Ospina‑Tascón et al. Ann.

Intensive Care (2020) 10:41

https://doi.org/10.1186/s13613-020-00658-8

RESEARCH Open Access

Diastolic shock index and clinical outcomes

in patients with septic shock
Gustavo A. Ospina‑Tascón1,2*, Jean‑Louis Teboul1,3,4, Glenn Hernandez1,5, Ingrid Alvarez1,
Alvaro I. Sánchez‑Ortiz1, Luis E. Calderón‑Tapia1, Ramiro Manzano‑Nunez1, Edgardo Quiñones1,
Humberto J. Madriñan‑Navia1, Juan E. Ruiz1, José L. Aldana1 and Jan Bakker1,5,6,7,8

Abstract 
Background:  Loss of vascular tone is a key pathophysiological feature of septic shock. Combination of gradual
diastolic hypotension and tachycardia could reflect more serious vasodilatory conditions. We sought to evaluate the
relationships between heart rate (HR) to diastolic arterial pressure (DAP) ratios and clinical outcomes during early
phases of septic shock.
Methods:  Diastolic shock index (DSI) was defined as the ratio between HR and DAP. DSI calculated just before start‑
ing vasopressors (Pre-VPs/DSI) in a preliminary cohort of 337 patients with septic shock (January 2015 to February
2017) and at vasopressor start (VPs/DSI) in 424 patients with septic shock included in a recent randomized controlled
trial (ANDROMEDA-SHOCK; March 2017 to April 2018) was partitioned into five quantiles to estimate the relative risks
(RR) of death with respect to the mean risk of each population (assumed to be 1). Matched HR and DAP subsamples
were created to evaluate the effect of the individual components of the DSI on RRs. In addition, time-course of DSI
and interaction between DSI and vasopressor dose (DSI*NE.dose) were compared between survivors and non-survi‑
vors from both populations, while ROC curves were used to identify variables predicting mortality. Finally, as explora‑
tory observation, effect of early start of vasopressors was evaluated at each Pre-VPs/DSI quintile from the preliminary
cohort.
Results:  Risk of death progressively increased at gradual increments of Pre-VPs/DSI or VPs/DSI (One-way ANOVA,
p < 0.001). Progressive DAP decrease or HR increase was associated with higher mortality risks only when DSI concom‑
itantly increased. Areas under the ROC curve for Pre-VPs/DSI, SOFA and initial lactate were similar, while mean arterial
pressure and systolic shock index showed poor performances to predict mortality. Time-course of DSI and DSI*NE.
dose was significantly higher in non-survivors from both populations (repeated-measures ANOVA, p < 0.001). Very
early start of vasopressors exhibited an apparent benefit at higher Pre-VPs/DSI quintile.
Conclusions:  DSI at pre-vasopressor and vasopressor start points might represent a very early identifier of patients
at high risk of death. Isolated DAP or HR values do not clearly identify such risk. Usefulness of DSI to trigger or to direct
therapeutic interventions in early resuscitation of septic shock need to be addressed in future studies.
Keywords:  Septic shock, Acute circulatory dysfunction, Diastolic shock index, Clinical outcomes

Background
Definition of shock incorporates the presence of low
arterial pressure in association with abnormalities in tis-
sue perfusion leading to abnormal oxygen metabolism by
*Correspondence: [email protected]
1
Department of Intensive Care Medicine, Fundación Valle del Lili - the cells [1]. Because the intimate relationship between
Universidad ICESI, Av. Simón Bolívar Cra. 98, Cali, Colombia blood pressure and flow, operational definitions of shock
Full list of author information is available at the end of the article include the fall of mean (MAP) and/or systolic arterial

© The Author(s) 2020. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and
the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material
in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material
is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the
permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat​iveco​
mmons​.org/licen​ses/by/4.0/.
Ospina‑Tascón et al. Ann. Intensive Care (2020) 10:41 Page 2 of 11

pressure (SAP) [1, 2]. Nevertheless, alterations of pulse relationships between very early HR:DAP ratios (i.e.,
wave could grossly mirror, in some extend, the underly- the diastolic shock index, or DSI, calculated just before
ing mechanisms of acute circulatory failure implied in or at the start of vasopressor support) and clinical out-
shock. For example, SAP results particularly important comes in patients with septic shock, hypothesizing that
to define cardiogenic shock [2], hemorrhagic [3] or any very early DSI values could promptly identify patients
type of shock with a hypovolemic component, since at at high risk of unfavorable outcomes, while persistence
very early stage of these conditions, SAP and pulse pres- of high DSI during the first hours of resuscitation could
sure (PP) fall while diastolic arterial pressure tends to be reflect more severe cardiovascular dysfunction.
sustained. However, hypotension observed during septic
shock results from a complex interaction between vaso-
dilation, relative and absolute hypovolemia, myocardial Materials and methods
dysfunction, and altered blood flow distribution [4]. In Study population
particular, vasodilation resulting from the failure of the A total of 761 patients were analyzed: a preliminary
vascular smooth muscle to constrict is one of the lead- cohort of 337 patients with sepsis requiring vasopres-
ing mechanisms associated with hypotension and tissue sor support (January 2015 to February 2017) from one
hypoperfusion in septic shock [5]. In these cases, dias- mixed-ICU in a university hospital in Colombia (Fun-
tolic arterial pressure (DAP) would better reflect vasodi- dación Valle del Lili, Cali, Colombia) and 424 patients
lation than SAP or MAP. with septic shock included in a recent randomized con-
In healthy people, DAP is mainly determined by vascu- trolled study (March 2017 to April 2018) conducted in
lar tone and it remains nearly constant from the ascend- 28 hospitals in 5 countries (Argentina, Chile, Colombia,
ing aorta to the peripheral vessels [6]. Thus, detection Ecuador, Uruguay), the ANDROMEDA-SHOCK trial
of low DAP at peripheral vessels should reflect systemic [21]. The respective ethical and research committee
vasodilation as long as aortic valve is competent. How- involving human beings approved the use of the data
ever, in general, DAP is not considered for definition of obtained in both the initial cohort (Protocol number
septic shock, and with few exceptions, its relationship 1238, IRB/EC approval number 099-2018, Fundación
with clinical outcomes has not been widely described [7]. Valle del Lili, Cali, Colombia) and the randomized con-
Important studies in patients with septic shock define trolled trial [21].
hypotension in terms of MAP and SAP values [8–10] Septic shock was defined in the ANDROMEDA-
assuming the pivotal role of MAP [11] or SAP, on organ SHOCK population according to the Third Interna-
perfusion [12–14], in addition to the prognostic value tional Consensus Definitions for Sepsis and Septic
of sustained low MAP values [15]. Nevertheless, evalua- Shock (Sepsis 3.0), which states septic shock as the
tion of the loss of vascular tone through the severity of combination of suspected infection accompanying
diastolic hypotension could have profound implications life-threatening organ dysfunction, requirement of
on therapeutic decisions since there are not robust clues vasopressor therapy to elevate MAP ≥ 65  mmHg and
to rapidly predict when hypotension will be sustainably lactate > 2  mmol/L despite adequate fluid resuscitation
corrected with fluid loading. Thus, rapid assessment of [22]. Meanwhile, patients from the preliminary cohort
severity of vasodilation could influence therapeutic deci- were included under the diagnostic criteria for septic
sions such as the early introduction of vasoactive agents shock stated in the Surviving Sepsis Campaign: Inter-
[16], which theoretically would avoid unnecessary fluid national Guidelines for Management of Severe Sep-
administration while promptly restoring tissue perfusion. sis and Septic Shock: 2012 [23], based on the previous
Remarkably, DAP should not be evaluated separately 2001 SCCM/ESICM/ACCP/ATS/SIS International Sep-
from heart rate. Acute reductions in arterial pres- sis Definitions Conference [24], valid during the period
sure are compensated by increased sympathetic activ- in which the database was constructed.
ity, although sometimes such compensation becomes Exclusion criteria for preliminary cohort covered
maladaptive. This was the original rationale to index- patients < 18-year old, pregnant women, patients with
ing SAP by heart rate (HR) during hemorrhagic shock liver failure (protrombin time > 15  s or international
and acute critical illness [17, 18], or indexing MAP by normalized ratio ≥ 1.5 and any hepatic encephalopa-
HR to detect myocardial hypoperfusion [19]. Likewise, thy), advanced liver cirrhosis (Child–Pugh C), acute/
as DAP depends on vascular tone and the duration of chronic atrial fibrillation, presence of ventricular
the cardiac cycle [20], a combination of DAP and HR arrhythmia, use of definitive/transitory pacemaker
could reflect the severity of circulatory dysfunction and those with do-not-resuscitate orders. Meanwhile,
during vasodilatory conditions. Thus, we evaluated the exclusion criteria for the ANDROMEDA-SHOCK pop-
ulation are detailed elsewhere [21].
 Ospina‑Tascón et al. Ann. Intensive Care (2020) 10:41 Page 3 of 11

Study design Statistical analysis

DSI was calculated as the quotient between HR and First, DSI values, calculated just before the start of vaso-
DAP registered just before the start of vasopressor ther- pressors (Pre-VPs/DSI) in the preliminary cohort or at
apy (Pre-VPs/DSI) in the preliminary cohort and at the the randomization point (VPs/DSI) in the ANDROM-
randomization point in the ANDROMEDA-SHOCK EDA-SHOCK population, were partitioned into five
population (< 4  h of septic shock diagnosis according to quantiles to estimate the relative risks (RR) of death in
inclusion criteria), i.e., VPs/DSI [21]. Then, DSI was sub- relation to the mean risk of their respective population
sequently calculated 2, 4, and 8  h after the introduction (assumed to be 1). The mean risk and 95% confidence
of vasopressor support in both populations. Time elapsed intervals at each DSI quintile were calculated after
from the first hypotension episode and the first fluid load adjustment for the covariables: age, SOFA score day-1,
with resuscitative intention was registered in the pre- APACHE II, initial arterial lactate, and volume of resus-
liminary cohort, while time elapsed from the diagnosis citation fluids received before start of vasopressors and
of septic shock up to randomization was recorded for the from vasopressor start up to 8 h after. Then, new parti-
ANDROMEDA-SHOCK population. Most of the initial tions were performed aiming to evaluate the effect of
measurements (i.e., pre-vasopressor and at the start of individual components of Pre-VPs/DSI or VPs/DSI (i.e.,
vasopressor) were obtained by non-invasive techniques DAP and HR) on the relative risk of death, as follows:
using an oscillometric brachial cuff, typically in those (a) into quintiles of progressively higher DAP; (b) into
patients admitted from the emergency room and general quintiles of progressively higher HR; (c) re-stratifying
wards. However, invasive pressures were registered later each original quintile of DAP into 5 sub-clusters of DSI
on, when an indwelling intra-arterial catheter was placed. to extract patients with similar DSI values and there-
The volume of resuscitation fluids was registered at Pre- fore, simultaneous increasing of HR and DAP.
VPs point, and then, 2, 4 and 8  h after in the prelimi- Second, repeated-measures ANOVA were used to
nary cohort, and at the VPs/DSI point, and 8  h after in evaluate differences in the time-course of DSI, mean
the ANDROMEDA-SHOCK population. Meanwhile, net arterial pressure, DAP, HR, pulse pressure, and vaso-
fluid balance was recorded at 8 and 24 h after the start of pressor doses between survivors and non-survivors
vasopressors in both populations. The HR-to-SAP ratio at day-90 in both preliminary and ANDROMEDA-
[18, 25] was also calculated at same time points. Multi- SHOCK populations. Similarly, the time-course of the
ple organ dysfunction was assessed using the Sequential product of DSI and dose of vasopressor (DSI*NE.dose)
Organ Failure Assessment Score (SOFA) [26], while ven- was compared between survivors and non-survivors at
tilator-free days and requirement of acute renal replace- day-90.
ment therapy were also registered. Third, receiver operating characteristic (ROC) curves
Finally, as a simple exploratory observation, the effect were used to identify the performance of variables at
of timing to start vasopressor support was evaluated in pre-VP point (for preliminary cohort) or at randomiza-
the preliminary cohort. A very early start of vasopres- tion point (for ANDROMEDA-SHOCK), and 8 h after,
sor was defined as the one started within the first hour of to predict mortality at day-28 and 90. Such variables
receiving the first fluid load with resuscitative intention were Pre-VPs/DSI (or VPs/DSI, in the case of patients
such as it was recently reported [27]. from ANDROMEDA-SHOCK), lactate, mean arterial
pressure, SOFA score, APACHE II, and systolic shock
index (HR:SAP ratio). In addition, the interaction or
General management product of DSI by the dose of vasopressor (DSI*NE.
Patients from the preliminary cohort followed an early dose) was also included at points where the patients
quantitative resuscitation protocol adapted from the were under vasopressor support.
Surviving Sepsis Campaign [23, 28], aimed in general to Fourth, the effect of very early start of vasopressors
target (a) MAP ≥ 65  mmHg; (b) urine output > 0.5  mL/ on mortality at day-90 in each quintile of Pre-VPs/DSI
kg/h; (c) ­ScvO2 ≥ 70%, when available; (d) normaliza- from the preliminary cohort was evaluated using a Chi
tion of lactate levels or decreasing of 20% every-2  h in square test and additionally, logistic regression models
lactate levels. A complete description of the resuscita- adjusted by SOFA score and initial lactate at each Pre-
tion protocol and general management in such cohort is VPs/DSI quintile. A Hosmer and Lemeshow test was
described elsewhere [29]. Meanwhile, patients collected used to assess the goodness of fit in each model.
from ANDROMEDA-SHOCK trial were randomly allo-
cated to peripheral perfusion-targeted resuscitation or
lactate level-targeted resuscitation following a protocol
described in detail elsewhere [30].
Ospina‑Tascón et al. Ann. Intensive Care (2020) 10:41 Page 4 of 11

Results arterial pressure or isolated diastolic arterial pressure

A total of 761 patients with septic shock were ana- and the systolic shock index showed poor performance
lyzed: 337 patients from a preliminary cohort (Addi- for such prediction. DSI and DSI*NE.dose at 8 h showed
tional file 1: Figure S1a) and 424 from the randomized again similar performances than SOFA score and lactate
controlled trial ANDROMEDA-SHOCK (Additional values, while mean arterial pressure, diastolic arterial
file  1: Figure S1b). A STROBE statement checklist for pressures and the systolic shock index depicted a poor
observational studies is provided in SDC Additional performance to predict mortality at day-90 (Additional
file 1: Table S1. Lengths of ICU and hospital stay were file 1: Figure S8b–S9b).
9 (4–16) and 14 (6–29) days, respectively, in the pre- Very early start of norepinephrine (i.e., norepinephrine
liminary cohort, while these were 6 (3–12) and 13 started within the first hour of the first fluid load with
(6–26) days in the ANDROMEDA-SHOCK. Over- resuscitative intention) was related with a lower mortal-
all mortality at days-28 and 90 were 38.3% and 43.0% ity in higher Pre-VPs/DSI (i.e., the Pre-VPs/DSI Quin-
in the preliminary cohort, and 39.2% and 43.9% in the tile-5) (Additional file 1: Tables S6, S7).
ANDROMEDA-SHOCK. General characteristics of
both preliminary cohort and ANDROMEDA-SHOCK Discussion
are presented in the Table 1. Our study retrieves four important findings: (a) pro-
Progressive increases in Pre-vasopressor DSI (Pre-VPs/ gressively higher DSI values calculated just before or at
DSI) or DSI at vasopressor start (VPs/DSI) were related the start of vasopressors are associated with a gradual
with gradual increases in the relative risk of death at increase in the risk of death in patients with septic shock;
day-90 in the preliminary and ANDROMEDA-SHOCK (b) isolated low DAP or high HR values do not clearly
populations (Fig. 1). Similar HR values were related with identify such risk; (c) non-survivors evolve with per-
progressively lower risk of death as long as DAP gradually sistently high DSI values while requiring higher doses
increases, and consequently, DSI values decrease (Fig. 2). of vasopressors and more resuscitation fluids than sur-
Likewise, similar DAP values were related with progres- vivors; (d) Pre-VPs/DSI and VPs/DSI showed similar
sively higher risk of mortality as long as HR gradually performance to SOFA score and initial lactate levels to
increases, and consequently, DSI also did (Fig.  3). Nev- predict mortality, while mean arterial pressure and sys-
ertheless, simultaneous increases in HR and DAP with tolic shock index did not.
subsequent similar DSI values were related with similar Vasodilation plays a key role in the development of
risk of death (Additional file  1: Figure S2). A complete hypotension and tissue hypoperfusion in septic shock
description for DSI, DAP and HR partitioning is pre- [5]. DAP reflects in part the vascular tone when aortic
sented in the Additional file 1: Tables S2, S3. Meanwhile, valve is competent. Nevertheless, the duration of the car-
a complete description of general demographics, hemo- diac cycle, the blood volume ejected to the aorta and the
dynamics, lactate, renal replacement and mechanical arterial compliance also influence DAP [20]. Thus, under
ventilation requirements, resuscitation and cumulative isovolemic conditions and constant arterial compliance,
fluids according to the Pre-VPs/DSI and VPs/DSI in the shortening diastolic times are associated with higher
preliminary cohort and ANDROMEDA-SHOCK popula- DAP while a prolonged diastole leads to an opposite
tions are presented in Additional file 1: Tables S4, S5. effect [20]. Consequently, simultaneous and opposite var-
There were significant differences in the time-course iations in DAP and HR could suggest more severe cardio-
of DSI between survivors and non-survivors at day-90 vascular dysfunction, with progressively high HR unable
in both populations (repeated-measures ANOVA, inter- to compensate DAP drops as a consequence of gradual
subjects difference p < 0.001) (Fig. 4). Similarly, the prod- decrease in vascular tone. Supporting this, our data sug-
uct of DSI and dose of norepinephrine (DSI*NE.dose) gest that such progressively opposite changes in HR and
remained significantly high in non-survivors from both DAP represent more severe circulatory dysfunction with
populations (repeated-measures ANOVA, inter-subjects proportional increases in the relative risk of death.
difference p < 0.001) (Fig.  4). Time-course of diastolic Persistently low MAP [15, 31] or DAP [7] have been
pressure, heart rate, mean arterial pressure, pulse pres- related to worse outcomes in septic shock, while new-
sure and systolic shock index for survivors and non-sur- onset prolonged sinus tachycardia as a consequence of
vivors are showed in Additional file 1: Figures S3–S7. sympathetic activity has been associated with increased
Pre-VPs/DSI from preliminary cohort or VPs/DSI major cardiovascular events, prolonged length of stay
from ANDROMEDA-SHOCK depicted similar perfor- [32], and higher mortality rates [33]. Nevertheless, iso-
mance to predict mortality at day-28 and 90 than other lated DAP or HR just before or at start of vasopressors
variables such as SOFA score and initial lactate levels was not clearly related with mortality in the prelimi-
(Additional file  1: Figure S8a–S9b). Conversely, mean nary cohort and ANDROMEDA-SHOCK populations.
 Ospina‑Tascón et al. Ann. Intensive Care (2020) 10:41 Page 5 of 11

Table 1  General characteristics

Preliminary Cohort Andromeda-Shock
(n = 337) (n = 424)

General characteristics
 Age, years 64 (51 to 74) 66 (52 to 76)
 Male sex, n (%) 188 (55.8) 226 (53.3)
 Weight, Kg 68 (59 to 76) 70 (59 to 80)
 APACHE II 16 (13 to 22) 21 (17 to 28)
 SOFA day-1 9 (7 to 12) 10 (7 to 12)
 Charlson Comorbidity Index 4 (2 to 5) 3 (1 to 5)
 Chronic hypertension, n (%) 73 (39.2) 176 (41.5)
Source of Infection
 Lung 120 (35.6) 128 (30.2)
 Genitourinary 60 (17.8) 87 (20.5)
 Abdominal 110 (32.6) 149 (35.1)
 Other 47 (13.9) 60 (14.2)
 Delay time antibiotics, hours 2 (− 2 to 5) 2 (1 to 2)
 Time from first fluid resuscitation load up to vasopressor start, hours 2 (0 to 3) –
 Time from hypotension up to vasopressor start, hours 3 (1 to 4) –
 Time from septic shock diagnosis up to randomization, min – 81 (0 to 180)
At vasopressor start
 SAP 92 (83 to 106) 100 (85 to 113)
 DAP 45 (40 to 51) 52 (45 to 60)
 MAP 63 (56 to 69) 66 (60 to 76)
 HR 104 (87 to 121) 103 (87 to 120)
 PP 46 (35 to 59) 45 (35 to 58)
 DSI 2.28 (1.83 to 2.74) 1.97 (1.58 to 2.48)
SvO2, %, n 71.7 (63.8 to 78.2), 196 73.0 (65.0 to 79.0), 401
Pv-aCO2, mmHg, n 5.0 (3.7 to 7.0), 195 7.0 (5.0 to 10.0), 398
CVP at VPs, mmHg, n 7 (4 to 12), 69 9 (6 to 13), 393
Lactate (initial), mmol/L, n 2.7 (1.6 to 4.9), 337 3.5 (2.7 to 5.4), 424
Fluids/VP/RRT​
 Volume of resuscitation fluids up to start of VP, mL 1200 (400 to 2000) 2000 (1200 to 2800)
 Volume of resuscitation fluids up to start of VP, mL/kg 16.3 (5.7 to 30.0) 27.8 (18.8 to 41.7)
 Volume of resuscitation fluids up to 8 h, mL 1050 (1000 to 2500) 1000 (0 to 2000)
Net fluid balance
 At 24 h 2700 (1200 to 4500) 1940 (900 to to 3350)
 Norepinephrine max. dose, µg/kg/min 0.26 (0.13 to 0.48) 0.26 (0.11 to 0.45)
 Acute RRT​ 94 (27.9) 72 (17.0)
Clinical outcomes
 ICU LOS 9 (4 to 16) 6 (3 to 12)
 Hospital LOS 14 (6 to 29) 13 (6 to 26)
 Mechanical ventilation-free days 20 (0 to 27) 16 (0 to 26)
 RRT-free days 28 (6 to 28) 28 (2 to 28)
 Mortality 28-day, n (%) 129 (38.3) 166 (39.2)
 Mortality 90-day, n (%) 145 (43.0) 186 (43.9)
*Including only patients receiving renal replacement therapy at least for one session
APACHE II Acute physiology and chronic health evaluation, SOFA sequential organ failure assessment, VP vasopressor, SAP systolic arterial pressure, DAP diastolic
arterial pressure, MAP mean arterial pressure, HR heart rate, PP pulse pressure, DSI diastolic shock index (HR:DAP ratio), SvO2 oxygen venous saturation, Pv-aCO2
venous-to-arterial carbon dioxide difference, CVP central venous pressure, RRT​ renal replacement therapy, ICU LOS Intensive Care Unit length of stay, Hospital LOS
hospital length of stay
Ospina‑Tascón et al. Ann. Intensive Care (2020) 10:41 Page 6 of 11

Fig. 1  Relative risk of death at day-90 according to pre-vasopressor diastolic shock index (Pre-VPs/DSI) or vasopressor start (VPs/DSI) partitions
in the preliminary and ANDROMEDA SHOCK populations. Diastolic shock index values obtained from just before the start of vasopressor (in
preliminary cohort) and at the start vasopressor support (in ANDROMEDA-SHOCK) were partitioned into 5 quantiles (Q1 to Q5). Distribution of
heart rate (HR) and diastolic pressure (DAP) (top) and their respective diastolic shock index distribution (middle) are presented through the quantile
distribution. Boxplots (top and middle) delineate the interquartile range, the median is shown as a line in the middle of the box, and tails represent
the 95% range. Coefficients derived from a logistical regression were used to calculate the cut-off value of the diastolic shock index (DSI) detecting
the mean risk of mortality of the entire population at 28 days. This point was used as the reference to calculate the adjusted relative risks, in such a
way that a relative risk of 1 represents the mean risk of the respective population (bottom). The mean risk and 95% confidence interval (error bars
at the bottom) for each percentile were calculated after multivariate adjustment (Cox proportional-hazards model) for the covariables: age, gender,
SOFA score day-1, initial arterial lactate and pH, and resuscitation fluids from VP to 8H. The gray zone represents the 95% confidence interval for the
Cox regression (continuous line) across the complete population, assuming the diastolic shock index as a continuous variable. Note that adjusted
relative risk of death increases as diastolic shock index also does through the quintile distribution

Indeed, at similar HR or DAP values, the risk of death levels, DSI could add some practical and valuable infor-
was increased only when DSI concomitantly increased mation about how to intervene the initial hemodynamic
(Figs.  2, 3; Additional file  1: Figure S2). In addition, condition in sepsis.
MAP, SAP and “systolic shock index” (or, HR:SAP ratio) Progressively high DSI values calculated just before
were not related with mortality in both populations and at the start of vasopressor support were related with
(Additional file 1: Figures S8, S9). We hypothesized that gradual increases in the risk of death. Patients in the
although MAP and SAP are used to operatively define higher quintiles of pre-VPs/DSI and VPs/DSI required
septic and another types of shock, initial MAP or SAP more renal replacement therapy, depicted higher lac-
does not reflect systemic vasodilation, which is a lead- tate values and also showed slower lactate decreases
ing mechanism in septic shock. Although DSI depicted over the first 8 h of resuscitation. They also required sig-
a similar AUC–ROC than SOFA score and initial lactate nificantly more resuscitation fluids and higher doses of
 Ospina‑Tascón et al. Ann. Intensive Care (2020) 10:41 Page 7 of 11

Fig. 2  Relative risk of death at day-90 according to diastolic arterial pressure (DAP) partition in the preliminary and ANDROMEDA SHOCK
populations. Diastolic arterial pressure (DAP) values from just before the start of vasopressor support were partitioned into 5 quantiles (Q1 to Q5).
Distribution of heart rate (HR) and diastolic pressure (DAP) (top) displays a progressive increasing of DAP values through the quantile partitioning
with their corresponding HR values, which remains similar from Q1 to Q5. The respective diastolic shock index distribution (middle) is presented
through the quantile distribution. The boxes (top) delineate the interquartile range, the median is shown as a line in the middle of the box, and
tails represent the 95% range. Boxplots/error bars (middle) represent medians and 95% confidence intervals of the diastolic shock index (DSI) at
each quantile. Relative risks’ distributions (bottom) were calculated as described in Fig. 1. Note that adjusted relative risk of death decreases as DAP
increases and subsequently DSI decreases, for similar HR values

vasopressors as reflected by the product of DSI and dose arterial pressure is observed from the ascending aorta to
of norepinephrine (DSI*NE.dose). We hypothesize that the peripheral vessels, DAP remains almost constant [6,
persistently higher DSI values reflect a lack of vascular 34], even during experimental endotoxemic conditions
tone requiring progressively higher doses of vasopres- in which a “vascular tone decoupling” from central-to-
sors with an inadequate restoration of tissue perfusion. peripheral circulation can occur [35]. Thus, DAP records
However, the observational nature of our study hinders obtained at peripheral circulation closely reflect central
the direct effect of variations in vasopressor dose or fluid DAP measurements even during severe inflammatory
loading on the DSI since the resuscitation maneuvers in conditions with increased vasodilation and altered arte-
each group were guided targeting MAP but not DAP. rial compliance. Although it could be argued that inva-
All arterial pressure measurements used for DSI cal- sive vs. non-invasive measurement methods to measure
culations in our study were obviously obtained at the arterial pressure could influence our results, the bias for
peripheral circulation (i.e., at brachial, femoral or radial DAP measurements is far lower than that observed for
sites). Although some disagreement in systolic or mean SAP [36]. Furthermore, although significant differences
Ospina‑Tascón et al. Ann. Intensive Care (2020) 10:41 Page 8 of 11

Fig. 3  Relative risk of death at day-28 according to heart rate (HR) partition the preliminary and ANDROMEDA SHOCK populations. Heart rate (HR)
values from just before the start of vasopressor support were partitioned into 5 quantiles (Q1 to Q5). Distribution of heart rate (HR) and diastolic
pressure (DAP) (top) displays a progressive increasing of HR values through the quantile partitioning with their corresponding DAP values, which
remains similar from Q1 to Q5. The respective diastolic shock index distribution (middle) is presented through the quantile distribution. The boxes
(top) delineate the interquartile range, the median is shown as a line in the middle of the box, and tails represent the 95% range. Boxplots/error bars
(middle) represent medians and 95% confidence intervals of the diastolic shock index (DSI) at each quantile. Relative risks’ distributions (bottom)
were calculated as described in Fig. 1. Note that adjusted relative risk of death increases as HR and subsequently DSI also increases, for similar DAP
values

in SAP or MAP are observed according to if invasive vs. Although also considered as “first line intervention”,
non-invasive method are used [36], DAP recordings are vasopressors are usually used as a rescue therapy when
closer at progressively lower DAP values [35]. Conse- initial fluid administration fails to correct hypotension
quently, all these considerations claim against the intro- or when arterial pressure is judged to be insufficient to
duction of considerable errors in DSI calculation when ensure an adequate tissue perfusion. Recent experi-
using invasive vs. non-invasive DAP values and also favor mental and observational data suggest that very early
the notion of DSI as a global marker of decreased vascu- start of vasopressor support could be beneficial [27, 39].
lar tone since DAP is less influenced by the reflection of Nevertheless, there are no clear signals indicating when
pulse waves. vasopressor support should be started. In this way, very
This study may have some important clinical implica- early signals of severe vasodilation should alert on its
tions. It is unlikely that severe hypotension as a result possible immediate requirement. Thus, DSI should not
of severe vasodilation could be reversed by simple fluid be interpreted as “another index of death”. Instead, a
administration and instead, unnecessary fluids with higher DSI value at presentation of severe cases of sep-
subsequent harmful accumulation can occur [37, 38]. sis could identify patients who might benefit from some
 Ospina‑Tascón et al. Ann. Intensive Care (2020) 10:41 Page 9 of 11

Fig. 4  Time-course of diastolic shock index (DSI) and the interaction between DSI and norepinephrine dose for survivors and non-survivors at
day-90 in the preliminary cohort and ANDROMEDA-SHOCK. Left panel, Top. Time-course of DSI for survivors and non-survivors at day-90 in
the preliminary cohort. Repeated-measures ANOVA, Time*Outcome day-90, p < 0.001. Inter-subjects difference, p < 0.001. Left panel, Bottom.
Time-course of interaction of DSI and norepinephrine dose for survivors and non-survivors at day-90 in the preliminary cohort. Repeated-measures
ANOVA, Time*Outcome day-90, p = 0.18. Inter-subjects’ difference, p < 0.001. Right panel, Top. Time-course of DSI for survivors and non-survivors
at day-90 in ANDROMEDA-SHOCK population. Repeated-measures ANOVA, Time*Outcome day-90, p = 0.34. Inter-subjects difference, p < 0.001.
Right panel, Bottom. Time-course of interaction of DSI and norepinephrine dose for survivors and non-survivors at day-90 in ANDROMEDA-SHOCK
population. Repeated-measures ANOVA, Time*Outcome day-90, p = 0.02. Inter-subjects’ difference, p < 0.001

very early interventions capable of modifying the course controlled trial, reinforce the strength of DSI as an early
of septic shock. Our data suggest some beneficial of very identifier of septic patients at high risk of death. Second,
early start of vasopressors in patients at the higher pre- we did not include a control group of normal subjects, so
VPs DSI. Nevertheless, sample size and the retrospective recognizing a DSI cutoff to identify abnormality could
nature of such observation simply pose a hypothesis to be be misleading. Third, although persistently high DSI val-
tested in the future. ues were consistently observed in non-survivors in both
Our study has several limitations. First, as previously the preliminary and ANDROMEDA-SHOCK groups,
mentioned, its retrospective nature might limit the con- there are no clues about whether it is possible to inter-
clusions since some confounding factors and poten- vene DSI course or even whether modifying DSI course
tial bias may not have been controlled. Nevertheless, might influence clinical outcomes. Nevertheless, this
observations from preliminary cohort, corroborated in could be an important research question as recent exper-
prospectively collected data from a recent randomized imental observations suggest that some early therapeutic
Ospina‑Tascón et al. Ann. Intensive Care (2020) 10:41 Page 10 of 11

interventions might modify the time-course of cardio- Competing interests

The authors declare that they have no competing interests.
vascular dysfunction in septic shock. Finally, despite the
apparent plausibility of DSI at very early stages of septic Author details
shock, our observations are limited to a relative small 1
 Department of Intensive Care Medicine, Fundación Valle del Lili - Universidad
ICESI, Av. Simón Bolívar Cra. 98, Cali, Colombia. 2 Traslational Medicine in Criti‑
sample of patients. Consequently, the potential util- cal Care and Experimental Surgery Laboratory (TransLab‑CCM), Universidad
ity of DSI in the clinical practice should be additionally ICESI, Cali, Colombia. 3 Service de Réanimation Médicale, Hôpital Bicêtre,
explored. Hôpitaux Universitaires Paris–Sud, Paris, France. 4 Assistance Publique Hôpitaux
de Paris, Université Paris–Sud, Paris, France. 5 Departamento de Medicina
Intensiva, Pontificia Universidad Católica de Chile, Santiago, Chile. 6 Depart‑
ment of Intensive Care Adults, Erasmus MC University Medical Center, Rotter‑
Conclusion dam, The Netherlands. 7 Department of Pulmonary and Critical Care, New York
DSI calculated just before or at the vasopressor start University, New York, USA. 8 Division of Pulmonary, Allergy, and Critical Care
Medicine, Columbia University Medical Center, New York, USA.
might identify patients with septic shock at high risk of
death. Isolated DAP or high HR is not clearly related with Received: 6 January 2020 Accepted: 4 April 2020
such risk. Whether the DSI could be used as a trigger
or to direct therapeutic interventions in septic shock or
sepsis-related cardiovascular dysfunction deserves future
research efforts. References
1. Cecconi M, De Backer D, Antonelli M, Beale R, Bakker J, Hofer C, et al. Con‑
sensus on circulatory shock and hemodynamic monitoring. Task force
Supplementary information of the European Society of Intensive Care Medicine. Intensive Care Med.
Supplementary information accompanies this paper at https​://doi. 2014;40(12):1795–815.
org/10.1186/s1361​3-020-00658​-8. 2. van Diepen S, Katz JN, Albert NM, Henry TD, Jacobs AK, Kapur NK, et al.
Contemporary management of cardiogenic shock: a Scientific Statement
From the American Heart Association. Circulation. 2017;136(16):e232–68.
Additional file 1. Additional tables and figures. 3. Stern SA, Dronen SC, Birrer P, Wang X. Effect of blood pressure on hemor‑
rhage volume and survival in a near-fatal hemorrhage model incorporat‑
ing a vascular injury. Ann Emerg Med. 1993;22(2):155–63.
Abbreviations 4. Vincent JL, De Backer D. Circulatory shock. N Engl J Med.
HR: Heart rate; DAP: Diastolic arterial pressure; MAP: Mean arterial pressure; 2013;369(18):1726–34.
SAP: Systolic arterial pressure; DSI: diastolic shock index; SOFA: Sequential 5. Siegel JH, Greenspan M, Del Guercio LR. Abnormal vascular tone, defec‑
Organ Failure Assessment Score; APACHE II: Acute Physiology and Chronic tive oxygen transport and myocardial failure in human septic shock. Ann
Health Evaluation. Surg. 1967;165(4):504–17.
6. Hamilton W. The patterns of the arterial pressure pulse. Am J Physiol
Acknowledgements Legacy Content. 1944;141(2):235–41.
The authors thank Dr. Sergio Prada (CIC, Fundación Valle del Lili-Universidad 7. Benchekroune S, Karpati PC, Berton C, Nathan C, Mateo J, Chaara M, et al.
Icesi, Cali, Colombia) and Dra. Yuri Takeuchi (Universidad Icesi - Fundación Valle Diastolic arterial blood pressure: a reliable early predictor of survival in
del Lili) for their unconditional support to this project. human septic shock. J Trauma. 2008;64(5):1188–95.
8. Mouncey PR, Osborn TM, Power GS, Harrison DA, Sadique MZ, Grieve RD,
Authors’ contributions et al. Trial of early, goal-directed resuscitation for septic shock. N Engl J
Literature search: GAOT, GH and JB; data acquisition: GAOT, IA, LECT, RMN, EQ, Med. 2015;372(14):1301–11.
HJMN, JERY, and JLA; data analysis and interpretation: GAOT, GH, AISO, LECT, 9. Peake SL, Delaney A, Bailey M, Bellomo R, Cameron PA, Cooper DJ, et al.
RMN, EQ, HJ-MN, JER, JLA, JLT, JB; critical review: GAOT, GH, JLT, DDB and JB; Goal-directed resuscitation for patients with early septic shock. N Engl J
conception, hypothesis delineation, and design of the study: GAOT, JLT, GH, Med. 2014;371(16):1496–506.
DDB, and JB. All authors read and approved the final manuscript. 10. Yealy DM, Kellum JA, Huang DT, Barnato AE, Weissfeld LA, Pike F, et al. A
randomized trial of protocol-based care for early septic shock. N Engl J
Funding Med. 2014;370(18):1683–93.
The current study received logistic support from the Centro de Investiga‑ 11. Sevransky JE, Nour S, Susla GM, Needham DM, Hollenberg S, Pronovost P.
ciones Clínicas - Fundación Valle del Lili, Cali – Colombia. Hemodynamic goals in randomized clinical trials in patients with sepsis: a
systematic review of the literature. Crit Care. 2007;11(3):R67.
Availability of data and materials 12. Peake SL, Moran JL, Leppard PI. N-Acetyl-l-cysteine depresses car‑
The datasets generated and/or analyzed during the current study are not pub‑ diac performance in patients with septic shock. Crit Care Med.
licly available as recommended by the local Ethical and research committee 1996;24(8):1302–10.
involving human beings (Fundación Valle del Lili, Cali, Colombia). Neverthe‑ 13. Tuchschmidt J, Fried J, Astiz M, Rackow E. Elevation of cardiac out‑
less, it could be available from the corresponding author on reasonable put and oxygen delivery improves outcome in septic shock. Chest.
request and under prior approval by such committee. 1992;102(1):216–20.
14. Bollaert PE, Charpentier C, Levy B, Debouverie M, Audibert G, Larcan A.
Ethics approval and consent to participate Reversal of late septic shock with supraphysiologic doses of hydrocorti‑
The ethical and research committee involving human beings approved the sone. Crit Care Med. 1998;26(4):645–50.
current study (Protocol number 1238, IRB/EC approval number 099-2018, 15. Varpula M, Tallgren M, Saukkonen K, Voipio-Pulkki LM, Pettilä V. Hemody‑
Fundación Valle del Lili, Cali, Colombia).The ANDROMEDA-SHOCK randomized namic variables related to outcome in septic shock. Intensive Care Med.
clinical trial was conducted at 28 hospitals in 5 countries (Argentina, Chile, 2005;31(8):1066–71.
Colombia, Ecuador, Uruguay). The institutional review board at each site 16. Hamzaoui O, Teboul JL. Temporary removal: importance of diastolic arte‑
approved the study and the use of data for post hoc analysis. rial pressure in septic shock: PRO. J Crit Care. 2018;51:238.
17. King RW, Plewa MC, Buderer NM, Knotts FB. Shock index as a marker for
Consent for publication significant injury in trauma patients. Acad Emerg Med. 1996;3(11):1041–5.
Not applicable.
 Ospina‑Tascón et al. Ann. Intensive Care (2020) 10:41 Page 11 of 11

18. Rady MY, Nightingale P, Little RA, Edwards JD. Shock index: a re-evalua‑ venous-arterial ­CO2 to arterial-venous ­O2 content difference ratio as
tion in acute circulatory failure. Resuscitation. 1992;23(3):227–34. markers of resuscitation in patients with septic shock. Intensive Care Med.
19. Buffington CW, Sivarajan M, Bashein G. The quotient of mean arterial 2015;41(5):796–805.
pressure and heart rate predicts hypoperfusion of collateral-dependent 30. Hernández G, Cavalcanti AB, Ospina-Tascón G, Zampieri FG, Dubin A,
myocardium. J Cardiothorac Anesth. 1989;3(1):65–9. Hurtado FJ, et al. Early goal-directed therapy using a physiological holistic
20. O’Rourke MF. Steady and pulsatile energy losses in the systemic circula‑ view: the ANDROMEDA-SHOCK-a randomized controlled trial. Ann Inten‑
tion under normal conditions and in simulated arterial disease. Cardio‑ sive Care. 2018;8(1):52.
vasc Res. 1967;1(4):313–26. 31. Dünser MW, Takala J, Ulmer H, Mayr VD, Luckner G, Jochberger S, et al.
21. Hernández G, Ospina-Tascón GA, Damiani LP, Estenssoro E, Dubin A, Arterial blood pressure during early sepsis and outcome. Intensive Care
Hurtado J, et al. Effect of a resuscitation strategy targeting peripheral per‑ Med. 2009;35(7):1225–33.
fusion status vs serum lactate levels on 28-day mortality among patients 32. Sander O, Welters ID, Foëx P, Sear JW. Impact of prolonged elevated
with septic shock: the ANDROMEDA-SHOCK Randomized Clinical Trial. heart rate on incidence of major cardiac events in critically ill patients
JAMA. 2019;321(7):654–64. with a high risk of cardiac complications. Crit Care Med. 2005;33(1):81–8
22. Singer M, Deutschman CS, Seymour CW, Shankar-Hari M, Annane D, (discussion 241–2).
Bauer M, et al. The third international consensus definitions for sepsis and 33. Park S, Kim DG, Suh GY, Park WJ, Jang SH, Hwang YI, et al. Significance of
septic shock (sepsis-3). JAMA. 2016;315(8):801–10. new-onset prolonged sinus tachycardia in a medical intensive care unit: a
23. Dellinger RP, Levy MM, Rhodes A, Annane D, Gerlach H, Opal SM, et al. prospective observational study. J Crit Care. 2011;26(5):534.e1-.e8.
Surviving Sepsis Campaign: international guidelines for manage‑ 34. O’Rourke MF. Pressure and flow waves in systemic arteries and the ana‑
ment of severe sepsis and septic shock, 2012. Intensive Care Med. tomical design of the arterial system. J Appl Physiol. 1967;23(2):139–49.
2013;39(2):165–228. 35. Hatib F, Jansen JR, Pinsky MR. Peripheral vascular decoupling in porcine
24. Levy MM, Fink MP, Marshall JC, Abraham E, Angus D, Cook D, et al. 2001 endotoxic shock. J Appl Physiol. 2011;111(3):853–60.
SCCM/ESICM/ACCP/ATS/SIS international sepsis definitions conference. 36. Lehman LW, Saeed M, Talmor D, Mark R, Malhotra A. Methods of blood
Crit Care Med. 2003;31(4):1250–6. pressure measurement in the ICU. Crit Care Med. 2013;41(1):34–40.
25. Allgöwer M, Burri C. Schockindex. DMW-Deutsche Medizinische Wochen‑ 37. Marik PE, Linde-Zwirble WT, Bittner EA, Sahatjian J, Hansell D. Fluid
schrift. 1967;92(43):1947–50. administration in severe sepsis and septic shock, patterns and out‑
26. Vincent JL, de Mendonça A, Cantraine F, Moreno R, Takala J, Suter PM, comes: an analysis of a large national database. Intensive Care Med.
et al. Use of the SOFA score to assess the incidence of organ dysfunction/ 2017;43(5):625–32.
failure in intensive care units: results of a multicenter, prospective study. 38. Silversides JA, Fitzgerald E, Manickavasagam US, Lapinsky SE, Nisenbaum
Working group on “sepsis-related problems” of the European Society of R, Hemmings N, et al. Deresuscitation of patients with iatrogenic fluid
Intensive Care Medicine. Crit Care Med. 1998;26(11):1793–800. overload is associated with reduced mortality in critical illness. Crit Care
27. Ospina-Tascón GA, Hernandez G, Alvarez I, Calderón-Tapia LE, Manzano- Med. 2018;46:1600–7.
Nunez R, Sánchez-Ortiz AI, et al. Effects of very early start of norepineph‑ 39. Byrne L, Obonyo NG, Diab SD, Dunster KR, Passmore MR, Boon AC, et al.
rine in patients with septic shock: a propensity score-based analysis. Crit Unintended consequences; fluid resuscitation worsens shock in an ovine
Care. 2020;24(1):52. model of endotoxemia. Am J Respir Crit Care Med. 2018;198:1043–54.
28. Rhodes A, Evans LE, Alhazzani W, Levy MM, Antonelli M, Ferrer R, et al.
Surviving sepsis Campaign: international guidelines for management of
sepsis and septic Shock: 2016. Intensive Care Med. 2017;43(3):304–77. Publisher’s Note
29. Ospina-Tascón GA, Umaña M, Bermúdez W, Bautista-Rincón DF, Springer Nature remains neutral with regard to jurisdictional claims in pub‑
Hernandez G, Bruhn A, et al. Combination of arterial lactate levels and lished maps and institutional affiliations.