The Journal of Maternal-Fetal & Neonatal Medicine

ISSN: 1476-7058 (Print) 1476-4954 (Online) Journal homepage: [Link]

Shock index and delta-shock index are superior

to existing maternal early warning criteria to
identify postpartum hemorrhage and need for
intervention

Jaden R. Kohn, Gary A. Dildy & Catherine S. Eppes

To cite this article: Jaden R. Kohn, Gary A. Dildy & Catherine S. Eppes (2017): Shock index and
delta-shock index are superior to existing maternal early warning criteria to identify postpartum
hemorrhage and need for intervention, The Journal of Maternal-Fetal & Neonatal Medicine, DOI:
10.1080/14767058.2017.1402882

To link to this article: [Link]

Accepted author version posted online: 07

Nov 2017.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

[Link]

Download by: [University of Florida] Date: 09 November 2017, At: 15:43

 Title: Shock index and delta-shock index are superior to existing maternal early warning criteria to
identify postpartum hemorrhage and need for intervention

Short Title: SI as PPH warning criterion

Authors: Jaden R. Kohn, BS1 Gary A. Dildy, MD2 Catherine S. Eppes, MD, MPH2

Affiliations:
1 Baylor College of Medicine, Houston, Texas.
2Department of Obstetrics and Gynecology, Division of Maternal-Fetal Medicine, Baylor College of

D
Medicine, Houston, Texas.

TE
Downloaded by [University of Florida] at 15:43 09 November 2017

Correspondence:
Catherine S. Eppes
Chief of Obstetrics, Ben Taub Hospital
Assistant Professor
Department of Obstetrics and Gynecology
Division of Maternal-Fetal Medicine
Baylor College of Medicine
EP
C
1504 Taub Loop 3rd floor, Houston, TX 77030
Telephone: 713-873-8794
AC

Email: cseppes@[Link]

Word Count of Abstract: 171

Word Count of Main Text: 3382

ST

Keywords: Blood Pressure; Heart Rate; Obstetric Labor; Postpartum Hemorrhage; Shock Index.
JU
 ABSTRACT

Objective: To determine whether shock index (SI) is superior to traditional vital signs in predicting

postpartum hemorrhage and need for intervention.

Methods: Retrospective case-control study in an academic tertiary-care county hospital. Forty-one

consecutive PPH cases and 41 controls were frequency-matched by mode of delivery and maternal

D
weight. We measured four criteria: heart rate, SBP, SI (HR/SBP), and delta-SI (peak SI – baseline

TE
SI). Using received operating characteristic curves, we compared the discrimination performance of
Downloaded by [University of Florida] at 15:43 09 November 2017

each criterion to predict PPH, transfusion, and surgical intervention, and identified thresholds with

EP
the strongest classification.

Results: SI ≤ 1.1 can be normal in peripartum. Peak SI and delta-SI were generally superior to HR
C
and SBP in predicting PPH, transfusion, and surgical intervention. SI≥1.143 and SI≥1.412 were
AC

strong initial and “critical” thresholds. Delta-SI was the strongest classifier overall; both SI and

delta-SI remain sensitive and specific when adjusted for potential confounders.
ST

Conclusion: SI and delta-SI appear to be superior to HR and SBP in predicting PPH and need for

intervention. Utility of delta-SI should be prospectively explored.

JU

BRIEF RATIONALE

Postpartum hemorrhage (PPH) is a significant cause of maternal mortality and morbidity in

the United States and worldwide, and delayed management is associated with adverse outcomes.

Early warning criteria with greater sensitivity to physiologic changes may improve the ability to

identify patients at risk for decompensation. We performed a case-control study to determine if

shock index is superior to heart rate (HR) or systolic blood pressure (SBP) in predicting postpartum

hemorrhage, transfusion, and surgical intervention, and to identify clinically-useful thresholds for
 shock index that classify patients who require further intervention. We found that SI up to 1.1 can

be normal in peripartum women. Peak SI and delta-SI were generally superior to HR and SBP in

predicting PPH, transfusion, and surgical intervention. Both SI and delta-SI remain sensitive and

specific when adjusted for potential confounders. SI≥1.143 and SI≥1.412 were strong initial and

“critical” thresholds. Delta-SI was the strongest classifier overall; its utility should be prospectively

explored.

D
INTRODUCTION

Postpartum hemorrhage (PPH) is a significant cause of maternal mortality and morbidity in

TE
Downloaded by [University of Florida] at 15:43 09 November 2017

the United States and worldwide.1, 2, 3 Delayed management of PPH is associated with adverse

outcomes.4, 5 To facilitate timely recognition and treatment, early warning criteria have been

EP
proposed, in which one or more abnormal parameters trigger escalation of care.6, 7 Non-obstetric

early warning systems, such as the systemic inflammatory response syndrome (SIRS) or modified
C
early warning score (MEWS), have poor predictive abilities in pregnant women due to the
AC

significant physiologic changes during pregnancy and labor, necessitating unique obstetric criteria.8,

In the United Kingdom, the National Health System incorporated the Modified Early
ST

Obstetric Warning System (MEOWS) into its maternal safety standards in 2012.10 A prospective

validation of MEOWS with 676 patients found that 30% of patients triggered evaluation and 13%
JU

experienced a complication, predicting maternal morbidity with 89% sensitivity and 79%

specificity.11 In the United States, the National Partnership for Maternal Safety proposed the

Maternal Early Warning Criteria.12 Proposed thresholds include heart rate (HR) <50 or >120

beats/minute, systolic blood pressure (SBP) <90 or >160 mmHg, diastolic blood pressure >100

mmHg, respiratory rate <10 or >30 breaths/minute, oxygen saturation <95%, and urine output

<35mL/hour for ≥2 hours.12

Additional early warning criteria with greater sensitivity to physiologic changes may
 improve the ability of early warning systems to identify patients at risk for decompensation at an

earlier time, limiting morbidity. Shock index (SI), the ratio of HR to SBP, is a clinically useful

criterion in multiple settings.13, 14, 15 In trauma, shock index predicts early hypovolemia and need for

transfusion or operative management,13 likely because it is an effective proxy for left ventricular

stroke work, which is altered during the physiologic response to hypovolemia.16 Additionally, shock

index was a better predictor of ruptured ectopic pregnancy than heart rate or systolic blood

D
pressure.14, 15 The normal range for SI in trauma and ectopic pregnancy has been considered to be

0.5 to 0.7.17

TE
Downloaded by [University of Florida] at 15:43 09 November 2017

We aimed (1) to determine if shock index is superior to HR or SBP in predicting postpartum

hemorrhage, transfusion, and surgical intervention, and (2) to identify clinically-useful thresholds

EP
for shock index that classify patients who require further intervention.

METHODS
C
We performed a retrospective case-control study of 82 women who delivered at an academic
AC

tertiary hospital. All consecutive cases of postpartum hemorrhage (PPH) were identified from the

institutional obstetric quality-monitoring database for a 12-month period (August 2014 through

September 2015, n=41).

ST

Controls were women with normal estimated blood loss who were identified from the labor

and delivery unit registry and frequency-matched to cases based on mode of delivery and maternal
JU

weight. The upper limit of normal blood loss was 500 mL for vaginal delivery and 1000 mL for

cesarean delivery, in accordance with the ACOG guidelines.18 During this time period, blood loss

was estimated by visual examination of the drapes. We excluded women with massive hemorrhage

(>5 liters) as most were due to morbidly adherent placenta, and proactively received transfusions

and surgical interventions while hemorrhaging, limiting vital sign accuracy as an indicator of

volume status. Our Institutional Review Board granted approval for this study. Analysis was

performed using STATA/IC 14.2 for Mac (StataCorp, College Station, TX); all analyses utilized a
 significance level of alpha=0.05.

We performed medical record review to extract data about the patient characteristics:

maternal age at delivery, maternal weight in kilograms (upon presentation to the facility or at last

antenatal visit), gestational age at delivery, type of gestation (singleton or twin), prior parity,

preeclampsia, use of anti-hypertensives, laboring on facility presentation, chorioamnionitis, and

mode of delivery (spontaneous vaginal, operative vaginal, and cesarean). We also extracted the

D
following outcomes: estimated blood loss (EBL, mL); change in hemoglobin (Hgb) and hematocrit

(Hct) from prior to delivery to immediately following delivery; change in Hgb and Hct from prior to

TE
Downloaded by [University of Florida] at 15:43 09 November 2017

delivery to the nadir during hospitalization; transfusion of packed red blood cells and number of

units; and surgical intervention required to manage hemorrhage (including laceration repair in the

EP
operating room, dilation & curettage, exploratory laparotomy, and hysterectomy). We compared

patient characteristics and outcomes between cases and controls using Student’s t-test,
C
Mann-Whitney-U (Wilcoxon rank-sum), and Pearson Chi-square as appropriate.
AC

We measured four vital sign parameters for use as maternal early warning criteria: heart rate

(HR, beats/minute), systolic blood pressure (SBP, mmHg), shock index (SI, HR / SBP). Instead of

assessing all vital signs during the labor and delivery admission, we assessed HR, SBP, and SI at
ST

four key peripartum moments: last antenatal visit prior to admission, upon facility admission,

immediately prior to delivery, and at the peak SI. We measured the peak SI at any time during
JU

hospitalization as this measure was utilized to examine shock index in the context of ectopic

pregnancy.13, 14, 15 To identify a range of normal, we calculated the 5th and 95th percentiles of HR,

SBP, and SI at each of the four peripartum moments for all controls (women with blood loss at or

below the upper limit of normal). We also measured a novel criterion, delta-SI, which was

calculated as the difference between peak SI and SI at the last antenatal visit or at facility

presentation if antenatal records were unavailable.

Using Student’s t-tests with unequal variances, we evaluated whether differences exist in 1)
 SI at each of the four peripartum moments and 2) delta-SI, between women with and without the

following outcomes or characteristics: PPH, transfusion, surgical intervention, preeclampsia, labor

on presentation, and vaginal versus cesarean delivery (stratified by labor on presentation).

We used receiver operating characteristic (ROC) curves to assess the performance of HR,

SBP, peak SI, and delta-SI in discriminating between women with and without PPH, transfusion,

and surgical intervention. ROC curves plot the true positive rate (sensitivity) against the false

D
positive rate (1-specificity) for all possible thresholds of a given criterion. The area under the curve

corresponds to the probability the criterion will correctly classify a random observation; an area

TE
Downloaded by [University of Florida] at 15:43 09 November 2017

under the curve > 0.5 indicates that the criterion is superior to chance.

Detailed output of the roctab function in STATA provides information about the proportion

EP
of observations that are correctly classified at each threshold on the ROC curve. Using this output,

we identified two thresholds of peak SI (1.143 and 1.412) and three of delta-SI (0.332, 0.559, 0.847)
C
that demonstrated superior performance for the outcomes of interest.
AC

Next, we compared classification performance of the five thresholds we identified to the

classification performance of proposed maternal early warning criteria (HR >120 beats/minute, SBP

<90 or >160 mmHg)12 and to the classification performance of the peak SI thresholds used in
ST

Nathan et al (peak SI ≥ 0.7, 0.9, 1.5, and 1.7).19

Logistic regression was performed to determine the odds of PPH, transfusion, and surgical
JU

intervention predicted by peak SI and delta-SI. We performed logistic regression without the

addition of covariates, and then repeated it while adjusting for potential confounders: maternal age,

maternal weight, gestational age at delivery, prior parity, laboring on presentation, preeclampsia,

chorioamnionitis, and mode of delivery.

Post-hoc power calculations were performed using the means and standard deviations of

peak SI and delta-SI in Table 3. For PPH, we had 91% power to detect a difference in peak SI and

99% power for delta-SI. For transfusion, we had 86% power to detect a difference in peak SI and
 97% power for delta-SI. For surgical intervention, we had only 35% power to detect a difference in

peak SI and 56% power for delta-SI, due to small sample size (n=12 cases).

RESULTS

Our institutional obstetric quality-monitoring database identified 41 consecutive cases of

PPH <5000 mL from August 2014 through September 2015. Cases were frequency-matched by

maternal weight and mode of delivery to 41 controls with normal estimated blood loss (EBL).

D
Characteristics and outcomes of all patients in our sample are shown in Table 1. There were

no significant differences between PPH cases and normal controls with respect to maternal age,

TE
Downloaded by [University of Florida] at 15:43 09 November 2017

maternal weight, gestational age at delivery, type of gestation, prior parity, preeclampsia, use of

anti-hypertensives, labor on facility presentation, or mode of delivery. Due to this, we concluded

EP
that frequency-matching between cases and controls was adequate. For normal controls, median

EBL was 350 mL for vaginal deliveries and 750 mL for cesarean deliveries; for PPH cases, median
C
EBL was 750 mL for vaginal deliveries and 1800 mL for cesarean deliveries. There were no
AC

significant differences between PPH cases and normal controls in the change in Hgb or Hct

immediately after delivery; however, there was a significantly difference in Hgb and in Hct from

prior to delivery to the nadir after delivery. Of the 41 PPH cases, 16 required a transfusion, 12
ST

required surgical intervention, and 9 required both transfusion and surgical intervention.

We used 5th and 95th percentile vital signs in the controls to determine the normal range for
JU

HR, SBP, and SI in our sample. Throughout peripartum, 5th and 95th percentiles for HR were

63–105 beats/minute, and for SBP were 104–166 mmHg. The proposed maternal early warning

criteria encompass nearly all of these values within its range of normal.12 For SI, the 5th and 95th

percentiles were 0.437–0.870 at admission, 0.497–1.096 immediately prior to delivery, and

0.775–1.140 at the peak, suggesting that SI ≤ 1.1 can be considered normal in the gravid patient.

Supplemental Table 1 shows the differences in mean peripartum SI and delta-SI between

women with and without PPH, transfusion, and surgical intervention, and various characteristics of
 interest. There were significant differences in peak SI only between women with and without PPH

and with and without transfusion. There were significant differences in delta-SI between women

with and without PPH, with and without transfusion, and with and without surgical intervention.

Compared to women without preeclampsia, women with preeclampsia had a lower mean SI

during the last antenatal visit, upon facility presentation, and immediately prior to delivery. There

was no difference in peak SI by preeclampsia – thus, women with preeclampsia tended to have a

D
greater increase in SI from baseline to peak, though delta-SI was not significant.

At facility presentation, there were no differences in mean peak SI or delta-SI between

TE
Downloaded by [University of Florida] at 15:43 09 November 2017

women who were and were not laboring. The mean intrapartum SI tended to be higher for women

who eventually delivered vaginally compared to women who delivered by cesarean, and this

EP
remained present with stratification by labor status on presentation.

Supplemental Table 2 shows the discrimination performance of the ROC curves for heart
C
rate, systolic blood pressure, peak SI, and delta-SI to classify women with PPH, need for
AC

transfusion, and surgical intervention. Notably, heart rate and systolic blood pressure alone were not

significantly better than chance or a random guess in predicting PPH, need for transfusion, or

surgical intervention (95% confidence interval for the area under the ROC curve contains values ≤
ST

0.500).Peak SI and delta-SI were strong predictors of PPH, transfusion, and need for surgical

intervention, and delta-SI demonstrated the strongest discriminatory performance (highest area
JU

under the curve) for all outcomes.

Table 2 compares the classification performance of the thresholds we identified (peak SI ≥

1.143 and 1.412, and delta-SI ≥ 0.332, 0.559, and 0.847) to classification performance of individual

vital signs used in a proposed maternal early warning criteria (HR >120 beats/minute, SBP <90 or

>160 mmHg).12 Peak SI and delta-SI demonstrated generally superior classification performance

compared to HR or SBP. SI ≥1.143 correctly classified 67-77% of all outcomes. SI ≥1.412

predicted PPH and need for transfusion with 100% specificity and predicted surgical intervention
 with 97% specificity. Compared to HR, SBP, and peak SI, delta-SI ≥0.332 was the best classifier of

PPH (72%), delta-SI ≥0.559 was the best classifier of need for transfusion (83%), and delta-SI

≥0.847 was the best classifier of surgical intervention (87%).

We compared the classification performance of the two peak SI thresholds we identified to

the peak SI thresholds suggested by Nathan et al (peak SI ≥ 0.7, 0.9, 1.5, and 1.7, shown in

Supplemental Table 3).19 Our thresholds correctly classify a larger proportion of observations.

D
Previously, Nathan et al. recommended SI ≥ 0.9 as the initial trigger in low-resource settings to

initiate transfer of care, and SI ≥ 1.7 to identify critically ill patients with high mortality risk. In our

TE
Downloaded by [University of Florida] at 15:43 09 November 2017

sample, the specificity of SI ≥ 0.9 for PPH was only 24%, thus 31/41 normal controls would have

triggered the criteria erroneously if implemented in a clinical early warning system, likely resulting

EP
in ‘alarm fatigue’. Conversely, SI ≥ 1.7 as a threshold in our sample failed to detect 39/41 women

with PPH, 16/18 women with PPH who required transfusion, and 11/12 women who required
C
surgical intervention; thus it likely also has low clinical utility. Peak SI ≥ 1.143 demonstrates
AC

improved classification, compared to previous values,19 as an initial threshold due to its

substantially higher specificity and peak SI ≥ 1.412 demonstrates much stronger classification

performance as a “critical level” threshold due to its improved sensitivity.

ST

We used logistic regression to adjust for factors that could potentially confound the

predictive relationship of peak SI and delta-SI for PPH, need for transfusion, and surgical
JU

intervention (Supplemental Table 4). We included maternal age, maternal weight, gestational age at

delivery, prior parity, preeclampsia, laboring on presentation, chorioamnionitis, and mode of

delivery as covariates in our adjusted model. Inclusion of potential confounders as covariates in the

logistic regression tended to strengthen the predictive odds of peak SI and delta-SI. Maternal age,

gestational age, preeclampsia, and cesarean delivery were significant covariates in some but not all

models, and the odds ratios for the peak SI or delta-SI thresholds were substantially greater than the

odds ratios for these covariates (data available upon request). Maternal age, gestational age, and
 presence of preeclampsia were associated with increased odds of PPH, transfusion, and surgical

intervention, while cesarean delivery was associated with decreased odds of PPH, transfusion, and

surgical intervention.

DISCUSSION

Main Findings

D
First, peak SI and delta-SI are generally superior to proposed maternal early warning criteria

as predictors of PPH, transfusion, and surgical intervention. We found that peak SI and delta-SI

TE
Downloaded by [University of Florida] at 15:43 09 November 2017

were higher in cases (women with PPH, transfusion, and surgical intervention) compared to controls

(women without these adverse outcomes). Next, peak SI and delta-SI demonstrated superior

EP
discrimination of these adverse outcomes compared to HR or SBP, shown by a greater area under

the ROC curve. Lastly, when testing the classification performance of peak SI and delta-SI
C
thresholds for PPH, transfusion, and surgical intervention, peak SI and delta-SI exhibited equivalent
AC

or superior performance to the HR or SBP thresholds recommended in proposed maternal early

warning criteria.

Additionally, our thresholds for peak SI correctly classified more observations than the
ST

thresholds suggested by Nathan et al.19,23 and El Ayadi et al.20 in studies to assess the usefulness of

shock index in women with PPH. For an initial threshold, we found that peak SI ≥ 1.143 was much
JU

stronger than peak SI ≥ 0.9, as 1.143 excluded a larger proportion of normal women and thus had a

much higher specificity. For a “critical” threshold, we found that peak SI ≥ 1.412 demonstrates

equal specificity to the more extreme peak SI ≥ 1.5 and peak SI ≥ 1.7, while remaining a more

conservative threshold, and thus retained higher sensitivity.

Notably, our novel criterion, delta-SI, may better account for variation unique to an

individual by assessing the change in shock index from the patient’s baseline. Compared to HR and

SBP thresholds in proposed maternal early warning criteria, we found delta-SI thresholds provided
 excellent classification for all adverse outcomes of interest. The best classifiers of PPH, transfusion,

and surgical intervention were delta-SI ≥ 0.332, delta-SI ≥ 0.559, and delta-SI ≥ 0.847,

respectively.

We note that shock index at the last antenatal visit and upon facility presentation were not

predictive of any future adverse outcome that we examined, and thus calculation of shock index at

these times is not useful for determining who may experience postpartum hemorrhage. However,

D
calculation of shock index at one or both of these times is essential for determining a baseline by

which to calculate delta-SI during labor, delivery, and the postpartum hospitalization.

TE
Downloaded by [University of Florida] at 15:43 09 November 2017

Finally, we found that peak SI and delta-SI remain sensitive and specific classifiers, even

when adjusted for potential confounders.

Strengths and Limitations

EP
C
Strengths of our study include performance assessment of existing vital sign values within
AC

proposed maternal early warning criteria, and comparison with novel thresholds. Additionally, our

study was appropriately powered to detect differences in peak SI and delta-SI between women with

and without PPH, transfusion, and surgical intervention. Lastly, our study proposes a novel criterion
ST

that may account for variation unique to an individual by assessing the change in shock index from

the patient’s baseline; this novel criterion demonstrated strong discrimination and classification
JU

performance.

Our study has important limitations. We identified trigger thresholds with strong

classification performance using a retrospective case-control design with a limited sample size, thus

prospective validation within a larger cohort is necessary to assess the frequency that these criteria

would trigger escalation of care. Additionally, at the time of this analysis, blood loss was estimated

and not quantified by weighed on our labor and delivery unit, and therefore the accuracy of the

blood loss measures are limited. We extracted data regarding utilization of anti-hypertensives for
 preeclampsia in our sample, but did not include this data in analysis due to the very limited sample

size. Finally, we excluded women with massive hemorrhage (>5 liters), as most of these cases were

due to morbidly adherent placenta and were concurrently and proactively transfused during ongoing

blood loss. As a result, hemodynamic repletion concurrent with hemorrhage prevented accurate

measurement of the “true” vital signs that would occur as a physiologic response.

D
Conclusions

In our study, we demonstrate that shock index up to 1.1 can be considered normal in

TE
Downloaded by [University of Florida] at 15:43 09 November 2017

peripartum gravidas, which is consistent with previous studies evaluating postpartum SI,21, 22, 23 and

higher than the upper limit of 0.9 suggested by other studies.24 Regardless, this reaffirms the

EP
necessity of unique SI thresholds for the obstetric patient, rather than using values calibrated for the

non-gravid individual, in which 0.5 to 0.7 is the upper limit of normal.17

C
Women with preeclampsia require unique consideration. No studies have previously
AC

evaluated shock index changes in patients with preeclampsia. We included these women in our

sample in order to assess whether early warning criteria incorporating shock index could apply to

women, regardless of preeclampsia status. In our study, there was a slightly higher proportion of
ST

women with preeclampsia in the postpartum hemorrhage cases compared to the normal controls, as

would be expected because preeclampsia is an independent risk factor for PPH,25, 26 though the
JU

difference was not statistically significant. We found that shock index was significantly lower in

women with preeclampsia during prenatal care, at admission, and intrapartum, likely due to the

elevated systolic blood pressure present with preeclampsia. Conversely, peak SI did not differ by

preeclampsia; this acute rise in heart rate and concomitant rise in shock index suggests that the

physiology of preeclampsia is associated with reduced capacity for hemodynamic compensation

during peripartum. As a result, women with preeclampsia had a slightly higher delta-SI.

Preeclampsia was often a significant covariate in logistic regression; however, inclusion as a

 covariate tended to strengthen the predictive odds of peak-SI and delta-SI thresholds for the adverse

outcomes of interest, suggesting that the utility of shock index as an early warning criteria does not

diminish with a diagnosis of preeclampsia.

Shock index and its derivative, delta-SI, are valuable indicators of PPH and maternal

deterioration requiring transfusion and/or surgical intervention. Our case-control study was

sufficiently powered to detect differences in peak SI and delta-SI between women with and without

D
these adverse outcomes. Peak SI and delta-SI are superior to HR and SBP as maternal early warning

criteria, and we identified new thresholds for peak SI and for delta-SI with improved classification

TE
Downloaded by [University of Florida] at 15:43 09 November 2017

performance (and thus potentially greater clinical utility) compared to previous SI thresholds. We

recommend a threshold of SI ≥ 1.143 as an initial threshold to trigger further evaluation, and SI

EP
≥1.412 as a threshold to trigger emergent intervention. Finally, we propose a novel criterion,

delta-SI, as a new metric that facilitates monitoring of hemodynamic changes unique to the
C
individual patient during peripartum, because delta-SI remained sensitive and specific even in the
AC

setting of preeclampsia and demonstrated the strongest performance of all classifiers examined.

Future studies should prospectively evaluate its utility in clinical care.

Acknowledgements: None.
ST

Conflicts of Interests: GAD is a co-inventor and patent holder of the Ebb Balloon Tamponade
JU

System (for postpartum hemorrhage) licensed to Clinical Innovations. JRK and CSE do not have
financial, personal, political, intellectual, or religious interests.

Contribution to authorship:
JRK: Contributed to the conception and design of the work, acquired and interpreted the data for the
work, drafted the manuscript and revised it for important intellectual content, provided final
approval for publication, and agreed to be accountable for all aspects of the work.

GAD: Contributed to the conception and design of the work, interpreted the data for the work,
 revised the manuscript for important intellectual content, provided final approval for publication,
and agreed to be accountable for all aspects of the work.

CSE: Contributed to the conception and design of the work, acquired and interpreted the data for the
work, drafted the manuscript and revised it for important intellectual content, provided final
approval for publication, and agreed to be accountable for all aspects of the work.

Ethical Approval: Institutional Review Board approval was granted for this work prior to any data

D
collection (Human Protocol 37476).

TE
Downloaded by [University of Florida] at 15:43 09 November 2017

Funding: There was no funding associated with this work.

REFERENCES
EP
1. Pregnancy Mortality Surveillance System | Pregnancy | Reproductive Health | CDC. 2016 January
C
21, 2016 [cited 2016 October 31]Available from:
AC

[Link]

2. Creanga AA, Berg CJ, Syverson C, Seed K, Bruce FC, Callaghan WM. Pregnancy-related mortality

in the United States, 2006-2010. Obstetrics and gynecology 2015, 125(1): 5-12.
ST

3. Say L, Chou D, Gemmill A, Tuncalp O, Moller AB, Daniels J, et al. Global causes of maternal death:

a WHO systematic analysis. The Lancet Global health 2014, 2(6): e323-333.
JU

4. California Partnership for Maternal Safety | California Maternal Quality Care Collaborative. 2016

[cited 2016 October 31]Available from:

[Link]

5. The AWHONN Postpartum Hemorrhage Project. 2016 [cited 2016 October 31]Available from:

[Link]

6. Maguire PJ, O'Higgins A, Power K, Turner MJ. The Irish Maternity Early Warning System

(IMEWS). Irish medical journal 2014, 107(10): 309.

 7. Friedman AM. Maternal early warning systems. Obstetrics and gynecology clinics of North America

2015, 42(2): 289-298.

8. Edwards SE, Grobman WA, Lappen JR, Winter C, Fox R, Lenguerrand E, et al. Modified obstetric

early warning scoring systems (MOEWS): validating the diagnostic performance for severe sepsis in women

with chorioamnionitis. American journal of obstetrics and gynecology 2015, 212(4): 536.e531-538.

9. Lappen JR, Keene M, Lore M, Grobman WA, Gossett DR. Existing models fail to predict sepsis in an

D
obstetric population with intrauterine infection. American journal of obstetrics and gynecology 2010, 203(6):

TE
573.e571-575.
Downloaded by [University of Florida] at 15:43 09 November 2017

10. Mackintosh N, Watson K, Rance S, Sandall J. Value of a modified early obstetric warning system

(MEOWS) in managing maternal complications in the peripartum period: an ethnographic study. BMJ

quality & safety 2014, 23(1): 26-34.

EP
11. Singh S, McGlennan A, England A, Simons R. A validation study of the CEMACH recommended
C
modified early obstetric warning system (MEOWS). Anaesthesia 2012, 67(1): 12-18.
AC

12. Mhyre JM, D'Oria R, Hameed AB, Lappen JR, Holley SL, Hunter SK, et al. The maternal early

warning criteria: a proposal from the national partnership for maternal safety. Obstetrics and gynecology

2014, 124(4): 782-786.

ST

13. Jaramillo S, Barnhart K, Takacs P. Use of the shock index to predict ruptured ectopic pregnancies.

International journal of gynaecology and obstetrics: the official organ of the International Federation of
JU

Gynaecology and Obstetrics 2011, 112(1): 68.

14. Birkhahn RH, Gaeta TJ, Bei R, Bove JJ. Shock index in the first trimester of pregnancy and its

relationship to ruptured ectopic pregnancy. Academic emergency medicine : official journal of the Society for

Academic Emergency Medicine 2002, 9(2): 115-119.

15. Birkhahn RH, Gaeta TJ, Van Deusen SK, Tloczkowski J. The ability of traditional vital signs and

shock index to identify ruptured ectopic pregnancy. American journal of obstetrics and gynecology 2003,

189(5): 1293-1296.
 16. Rady MY, Nightingale P, Little RA, Edwards JD. Shock index: a re-evaluation in acute circulatory

failure. Resuscitation 1992, 23(3): 227-234.

17. Rady MY, Rivers EP, Martin GB, Smithline H, Appelton T, Nowak RM. Continuous central venous

oximetry and shock index in the emergency department: use in the evaluation of clinical shock. The

American journal of emergency medicine 1992, 10(6): 538-541.

18. ACOG Practice bulletin No. 76: Postpartum hemorrhage. Obstetrics and gynecology 2006, 108:

D
1039-1047.

TE
19. Nathan HL, El Ayadi A, Hezelgrave NL, Seed P, Butrick E, Miller S, et al. Shock index: an
Downloaded by [University of Florida] at 15:43 09 November 2017

effective predictor of outcome in postpartum haemorrhage? BJOG : an international journal of obstetrics

and gynaecology 2015, 122(2): 268-275.

EP
20. El Ayadi AM, Nathan HL, Seed PT, Butrick EA, Hezelgrave NL, Shennan AH, et al. Vital Sign

Prediction of Adverse Maternal Outcomes in Women with Hypovolemic Shock: The Role of Shock Index.
C
PLoS One 2016, 11(2): e0148729.
AC

21. Taylor D, Fleischer A, Rosen L. Determining the Normal Values of Shock Index, Pulse Pressure,

and ROPE in the Immediate Postpartum Period. Obstetrics & Gynecology 2016, 127(Supplement 1): 38S.

22. Le Bas A, Chandraharan E, Addei A, Arulkumaran S. Use of the "obstetric shock index" as an
ST

adjunct in identifying significant blood loss in patients with massive postpartum hemorrhage. International

journal of gynaecology and obstetrics: the official organ of the International Federation of Gynaecology and
JU

Obstetrics 2014, 124(3): 253-255.

23. Taylor D, Fleischer A, Meirowitz N, Rosen L. Shock index and vital-sign reference ranges during

the immediate postpartum period. International journal of gynaecology and obstetrics: the official organ of

the International Federation of Gynaecology and Obstetrics 2017.

24. Nathan HL, Cottam K, Hezelgrave NL, Seed PT, Briley A, Bewley S, et al. Determination of

Normal Ranges of Shock Index and Other Haemodynamic Variables in the Immediate Postpartum Period: A

Cohort Study. PLoS One 2016, 11(12): e0168535.

 25. von Schmidt auf Altenstadt JF, Hukkelhoven C, van Roosmalen J, Bloemenkamp KWM.

Pre-Eclampsia Increases the Risk of Postpartum Haemorrhage: A Nationwide Cohort Study in The

Netherlands. PLoS One 2013, 8(12): e81959.

26. Eskild A, Vatten LJ. Abnormal bleeding associated with preeclampsia: a population study of

315,085 pregnancies. Acta obstetricia et gynecologica Scandinavica 2009, 88(2): 154-158.

TABLE AND FIGURE LEGENDS

D
Table 1. Patient characteristics and outcomes

TE
Downloaded by [University of Florida] at 15:43 09 November 2017

Table 2. Classification performance to predict adverse outcomes (strongest classifier in bold)

EP
Supplemental Table 1. Peripartum SI and delta-SI compared for various groups (mean ± standard
C
deviation)
AC

Supplemental Table 2. Discrimination performance of classifiers for PPH, transfusion, and surgical
intervention (area under the ROC curve)
ST

Supplemental Table 3. Classification performance of our SI thresholds compared to SI thresholds

identified in Nathan et al.
JU

Supplemental Table 4. Odds of adverse maternal outcomes using SI and delta-SI (OR, 95% CI).

Supplemental Figure 1. Receiver operating characteristic (ROC) curve assessing the discrimination
performance of peak SI for postpartum hemorrhage (PPH).

Supplemental Figure 2. Receiver operating characteristic (ROC) curve assessing the discrimination
 Downloaded by [University of Florida] at 15:43 09 November 2017

JU
ST
AC
C
performance of delta-SI for postpartum hemorrhage (PPH).

EP
TE
D
 Table 1. Patient characteristics and outcomes

Patient Characteristics Cases (n = 41) Controls (n = 41)

Maternal age (years, mean±SD) 29.2 ± 6.5 27.9 ± 6.5 p = 0.37
Maternal weight (kg, mean±SD) 81.2 ± 18.1 82.4 ± 14.4 p = 0.74
Gestational age at delivery (n, %)
Full term (≥37 weeks) 38, 93% 32, 78% p = 0.06
Pre term (<37 weeks) 03, 07% 09, 22%
Singleton gestation (n, %) 39, 95% 41, 100% p = 0.15
Parity (n, %)

D
P0 15, 37% 13, 32% p = 0.36
P1 – P3 19, 46% 26, 63%

TE
P≥4 07, 17% 02, 05%
Downloaded by [University of Florida] at 15:43 09 November 2017

Preeclampsia (n, %) 12, 29% 06, 15% p = 0.11

Anti-hypertensive use (n, %) 07, 17% 04, 10% p = 0.73
Laboring on presentation (n, %) 22, 54% 24, 58% p = 0.66
Chorioamnionitis (n, %)
Mode of delivery (n, %)
Spontaneous vaginal delivery
08, 20%

17, 41%
EP 04, 10%

16, 39%
p = 0.21

p = 0.08
C
Operative vaginal delivery 11, 27% 04, 10%
Cesarean delivery 13, 32% 21, 51%
AC

Patient Outcomes Cases (n = 41) Controls (n = 41)

Blood loss (mL, median & IQR)
Spontaneous vaginal delivery 1050 (750-1450) 350 (325-350) p < 0.01
Operative vaginal delivery
ST

Cesarean delivery 1800 (1500-2400) 750 (600-900)

Change in labs (median & IQR)
Pre-delivery – post-delivery, immediate
∆ Hemoglobin (g/dL) -1.30 (0.70-2.10) -1.10 (0.35-2.05) p = 0.38
JU

∆ Hematocrit (%) -4.10 (1.65-6.65) -3.00 (1.45-6.05) p = 0.35

Pre-delivery – post-delivery, nadir
∆ Hemoglobin (g/dL) -2.70 (1.25-4.00) -1.20 (0.30-2.25) p < 0.01
∆ Hematocrit (%) -7.90 (5.75-11.30) -3.35 (1.55-6.35) p < 0.01
Transfusion (n, %) 16, 39% 3, 7% p < 0.01
Surgical Interventions (n, %)
Dilation & curettage 6, 15% 0, 0% p < 0.01
Laceration repair in the OR 3, 07% 0, 0%
Exploratory laparotomy 1, 02% 0, 0%
Labial hematoma drainage 1, 02% 0, 0%
Hysterectomy 1, 02% 0, 0%
 n: number of observations in sample group, SD: standard deviation, kg: kilograms, %: proportion with given
characteristic in sample group, mL: milliliters, IQR: interquartile range, ∆: change.
Table 2. Classification performance to predict adverse outcomes (strongest classifier in
bold)

Outcome Threshold Correct Classification Sensitivity Specificity

PPH ∆SI ≥ 1.143 67% 41% 93%
∆SI ≥ 1.412 55% 10% 100%
Delta-SI ≥ 0.332 72% 68% 76%
Delta-SI ≥ 0.559 60% 22% 98%

D
Delta-SI ≥ 0.847 54% 07% 100%
HR > 120 53% 12% 94%

TE
Downloaded by [University of Florida] at 15:43 09 November 2017

SBP < 90 52% 04% 98%

SBP > 160 50% 03% 97%
Transfusion ∆SI ≥ 1.143 77% 53% 84%

EP
∆SI ≥ 1.412 82% 21% 100%
Delta-SI ≥ 0.332 72% 90% 67%
Delta-SI ≥ 0.559 83% 37% 95%
Delta-SI ≥ 0.847 81% 16% 100%
C
HR > 120 71% 07% 90%
SBP < 90 80% 10% 99%
AC

SBP > 160 78% 04% 98%

Surgical ∆SI ≥ 1.143 73% 42% 79%
Intervention ∆SI ≥ 1.412 85% 17% 97%
Delta-SI ≥ 0.332 63% 83% 60%
ST

Delta-SI ≥ 0.559 83% 33% 91%

Delta-SI ≥ 0.847 87% 17% 99%
HR > 120 79% 11% 91%
SBP < 90 85% 06% 97%
JU

SBP > 160 84% 03% 97%

SI: shock index, Delta-SI: change in shock index from baseline value, PPH: postpartum hemorrhage.
 Supplemental Table 1. Peripartum SI and delta-SI values for various groups (mean ±

standard deviation, n for the group)

OUTCOMES Antenatal SI Admission SI Intrapartum SI Peak SI Delta-SI
PPH Present(n=41) 0.696 ± 0.103 0.656 ± 0.123 0.798 ± 0.180 1.131 ± 0.253 0.437 ± 0.247
Absent(n=41) 0.743 ± 0.114 0.708 ± 0.149 0.772 ± 0.157 0.979 ± 0.142 0.236 ± 0.162
p-value p = 0.13 p = 0.09 p = 0.51 p < 0.01 p < 0.01
Transfusion Present(n=19) 0.658 ± 0.101 0.644 ± 0.144 0.814 ± 0.238 1.227 ± 0.291 0.545 ± 0.283
Absent(n=63) 0.737 ± 0.109 0.694 ± 0.137 0.777 ± 0.142 1.003 ± 0.160 0.274 ± 0.171
p-value p = 0.07 p = 0.19 p = 0.55 p < 0.01 p < 0.01

D
Surgical Present(n=12) 0.659 ± 0.076 0.643 ± 0.142 0.823 ± 0.204 1.197 ± 0.331 0.534 ± 0.338
intervention Absent(n=70) 0.730 ± 0.112 0.689 ± 0.139 0.779 ± 0.162 1.030 ± 0.185 0.303 ± 0.191

TE
Downloaded by [University of Florida] at 15:43 09 November 2017

p-value p = 0.16 p = 0.31 p = 0.49 p = 0.12 p = 0.04

CHARACTERISTICS Antenatal SI Admission SI Intrapartum SI Peak SI Delta-SI
Pre-eclampsiaPresent(n=18) 0.636 ± 0.105 0.605 ± 0.182 0.706 ± 0.114 1.079 ± 0.282 0.419 ± 0.316

EP
Absent(n=64) 0.748 ± 0.101 0.704 ± 0.117 0.811 ± 0.176 1.048 ± 0.198 0.314 ± 0.198
p-value p < 0.01 p = 0.04 p = 0.02 p = 0.60 p = 0.09
Laboring Present(n=46) 0.728 ± 0.104 0.686 ± 0.113 0.813 ± 0.178 1.054 ± 0.210 0.339 ± 0.218
on Absent(n=36) 0.721 ± 0.120 0.677 ± 0.169 0.749 ± 0.150 1.056 ± 0.231 0.335 ± 0.250
C
Presentation p-value p = 0.81 p = 0.78 p = 0.10 p = 0.96 p = 0.94
AC

Mode of Vaginal 0.752 ± 0.098 0.719 ± 0.115 0.845 ± 0.168 1.109 ± 0.231 0.364 ± 0.255
Delivery (n=48)
Cesarean 0.701 ± 0.118 0.630 ± 0.155 0.685 ± 0.114 0.979 ± 0.174 0.298 ± 0.190
(n=34)
p-value p = 0.10 p < 0.01 p < 0.01 p < 0.01 p = 0.20
ST

Laboring on presentation
Vaginal (n=32) 0.765 ± 0.090 0.708 ± 0.086 0.864 ± 0.174 1.092 ± 0.206 0.360 ± 0.214
Cesarean 0.679 ± 0.104 0.636 ± 0.151 0.676 ± 0.103 0.966 ± 0.199 0.291 ± 0.227
JU

(n=14)
p-value p = 0.03 p = 0.05 p < 0.01 p = 0.06 p = 0.33
Not laboring on presentation
Vaginal 0.727 ± 0.112 0.743 ± 0.160 0.806 ± 0.154 1.142 ± 0.280 0.374 ± 0.330
(n=16)
Cesarean 0.717 ± 0.128 0.625 ± 0.160 0.692 ± 0.124 0.988 ± 0.159 0.303 ± 0.164
(n=20)
p-value p = 0.85 p = 0.04 p = 0.03 p = 0.04 p = 0.41

SI: shock index, delta-SI: change in shock index from baseline value, n: number of observations in sample group, PPH:
postpartum hemorrhage.
Supplemental Table 2. Discrimination performance of classifiers for PPH, transfusion, and
 surgical intervention (area under the ROC curve)

Antenatal Admission Intrapartum Peak SI Delta-SI HR SBP

SI SI SI
PPH 0.374 0.378 0.520 0.677 0.767 0.538 0.567
Transfusion 0.311 0.360 0.487 0.760 0.834 0.472 0.499
Surgical 0.296 0.358 0.541 0.649 0.748 0.516 0.585
intervention

ROC curve: receiver operating characteristic curve, AUC: area under the curve (probability an observation will be

D
correctly classified), SI: shock index, HR: heart rate, SBP: systolic blood pressure, delta-SI: change in shock index from
baseline value, PPH: postpartum hemorrhage.

TE
Downloaded by [University of Florida] at 15:43 09 November 2017

Bold indicates that the 95% confidence interval for the area under the ROC curve does not contain 0.500 (the value is
statistically greater than 0.5 – i.e. the classifier is better than chance).

EP
HR and SBP combine all of the following vital sign values: upon presentation to the facility, intrapartum (last taken
before delivery), and at peak SI.
C
AC

Supplemental Table 3. Classification performance of our SI thresholds compared to the SI thresholds

identified in Nathan et al.

Outcome Threshold Correctly Classified Sensitivity Specificity

ST

PPH SI ≥ 1.143 67% 41% 93%

SI ≥ 1.412 55% 10% 100%
SI ≥ 0.700 51% 100% 2%
SI ≥ 0.900 55% 85% 24%
JU

SI ≥ 1.500 52% 5% 100%

SI ≥ 1.700 52% 5% 100%
Transfusion SI ≥ 1.143 77% 53% 84%
SI ≥ 1.412 82% 21% 100%
SI ≥ 0.700 24% 100% 2%
SI ≥ 0.900 40% 95% 24%
SI ≥ 1.500 79% 11% 100%
SI ≥ 1.700 79% 11% 100%
 Surgical SI ≥ 1.143 73% 42% 79%
Intervention SI ≥ 1.412 85% 17% 97%
SI ≥ 0.700 16% 100% 1%
SI ≥ 0.900 34% 100% 23%
SI ≥ 1.500 85% 8% 99%
SI ≥ 1.700 85% 8% 99%

SI: shock index, PPH: postpartum hemorrhage.

D
TE
Downloaded by [University of Florida] at 15:43 09 November 2017

EP
C
AC
ST
JU
 Supplemental Table 4. Odds of adverse maternal outcomes using SI and delta-SI (OR with
95% CI).

Outcome Threshold Unadjusted OR Adjusted OR

PPH SI ≥ 1.143 12.5 (2.6-59.0) 32.8 (3.6-297.2)
SI ≥ 1.412 - -
Delta-SI ≥ 0.332 6.0 (2.3-15.6) 6.1 (1.9-19.7)
Delta-SI ≥ 0.559 11.3 (1.4-93.5) 24.2 (1.3-467.1)
Delta-SI ≥ 0.847 - -

D
Transfusion SI ≥ 1.143 3.9 (1.2-12.0) 4.6 (1.0-21.0)
SI ≥ 1.412 - -

TE
Downloaded by [University of Florida] at 15:43 09 November 2017

Delta-SI ≥ 0.332 10.7 (2.8-40.7) 8.0 (1.8-35.3)

Delta-SI ≥ 0.559 11.7 (2.6-51.6) 34.8 (3.0-397.0)
Delta-SI ≥ 0.847 - -

EP
Surgical SI ≥ 1.143 3.1 (0.9-11.4) 2.3 (0.3-17.5)
Intervention SI ≥ 1.412 13.8 (1.1-166.5) 60.0 (1.8-2013.8)
Delta-SI ≥ 0.332 8.0 (1.6-39.2) 5.0 (0.6-38.6)
Delta-SI ≥ 0.559 5.3 (1.2-23.0) 10.7 (0.9-130.4)
C
Delta-SI ≥ 0.847 13.8 (1.1-166.5) 60.0 (1.8-2013.8)
AC

SI: Shock index, Delta-SI: change in shock index from baseline value, OR: Odds ratio, 95% CI: Confidence interval,
PPH: postpartum hemorrhage.

“-” indicates that the threshold has 100% specificity for the adverse outcome of interest; thus, it is a perfect predictor of
ST

the outcome and is omitted from the model.

2
Kohn
JU

SI as PPH warning criterion

21
Kohn
SI as PPH warning criterion