Hg above the patient’s systolic blood pressure. The pneumatic cuff was kept inflated for 3 min while the StO₂ values were recorded over six time points at 30-second intervals. After the 3-minute ischemic period had elapsed, the cuff was deflated rapidly within 3 s. The hyper- emic phase was monitored for the highest StO₂ measurement reached and the time taken to achieve this value. Typically, the StO₂ values would decrease slowly from the peak back to a new baseline value. When the StO₂ had reached a new stable value without fluctuating N2%, this value was recorded as the new post-VOT StO_2Baseline. Other dy- namic parameters resulting from this ischemic challenge were also ob- tained: DeStO₂, ReStO₂, and ?StO₂ (see Fig. 1).

inferior vena cava ultrasonography

All ultrasonographical evaluation (IVC assessment and echocardiog- raphy) was performed by a single investigator (NYYN), who received formal ultrasound training prior to study commencement. Bedside ul- trasonography of the IVC was performed at initial presentation, 1 h and 2 h later. A curviLinear probe was used to obtain images of the IVC using the sub-xiphoid long-axis view with the patient in supine posi- tion. IVC diameters in M-mode were measured 20 mm caudal from the junction of IVC and hepatic vein. The IVC collapsibility index was cal- culated using the formula: ([max IVC diameter – min IVC diameter] / max IVC diameter) x 100.

Bedside echocardiography

Transthoracic 2D-echocardiography was performed at the bedside with the patient in semi-recumbent position at initial presentation. Two acoustic windows were used: parasternal long and short axes, and Apical 4-chamber views. The recordings were saved for subsequent review. Contractility based on ejection fraction (EF) was classified using visual estimation into one of four categories: normal, mild to moder- ately depressed, severely depressed, and hyperdynamic, by 2 indepen- dent reviewers (WSK and MTC) blinded to the clinical data. Both reviewers were board-certified emergency medicine physicians accredited in emergency point-of-care ultrasonography. Disagreements were resolved through consensus discussion.

Images of the IVC ultrasound and echocardiographic recordings were made using the SonoSite Edge II (Fujifilm SonoSite, Inc., Bothell, WA) and Terason (Teratech Corporation, Burlington, MA) ultrasound scanners, depending on availability.

Outcome measures

The primary outcome measure is sepsis represented by a composite of proportion of patients requiring intensive care unit (ICU) or high de- pendency (HD) stay, hypotensive episodes requiring fluid resuscitation or vasopressor use, and deterioration of clinical status necessitating the escalation of antibiotics. Secondary outcome measures include dis- charge status, mortality, and length of ICU and hospital stay.

Statistical analysis

Patients who developed the primary outcome measure were com- pared with those who did not. Results were analyzed using Stata 14 (StataCorp LP, College Station, TX). Categorical data were analyzed

Fig. 1. Vascular Occlusion Test (VOT) and derived tissue oxygenation parameters. StO₂, tissue oxygen saturation; StO_2Baseline, resting StO₂ before VOT; DeStO₂, desaturation slope; ReStO₂, resaturation slope; Post-VOT StO_2Baseline, new baseline StO₂ reached after VOT; StO₂ difference between StO_2Max – StO_2Baseline

using Chi-squared test or Fisher’s exact test, where appropriate. Kruskal-Wallis and Mann-Whitney U tests were used for non- parametric variables; analysis of variance and Student’s t-test for para- metric variables. Medians were reported with interquartile ranges (IQR) and means with standard deviations (SD). Multivariate logistic re- gression was performed to adjust for effects of macrohemodynamics and complete cell count levels on tissue oxygenation parameters.

Sample size was calculated based on the estimation of proportion of detection of patients who have the composite primary outcome. By es- timating the true proportion to be 10%, at an alpha of 5% and confidence interval of 95%, we estimate that at least 139 subjects are required for the study. A P value of b0.05 was set for statistical significance.

Results

A total of 546 patients were screened with 184 patients recruited over the study period (Fig. 2). The patients were subsequently divided into 3 groups: group A consisted of patients who had final diagnoses

unrelated to infectious causes (n = 63); group B comprised patients with infections but did not achieve any composite primary outcome (n = 107); group C included patients with infections and any of the composite primary outcome of ICU or HD transfer, hypotensive epi- sodes requiring fluid resuscitation or vasopressor use, and escalation of antibiotics due to worsening clinical status (n = 14).

There was no difference in baseline demographics between the 3 groups (Table 1). Group C, however, has significantly larger proportion of patients with Renal impairment compared to the other groups. In ad- dition, this same group of patients had statistically significantly higher temperature and heart rate, lower systolic and diastolic blood pressures, higher white cell count with neutrophilia and lower hemoglobin and Platelet levels.

Between the non-infective group (group A) and infective group without positive composite primary outcomes (group B), the tissue ox- ygenation readings did not show significant differences (Table 2). Al- though the ?StO₂ (maximum StO₂ achieved – baseline StO₂) and resaturation slope achieved statistical significance between these 2

Fig. 2. Recruitment flowchart. DNR, do-not-resuscitate; VOT, vascular occlusion test.

Table 1 Baseline demographics, comorbidities and clinical parameters.
	Group A	Group B	Group C	P value
	(n = 63)	(n = 107)	(n = 14)
Demographics
Age, mean (SD)	54.98 (18.11)	54.36 (18.20)	65.29 (15.76)	0.103
Gender, male, n (%) Race, n (%)	27 (42.86)	61 (57.01)	7 (50.00)	0.202 0.924^a
Chinese	35 (55.56)	57 (53.27)	6 (42.86)
Malay	16 (25.40)	28 (26.17)	4 (28.57)
Indian	8 (12.70)	17 (15.89)	3 (21.43)
Other	4 (6.35)	5 (4.67)	1 (7.14)
Comorbidities, n (%) Asthma	6 (9.52)	20 (18.69)	1 (7.14)	0.206^a
Chronic obstructive pulmonary disease	2 (3.17)	2 (1.87)	0 (0.0)	0.729^a
Stroke	4 (6.35)	8 (7.48)	0 (0.00)	0.900^a
Atrial fibrillation	7 (11.11)	10 (9.35)	2 (14.29)	0.693^a
Diabetes mellitus	20 (31.75)	35 (32.71)	3 (21.43)	0.751^a
Hypertension	29 (46.03)	41 (38.32)	9 (64.29)	0.151
Dyslipidemia	29 (46.03)	37 (34.58)	7 (50.00)	0.241
Ischemic heart disease	19 (30.16)	18 (16.82)	3 (21.43)	0.126
Cardiomyopathy	2 (3.17)	1 (0.93)	1 (7.14)	0.178^a
Renal impairment	7 (11.11)	6 (5.61)	6 (42.86)	b0.001
Liver cirrhosis	1 (1.59)	3 (2.80)	0 (0.00)	1.000^a
Cancer	6 (9.52)	11 (10.28)	2 (14.29)	0.859^a
Clinical parameters, median (IQR) Temperature, ^oC	36.8 (36.6-37.5)	37.5 (36.9-38.4)	38.15 (36.8-38.9)	b0.001
Systolic blood pressure, mm Hg	133 (117-151)	127 (116-144)	115.5 (108-126)	0.009
Diastolic blood pressure, mm Hg	76 (66-85)	74 (67-84)	65.5 (58-73)	0.020
Mean arterial pressure, mm Hg	95 (85-108)	93 (84-103)	82 (75-91)	0.006
SpO₂, %	98 (97-99)	98 (96-99)	97.5 (96-98)	0.099
Respiratory rate, breaths per minute	20 (18-23)	20 (18-21)	19 (18-20)	0.442
Heart rate, beats per minute	97 (80-107)	100 (89-110)	109.5 (98-115)	0.028
White cell count, x10⁹/L^b	8.71 (7.27-10.98)	9.49 (7.11-13.15)	14.82 (10.92-13.15)	b0.001
Absolute Neutrophil count (ANC), x10⁹/L	5.54 (4.59-7.87)	6.8 (4.74-10.3)	12.51 (8.36-20.23)	b0.001
% ANC	65.4 (59.9-74.2)	74.7 (64.6-81.7)	89.75 (79.1-92.5)	b0.001
Hemoglobin, g/dL	13.4 (11.7-14.7)	13.7 (12.1-14.8)	12.15 (11.1-13.5)	0.041
Hematocrit, %	39.1 (35.9-43.4)	39.5 (35.3-42.9)	35.65 (33.4-37.9)	0.029
Platelet, x10⁹/L	250 (216-318)	240 (174-296)	202 (132-264)	0.049

Group A, non-infective patients; Group B, infective patients without any of the composite primary outcomes; Group C, infective patients with any one of the composite primary outcomes. SD, standard deviation; IQR, interquartile range.

^a Test of statistical significance using Fisher’s exact test.

^b Complete blood count results available in 182 patients.

groups at 1 h, this difference was not consistently seen at baseline and at 2 h. The ?StO₂ remained significantly different between the infective group with Composite outcome and the other 2 groups (Table 2) across the 3 time points of measurement.

A larger proportion of patients with positive outcomes had de- pressed EF whereas collapsibility of inferior vena cava did not predict clinical deterioration (Table 3). In a subgroup analysis of patients with depressed EF (mild to severe), the median baseline StO₂ was not statis- tically different between Group B and C (78% [IQR 75 to 83%] vs. 84% [IQR 74 to 87%], p = 0.363) and the trend towards a lesser ?StO₂ remained in Group C compared to B (10% [IQR 8-11%] vs. 13% [IQR 10 to 16%], p = 0.05). The StO₂ parameters did not change significantly at baseline, and 2 h after fluids and antibiotics were administered (re- sults not shown).

No mortality was observed in our study cohort. Among the patients in group C, 71.4% (10/14) developed hypotension, 28.6% (4/14) required admission to the ICU, 64.3% (9/14) received fluid resuscitation, and 21.4% (3/14) required vasopressors, while 57.1% (8/14) required antibi- otic escalation. The median length of stay in the ICU for patients in group C was 0.5 days (IQR 0 to 2 days) compared with 0 days among groups A and B (p b 0.001). Admission to a rehabilitation facility upon hospital discharge was required in 21.4% (3/14) of patients in group C compared to 2.35% (4/170) of patients in groups A and B who required further re- habilitation (p = 0.010). Patients in group C also had a longer median length of stay (9.5 days [IQR 7 to 19]) compared to the other 2 groups (2 days [IQR 0 to 4 days], p b 0.001).

Among the infective patients, after multivariate logistic regression to adjust for macrohemodynamics and complete cell count levels; time taken to reach new baseline StO₂, ?StO₂ and EF along with temperature, systolic blood pressure, white cell count and Hemoglobin levels remained significant predictors of our composite primary outcomes (Table 4).

Discussion

In this prospective observational study of ED patients presenting with a myriad of symptoms, our results show that a combination of NIRS technology and bedside echocardiography may have the potential in detecting patients with occult sepsis before their overt clinical mani- festation. In our study population, septic patients who achieved the composite outcome of ICU or HD transfer, hypotensive episodes requir- ing fluid resuscitation or vasopressor use, and/or escalation of antibi- otics due to worsening clinical status (Group C) showed significantly lower ?StO₂ values at initial presentation (time 0) compared to non- septic patients (Group A) and patients who did not achieve our compos- ite outcome (Group B). In addition, infective patients with a concomi- tant depressed ejection fraction (EF) are associated with poorer outcomes. To the best of our knowledge, this is the first study that com- bines the use of both tissue oxygenation parameters and point-of-care ultrasonography in the ED to predict clinical outcomes.

Patients with infections and sepsis present to the ED at varying stages of their illnesses. ED physicians are often able to recognize pa- tients in severe sepsis from abnormal macrohemodynamics. The

Table 2

Tissue oxygenation parameters at baseline (0 h), 1 hour (1 h) and 2 hours (2 h) after enrolment.

StO₂ parameters	Groups			P values
	A	B	C	A vs. B	(A + B) vs. C
At time = 0 h
No. of observations	63	107	14
Baseline StO₂, %	79 (76-83.5)	78 (76-83)	84 (77-87)	0.738	0.120
?StO₂, %	13 (9-16)	13 (10-15)	10 (7-11)	0.864	0.003
Desaturation slope, %/min	-11.67 (-13.67 to -9)	-11 (-13.33 to -9.33)	-12 (-12.67 to -9.67)	0.584	0.970
Resaturation slope, %/sec	1.34 (0.91-1.88)	1.43 (1.00-1.71)	1.07 (0.83-1.36)	0.971	0.031
New baseline StO₂, %	81 (76-85.5)	79 (77-83)	82.5 (77-85)	0.570	0.413
Time to new baseline, sec	155.5 (113-184.5)	150 (120-183)	143 (128-196)	0.332	0.845
StO₂ parameters	Groups			P values
	A	B	C	A vs. B	(A + B) vs. C
At time = 1 h
No. of observations	56	90	11
Baseline StO₂, %	80 (74-83)	81 (76-85)	83 (80-86)	0.192	0.134
?StO₂, %	13 (10-17.5)	11.5 (8-15)	8 (1-10)	0.040	0.003
Desaturation slope, %/min	-11.33 (-14.5 to -9.33)	-10.33 (-12.33 to -8.67)	-11.33 (-14.33 to -8.33)	0.159	0.818
Resaturation slope, %/sec	1.64 (1.04-1.93)	1.17 (0.90-1.58)	0.95 (0.70-1.69)	0.025	0.188
New baseline StO₂, %	81 (75-85.5)	81 (76-85)	80 (80-85)	0.909	0.812
Time to new baseline, sec	140 (118.5-189)	149.5 (123-171)	117 (97-140)	0.987	0.026
StO₂ parameters	Groups			P values
	A	B	C	A vs. B	(A+B) vs. C
At time = 2 h
No. of observations	51	79	9
Baseline StO₂, %	80 (75-85)	79 (75-85)	81 (76-84)	0.809	0.601
?StO₂, %	12.5 (8-16)	12 (8-15)	8 (6-10)	0.123	0.020
Desaturation slope, %/min	-11.33 (-15.33 to -8.33)	-10.67 (-12.33 to -9)	-11 (-13.33 to -7)	0.333	0.801
Resaturation slope, %/sec	1.26 (1-1.89)	1.20 (1.03-1.6)	1.05 (0.91-1.24)	0.418	0.178
New baseline StO₂, %	81 (76-86)	80 (77-85)	81 (75-82)	0.747	0.578
Time to new baseline, sec	153 (110-172)	144.5 (122-183)	140 (116-150)	0.907	0.427

Medians with IQR are shown.

StO₂, tissue oxygen saturation; ?StO₂, difference between StO_2Max – StO_2Baseline.

challenge is to identify patients with occult sepsis, who may not be he- modynamically compromised and have no obvious abnormal clinical signs on initial review. These patients often deteriorate unpredict- ably within a short time span; not infrequently with wrong siting of care of such patients in the general ward instead of an area with higher acuity. NIRS technology, which is noninvasive and provides a glimpse at the microcirculation, has shown some promise in

assessing tissue oxygenation in sepsis and thus useful for prognos- tication of Disease progression [20-22]. However, current available studies that utilize tissue oxygenation monitors are limited and mostly conducted in the ICU setting, with paucity of evidence among ED patients [21].

Existing evidence support the consensus for the incorporation of a VOT to increase the diagnostic sensitivity of StO₂ measurements in

Table 3

Inferior vena cava collapsibility and ejection fraction readings.

Other noninvasive markers, n (%) Group A Group B Group C P value IVC collapsibility at 0 h N = 51 N = 87 N = 12 0.733^a

– Less than 50% 43 (84.31) 68 (78.16) 10 (83.33)

– More than or equal to 50% 8 (15.69) 19 (21.84) 2 (16.67)

IVC collapsibility at 1 h N = 52 N = 78 N = 10 0.287^a

– Less than 50% 45 (86.54) 60 (76.92) 7 (70.0)

– More than or equal to 50% 7 (13.46) 18 (23.08) 3 (30.0)

IVC collapsibility at 2 h N = 46 N = 68 N = 8 0.168^a

– Less than 50% 41 (89.13) 52 (76.47) 6 (75.0)

– More than or equal to 50% 5 (10.87) 16 (23.53) 2 (25.0)

Ejection fraction at 0 h N = 62 N = 102 N = 14 0.033^a

– Normal 43 (69.35) 76 (74.51) 7 (50.0)

– Mild to moderately depressed 12 (19.35) 12 (11.76) 6 (42.86)

– Severely depressed 6 (9.68) 5 (4.60) 1 (7.14)

– Hyperdynamic 1 (1.61) 9 (8.82) 0 (0.0)

N, number of observations.

Group A, non-infected patients; Group B, infected patients without any of the composite primary outcomes; Group C, infected patients with any one of the composite primary outcomes.

^a Test of statistical significance using Fisher’s exact test.

Table 4

Multivariate logistic regression for infective groups (groups B and C).

Variable	Adjusted odds ratio	95% CI	P value
Temperature	3.05	1.16 to 8.02	0.023
Systolic blood pressure	0.89	0.82 to 0.97	0.006
White cell count	1.10	1.03 to 1.19	0.006
Hemoglobin level	0.45	0.25 to 0.80	0.007
?StO₂	0.76	0.59 to 0.98	0.032
Desaturation slope	0.68	0.45 to 1.01	0.055
Time to new baseline (sec)	1.03	1.01 to 1.06	0.016
Depressed ejection fraction^a	33.9	2.19 to 523.64	0.012

Group B, infected patients without any of the composite primary outcomes; Group C, in- fected patients with any one of the composite primary outcomes.

StO₂, tissue oxygen saturation; ?StO₂, difference between StO_2Max – StO_2Baseline.

^a Depressed ejection fraction comprised patients with mild, moderate and severely depressed EF. No patient in group C had hyperdynamic EF.

septic patients for evaluation of outcomes [20-22]. Prior studies evaluat- ing NIRS have shown that two VOT-derived parameters, namely ?StO₂

[23] and ReStO₂ [20], are promising markers in reliably differentiating patients with severe sepsis or septic shock from those without shock. Our study showed similar results with regard to ?StO₂; patients with in- fection and subsequent adverse clinical outcomes had lower ?StO₂ values at baseline, 1 h and 2 h later. Several investigators have postu- lated that the lower ?StO₂ after a VOT may be due to underlying micro- vascular alterations present in sepsis, giving rise to a lower hyperemic reaction after an ischemic challenge [24, 26-30].

In contrast, the gradient of the resaturation slope was significantly lower (i.e. takes a longer time to re-saturate) at baseline for infective pa- tients with adverse outcomes but did not show any differences between groups at subsequent measurements. The difference in the recovery slope at baseline corresponded to prior studies [20], and resonates with the hypothesis that microcirculatory dysfunction exists among pa- tients with severe sepsis and reflects a reduced reserve capacity in this group of patients.

Although a higher desaturation gradient showed a trend towards lower odds of adverse clinical outcomes in multivariate analysis (ad- justed OR 0.68; 95% CI 0.45 to 1.01, p = 0.055) that is consistent with previous studies [10], the results did not reach statistical significance. This association should be further evaluated in larger studies.

Similar to existing critical care evidence, the assessment of IVC in predicting outcomes was not useful in predicting clinical deterioration. Fluid resuscitation helps to increase stroke volume hence helping with the circulatory failure in sepsis. Recent studies [25] have shown that the variation in IVC diameter has limited usage in predicting fluid re- sponsiveness. This is congruent with our study findings that IVC collaps- ibility does not predict clinical deterioration.

The circulatory system is highly complex in sepsis, with interaction between cardiac function and the microcirculation. Depressed EF using point-of-care ultrasound evaluation is associated with poorer clinical outcomes in our study cohort. This is congruous with previous studies evaluating the usefulness of echocardiography in sepsis, which reported an incidence of reduced EF ranging from 20% to 60% in septic shock patients [31-34], suggesting the importance of EF assessment in predicting mortality in septic shock [34]. Individuals with cardiac sup- pression have impaired neurohormonal compensatory mechanisms, thus resulting in less reserve capacity and ability in autoregulating blood flow and oxygenation [35]. Both ?StO₂ and depressed EF remain significant predictors of adverse clinical outcomes after adjusting for macrohemodynamics and complete cell counts (Table 4). A previous study by Podbregar and Mozina [36] in patients with heart failure showed that septic shock with concomitant cardiac failure resulted in higher StO₂ compared to patients with cardiac failure and no sepsis. We did not find similar results in our study cohort; the baseline StO₂ among patients with depressed EF was not statistically different among patients with or without adverse outcomes. However, we are

limited by the same sample size and further studies would be required for more conclusive results.

There have been recent observations regarding an association be- tween hyperdynamic circulation and sepsis; some studies have shown that patients with hyperdynamic circulation in sepsis were associated with decreased mortality [37]. The patients in our study who had worse outcomes were found to have either depressed or normal EF. This is an interesting hypothesis-generating association that may be evaluated in future studies.

Limitations

Our study has a few limitations. First, in an attempt to detect occult sepsis, the study team had decided to use broad inclusion criteria. This resulted in an overall positive event rate of 7.6% (14/184). However, the number of positive outcomes corresponded with our predicted number based on our sample size calculation (10% of 139 patients, 14 patients).

Second, our primary outcome is a pragmatic measure of worsening infection and/or sepsis by evaluating for escalation of treatment and care required, notably the need for ICU or HD stay, fluid resuscitation or vasopressor use, and escalation of antibiotics. We believe that this is a more clinically important outcome as it reflects healthcare resource utilization.

Third, our study was limited by a small sample size. The lack of sta- tistical significance in ?StO₂ in the subgroup of patients with cardiac failure with adverse outcomes and the desaturation gradient was likely a result of this small sample size. Likewise, the imprecise odds ratio es- timate for depressed EF. Despite this, we were able to generate new hy- potheses based on these results and observed trends for larger studies and better planning of subgroup analyses in the future.

Forth, we were also unable to establish appropriate cut-off values for the tissue oxygenation parameters for clinical use. This is the current existing dilemma in other studies evaluating tissue oxygenation param- eters. Fourth, due to availability of investigators, convenience sampling was employed, thus possibly subjecting our recruitment to selection bias.

Finally, follow-up measurements at 1- and 2-hour time points were not always possible due to the frequent movement of the ED patients, such as transferring for radiological imaging or to intensive care units for further management. Mandating stay in ED for completion of this study was not enforced due to ethical considerations. However, we were able to achieve completed readings in 75.5% of the recruited patients.

Conclusion

The combined use of NIRS in evaluation of tissue oxygenation pa- rameters and bedside point-of-care ultrasonography to assess cardiac function could potentially detect occult sepsis and predict adverse clin- ical outcomes in ED patients. A lower ?StO₂ value and depressed EF could possibly risk stratify undifferentiated patients. Larger studies are warranted in the future to investigate and determine useful cut-off ref- erence values as part of a non-invasive Resuscitation protocol for septic patients.

Declarations

Ethics approval and consent to participate: Ethics approval was obtained from the local ethics institutional review board, National Healthcare Group Domain-Specific Review Board (NHG DSRB), Singapore. Informed consent to participate was obtained from the patient or their legally appointed representative (if they are deemed to lack mental capacity for consent).

Consent for publication: Not applicable.

Availability of data and material: The datasets generated and/or ana- lyzed during the current study are not publicly available due to data protection regulations but may be available from the corresponding au- thor on reasonable request.

Competing interests: The authors declare no conflict of interest.

Funding: No funding was received for this study.

Authors’ contributions: Ng NYY, Ang HHE, Tan JCL, Kuan WS and Chua MT planned and designed the study methodology. Ng NYY recruited patients and collected data. Chua MT and Kuan WS analyzed the data collected and interpreted the results from the study. All authors contrib- uted to the writing of the manuscript, read and approved the final manuscript.

Acknowledgements: None.

References

AJEM

Evaluation for occult sepsis incorporating NIRS and emergency sonography

a b s t r a c t

Introduction

Methods