Tests for diagnosis of postpartum haemorrhage at vaginal birth (2025)

Abstract

Background

Postpartum haemorrhage (PPH) is the leading cause of maternal mortality worldwide. Accurate diagnosis of PPH can prevent adverse outcomes by enabling early treatment.

Objectives

What is the accuracy of methods (index tests) for diagnosing primary PPH (blood loss ≥ 500 mL in the first 24 hours after birth) and severe primary PPH (blood loss ≥ 1000 mL in the first 24 hours after birth) (target conditions) in women giving birth vaginally (participants) compared to weighed blood loss measurement or other objective measurements of blood loss (reference standards)?

Search methods

We searched CENTRAL, MEDLINE, Embase, Web of Science Core Collection, ClinicalTrials.gov, and the World Health Organization International Clinical Trials Registry Platform to 24 May 2024.

Selection criteria

We included women who gave birth vaginally in any setting. Study types included diagnostic cohort studies and cross‐sectional studies that reported 2 x 2 data (number of true positive, false positive, false negative, and true negative results) or where the 2 x 2 data could be derived from test accuracy estimates.

Eligible index tests included: visual estimation; calibrated blood collection devices; approach with calibrated drape and observations; blood loss estimation using the SAPHE (Signalling a Postpartum Hemorrhage Emergency) Mat; blood loss field image analysis and other technologies; uterine atony assessment; clinical variables (e.g. heart rate, blood pressure, shock index); early warning charts; haemoglobin levels; and predelivery fibrinogen levels.

Eligible reference standards included objective methods such as: gravimetric blood loss measurement, which involves weighing collected blood, as well as weighing blood‐soaked pads, gauze and sheets, and subtracting their dry weight; calibrated devices to measure blood volume (volumetric blood loss measurement); the alkaline‐haematin method of blood loss estimation; and blood extracted using machine‐extraction and measured spectrometrically as oxyhaemoglobin.

Data collection and analysis

At least two review authors, working independently, undertook study screening, selection, data extraction, assessment of risk of bias, and assessment of the certainty of the evidence. We resolved any differences through consensus or with input from another author.

We generated 2 x 2 tables of the true positives, true negatives, false positives, and false negatives to calculate the sensitivity, specificity, and 95% confidence intervals for each index test. We presented sensitivity and specificity estimates from studies in forest plots. Where possible, we conducted meta‐analyses for each index test and reference standard combination for each target condition.

We examined heterogeneity by visual inspection of the forest plots.

Main results

Our review included 18 studies with a total of 291,040 participants. Fourteen studies evaluated PPH and seven studies evaluated severe PPH. Most studies were conducted in a hospital setting (16 of 18).

There were four studies at high risk of bias for the patient selection domain and 14 studies at low risk. For the index test domain, 10 studies were at low risk of bias, seven studies at high risk, and one study at uncertain risk. For the reference standard domain, one study was at high risk of bias and 17 studies at low risk. For the flow and timing domain, three studies were at high risk of bias and 15 studies at low risk. The applicability concerns were low for all studies across the domains.

In the abstract, we have prioritised reporting results for common, important thresholds for index tests or where the certainty of the evidence for the sensitivity estimate was at least moderate. Full results are in the main body of the review.

PPH (blood loss ≥ 500 mL)

For PPH, visual estimation with gravimetric blood loss measurement as the reference standard had 48% sensitivity (95% confidence interval (CI) 44% to 53%; moderate certainty) and 97% specificity (95% CI 95% to 99%; high certainty) (4 studies, 196,305 participants).

Visual estimation with volumetric blood loss measurement as the reference standard showed 22% sensitivity (95% CI 12% to 37%; moderate certainty) and 99% specificity (95% CI 97% to 100%; moderate certainty) (2 studies, 514 participants).

The diagnostic approach with calibrated drape plus observations, with gravimetric blood loss measurement as the reference standard for PPH, showed 93% sensitivity (95% CI 92% to 94%; high certainty) and 95% specificity (95% CI 95% to 96%; high certainty) (2 studies, 53,762 participants).

A haemoglobin level of less than 10 g/dL with gravimetric blood loss measurement as the reference standard showed 37% sensitivity (95% CI 30% to 44%; high certainty) and 79% specificity (95% CI 76% to 82%; high certainty) (1 study, 1058 participants).

Severe PPH (blood loss ≥ 1000 mL)

For severe PPH, visual estimation, with volumetric plus gravimetric blood loss measurement as the reference standard, showed 9% sensitivity (95% CI 0% to 41%; low certainty) and 100% specificity (95% CI 99% to 100%; moderate certainty) (1 study, 274 participants).

A shock index level of 1.0 or higher (commonly used as a threshold for severe PPH) up to two hours after birth, with gravimetric blood loss measurement as the reference standard, showed 30% sensitivity (95% CI 27% to 33%; moderate certainty) and 93% specificity (95% CI 92% to 93%; moderate certainty) (1 study, 30,820 participants).

A haemoglobin level of less than 10 g/dL, with gravimetric blood loss measurement as the reference standard, showed 71% sensitivity (95% CI 51% to 87%; moderate certainty) and 77% specificity (95% CI 75% to 80%; high certainty) (1 study, 1058 participants).

Authors' conclusions

Visual estimation of blood loss to diagnose PPH showed low sensitivity and is likely to miss the diagnosis in half of women giving birth vaginally. A diagnostic approach using a calibrated drape to objectively measure blood loss plus clinical observations showed high sensitivity and specificity for diagnosing PPH. Other index tests showed low to moderate sensitivities in diagnosing PPH and severe PPH.

Future research should determine the accuracy of diagnostic tests in non‐hospital settings and consider combining index tests to increase the sensitivity of PPH diagnosis.

Funding

Bill and Melinda Gates Foundation

Registration

PROSPERO (CRD42024541874)

Plain language summary

How accurate are tests that diagnose excessive bleeding (postpartum haemorrhage) after women give birth vaginally?

Key messages

• Our study found that using visual estimation to diagnose postpartum haemorrhage (PPH) was inaccurate. Using a plastic drape to collect and measure the volume of blood loss – alongside observations including heart rate, blood pressure, tone of the womb, and flow of blood – showed high accuracy.

• Other tests, including blood tests and measurements such as heart rate and blood pressure alone, showed varying levels of accuracy.

What is postpartum haemorrhage (PPH)?

The World Health Organization defines PPH as blood loss of 500 mL or more in the first 24 hours after delivery in women who gave birth vaginally. Severe PPH is when 1000 mL or more of blood is lost in the same time period.

Why is an accurate diagnosis of PPH important?

PPH is the commonest cause of mothers dying worldwide. Accurately diagnosing it can help in early treatment.

What are the tests used to diagnose PPH?

The commonest test used to diagnose PPH and severe PPH is visual estimation, where a birth attendant estimates the volume of blood lost by looking at the extent to which bedsheets and pads are soaked with blood. Other tests include measuring the volume of blood lost using a plastic drape, tray or bowl with markings indicating the volume. Blood can also be collected and weighed. This weight is converted to a volume using a formula. Other tests include measuring changes in (1) the levels of certain chemicals in the birthing woman's blood, or (2) variables such as her heart rate and blood pressure.

What did we want to find out?

We wanted to identify different tests and methods used to diagnose PPH and severe PPH and find out how accurate they are.

What did we do?

We looked for studies that assessed the accuracy of tests used to diagnose PPH and severe PPH when compared with a reliable standard such as weighed blood loss or the measured volume of blood loss. Women of any age who gave birth vaginally in any setting (hospitals, delivery units in the community, home births) were included.

We included studies which provided data that we could use to determine two measures of test accuracy: (1) sensitivity (percentage of women with the condition who are correctly identified by the test); and (2) specificity (percentage of women without the condition who are correctly excluded by the test). Where appropriate, we combined the results across these studies. We excluded studies that did not provide this type of information.

Examples of tests we looked for included the assessor: looking at the blood loss and estimating the blood volume (visual estimation); measuring the volume of blood loss in a drape or other collecting device with markings indicating the volume (volumetric method); weighing blood loss using scales (gravimetric method); measuring changes in variables such as heart rate and blood pressure, or changes in blood levels of chemicals such as haemoglobin and fibrinogen. We also tried to find studies which combined the tests mentioned above and tests using new technologies such as camera systems and artificial intelligence.

What did we find?

We found 18 studies with a total of 291,040 participants.

Fourteen studies assessed diagnostic tests for PPH and seven studies assessed tests for severe PPH. Most of the studies were performed in hospitals.

Visual estimation had poor sensitivity. In one of our analyses, we found that visual estimation will only diagnose 48 out of 100 women who have PPH, but 52 women with PPH will be missed (i.e. will be false negatives). These missed cases may not receive treatment and so may suffer avoidable harm and may even die. Of every 100 women without PPH, three will be wrongly diagnosed as having it (i.e. will be false positives). These incorrectly diagnosed women may receive treatment which they don't require and may suffer harm as a result.

A diagnostic approach which used a calibrated drape to measure the volume of blood loss, along with observations such as heart rate, blood pressure, the tone of the womb, and flow of blood to diagnose PPH had very good sensitivity and specificity. Using this approach will diagnose 93 out of 100 women who have PPH, and only seven women with PPH will be missed (i.e. will be false negatives). Of every 100 women without PPH, five will be wrongly diagnosed as having it (i.e. will be false positives).

Other tests showed varying levels of accuracy for diagnosing PPH and severe PPH.

What are the limitations of the evidence?

The studies we found were mainly performed in hospitals. We would have liked more information on how accurate the tests are in other settings, such as in the community and at home.

We would also have liked to know how accurate the other tests are when used in combination, particularly the combination of measured blood loss, changes in factors such as heart rate and blood pressure, and changes in blood chemical levels.

How current is this evidence?

This evidence is current to 24 May 2024.

Summary of findings

Summary of findings 1. Summary of findings 1. Diagnostic accuracy of tests for diagnosing postpartum haemorrhage in women giving birth vaginally.

Participants: women who gave birth vaginally
Setting: hospital in 13 studies (Brant 1967; Devall 2024; Duthie 1991; Hazarika 2022; Larsson 2006; Lertbunnaphong 2016; Pacagnella 2022; Prasertcharoensuk 2000; Razvi 1996; Rubenstein 2020; Gallos 2024; Wilcox 2017; Yunas 2024) and home birth in one study (Anger 2019)
Index tests: visual estimation was evaluated in 11 studies (Brant 1967; Devall 2024; Duthie 1991; Hazarika 2022; Larsson 2006; Lertbunnaphong 2016; Prasertcharoensuk 2000; Razvi 1996; Rubenstein 2020; Gallos 2024; Yunas 2024); the approach with a calibrated drape use was evaluated in two studies (Devall 2024; Yunas 2024); the SAPHE Mat was evaluated in one study (Wilcox 2017); heart rate and shock index were evaluated in one study (Pacagnella 2022), and a haemoglobin drop of 2 g/dL or more, a haemoglobin level of less than 10 g/dL, and a haemoglobin level of less than 7 g/dL were evaluated in one study (Anger 2019)
Reference standards: gravimetric blood loss measurement was used in six studies (Anger 2019; Devall 2024; Hazarika 2022; Gallos 2024; Wilcox 2017; Yunas 2024); volumetric blood loss measurement was used in two studies (Lertbunnaphong 2016; Prasertcharoensuk 2000); volumetric plus gravimetric blood loss measurement was used in two studies (Rubenstein 2020; Pacagnella 2022); the alkaline‐haematin method of blood loss measurement was used in three studies (Duthie 1991; Larsson 2006; Razvi 1996); and blood loss measured using the oxyhaemoglobin method was used in one study (Brant 1967).
Target condition: the target condition was postpartum haemorrhage (500 mL) in 14 studies
Included studies: 14 studies; 252,475 total participants
Index test and reference standardNumber of studies (participants)Sensitivity; % (95% CI)Specificity; % (95% CI)Certainty of evidence (GRADE)
Visual estimation with gravimetric blood loss measurement as the reference standard*Total
4 (196,305)
Sensitivity
4 (30,377)
Specificity
4 (165,928)
48 (44 to 53)97 (95 to 99)Sensitivity
⨁⨁⨁◯
Moderate
Downgraded due to inconsistency: wide range of estimates and the CIs showed little overlap
Specificity
⨁⨁⨁⨁
High
Visual estimation with volumetric blood loss measurement as the reference standardTotal
2 (514)
Sensitivity
2 (89)
Specificity
2 (425)
22 (12 to 37)99 (97 to 100)Sensitivity
⨁⨁⨁◯
Moderate
Downgraded due to risk of bias – included studies at high risk of bias
Specificity
⨁⨁⨁◯
Moderate
Downgraded due to risk of bias – included studies at high risk of bias
Visual estimation with volumetric plus gravimetric blood loss measurement as the reference standardTotal
1 (274)
Sensitivity
1 (73)
Specificity
1 (201)
14 (7 to 24)98 (95 to 99)Sensitivity
⨁⨁⨁◯
Moderate
Downgraded due to risk of bias – included study at high risk of bias
Specificity
⨁⨁⨁◯
Moderate
Downgraded due to risk of bias – included study at high risk of bias
Visual estimation with the alkaline‐haematin method of blood loss measurement as the reference standard3 (199)Range 0% to 100%Range 97% to 100%Sensitivity
⨁◯◯◯
Very low
Downgraded due to risk of bias, inconsistency (wide range of estimates) and imprecision (wide CI for summary estimate)
Specificity
⨁⨁⨁◯
Moderate
Downgraded due to risk of bias – included studies at high risk of bias
Visual estimation with the oxyhaemoglobin method of blood loss measurement as the reference standard1 (57)28 (14 to 47)100 (86 to 100)Sensitivity
⨁⨁◯◯
Low
Downgraded due to imprecision (very serious) – fewer than 100 participants
Specificity
⨁⨁◯◯
Low
Downgraded due to imprecision (very serious) – fewer than 100 participants
Approach with calibrated drape plus observations with gravimetric blood loss measurement as the reference standardTotal
2 (53,762)
Sensitivity
2 (4267)
Specificity
2 (49,495)
93 (92 to 94)95 (95 to 96)Sensitivity
⨁⨁⨁⨁
High
Downgraded due to inconsistency – CIs did not overlap. Upgraded due to large effect
Specificity
⨁⨁⨁⨁
High
The SAPHE Mat with gravimetric blood loss measurement as the reference standard1 (36)100 (40 to 100)91 (75 to 98)Sensitivity
⨁◯◯◯
Very low
Downgraded due to risk of bias and imprecision (very serious – fewer than 100 participants and wide CIs for the summary estimates)
Specificity
⨁◯◯◯
Very low
Downgraded due to risk of bias and imprecision (very serious – fewer than 100 participants)
Heart rate 91.2 bpm (21 to 40 min after birth) with volumetric plus gravimetric blood loss measurement as the reference standard1 (270)59 (53 to 65)73 (68 to 78)Sensitivity
⨁⨁⨁◯
Moderate
Downgraded due to risk of bias – included study at high risk of bias
Specificity
⨁⨁⨁◯
Moderate
Downgraded due to risk of bias – included study at high risk of bias
Shock index 0.805 (21 to 40 min after birth) with volumetric plus gravimetric blood loss measurement as the reference standard1 (270)43 (37 to 49)76 (71 to 81)Sensitivity
⨁⨁⨁◯
Moderate
Downgraded due to risk of bias – included study at high risk of bias
Specificity
⨁⨁⨁◯
Moderate
Downgraded due to risk of bias – included study at high risk of bias
A drop in Hb of 2 g/dL or more with gravimetric blood loss measurement as the reference standard1 (1058)36 (30 to 43)85 (83 to 88)Sensitivity
⨁⨁⨁⨁
High
Specificity
⨁⨁⨁⨁
High
Hb level of less than 10 g/dL with gravimetric blood loss measurement as the reference standard1 (1058)37 (30 to 44)79 (76 to 82)Sensitivity
⨁⨁⨁⨁
High
Specificity
⨁⨁⨁⨁
High
Hb level of less than 7 g/dL with gravimetric blood loss measurement as the reference standard1 (1058)2 (1 to 6)100 (99 to 100)Sensitivity
⨁⨁⨁⨁
High
Specificity
⨁⨁⨁⨁
High
*Interpretation of the result for visual estimation with gravimetric blood loss measurement as the reference standard (an illustrative example): the interpretation of this result is that visual estimation will diagnose 48 out of 100 women who have PPH, but 52 women with PPH will be missed (i.e. will be false negatives). These missed cases may not receive treatment and so may suffer avoidable morbidity and mortality. Of every 100 women without PPH, three will be wrongly diagnosed as having it (i.e. will be false positives). These wrongly diagnosed women may receive treatment which they don't require and may suffer harm as a result.

Open in a new tab

bpm: beats per minute; CI: confidence interval; FN: false negatives; FP: false positives; Hb: haemoglobin; min: minutes; PPH: postpartum haemorrhage; TN: true negatives; TP: true positives

GRADE certainty of evidence grades:

  • High certainty: the true effect is close to the estimate

  • Moderate certainty: the true effect is likely to be close to the estimate, but it could be different

  • Low certainty: the true effect is likely to be different from the estimate

  • Very low certainty: the true effect is very likely to be different from the estimate

We did not downgrade for publication bias as our literature search was broad and comprehensive. We also contacted researchers and experts in the field to identify unpublished studies.

Summary of findings 2. Summary of findings 2. Diagnostic accuracy of tests for diagnosing severe postpartum haemorrhage in women giving birth vaginally.

Participants: women who gave birth vaginally
Setting: hospital in five studies (Drew 2021; Madar 2024; Niepraschk‐von Dollen 2016; Pacagnella 2022; Rubenstein 2020), primary healthcare unit in one study (Ushida 2020), and home birth in one study (Anger 2019)
Index tests: visual estimation was evaluated in one study (Rubenstein 2020); heart rate was evaluated in one study (Pacagnella 2022); shock index was evaluated in four studies (Drew 2021; Madar 2024; Pacagnella 2022; Ushida 2020); a reduction in total haemoglobin (SpHb) of 7 g/L was evaluated in one study (Drew 2021); a haemoglobin drop of 2 g/dL or more, a haemoglobin level of less than 10 g/dL, and a haemoglobin level of less than 7 g/dL were evaluated in one study (Anger 2019); and predelivery fibrinogen levels of 4.080 g/L and 4.46 g/L were evaluated in one study (Niepraschk‐von Dollen 2016).
Reference standards: gravimetric blood loss measurement was used in three studies (Anger 2019; Drew 2021; Ushida 2020); volumetric blood loss measurement was used in two studies (Madar 2024; Niepraschk‐von Dollen 2016); and volumetric plus gravimetric blood loss measurement was used in two studies (Pacagnella 2022; Rubenstein 2020).
Target condition: the target condition was severe PPH (1000 mL) in 7 studies
Included studies: 7 studies; 37,068 total participants
Index test and reference standardNumber of studies (participants)Sensitivity; % (95% CI)Specificity; % (95% CI)Certainty of evidence (GRADE)
Visual estimation with volumetric plus gravimetric blood loss measurement as the reference standard*1 (274)9 (0 to 41)100 (99 to 100)Sensitivity
⨁⨁◯◯
Low
Downgraded due to risk of bias and imprecision – wide CI of estimate
Specificity
⨁⨁⨁◯
Moderate
Downgraded due to risk of bias
Heart rate 105.2 bpm (21 to 40 min after birth) with volumetric plus gravimetric blood loss measurement as the reference standard1 (270)55 (49 to 61)90 (87 to 94)Sensitivity
⨁⨁◯◯
Low
Downgraded due to risk of bias and imprecision – wide CI of estimate
Specificity
⨁⨁⨁◯
Moderate
Downgraded due to risk of bias – included study at high risk of bias
Shock index ≥ 1.0 (21 to 40 min after birth) with volumetric plus gravimetric blood loss measurement as the reference standard1 (270)32 (14 to 55)97 (94 to 99)Sensitivity
⨁⨁◯◯
Low
Downgraded due to risk of bias and imprecision – wide CI of estimate
Specificity
⨁⨁⨁◯
Moderate
Downgraded due to risk of bias – included study at high risk of bias
Shock index ≥ 1.0 (maximum value up to 2 hours after birth) with gravimetric blood loss measurement as the reference standard1 (30,820)30 (27 to 33)93 (92 to 93)Sensitivity
⨁⨁⨁◯
Moderate
Downgraded due to risk of bias
Specificity
⨁⨁⨁◯
Moderate
Downgraded due to risk of bias
A reduction in SpHb of 7 g/L with gravimetric blood loss measurement as the reference standard1 (66)83 (52 to 98)62 (49 to 76)Sensitivity
⨁◯◯◯
Very low
Downgraded due to risk of bias and imprecision (very serious – fewer than 100 participants and wide CIs for the summary estimates)
Specificity
⨁◯◯◯
Very low
Downgraded due to risk of bias and imprecision (very serious – fewer than 100 participants)
A drop in Hb of 2 g/dL or more with gravimetric blood loss measurement as the reference standard1 (1058)68 (48 to 84)83 (80 to 85)Sensitivity
⨁⨁⨁◯
Moderate
Downgraded due to imprecision – wide CI of estimate
Specificity
⨁⨁⨁⨁
High
Hb level of less than 10 g/dL with gravimetric blood loss measurement as the reference standard1 (1058)71 (51 to 87)77 (75 to 80)Sensitivity
⨁⨁⨁◯
Moderate
Downgraded due to imprecision – wide CI of estimate
Specificity
⨁⨁⨁⨁
High
Hb level of less than 7 g/dL with gravimetric blood loss measurement as the reference standard1 (1058)11 (2 to 28)99 (99 to 100)Sensitivity
⨁⨁⨁◯
Moderate
Downgraded due to imprecision – wide CI of estimate
Specificity
⨁⨁⨁⨁
High
Predelivery fibrinogen level of 4.080 g/L with volumetric blood loss measurement as the reference standard1 (689)54 (33 to 74)78 (75 to 81)Sensitivity
⨁◯◯◯
Very low
Downgraded due to risk of bias (very serious) and imprecision – wide CI for the summary estimate
Specificity
⨁⨁◯◯
Low
Downgraded due to risk of bias (very serious)
Predelivery fibrinogen level of 4.46 g/L with volumetric blood loss measurement as the reference standard1 (689)71 (49 to 87)59 (55 to 62)⨁◯◯◯
Very low
Downgraded due to risk of bias (very serious) and imprecision – wide CI for the summary estimate
Specificity
⨁⨁◯◯
Low
Downgraded due to risk of bias (very serious)
*Interpretation of the result for visual estimation with volumetric plus gravimetric blood loss measurement as the reference standard (an illustrative example): the interpretation of this result is that visual estimation will diagnose nine out of 100 women who have PPH, but 91 women with PPH will be missed (i.e. will be false negatives). These missed cases may not receive treatment and so may suffer avoidable morbidity and mortality. Of every 100 women without PPH, none will be wrongly diagnosed as having it (i.e. will be false positives).

Open in a new tab

bpm: beats per minute; CI: confidence interval; FN: false negatives; FP: false positives; Hb: haemoglobin; min: minutes; PPH: postpartum haemorrhage; SpHb: total haemoglobin; TN: true negatives; TP: true positives

GRADE certainty of evidence grades:

  • High certainty: the true effect is close to the estimate

  • Moderate certainty: the true effect is likely to be close to the estimate, but it could be different

  • Low certainty: the true effect is likely to be different from the estimate

  • Very low certainty: the true effect is very likely to be different from the estimate

We did not downgrade for publication bias as our literature search was broad and comprehensive. We also contacted researchers and experts in the field to identify unpublished studies.

Background

Target condition being diagnosed

Postpartum haemorrhage (PPH) is a serious obstetric emergency. It is a treatable condition if diagnosed promptly and accurately. Despite this, it accounts for more maternal deaths worldwide than any other cause [1].

The accepted definition of primary PPH at vaginal birth is blood loss of 500 mL or more in the first 24 hours. Severe postpartum haemorrhage is defined as blood loss of 1000 mL or more in the first 24 hours.

Accurate diagnosis of PPH allows for available treatments to be administered. A recent large, randomised control trial highlighted the importance of early and accurate detection in reducing the adverse effects of PPH [2].

Index test(s)

The methods and tools used to diagnose PPH vary greatly. These include: visual estimation of blood loss; objectively measuring blood loss using, for example, calibrated collection devices; assessment of clinical variables such as heart rate and blood pressure; and blood tests. Certain methods of diagnosing PPH, such as visual estimation of blood loss, have been shown to be inaccurate [3]. Diagnostic methods which use an objective rather than subjective measurement of blood loss are likely to be more reliable and accurate [3].

Clinical pathway

Women who give birth vaginally may experience primary PPH at any point in the first 24 hours after delivery. The commonest method used to estimate blood loss of 500 mL or more is visual estimation. Direct methods of blood loss estimation include gravimetric and volumetric approaches. These methods are usually employed by the birth attendant, who may be a midwife, doctor, or student. The timing of the methods can vary. Indirect methods include the use of clinical variables such as heart rate and blood pressure. In some situations, measurements taken before birth – for example, of haemoglobin levels – are compared to levels postpartum to diagnose PPH. Diagnosis of PPH allows for prompt treatment initiation which can reduce the risk of adverse outcomes. This review identified and assessed the accuracy of various index tests used to diagnose PPH and severe PPH.

Rationale

Early and accurate diagnosis of PPH will allow prompt treatment and management. This is particularly important for low‐ and middle‐income countries (LMIC) where early diagnosis and management would give the best chance of maternal survival in the context of limited resources [2].

Identifying the range of available tools and methods to diagnose and detect PPH and determining their accuracy will allow healthcare providers to prioritise the use of effective ones. There is currently no accepted gold standard method or tool for diagnosing PPH.

The aim of this systematic review was to identify tools and methods to diagnose PPH and severe PPH, and to assess their accuracy.

Objectives

What is the accuracy of methods (index tests) for diagnosing primary PPH (blood loss ≥ 500 mL in the first 24 hours after birth) and severe primary PPH (blood loss ≥ 1000 mL in the first 24 hours after birth) (target conditions) in women giving birth vaginally (participants) compared to weighed blood loss measurement or other objective measurements of blood loss (reference standards)?

Methods

Criteria for considering studies for this review

Types of studies

We included studies that reported 2 x 2 data (number of true positive, false positive, false negative, and true negative results) or where the 2 x 2 data could be derived from the available study data on any of the index tests, singularly or in combination, and with a reference standard of weighed blood loss or another objective assessment of blood loss in women giving birth vaginally. This allowed us to create 2 x 2 tables to calculate sensitivity and specificity. Study types included diagnostic cohort studies and cross‐sectional studies.

We excluded case reports, case series, and diagnostic case‐control (two‐gate) studies.

Participants

We included women who gave birth vaginally in any setting. If studies provided data on a mixed population (i.e. women giving birth vaginally and by caesarean section), we included the vaginal delivery data if these data were presented separately. There was no age restriction for participants.

Index tests

We considered including studies which evaluated at least one of the index tests. Index tests used to diagnose PPH include the following.

  • Visual estimation of blood loss – this involves the attendant estimating the blood loss by looking at the extent to which drapes and sheets are soaked with blood, along with the visual estimation of any blood collected in non‐calibrated collection devices such as drapes, bowls, cylinders, and jugs.

  • Calibrated collection devices (volumetric) – examples include a plastic drape with a blood collection pouch that is placed under the woman giving birth. Blood is collected in the drape and graduations in the collection pouch indicate the volume of blood gathered. Calibrated trays, bowls, cylinders, and jugs can also be used.

  • Diagnostic approach using calibrated drape plus observations – cumulative blood loss volume reading taken using the calibration lines on the drape, with the assessment of uterine tone every 15 minutes. Blood pressure and pulse measurements at least once in the first hour and repeated if abnormal or if there is ongoing bleeding. Abnormal observations include: tachycardia (> 100 beats per minute (bpm) or an increase of 20 bpm from baseline), decreasing systolic blood pressure (< 100 mmHg or a decrease of 20 mmHg from baseline), soft uterine tone, heavy vaginal blood flow, large clots being expelled, or a constant trickle of blood. PPH diagnosed and treated if any of the following three criteria are met: (1) Clinical judgement – heavy vaginal blood loss, large blood clots, a constant trickle, or some other concern in the healthcare provider’s assessment; (2) Blood loss of 300 mL to less than 500 mL observed in the drape, plus one abnormal clinical observation or vital sign (heart rate, blood pressure, uterine tone, vaginal flow of blood); (3) Blood loss of 500 mL or more observed in the drape (regardless of other observations or vital signs) (Devall 2024 [4]; Yunas 2024 [5]).

  • Estimation of blood loss using the SAPHE (Signalling a Postpartum Hemorrhage Emergency) Mat. Each square on the Mat absorbs up to 50 mL of blood. The total volume of blood is estimated by visual assessment of the number of squares which have absorbed blood (Wilcox 2017 [6]).

  • Blood loss field image analysis and other technologies.

  • Assessment of uterine atony – assessed by palpation of the uterus as part of an abdominal clinical examination.

  • Clinical variables such as heart rate and blood pressure – PPH may be diagnosed by increases in heart rate (tachycardia, heart rate > 100 bpm) and fall in blood pressure (systolic blood pressure < 90 mmHg).

  • Shock index – this is defined as the heart rate divided by systolic blood pressure. Normal values range from 0.52 to 0.89 in the first hour postpartum [7].

  • Early warning charts – these use combinations of clinical variables to determine likely blood loss.

  • Change in haemoglobin (Hb) levels – this relies on comparisons of haemoglobin levels before birth and at time points after birth to determine blood loss. Haemoglobin can be measured by venous blood sampling or by using devices such as the HemoCue handheld device (HemoCue, Ängelholm, Sweden). Changes in haemoglobin level can also be measured using continuous haemoglobin monitoring using a non‐invasive disposable total haemoglobin (SpHb) probe placed on the patient’s index finger and connected to a continuous non‐invasive total Hb monitor.

  • Haemoglobin levels postpartum – low levels can indicate blood loss.

  • Predelivery clotting profile (e.g. fibrinogen level).

Target conditions

The target conditions are primary PPH (blood loss of 500 mL or more in the first 24 hours) in women giving birth vaginally, and severe primary PPH (blood loss of 1000 mL or more in the first 24 hours) in women giving birth vaginally.

Reference standards

The preferred reference standard was gravimetric blood loss measurement for diagnosing PPH (blood loss ≥ 500 mL) and severe PPH (blood loss ≥ 1000 mL). This method involves the use of scales to weigh collected blood loss as well as weighing blood‐soaked gauze, pads, and sheets, and subtracting their dry weight. Other objective blood loss measurement methods were also used as reference standards. These included volumetric methods using calibrated blood collection drapes, bags, or cylinders; the alkaline‐haematin method of blood loss estimation, as described by Newton and colleagues [8]; and blood extracted using the machine‐extraction method and measured spectrometrically as oxyhaemoglobin, as described by Brant 1967 [9]. Combinations of objective measurements were also used as reference standards, such as the volumetric plus gravimetric approach.

Search methods for identification of studies

Electronic searches

The search strategies were developed in collaboration with the Cochrane Information Specialist, who also ran the searches (Supplementary material 1).

We identified relevant trials through systematic searches of these databases:

  • Cochrane Central Register of Controlled Trials (CENTRAL);

  • MEDLINE (Ovid);

  • Embase (Ovid);

  • Web of Science Core Collection (Clarivate).

We searched databases from their inception to 24 May 2024 and did not impose any restriction on language of publication or publication status.

We searched ClinicalTrials.gov and the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) for ongoing and unpublished trials.

Searching other resources

We undertook handsearching of relevant systematic reviews published since 2018.

We considered abstracts for inclusion if they contained sufficient information to enable data extraction and quality assessment.

We checked the reference lists of included studies and other relevant systematic reviews to identify further potentially eligible studies.

We contacted experts in the field and trial authors where necessary for any additional information. We also examined any relevant retraction statements and errata for included studies.

Data collection and analysis

Selection of studies

At least two review authors independently screened the titles and abstracts of studies identified by the searches to determine if they satisfied the inclusion criteria. We retrieved the full texts of potentially eligible studies for detailed screening, and selected the eligible studies. We resolved any disagreements through consensus or by consulting another review author if required.

We include a PRISMA flowchart to detail the record screening process and selection of included studies (Figure 1).

1.

Tests for diagnosis of postpartum haemorrhage at vaginal birth (1)

Open in a new tab

Data extraction and management

A data extraction form was developed, piloted, and used by at least two review authors, working independently, to extract data from the included studies. We resolved any disagreements through discussion and by consulting another review author if required. We entered the study data into Review Manager software [10].

We extracted the following information.

  • Study characteristics

    • Study author and date

    • Study design

    • Study country and continent

    • Study income setting

    • Study healthcare setting

    • Funding sources and reported conflicts of interest

  • Participants

    • Total number of participants

    • Inclusion criteria

    • Exclusion criteria

    • Demographic details: maternal age, parity, body mass index (BMI), gestational age, history of PPH, admission haemoglobin

  • Index test

    • Type of index test

    • Positivity threshold for index test

    • Timing of index test

    • Who performed the index test

  • Reference standard

    • Type of reference standard

    • Positivity threshold for reference standard

    • Timing of reference standard

    • Who performed the reference standard

  • Accuracy data

    • True positive (TP), false positive (FP), false negative (FN), and true negative (TN) results for each index test, threshold and reference standard combination

Assessment of methodological quality

At least two review authors, working independently, assessed the methodological quality of the included studies using the Quality Assessment of Diagnostic Accuracy Studies version 2 (QUADAS‐2) tool to assess risk of bias and applicability [11]. We also involved a third review author, with an affiliation separate to the authors, to independently evaluate the methodological quality of the studies (Acknowledgements). The tool allows assessment across the following four domains: patient selection, index test, reference standard, and flow and timing. The risk of bias domains can be rated as low, high, or unclear. The patient selection, index test, and reference standard domains can also be rated as low, high, or unclear with respect to concerns regarding applicability.

We classified studies as overall low risk of bias if assessed as low risk in all four domains or low risk in three domains with a single domain rated as unclear. We modified the tool to our review question and adapted the signalling questions (Supplementary material 8). We resolved any discrepancies through discussion or sought input from another review author, as needed.

Statistical analysis and data synthesis

We estimated sensitivities and specificities from each included study along with 95% confidence intervals (CI). We presented sensitivity and specificity estimates from individual studies in forest plots. Where possible, we conducted meta‐analyses separately for each index test and reference standard combination and target condition. Since we used a common threshold for each analysis, we jointly synthesised sensitivity and specificity using the bivariate model to obtain summary estimates of sensitivity and specificity (summary points) with corresponding 95% CIs [12].

We performed meta‐analyses using STATA [13] and the STATA METADTA user written function [14]. METADTA defaults to univariate fixed‐effect logistic regression when there are fewer than four studies. This was not always appropriate given the observed between‐study variation in estimates of sensitivity and specificity. Therefore, when appropriate, we fitted univariate random‐effects logistic regression models using the meqrlogit command. When there were only three studies with sparse data for robust estimation of sensitivity, we did not perform meta‐analysis. Results were presented visually using Review Manager [10] and STATA [13].

When we calculated 2 x 2 tables from sensitivities, specificities, total number of participants, and disease prevalence quoted in study manuscripts, we rounded the calculations to the nearest whole number of participants. Due to the rounding, there were slight discrepancies when the forest plots were generated with the calculated 2 x 2 data. In the narrative of the results and summary of findings tables, we have maintained the quoted values according to the study manuscripts, but used the calculated values for the forest plots.

Investigations of heterogeneity

Due to limited data, we only assessed heterogeneity through visual inspection of the forest plots.

Sensitivity analyses

In our protocol, we stated that we would consider undertaking sensitivity analyses by restricting the primary analysis to studies at low risk of bias. However, we were unable to undertake sensitivity analyses due to the limited number of studies.

Assessment of reporting bias

We did not explore reporting bias as there is no consensus on the best approach for this in reviews of diagnostic test accuracy.

Assessment of the certainty of the evidence and summary of findings

We generated summary of findings (SoF) tables for PPH and severe PPH, and assessed the certainty of the evidence using the GRADE (Grading of Recommendations, Assessment, Development and Evaluations) approach [15]. At least two review authors rated the certainty of the evidence. We resolved disagreements through consensus or discussion with another author. The GRADE assessment rates the certainty of the evidence for the following five domains: risk of bias, indirectness, inconsistency, imprecision, and publication bias. We rated concerns about each of the five domains as 'not serious', 'serious', or 'very serious'. The imprecision domain could also be given an 'extremely serious' concern rating.

Each outcome received an overall certainty of evidence rating as follows.

  • High certainty: the true effect is close to the estimate.

  • Moderate certainty: the true effect is likely to be close to the estimate, but it could be different.

  • Low certainty: the true effect is likely to be different from the estimate.

  • Very low certainty: the true effect is very likely to be different from the estimate.

Results

Results of the search

Search results

Figure 1 shows the results of our searches. We identified 8538 records through our database searches. A further three sets of data were made available to us by the authors of unpublished studies and another study was highlighted to us by the Cochrane Peer Review Team. We screened 8460 records after the removal of duplicates. We excluded 8262 of these records as they did not meet the inclusion criteria. Amongst these, we identified four studies as ongoing (Supplementary material 5) and seven as 'awaiting classification', as we were unable to find the full texts (Supplementary material 4). We reviewed the full texts of 198 articles and excluded 180 with reasons provided (Supplementary material 3). We selected 18 studies to include in our review (Table 3, Supplementary material 2, and Figure 1).

1. Characteristics of included studies.
Study and settingPopulationIndex test(s) and thresholdsReference standard(s) and thresholds
Anger 2019
Retrospective secondary analysis of RCT
Pakistan
LMIC
Home births
(n = 1058)
Age < 25 years 263 (24.9%); 25 to 34 years 697 (65.9%); > 35 years 98 (9.3%); Nulliparous 211 (19.9%)Hb drop of 2 g/dL or more (pre‐ to postpartum)
Hb less than 10 g/dL (postpartum)
Hb less than 7 g/dL (postpartum)
Hb was measured using the HemoCue handheld device
Gravimetric blood loss measurement (500 mL)
Gravimetric blood loss measurement (1000 mL)
(Blood was collected in a jug and weighed; the dry weight was subtracted)
Brant 1967
Prospective study
UK
HIC
Hospital
(n = 57)
Visual estimation (500 mL)Blood loss measured spectrometrically as oxyhaemoglobin (500 mL)
(Blood was extracted in a washing machine and measured spectrometrically as oxyhaemoglobin.)
Devall 2024 (unpublished)
Retrospective secondary analysis of cluster‐RCT
Kenya, Nigeria, Tanzania, and South Africa
LMIC
Hospital
(n = 157,773 for visual estimation group; n = 48,678 for calibrated drape group)
Median age (IQR) calibrated drape group 26 years (21 to 31);
Median age (IQR) visual estimation group 26 years (21 to 30);
Median gestational age (IQR) calibrated drape group 39 weeks (37 to 40);
Median gestational age (IQR) visual estimation group 38 weeks (37 to 39);
PPH in previous pregnancy calibrated drape group 1440 (3.0%);
PPH in previous pregnancy visual estimation group 4634 (2.9%)
Visual estimation (500 mL)
Diagnostic approach using calibrated drape plus observations recorded on a chart. PPH was diagnosed if: 1. there was concern based on clinical judgement, or 2. ≥ 300 mL to < 500 mL of blood was collected in the drape plus one abnormal clinical observation or vital sign (heart rate, blood pressure, uterine tone, vaginal flow of blood), or 3. ≥ 500 mL of blood was collected in the drape (regardless of other observations or vital signs)
Gravimetric blood loss measurement (500 mL)
(The drape with the collected blood was weighed on a digital scale.)
Drew 2021
Prospective study
Canada
HIC
Hospital
(n = 66)
Mean age (SD) 32.1 years (4); mean BMI (SD) 28.2 (4.1); mean baseline SBP (SD) 117 mmHg (10); mean baseline HR (SD) 86 bpm (13); mean baseline shock index (SD) 0.74 (0.12); mean baseline SpHb (SD) 115 g/dL (10)
Included term labouring women (gestational age > 37 weeks) undergoing spontaneous or instrumental vaginal delivery
Excluded women who refused consent, those with cardiac rhythm abnormalities, cardiac disease, jaundice, peripheral vascular disease, hypertension, pre‐eclampsia, abnormal haemoglobin, on anti‐hypertensive drugs and drugs affecting heart rate
Shock index (≥ 0.9) (maximum shock index value within first hour of birth)
Reduction in SpHb of 7 g/L
(Continuous Hb monitoring (SpHb) was performed using a non‐invasive disposable SpHb probe placed on the patient’s index finger and connected to a continuous non‐invasive total haemoglobin monitor.)
Gravimetric blood loss measurement (1000 mL)
(Blood was collected in a drape and weighed with soiled pads, sponges, and linen. Known dry weights of items were subtracted from the weights of soiled items. Care was taken to avoid amniotic fluid contaminating the collecting system.)
Duthie 1991
Prospective study
Hong Kong
HIC
Hospital
(n = 62)
Mean age (SD) of primigravida 24.7 years (0.7); multipara 28.4 years (0.6); mean gestation (SD) primigravida 39.3 weeks (0.2); multigravida 39.2 weeks (0.2)
Included uncomplicated singleton pregnancies with spontaneous labour
Excluded forceps or ventouse deliveries
Visual estimation (500 mL)Alkaline‐haematin method of blood loss measurement (500 mL)
Gallos 2024 (unpublished)
Retrospective secondary analysis of randomised, double‐blind, non‐inferiority trial
Argentina/Egypt/India/Kenya/Nigeria/Singapore/South Africa/Thailand/Uganda/UK
Hospital
(n = 29,421)
Median age (IQR 25 years (22 to 30);
Median gestational age (IQR) 39 weeks (28 to 40);
Nulliparous 12,887 (43.8%)
Included women who expected to give birth vaginally and who had a singleton pregnancy and cervical dilatation of 6 cm or less
Excluded women who were in an advanced stage of labour (cervical dilatation > 6 cm), were too distressed to provide informed consent, had known allergies to carbetocin, oxytocin homologues, or excipients, or had a serious cardiovascular disorder, serious hepatic or renal disease, or epilepsy.
Visual estimation (500 mL)Gravimetric blood loss measurement (500 mL)
(After childbirth, a blood collection drape was positioned beneath the woman, and the volume of blood collected was weighed after one hour, or after two hours if the bleeding persisted beyond the initial hour.)
Hazarika 2022
Prospective study
India
LMIC
Hospital
(n = 100)
Mean age (SD) 24.91 years (3.74); 64% of patients were primigravida and 36% multigravida.
Low‐risk case mix
Included primigravidae, second gravidae, singleton pregnancy, no contraindications for vaginal delivery, and participants free of medical/obstetrical complications
Excluded women in active labour (cervical dilatation > 4 cm) who had difficulty in giving consent, women with history of antepartum haemorrhage, women delivering by caesarean section, women with anaemia, pregnancy‐induced hypertension, chronic hypertension, gestational diabetes mellitus, overt diabetes and bleeding disorders
Visual estimation (500 mL)Gravimetric blood loss measurement (500 mL)
(Blood was collected in drape and weighed along with blood‐soaked pads and gauze pieces. The dry weight of the pads and gauze pieces was subtracted.)
Larsson 2006
Prospective study
Sweden
HIC
Hospital
(n = 26)
Included primiparas and multiparasVisual estimation (500 mL)Alkaline‐haematin method of blood loss measurement (500 mL)
Lertbunnaphong 2016
Prospective study
Thailand
LMIC
Hospital
(n = 286)
Mean age (SD) 27.1 years (5.9) and the majority were nulliparous (46.2%)
Low‐risk case mix
Included pregnant Thai women aged ≥ 18 years who were admitted in the early phase of labour
Excluded women who had painful contractions, were near delivery, were under the sedative effect of morphine or its derivatives, had bleeding disorders, dependence on bleeding‐related medication, a caesarean delivery, foetal anomalies or a stillbirth.
Visual estimation (500 mL)Volumetric blood loss measurement (500 mL)
(Blood was collected in the plastic pouch of the drape and poured into a standard cylinder and measured.)
Madar 2024
Retrospective secondary analysis of RCT
France
HIC
Hospital
(n = 3486 for shock index at 15 minutes; n = 3517 for shock index at 30 minutes)
Shock index at 15‐minute group: age ≥ 35 years 643 (18.5%); BMI < 18.5 243 (7.0%); BMI 18.5 to 24.9 2270 (65.6%); BMI 25 to 29.9 642 (18.6%); BMI ≥ 30 305 (8.8%); Nulliparous 1836 (52.7%)
Shock index at 30‐minute group: age ≥ 35 years 648 (18.4%); BMI < 18.5 245 (7.0%); BMI 18.5 to 24.9 2291 (65.7%); BMI 25 to 29.9 646 (18.5%); BMI ≥ 30 306 (8.8%); Nulliparous 1871 (53.2%)
Included women aged ≥ 18 years with vaginal delivery of a singleton live foetus at 35 or more weeks
Excluded women with placenta praevia, coagulation disorders, a previous episode of venous or arterial thrombosis, history of epilepsy or seizure, in utero foetal death, multiple gestation or poor comprehension of oral French
Shock index ≥ 0.7 (15 min after birth);
Shock index ≥ 0.9 (15 min after birth);
Shock index ≥ 1.1(15 min after birth);
Shock index ≥ 0.7 (30 min after birth);
Shock index ≥ 0.9 (30 min after birth);
Shock index ≥ 1.1 (30 min after birth)
Volumetric blood loss measurement (collector bag) (1000 mL)
(Blood loss was quantified using a graduated collector bag placed just after delivery.)
Niepraschk‐von Dollen 2016
Prospective study
Germany
HIC
Hospital
(n = 689)
Median age 29 years; average parity was 2 (range 1 to 8), and average gestational age at delivery was 40 weeks (range 37 to 42 weeks)
Included live singleton pregnancy after 37 gestational weeks, age of at least 18 years or parental consent to participate, and comprehension of the German language
Excluded women who underwent caesarean delivery during labour, if blood loss during delivery was not measured, or if blood samples were not taken or were taken more than 72 hours before delivery
Pre‐delivery fibrinogen level 4.080 g/L
Predelivery fibrinogen level 4.46 g/L
(Fibrinogen levels were measured using venous blood samples.)
Volumetric blood loss measurement (calibrated drape) (1000 mL)
(Blood was measured using a calibrated transparent collecting drape. (Brenner Medical Munich, Germany))
Pacagnella 2022
Prospective study
Brazil
LMIC
Hospital
(n = 270)
Mean age (SD) 24.67 years (6.19); mean BMI (SD) 28.85 kg/m2 (4.61); mean gestation 38.93 (SD) weeks (1.47); 45.9% of patients were primiparous and 54.1% multiparous; 67.7% of patients were white and 32.3% non‐white.
Included women giving birth vaginally
Excluded women who underwent caesarean delivery, were below 34 weeks of gestational age, had hypertension, hyper‐ or hypothyroidism without treatment, any cardiac disease, infections with fever or sepsis, a history of coagulopathy, or if childbirth occurred after 7 pm.
Heart rate of 91.2 bpm (21 to 40 minutes after birth) for blood loss ≥ 500 mL in 2 hours
Heart rate of 105.2 bpm (at 21 to 40 minutes) for blood loss ≥ 1000 mL in 2 hours
Shock index of 0.805 (21 to 40 minutes after birth) for blood loss ≥ 500 mL in 2 hours
Shock index ≥ 0.8 (21 to 40 minutes after birth) for blood loss ≥ 1000 mL in 2 hours
Shock index ≥ 0.9 (21 to 40 minutes after birth) for blood loss ≥ 1000 mL in 2 hours
Shock index ≥ 1.0 (21 to 40 minutes after birth) for blood loss ≥ 1000 mL in 2 hours
(The study reported results for ranges of heart rate and shock index at different time points and for blood loss ≥ 500 mL and ≥ 1000 mL at both 2 hours and 24 hours after birth.
Based on the authors' reports that most blood loss occurred in the first 40 minutes after birth, we used the results for index test results for the 21‐ to 40‐minute category with blood loss measured in the first 2 hours after birth.
Volumetric plus gravimetric blood loss measurement (500 mL)
Volumetric plus gravimetric blood loss measurement (1000 mL)
(Total blood loss was measured using the volume of blood collected in a calibrated drape (Maternova, Providence, RI, USA), along with the weight of blood‐soaked compresses, with the dry weight subtracted which was then converted to a volume.)
Prasertcharoensuk 2000
Prospective study
Thailand
LMIC
Hospital
(n = 228)
Included vaginal deliveryVisual estimation (500 mL)Volumetric blood loss measurement (500 mL)
Razvi 1996
Prospective study
Singapore
HIC
Hospital
(n = 111)
Included normal singleton pregnancies at term in spontaneous labour, who did not require augmentation with syntocinon in labour and who progressed to normal vaginal delivery.Visual estimation (500 mL)Alkaline‐haematin method of blood loss measurement (500 mL)
Rubenstein 2020
Prospective study
USA
HIC
Hospital
(n = 274)
Mean age (SD) 30.4 years (5.2); mean weight (SD) 172 lbs [pounds] (30.8); mean BMI (SD) 24.2 (4.5); mean admission haemoglobin (SD) 12.1 g/dL (1.0)
A total of 30 patients (10.9%) were categorised as high risk, 86 (31.4%) as medium risk, and 155 (56.6%) as low risk.
Included vaginal delivery
Visual estimation (500 mL)
Visual estimation (1000 mL)
Volumetric plus gravimetric blood loss measurement (500 mL)
Volumetric plus gravimetric blood loss measurement (1000 mL)
(The Triton QBL (Gauss Surgical, Inc., Menlo Park, CA) device was used to quantify blood loss by batch weighing all blood‐containing sponges, towels, pads and other supplies and automatically subtracting their dry weights. It also accounts for collected fluids and subtracts the measured amount of amniotic fluid using a calibrated V‐Drape.)
Ushida 2020
Retrospective cohort study
Japan
HIC
Primary healthcare unit
(n = 30,820)
Median age (IQR) 31 years (28 to 34); Median gestational age at delivery (IQR) 39.3 weeks (38.4 to 40.0); Primipara (%) 55.2%; median pre‐pregnancy BMI (IQR) 20.1 (18.8 to 21.9); median BMI at delivery (IQR) 24.5 (22.9 to 26.5)
Low‐risk case mix
Included women with complete data on vital signs and blood loss from admission to 2 hours postpartum
Excluded preterm delivery (< 37 gestational weeks), post‐term delivery (≥ 42 gestational weeks), caesarean section, multiple pregnancy, stillbirth, major congenital or chromosomal abnormalities, transfer to a higher‐level facility, and missing data on vital signs and blood loss
Shock index (≥ 0.7);
Shock index (≥ 0.8);
Shock index (≥ 0.9);
Shock index (≥ 1.0);
Shock index (≥ 1.1);
Shock index (≥ 1.2);
Shock index (≥ 1.3);
Shock index (≥ 1.4);
Shock index (≥ 1.5)
(maximum shock index value taken up to 2 hours after birth)
Gravimetric blood loss measurement (1000 mL)
(Blood loss did not include amniotic fluid; dry weights of gauze and pads were subtracted.)
Wilcox 2017
Prospective study
USA
HIC
Hospital
(n = 36)
Excluded women under the age of 18 years, who did not understand either English or Spanish, or if they were undergoing a caesarean section.Blood loss volume visually estimated using the SAPHE (Signalling a Postpartum Hemorrhage Emergency) Mat (500 mL)Gravimetric blood loss measurement (500 mL)
The volume of blood loss was estimated by measuring the difference in weight of the SAPHE Mat before and after birth and converting it to a volume by dividing this by the density of blood (1.06 g/mL).
Yunas 2024 (unpublished)
Retrospective secondary analysis of before‐and‐after comparative study
Pakistan
LMIC
Hospital
(n = 9011 for visual estimation group; n = 5084 for calibrated drape group)
Median age (IQR) calibrated drape group 30 years (26 to 32);
Median age (IQR) visual estimation group 30 years (25 to 30);
Median gestational age (IQR) calibrated drape group 37 weeks (36 to 38);
Median gestational age (IQR) visual estimation group 37 weeks (36 to 38);
PPH in previous pregnancy calibrated drape group 79 (1.6%);
PPH in previous pregnancy visual estimation group 191 (2.1%)
Visual estimation (500 mL)
Diagnostic approach using calibrated drape plus observations recorded on a chart. PPH was diagnosed if: 1. there was concern based on clinical judgement, or 2. ≥ 300 mL to < 500 mL of blood was collected in the drape plus one abnormal clinical observation or vital sign (heart rate, blood pressure, uterine tone, vaginal flow of blood), or 3. ≥ 500 mL of blood was collected in the drape (regardless of other observations or vital signs).
Gravimetric blood loss measurement (500 mL)
(The drape with the collected blood was weighed on a digital scale.)

Open in a new tab

BMI: body mass index; bpm: beats per minute; Hb: haemoglobin; HIC: high‐income country; HR: heart rate; IQR: interquartile range; LMIC: low‐ or middle‐income country; PPH: postpartum haemorrhage; RCT: randomised control trial; SBP: systolic blood pressure; SD: standard deviation; SpHb: total haemoglobin

Characteristics of the included studies

Table 3 shows the characteristics of the included studies. We included 18 studies with a total of 291,040 participants for our analysis. Most studies were published as full texts (Anger 2019 [16]; Brant 1967; Drew 2021 [17]; Duthie 1991 [18]; Hazarika 2022 [19]; Larsson 2006 [20]; Lertbunnaphong 2016 [21]; Madar 2024 [22]; Niepraschk‐von Dollen 2016 [23]; Prasertcharoensuk 2000 [24]; Razvi 1996 [25]; Rubenstein 2020 [26]; Ushida 2020 [27]; Wilcox 2017; Pacagnella 2022 [28]), and three studies were unpublished data sets provided by the author teams (Devall 2024; Gallos 2024 [29]; Yunas 2024).

Most studies were conducted in a range of single countries (16 out of 18), whereas two studies were conducted in multiple countries (Devall 2024; Gallos 2024). The single country representations included Brazil (Pacagnella 2022), Canada (Drew 2021), France (Madar 2024), Germany (Niepraschk‐von Dollen 2016), Hong Kong (Duthie 1991), India (Hazarika 2022), Japan (Ushida 2020), Pakistan (Anger 2019; Yunas 2024), Singapore (Razvi 1996), Sweden (Larsson 2006), Thailand (Lertbunnaphong 2016; Prasertcharoensuk 2000), the UK (Brant 1967), and the USA (Rubenstein 2020; Wilcox 2017).

Participants were recruited in the hospital setting in 16 studies (Brant 1967; Devall 2024; Drew 2021; Duthie 1991; Hazarika 2022; Larsson 2006; Lertbunnaphong 2016; Madar 2024; Niepraschk‐von Dollen 2016; Pacagnella 2022; Prasertcharoensuk 2000; Razvi 1996; Rubenstein 2020; Gallos 2024; Wilcox 2017; Yunas 2024). Home births were the setting in one study (Anger 2019), and a primary healthcare unit was the setting in another study (Ushida 2020).

Methodological quality of included studies

Our assessment of risk of bias and applicability concerns for the included studies are summarised in Figure 2 and Figure 3. The individual study assessments are presented in Supplementary material 9.

2.

Tests for diagnosis of postpartum haemorrhage at vaginal birth (2)

Open in a new tab

3.

Tests for diagnosis of postpartum haemorrhage at vaginal birth (3)

Open in a new tab

Patient selection

We assessed four studies (22%) as high risk of bias for the patient selection domain as they did not avoid inappropriate exclusions. Three of these imposed language restrictions on inclusion of participants (Madar 2024; Niepraschk‐von Dollen 2016; Wilcox 2017), and one excluded women who gave birth after 7 pm (Pacagnella 2022). We assessed the remaining 14 (78%) studies as low risk of bias for this domain. Applicability concerns were low for all the included studies.

Index test

We assessed 10 studies (56%) as low risk of bias for the index test domain (Anger 2019; Brant 1967; Devall 2024; Hazarika 2022; Lertbunnaphong 2016; Madar 2024; Razvi 1996; Rubenstein 2020; Gallos 2024; Yunas 2024). Due to the index threshold used not being prespecified, we assessed seven (39%) studies as high risk of bias (Drew 2021; Duthie 1991; Larsson 2006; Niepraschk‐von Dollen 2016; Pacagnella 2022; Ushida 2020; Wilcox 2017). We assessed a single study (6%) as unclear risk of bias as it was unclear if the index threshold was prespecified and if the index test results were interpreted without knowledge of the results of the reference standard (Prasertcharoensuk 2000). The applicability concerns were low for all the studies.

Reference standard

We assessed a single study (6%) as high risk of bias for the reference standard domain, as the reference standard results were interpreted with knowledge of the index test results (Lertbunnaphong 2016). We assessed the remaining 17 studies (94%) as low risk of bias. The applicability concerns for all the studies were low.

Flow and timing

Our assessment of the flow and timing domain identified three studies (17%) at high risk of bias as not all participants were included in the analysis (Madar 2024; Niepraschk‐von Dollen 2016; Rubenstein 2020). We assessed the remaining 15 studies (83%) as low risk of bias for this domain (Anger 2019; Brant 1967; Devall 2024; Drew 2021; Duthie 1991; Hazarika 2022; Larsson 2006; Lertbunnaphong 2016; Razvi 1996; Ushida 2020; Gallos 2024; Wilcox 2017; Yunas 2024; Pacagnella 2022; Prasertcharoensuk 2000).

Findings

Key results are presented in the summary of findings tables (Table 1; Table 2), with summary sensitivities and specificities for the analyses. Figure 4 and Figure 5 show the study forest plots for PPH and severe PPH, respectively.

4.

Tests for diagnosis of postpartum haemorrhage at vaginal birth (4)

Open in a new tab

5.

Tests for diagnosis of postpartum haemorrhage at vaginal birth (5)

Open in a new tab

Diagnosis of PPH

Figure 12 shows the summary receiver operating characteristics (SROC) plot with all the studies evaluating visual estimation for diagnosing PPH (500 mL).

6.

Tests for diagnosis of postpartum haemorrhage at vaginal birth (6)

Open in a new tab

Accuracy of visual estimation with gravimetric blood loss measurement as the reference standard

From the four studies that evaluated visual estimation with gravimetric blood loss measurement as the reference standard for the diagnosis of PPH (Devall 2024; Hazarika 2022; Gallos 2024; Yunas 2024), we calculated a summary sensitivity of 48% (95% CI 44% to 53%; moderate‐certainty evidence) and summary specificity of 97% (95% CI 95% to 99%; high‐certainty evidence) (Table 1).

The interpretation of this result is that visual estimation will diagnose 48 out of 100 women who have PPH, but 52 women with PPH will be missed (i.e. will be false negatives). These missed cases may not receive treatment and so may suffer avoidable morbidity and mortality. Of every 100 women without PPH, three will be wrongly diagnosed as having it (i.e. will be false positives). These wrongly diagnosed women may receive treatment which they don't require and may suffer harm as a result.

Due to the low number of studies, we did not undertake subgroup analyses to investigate heterogeneity. Visual inspection of the forest plots indicated a high level of heterogeneity (Figure 6; Figure 7), as there was a wide range of estimates and the confidence intervals showed little overlap.

7.

Tests for diagnosis of postpartum haemorrhage at vaginal birth (7)

Open in a new tab

8.

Open in a new tab

Accuracy of visual estimation with volumetric blood loss measurement as the reference standard

In the two studies that evaluated visual estimation with volumetric blood loss measurement as the reference standard (Lertbunnaphong 2016; Prasertcharoensuk 2000), the summary sensitivity was 22% (95% CI 12% to 37%; moderate‐certainty evidence) and summary specificity was 99% (95% CI 97% to 100%; moderate‐certainty evidence) (Table 1).

The interpretation of this result is that visual estimation will diagnose 22 out of 100 women who have PPH, but 78 women with PPH will be missed (i.e. will be false negatives). Of every 100 women without PPH, one will be wrongly diagnosed as having it (i.e. will be a false positive).

Visual inspection of the forest plots showed a low level of heterogeneity as the 95% CIs overlapped (Figure 8; Figure 9).

9.

Tests for diagnosis of postpartum haemorrhage at vaginal birth (9)

Open in a new tab

10.

Open in a new tab

Accuracy of visual estimation with volumetric plus gravimetric blood loss measurement as the reference standard

In the single study which evaluated visual estimation with volumetric plus gravimetric blood loss measurement as the reference standard (Rubenstein 2020), the summary sensitivity was 14% (95% CI 7% to 24%; moderate‐certainty evidence) and summary specificity was 98% (95% CI 95% to 99%; moderate‐certainty evidence) (Table 1).

The interpretation of this result is that visual estimation will diagnose 14 out of 100 women who have PPH, but 86 women with PPH will be missed (i.e. will be false negatives). Of every 100 women without PPH, two will be wrongly diagnosed as having it (i.e. will be false positives).

Accuracy of visual estimation with the alkaline‐haematin method of blood loss measurement as the reference standard

Three studies evaluated visual estimation with the alkaline haematin method of blood loss measurement as the reference standard (Duthie 1991; Larsson 2006; Razvi 1996). The sensitivities ranged from 0% (95% CI 0% to 28%) to 100% (95% CI 16% to 100%) (very low‐certainty evidence), and the specificities ranged from 97% (95% CI 92% to 99%) to 100% (95% CI 93% to 100%) (moderate‐certainty evidence) (Table 1). We did not undertake meta‐analysis because there were sparse data for robust estimation of sensitivity. Visual inspection of the forest plots indicated a high level of heterogeneity (Figure 4).

Accuracy of visual estimation with the oxyhaemoglobin method of blood loss measurement as the reference standard

The single study which evaluated visual estimation with the oxyhaemoglobin method of blood loss measurement as the reference standard (Brant 1967), showed a sensitivity of 28% (95% CI 14% to 47%; low‐certainty evidence) and specificity of 100% (95% CI 86% to 100%; low‐certainty evidence) (Table 1).

The interpretation of this result is that visual estimation will diagnose 28 out of 100 women who have PPH, but 72 women with PPH will be missed (i.e. will be false negatives). Of every 100 women without PPH, none be wrongly diagnosed as having it (i.e. will be false positives).

Accuracy of diagnostic approach using a calibrated drape plus observations with gravimetric blood loss measurement as the reference standard

In the two studies that evaluated a diagnostic approach using a calibrated drape plus observations (Devall 2024; Yunas 2024), with gravimetric blood loss measurement as the reference standard, the summary sensitivity was calculated as 93% (95% CI 92% to 94%; high‐certainty evidence) and the summary specificity as 95% (95% CI 95% to 96%; high‐certainty evidence) (Table 1). Visual inspection of the forest plots showed high heterogeneity as the confidence intervals for sensitivity did not overlap (Figure 10; Figure 11).

11.

Tests for diagnosis of postpartum haemorrhage at vaginal birth (11)

Open in a new tab

12.

Open in a new tab

The interpretation of this result is that using the diagnostic approach with a calibrated drape plus observations will diagnose 93 out of 100 women who have PPH, and only seven women with PPH will be missed (i.e. will be false negatives). Of every 100 women without PPH, five will be wrongly diagnosed as having it (i.e. will be false positives).

Accuracy of the SAPHE Mat with gravimetric blood loss measurement as the reference standard

The single study that evaluated the SAPHE Mat (Wilcox 2017), with gravimetric blood loss measurement as the reference standard, provided a sensitivity of 100% (95% CI 40% to 100%; very low‐certainty evidence) and specificity of 91% (95% CI 75% to 98%; very low‐certainty evidence) (Table 1).

The interpretation of this result is that using the SAPHE Mat will diagnose all women who have PPH and no women with PPH will be missed (i.e. will be false negatives). Of every 100 women without PPH, nine will be wrongly diagnosed as having it (i.e. will be false positives).

Accuracy of heart rate 91.2 bpm with volumetric plus gravimetric blood loss measurement as the reference standard

The single study that evaluated a heart rate value of 91.2 bpm (between 21 and 40 minutes postpartum) (Pacagnella 2022), with volumetric plus gravimetric blood loss measurement as the reference standard, provided a sensitivity of 59% (95% CI 53% to 65%; moderate‐certainty evidence) and specificity of 73% (95% CI 68% to 78%; moderate‐certainty evidence) (Table 1).

The interpretation of this result is that using a heart rate value of 91.2 bpm between 21 and 40 minutes postpartum will diagnose 59 out of 100 women who have PPH, but 41 women with PPH will be missed (i.e. will be false negatives). Of every 100 women without PPH, 27 will be wrongly diagnosed as having it (i.e. will be false positives).

Accuracy of shock index level 0.805 with volumetric plus gravimetric blood loss measurement as the reference standard

The single study that evaluated a shock index level of 0.805 (between 21 and 40 minutes postpartum) (Pacagnella 2022), with volumetric plus gravimetric blood loss measurement as the reference standard, provided a sensitivity of 43% (95% CI 37% to 49%; moderate‐certainty evidence) and specificity of 76% (95% CI 71% to 81%; moderate‐certainty evidence) (Table 1).

The interpretation of this result is that using a shock index level of 0.805 between 21 and 40 minutes postpartum will diagnose 43 out of 100 women who have PPH, but 57 women with PPH will be missed (i.e. will be false negatives). Of every 100 women without PPH, 24 will be wrongly diagnosed as having it (i.e. will be false positives).

Accuracy of a drop in haemoglobin of 2 g/dL or more with gravimetric blood loss measurement as the reference standard

The single study that evaluated a drop in haemoglobin of 2 g/dL or more (Anger 2019), with gravimetric blood loss measurement as the reference standard, provided a sensitivity of 36% (95% CI 30% to 43%; high‐certainty evidence) and specificity of 85% (95% CI 83% to 88%; high‐certainty evidence) (Table 1).

The interpretation of this result is that using a drop in haemoglobin of 2 g/dL or more will diagnose 36 out of 100 women who have PPH, but 64 women with PPH will be missed (i.e. will be false negatives). Of every 100 women without PPH, 15 will be wrongly diagnosed as having it (i.e. will be false positives).

Accuracy of a haemoglobin level of less than 10 g/dL with gravimetric blood loss measurement as the reference standard

The single study that evaluated a haemoglobin level of less than 10 g/dL (Anger 2019), with gravimetric blood loss measurement as the reference standard, provided a sensitivity of 37% (95% CI 30% to 44%; high‐certainty evidence) and specificity of 79% (95% CI 76% to 82%; high‐certainty evidence) (Table 1).

The interpretation of this result is that using a haemoglobin level of less than 10 g/dL will diagnose 37 out of 100 women who have PPH, but 63 women with PPH will be missed (i.e. will be false negatives). Of every 100 women without PPH, 21 will be wrongly diagnosed as having it (i.e. will be false positives).

Accuracy of a haemoglobin level of less than 7 g/dL with gravimetric blood loss measurement as the reference standard

The single study that evaluated a haemoglobin level of less than 7 g/dL (Anger 2019), with gravimetric blood loss measurement as the reference standard, provided a sensitivity of 2% (95% CI 1% to 6%; high‐certainty evidence) and specificity of 100% (95% CI 99% to 100%; high‐certainty evidence) (Table 1).

The interpretation of this result is that using a haemoglobin level of less than 7 g/dL will diagnose two out of 100 women who have PPH, but 98 women with PPH will be missed (i.e. will be false negatives). Of every 100 women without PPH, none will be wrongly diagnosed as having it (i.e. will be false positives).

Diagnosis of severe PPH

Accuracy of visual estimation with volumetric plus gravimetric blood loss measurement as the reference standard

The results from the single study that evaluated visual estimation (Rubenstein 2020), with volumetric plus gravimetric blood loss measurement using the Triton system as the reference standard, for the diagnosis of severe PPH provided a sensitivity of 9% (95% CI 0% to 41%; low‐certainty evidence) and a specificity of 100% (95% CI 99% to 100%; moderate‐certainty evidence) (Table 2).

The interpretation of this result is that visual estimation will diagnose nine out of 100 women who have severe PPH, but 91 women with severe PPH will be missed (i.e. will be false negatives). Of every 100 women without severe PPH, none will be wrongly diagnosed as having it (i.e. will be false positives).

Accuracy of heart rate 105.2 bpm with volumetric plus gravimetric blood loss measurement as the reference standard

The single study that evaluated a heart rate of 105.2 bpm (at 21 to 40 minutes postpartum) (Pacagnella 2022), with volumetric plus gravimetric blood loss measurement as the reference standard, provided a sensitivity of 55% (95% CI 49% to 61%; low‐certainty evidence) and specificity of 90% (95% CI 87% to 94%; moderate‐certainty evidence) (Table 2).

The interpretation of this result is that using a heart rate of 105.2 bpm at 21 to 40 minutes postpartum will diagnose 55 out of 100 women who have severe PPH, but 45 women with severe PPH will be missed (i.e. will be false negatives). Of every 100 women without severe PPH, 10 will be wrongly diagnosed as having it (i.e. will be false positives).

Accuracy of shock index levels at 15 minutes and 30 minutes with volumetric blood loss measurement as the reference standard

The single study that evaluated shock index levels at 15 minutes and 30 minutes after birth (Madar 2024), with volumetric blood loss measurement as the reference standard, reported sensitivities and specificities as shown in Table 4. The authors reported the area under the curve (AUC) as 0.66 (lower limit of the 95% confidence interval (LCI) 0.60) for 15 minutes and 0.68 (LCI 0.61) for 30 minutes.

2. Shock index thresholds for detection of severe postpartum haemorrhage (1000 mL).
StudyShock index valueTimingSensitivity (95% CI)Specificity (95% CI)
Madar 2024Shock index ≥ 0.715 minutes after birth78.7% (67.7–87.3)41.1% (39.4–42.7)
Madar 2024Shock index ≥ 0.915 minutes after birth40.0% (28.9–52.0)85.9% (84.7–87.0)
Madar 2024Shock index ≥ 1.115 minutes after birth14.7% (7.6–24.7)97.0% (96.3–97.5)
Madar 2024Shock index ≥ 0.730 minutes after birth80.3% (69.1–88.8)41.3% (39.6–42.9)
Madar 2024Shock index ≥ 0.930 minutes after birth38.0% (26.8–50.3)86.3% (85.1–87.4)
Madar 2024Shock index ≥ 1.130 minutes after birth18.3% (10.1–29.3)97.3% (96.7–97.8)
Pacagnella 2022Shock index ≥ 0.8Between 21 and 40 minutes after birth63.6%74.2%
Pacagnella 2022Shock index ≥ 0.9Between 21 and 40 minutes after birth40.9%88.9%
Pacagnella 2022Shock index ≥ 1.0Between 21 and 40 minutes after birth31.8%96.7%
Drew 2021Shock index ≥ 0.9Maximum value within first hour after birth92% (62–99)56% (39–70)
Ushida 2020Shock index ≥ 0.7Maximum value up to two hours after birth84.2% (82.0–86.3)35.5% (35.0–36.1)
Ushida 2020Shock index ≥ 0.8Maximum value up to two hours after birth64.9% (62.1–67.7)62.8% (62.3–63.4)
Ushida 2020Shock index ≥ 0.9Maximum value up to two hours after birth45.6% (42.6–48.5)82.6% (82.1–83.0)
Ushida 2020Shock index ≥ 1.0Maximum value up to two hours after birth29.9% (27.2–32.6)92.7% (92.4–93.0)
Ushida 2020Shock index ≥ 1.1Maximum value up to two hours after birth19.2% (16.9–21.5)97.1% (96.9–97.2)
Ushida 2020Shock index ≥ 1.2Maximum value up to two hours after birth10.3% (8.5–12.1)98.7% (98.6–98.8)
Ushida 2020Shock index ≥ 1.3Maximum value up to two hours after birth5.6% (4.2–6.9)99.4% (99.3–99.5)
Ushida 2020Shock index ≥ 1.4Maximum value up to two hours after birth3.5% (2.4–4.6)99.7% (99.7–99.8)
Ushida 2020Shock index ≥ 1.5Maximum value up to two hours after birth2.2% (1.4–3.1)99.9% (99.8–99.9)

Open in a new tab

CI: confidence interval

Accuracy of shock index levels between 21 and 40 minutes with volumetric plus gravimetric blood loss measurement as the reference standard

The single study that evaluated shock index levels between 21 and 40 minutes after birth (Pacagnella 2022), with volumetric plus gravimetric blood loss measurement as the reference standard, reported sensitivities and specificities as shown in Table 4.

The sensitivity for a shock index level of 1.0 or higher, which is a commonly used cut‐off [30], was 32% (95% CI 14% to 55%; low‐certainty evidence) with a specificity of 97% (95% CI 94% to 99%; moderate‐certainty evidence) (Table 2).

The interpretation of this result is that using a shock index level of 1.0 or higher at 21 to 40 minutes after birth will diagnose 32 out of 100 women who have severe PPH, but 68 women with severe PPH will be missed (i.e. will be false negatives). Of every 100 women without severe PPH, three will be wrongly diagnosed as having it (i.e. will be false positives).

Accuracy of shock index levels with gravimetric blood loss measurement as the reference standard

The sensitivity and specificity for the maximum shock index value within the first hour of birth of 0.9 or higher, with gravimetric blood loss measurement as the reference standard, as reported by Drew 2021, is shown in Table 4.

The single study that evaluated maximum shock index values up to two hours after birth (Ushida 2020), with gravimetric blood loss measurement as the reference standard, reported sensitivities and specificities as shown in Table 4. The authors reported the AUC as 0.699 (95% CI 0.682 to 0.716). The sensitivity for a shock index level of 1.0 or higher, which is a commonly used cut‐off [30], was 30% (95% CI 27% to 33%; moderate‐certainty evidence) with a specificity of 93% (95% CI 92% to 93%; moderate‐certainty evidence) (Table 2).

The interpretation of this result is that using a maximum shock index value of 1.0 or higher up to two hours after birth will diagnose 30 out of 100 women who have severe PPH, but 70 women with severe PPH will be missed (i.e. will be false negatives). Of every 100 women without severe PPH, seven will be wrongly diagnosed as having it (i.e. will be false positives).

Accuracy of a reduction in total haemoglobin (SpHb) of 7 g/L with gravimetric blood loss measurement as the reference standard

The results from the single study that evaluated a reduction in SpHb of 7 g/L (Drew 2021), with gravimetric blood loss measurement as the reference standard, reported a sensitivity of 83% (95% CI 52% to 98%; very low‐certainty evidence) and a specificity of 62% (95% CI 49% to 76%; very low‐certainty evidence) (Table 2).

The interpretation of this result is that using a reduction in SpHb of 7 g/L will diagnose 83 out of 100 women who have severe PPH, but 17 women with severe PPH will be missed (i.e. will be false negatives). Of every 100 women without severe PPH, 38 will be wrongly diagnosed as having it (i.e. will be false positives).

Accuracy of a drop in haemoglobin of 2 g/dL or more with gravimetric blood loss measurement as the reference standard

The results from the single study that evaluated a drop in haemoglobin of 2 g/dL or more (Anger 2019), with gravimetric blood loss measurement as the reference standard, provided a sensitivity of 68% (95% CI 48% to 84%; moderate‐certainty evidence) and specificity of 83% (95% CI 80% to 85%; high‐certainty evidence) (Table 2).

The interpretation of this result is that using a drop in haemoglobin of 2 g/dL or more will diagnose 68 out of 100 women who have severe PPH, but 32 women with severe PPH will be missed (i.e. will be false negatives). Of every 100 women without severe PPH, 17 will be wrongly diagnosed as having it (i.e. will be false positives).

Accuracy of a haemoglobin level of less than 10 g/dL with gravimetric blood loss measurement as the reference standard

The single study that evaluated a haemoglobin level of less than 10 g/dL (Anger 2019), with gravimetric blood loss measurement as the reference standard, provided a sensitivity of 71% (95% CI 51% to 87%; moderate‐certainty evidence) and specificity of 77% (95% CI 75% to 80%; high‐certainty evidence) (Table 2).

The interpretation of this result is that using a haemoglobin level of less than 10 g/dL will diagnose 71 out of 100 women who have severe PPH, but 29 women with severe PPH will be missed (i.e. will be false negatives). Of every 100 women without severe PPH, 23 will be wrongly diagnosed as having it (i.e. will be false positives).

Accuracy of a haemoglobin level of less than 7 g/dL with gravimetric blood loss measurement as the reference standard

The single study that evaluated a haemoglobin level of less than 7 g/dL (Anger 2019), with gravimetric blood loss measurement as the reference standard, provided a sensitivity of 11% (95% CI 2% to 28%; moderate‐certainty evidence) and specificity of 99% (95% CI 99% to 100%; high‐certainty evidence) (Table 2).

The interpretation of this result is that using a haemoglobin level of less than 7 g/dL will diagnose 11 out of 100 women who have severe PPH, but 89 women with severe PPH will be missed (i.e. will be false negatives). Of every 100 women without severe PPH, one will be wrongly diagnosed as having it (i.e. will be false positives).

Accuracy of predelivery fibrinogen levels with volumetric blood loss measurement as the reference standard

The results from the single study that evaluated predelivery fibrinogen levels of 4.080 g/L and 4.46 g/L (Niepraschk‐von Dollen 2016), with volumetric blood loss measurement as the reference standard, provided sensitivities of 54% (95% CI 33% to 74%) and 71% (95% CI 49% to 87%), respectively (very low‐certainty evidence) and specificities of 78% (95% CI 75% to 81%) and 59% (95% CI 55% to 62%), respectively (low‐certainty evidence) (Table 2).

The interpretation of these results is that using a predelivery fibrinogen level of 4.080 g/L will diagnose 54 out of 100 women who have severe PPH, but 46 women with severe PPH will be missed (i.e. will be false negatives). Of every 100 women without severe PPH, 22 will be wrongly diagnosed as having it (i.e. will be false positives). Using a predelivery fibrinogen level of 4.46 g/L will diagnose 71 out of 100 women who have severe PPH, but 29 women with severe PPH will be missed (i.e. will be false negatives). Of every 100 women without severe PPH, 41 will be wrongly diagnosed as having it (i.e. will be false positives).

Discussion

Summary of main results

We identified 18 studies evaluating a range of index tests to diagnose PPH and severe PPH in women giving birth vaginally. There were 14 studies evaluating index tests for the diagnosis of PPH and seven studies for the diagnosis of severe PPH.

Our assessment of methodological quality identified eight studies as being at low overall risk of bias and 10 studies as being at high overall risk. There were no applicability concerns for any of the studies.

Visual estimation showed low summary sensitivities for the diagnosis of PPH, regardless of the reference standard used. The summary sensitivity of 48% was for visual estimation with gravimetric blood loss measurement as the reference standard.

For the diagnosis of PPH, two studies evaluated a diagnostic approach using a calibrated drape plus observations against a gravimetric blood loss measurement reference standard. The summary estimates for sensitivity and specificity for this outcome were both high (93% and 95%, respectively).

The use of a SAPHE Mat as the index test had both high sensitivity (100%) and specificity (91%) for diagnosing PPH, but it was only reported in a single study with few participants (36), and the certainty of evidence for the sensitivity was very low.

The single study which reported results for a heart rate cut‐off value of 91.2 bpm and a shock index cut‐off value of 0.805, between 21 and 40 minutes after birth, showed low sensitivities (59% and 43%, respectively).

Other index tests used to diagnose PPH showed low summary sensitivities. The single study which evaluated a haemoglobin drop of 2 g/dL or more, a haemoglobin level of less than 10 g/dL and a haemoglobin level of less than 7 g/dL as index tests, showed a sensitivity of only 37% (specificity 79%) for the haemoglobin level of less than 10 g/dL.

For the diagnosis of severe PPH, the single study which evaluated visual estimation (with a reference standard of volumetric plus gravimetric blood loss measurement) gave a low sensitivity of 9%.

The single study which reported results for a heart rate cut‐off value of 105.2 bpm, 21 to 40 minutes after birth, showed a low sensitivity of 55%.

The results for the different thresholds of shock index to diagnose severe PPH showed that as sensitivity increased, the specificity decreased.

The sensitivity of a shock index level of 1.0 or higher, which is used as a threshold for severe PPH [30] (with gravimetric blood loss measurement as the reference standard), showed low sensitivity (30%), but high specificity (93%) when measured as a maximum value up to two hours after birth. When shock index was measured between 21 and 40 minutes after birth (with a reference standard of volumetric plus gravimetric blood loss measurement), the sensitivity of a threshold of 1.0 or higher was low, at 32%, with high specificity of 97%.

The study which evaluated a drop of total haemoglobin (SpHb) of 7 g/L as an index test to diagnose severe PPH showed moderate sensitivity and specificity (83% and 62%). The certainty of evidence for both sensitivity and specificity was very low.

The single study which evaluated a haemoglobin drop of 2 g/dL or more, a haemoglobin level of less than 10 g/dL, and a haemoglobin level of less than 7 g/dL as index tests to diagnose severe PPH showed a sensitivity of 71% (specificity 77%) for the haemoglobin level of less than 10 g/dL.

The single study which evaluated predelivery fibrinogen levels as index tests to identify severe PPH had a sensitivity of 71% for a fibrinogen level of 4.46 g/dL, but the certainty of evidence was very low.

Due to the low number of identified studies for each combination of index test and reference standard, we were unable to formally investigate heterogeneity beyond visual examination of the forest plots. As a result, we were also unable to perform subgroup analyses.

Comparison with previous research

Our findings are supported by the results of previous reviews. Schorn undertook a literature review, and concluded that visual estimation was inaccurate and that gravimetric estimation of blood loss was accurate and feasible [31]. Ruiz and colleagues recently published a meta‐analysis in which they concluded that quantification of blood loss by any method was superior to visual estimation [32]. Our review extends the findings of these previous studies, as neither were reviews of diagnostic test accuracy.

Makino and colleagues concluded that the predictive performance of shock index to predict severe PPH was inconsistent, which is in keeping with our findings [33].

Differences from the protocol

This review was prospectively registered with PROSPERO (CRD42024541874).

We did not compare the diagnostic accuracy of the index tests due to lack of comparative accuracy studies. We included two studies that each evaluated two index tests (Devall 2024; Yunas 2024). However, the studies were not comparative because the two index tests were not evaluated in the same cohort in the studies, i.e. unpaired and not randomised, and we regarded both studies as studies of single test evaluations. For this reason, we also did not apply the QUADAS‐C tool, a tool designed primarily for assessment of fully paired or randomised comparative accuracy studies.

We included unpublished data to reduce publication and reporting bias. Furthermore, the unpublished studies provided large data sets and added robust data. We judged that omitting these data would have weakened the review and left an inappropriate emphasis on visual estimation, which is known to be inaccurate. The inclusion of the unpublished data allowed the consideration of accuracy data on a diagnostic approach using a calibrated drape plus observations, which may present a feasible, cost‐effective, and objective method of diagnosing PPH, particularly in low‐resource settings. We acknowledge that including unpublished studies is a deviation from the protocol.

Strengths and weaknesses of the review

Strengths of this review include the broad search strategy we employed to search several databases. This comprehensive search was not limited by language or date restrictions. We also obtained large data sets from three unpublished studies. We included unpublished data to minimise publication and reporting bias.

Limitations of our review include the fact that most eligible studies were restricted to the hospital setting. Only one study was undertaken in a primary healthcare unit and another in the home birth setting. The highest burden of PPH is in the non‐hospital setting, for which we found very little test accuracy data.

The limited number of studies available for each index test and reference standard combination resulted in few meta‐analyses. Our largest meta‐analysis contained four studies. The other two meta‐analyses contained two studies in each. As a result, we were unable to formally assess heterogeneity beyond forest plot visualisation or undertake subgroup analyses.

There may be other subgroup effects, such as vaginal delivery after a previous caesarean section, which may result in different PPH risk which we did not look at as part of this review. As mentioned, subgroup analysis may still not have been possible due to the limited number of studies available.

We were unable to find any studies which evaluated a calibrated blood collection drape or device as an index test with an objective reference standard for the diagnosis of severe PPH.

We were unable to identify any studies, meeting our inclusion criteria, which combined blood loss, clinical variables such as shock index, and the clotting status of the participants as a single index test.

Our search strategy was broad and aimed to capture newer methods to diagnose PPH, such as the use of camera and artificial intelligence technologies. We did find studies which included these methods, but were unable to include them as they did not allow us to extract 2 x 2 data, or they had diagnostic case‐control (two‐gate) designs.

Index tests such as visual estimation of blood loss are unlikely to be used in complete isolation to diagnose PPH. Birth attendants often incorporate other variables, such as the clinical appearance of the participant, and results of observations such as blood pressure and heart rate to make the final diagnosis. This may contribute to the heterogeneity seen.

Another limitation was the manner in which several studies presented data. In constructing the 2 x 2 tables, we often had to calculate the data using the quoted sensitivities, specificities, and disease prevalence. This required us to round the calculated values to the nearest whole number of participants. This rounding accounts for the discrepancies between the sensitivity and specificity values in the study manuscripts and those on our forest plots.

Applicability of findings to the review question

We evaluated the accuracy of a variety of index tests to diagnose PPH and severe PPH at vaginal birth. Our findings identified some index tests that have high sensitivity (approach with calibrated drape plus observations) and others which have low sensitivity (visual estimation). The review included studies conducted in at least 16 different countries, with representation from low‐ and middle‐income countries (LMICs) and high‐income countries (HICs). Most studies were conducted in the hospital setting and there were limited data from the community and home birth settings.

Authors' conclusions

Implications for practice

When prioritising the use of a test to diagnose postpartum haemorrhage (PPH), maximising sensitivity will reduce the number of false negatives and, therefore, the number of missed diagnoses. This is crucial in reducing morbidity and mortality from this time‐critical obstetric emergency.

Visual estimation of blood loss showed low sensitivity. The upper limit of the confidence interval (CI) for the most favourable summary sensitivity for visual estimation as an index test to diagnose PPH was 53%. This suggests that, at its best, visual estimation is likely to miss the diagnosis of PPH in half of the women who suffer from PPH.

The use of a diagnostic approach with a calibrated drape plus observations to diagnose PPH showed high sensitivity and specificity with high‐certainty evidence.

The diagnostic approach using the calibrated drape meets the requirement for it to be used as a reference standard, like gravimetric blood loss measurement. The two large studies, conducted in Africa and South Asia, which contributed this evidence were both multicentre trials with standardised data collection. Further confirmatory evidence may be needed in settings other than low‐ and middle‐income countries (LMICs).

The approach using a calibrated drape may also present a more convenient alternative to weighing blood loss in resource‐limited settings. An affordable and available calibrated drape, which can be part of a birth kit, will also allow more data to be gathered on PPH diagnosis in non‐hospital settings. The calibrated drape was used in some studies as the reference standard due to its accuracy.

The reference standard used to assess the calibrated drape approach was gravimetric weighed blood loss measurement, which itself is an accurate way of diagnosing PPH. The gravimetric approach, which involves weighing collected blood loss using a weighing scale, is available in resource‐limited settings, and midwives and local birth attendants can be trained in its use. Its use as an accurate test for PPH should be prioritised above visual estimation.

The prioritisation of objective index tests, such as the calibrated drape approach, will have to be cost‐effective. This is particularly relevant to LMICs, which carry the highest burden of PPH adverse outcomes and suffer from limited resources. The 'Signalling a Postpartum Haemorrhage Emergency' (SAPHE) Mat showed high sensitivity and specificity, but its use may be limited by cost, and the certainty of evidence was very low. An affordable calibrated drape may offer a cost‐effective, accurate, and viable option.

Other index tests – including heart rate, shock index, changes in haemoglobin (Hb) levels, and predelivery fibrinogen levels – showed low to moderate sensitivities. These tests may still have a role to play in diagnosing PPH if used in combination with each other, along with tests to objectively measure blood loss.

Implications for research

Test accuracy is known to vary with the setting. Further research should focus on evaluating the accuracy of index tests in non‐hospital settings, such as rural clinics and home births. The use of cost‐effective, accurate, and objective measures, such as a calibrated drape as part of a birth kit, may allow easier evaluation of index test accuracy in non‐hospital settings.

Index tests such as the heart rate, shock index, haemoglobin levels, and predelivery fibrinogen levels, did not show high sensitivity, but they may have a role to play in the diagnosis of PPH as part of combination diagnostic models. Further studies should aim to evaluate the accuracy of models that combine objective measures of blood loss, clinical variables such as shock index, and the clotting status of participants.

Novel strategies such as the use of cameras and artificial intelligence systems for diagnosing PPH should also be assessed as part of diagnostic test accuracy studies.

Supporting Information

Supplementary materials are available with the online version of this article: 10.1002/14651858.CD016134.

Supplementary materials are published alongside the article and contain additional data and information that support or enhance the article. Supplementary materials may not be subject to the same editorial scrutiny as the content of the article and Cochrane has not copyedited, typeset or proofread these materials. The material in these sections has been supplied by the author(s) for publication under a Licence for Publication and the author(s) are solely responsible for the material. Cochrane accordingly gives no representations or warranties of any kind in relation to, and accepts no liability for any reliance on or use of, such material.

Supplementary material 1 Search strategies

CD016134-SUP-01-searchStrategy.html (36KB, html)

Supplementary material 2 Characteristics of included studies

CD016134-SUP-02-characteristicsOfIncludedStudies.html (152.9KB, html)

Supplementary material 3 Characteristics of excluded studies

CD016134-SUP-03-characteristicsOfExcludedStudies.html (224.3KB, html)

Supplementary material 4 Characteristics of studies awaiting classification

CD016134-SUP-04-characteristicsOfAwaitingStudies.html (37.6KB, html)

Supplementary material 5 Characteristics of ongoing studies

CD016134-SUP-05-characteristicsOfOngoingStudies.html (32.1KB, html)

Supplementary material 6 Analyses

CD016134-SUP-06-analyses.html (290.9KB, html)

Supplementary material 7 Data package

CD016134-SUP-07-dataPackage.zip (50KB, zip)

Supplementary material 8 QUADAS‐2 template with signalling questions

CD016134-SUP-08-other.html (40.3KB, html)

Supplementary material 9 QUADAS‐2 assessments for included studies

CD016134-SUP-09-other.html (26.4KB, html)

New

Additional information

Acknowledgements

We would like to thank Jennifer Harrison for her helpful feedback on the plain language summary (consumer involvement), Charlene Bridges (Cochrane Information Specialist) for helping with the search strategies and running the searches, and Sofia Tsokani (Cochrane editorial support) for preliminary feedback on the analysis section at the protocol development stage. We would also like to thank James Martin for providing the data from two unpublished studies included in our review (Devall 2024; Yunas 2024) and Ioannis Gallos for providing data from the third unpublished study (Gallos 2024).

We are grateful to Dr Sufyan Hussain, Honorary Reader, King’s College London, for independently evaluating the methodological quality of the studies.

Editorial and peer‐reviewer contributions

The following people conducted the editorial process for this article:

  • Sign‐off Editors (final editorial decision): Gianni Virgili, Department NEUROFARBA, University of Florence, Italy, and Professor Zarko Alfirevic, University of Liverpool, UK;

  • Managing Editor (selected peer reviewers, provided editorial guidance to authors, edited the article): Liz Bickerdike, Cochrane Central Editorial Service;

  • Editorial Assistant (conducted editorial policy checks, collated peer‐reviewer comments, and supported editorial team): Lisa Wydrzynski, Cochrane Central Editorial Service;

  • Copy Editor (copy editing and production): Faith Armitage, Cochrane Central Production Service;

  • Peer‐reviewers (provided comments and recommended an editorial decision): Anderson Borovac‐Pinheiro, Madre Theodora Hospital, Rodolfo de Carvalho Pacagnella, Dr Sadia Malick FRCOG, Consultant Gynaecologist, King Faisal Specialist Hospital & Research Centre (KFSH&RC) Riyadh, KSA (clinical/content review), Sofia Tsokani, Cochrane Editorial & Methods Department (statistical review), Jo Platt, Central Editorial Information Specialist (search review).

Contributions of authors

Conception and design: IY, IG, YT, AC
Co‐ordination: IY
Search and selection of studies for inclusion: IY, AD, AC, (MP and JA when other authors were conflicted)
Data extraction: IY, AD, AC, (MP and JA when other authors were conflicted)
Assessment of the risk of bias: IY, AD, AC, SH (MP and JA when other authors were conflicted)
Analysis of data: IY, AC, YT
Assessment of the certainty of evidence: IY, AD, AC, (MP and JA when other authors were conflicted)
Data interpretation: IY, AC
Writing of first draft: IY

Declarations of interest

Idnan Yunas: no relevant interests were disclosed.

Ioannis Gallos: no relevant interests were disclosed.

Adam Devall: no relevant interests were disclosed.

Marcelina Podesek: no relevant interests were disclosed.

John Allotey: no relevant interests were disclosed.

Yemisi Takwoingi was not involved in the editorial process.

Arri Coomarasamy: Bill and Melinda Gates Foundation (Grant / Contract)

Sources of support

Internal sources

  • No sources of support provided

External sources

  • No sources of support provided

Registration and protocol

The protocol for this review was prospectively registered with PROSPERO (CRD42024541874).

Data, code and other materials

Supplementary material 6; Supplementary material 7

References

  • 1.Say L, Chou D, Gemmill A, Tuncalp O, Moller AB, Daniels J, et al. Global causes of maternal death: a WHO systematic analysis. Lancet Global Health 2014;2(6):e323-33. [DOI: 10.1016/S2214-109X(14)70227-X] [DOI] [Google Scholar]
  • 2.Gallos I, Devall A, Martin J, Middleton L, Beeson L, Galadanci H, et al. Randomized trial of early detection and treatment of postpartum hemorrhage. New England Journal of Medicine 2023;389(1):11-21. [DOI: 10.1056/NEJMoa2303966] [DOI] [Google Scholar]
  • 3.Lertbunnaphong T, Lapthanapat N, Leetheeragul J, Hakularb P, Ownon A. Postpartum blood loss: visual estimation versus objective quantification with a novel birthing drape. Singapore Medical Journal 2016;57(6):325-8. [DOI: 10.11622/smedj.2016107] [DOI] [Google Scholar]
  • 4.Devall AJ, Gallos ID, Martin JT, Price MJ, Middleton LJ, Beeson LE, et al. Data (as supplied 12 June 2024). Data on file.
  • 5.Yunas I, Devall AJ, Martin J, Beeson LE, Sindhu K, Sheikh L, et al. Data (as supplied 12 June 2024). Data on file.
  • 6.Wilcox L, Ramprasad C, Gutierrez A, Oden M, Richards-Kortum R, Sangi-Haghpeykar H, et al. Diagnosing postpartum hemorrhage: a new way to assess blood loss in a low-resource setting. Maternal and Child Health Journal 2017;21(3):516-23. [DOI: 10.1007/s10995-016-2135-5] [DOI] [Google Scholar]
  • 7.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. [DOI: 10.1371/journal.pone.0168535] [DOI] [Google Scholar]
  • 8.Newton J, Barnard G, Collins W. Rapid method for measuring menstrual blood-loss using automatic extraction. Contraception 1977;16(3):269-82. [DOI: Doi 10.1016/0010-7824(77)90026-9] [Google Scholar]
  • 9.Brant HA. Precise estimation of postpartum haemorrhage: difficulties and importance. British Medical Journal 1967;1(5537):398-400. [DOI: 10.1136/bmj.1.5537.398] [DOI] [Google Scholar]
  • 10.Review Manager (RevMan). Version 7.2.0. The Cochrane Collaboration, 2024. Available at revman.cochrane.org.
  • 11.Whiting PF, Rutjes AW, Westwood ME, Mallett S, Deeks JJ, Reitsma JB, et al. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Annals of Internal Medicine 2011;155(8):529-36. [DOI: 10.7326/0003-4819-155-8-201110180-00009] [DOI] [Google Scholar]
  • 12.Reitsma JB, Glas AS, Rutjes AW, Scholten RJ, Bossuyt PM, Zwinderman AH. Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews. Journal of Clinical Epidemiology 2005;58(10):982-90. [DOI: 10.1016/j.jclinepi.2005.02.022] [DOI] [Google Scholar]
  • 13.StataCorp. 2023. Stata Statistical Software: Release 18. College Station, TX: StataCorp LLC. StataCorp LLC, 2023.
  • 14.Nyaga VN, Arbyn M. Metadta: a Stata command for meta-analysis and meta-regression of diagnostic test accuracy data - a tutorial. Archives of Public Health 2022;80(1):95. [DOI: 10.1186/s13690-021-00747-5] [DOI] [Google Scholar]
  • 15.Schünemann H, Brożek J, Guyatt G, Oxman A, editor(s). Handbook for grading the quality of evidence and the strength of recommendations using the GRADE approach (updated October 2013). GRADE Working Group, 2013. Available from gdt.guidelinedevelopment.org/app/handbook/handbook.html.
  • 16.Anger H, Durocher J, Dabash R, Winikoff B. How well do postpartum blood loss and common definitions of postpartum hemorrhage correlate with postpartum anemia and fall in hemoglobin? PloS One 2019;14(8):p.e0221216. [DOI: 10.1371/journal.pone.0221216] [DOI] [Google Scholar]
  • 17.Drew T, Carvalho JC, Subramanian C, Yoon EW, Downey K, Thorneloe B, et al. The association of shock index and haemoglobin variation with postpartum haemorrhage after vaginal delivery: a prospective cohort pilot study. International Journal of Obstetric Anesthesia 2021;45:67-73. [DOI: 10.1016/j.ijoa.2020.10.010] [DOI] [Google Scholar]
  • 18.Duthie SJ, Ven D, Yung GL, Guang DZ, Chan SY, Ma HK. Discrepancy between laboratory determination and visual estimation of blood loss during normal delivery. European Journal of Obstetrics, Gynecology, and Reproductive Biology 1991;38(2):119-24. [DOI: 10.1016/0028-2243(91)90188-q] [DOI] [Google Scholar]
  • 19.Hazarika N, Kaur AP, Madan A. A comparative study of postpartum blood loss by visual estimation method and gravimetric method. Asian Journal of Pharmaceutical and Clinical Research 2022;15(7):127-30. [DOI: 10.22159/ajpcr.2022.v15i7.44589] [DOI] [Google Scholar]
  • 20.Larsson C, Saltvedt S, Wiklund I, Pahlen S, Andolf E. Estimation of blood loss after cesarean section and vaginal delivery has low validity with a tendency to exaggeration. Acta Obstetricia et Gynecologica Scandinavica 2006;85(12):1448-52. [DOI: 10.1080/00016340600985032] [DOI] [Google Scholar]
  • 21.Lertbunnaphong T, Lapthanapat N, Leetheeragul J, Hakularb P, Ownon A. Postpartum blood loss: visual estimation versus objective quantification with a novel birthing drape. Singapore Medical Journal 2016;57(6):325-28. [DOI: 10.11622/smedj.2016107] [DOI] [Google Scholar]
  • 22.Madar H, Deneux-Tharaux C, Sentilhes L; TRAAP Study Group. Shock index as a predictor of postpartum haemorrhage after vaginal delivery: secondary analysis of a multicentre randomised controlled trial. British Journal of Obstetrics and Gynaecology 2024;131(3):343-52. [DOI: 10.1111/1471-0528.17634] [DOI] [Google Scholar]
  • 23.Niepraschk-von Dollen K, Bamberg C, Henkelmann A, Mickley L, Kaufner L, Henrich W, et al. Predelivery maternal fibrinogen as a predictor of blood loss after vaginal delivery. Archives of Gynecology and Obstetrics 2016;294(4):745-51. [DOI: 10.1007/s00404-016-4031-z] [DOI] [Google Scholar]
  • 24.Prasertcharoensuk W, Swadpanich U, Lumbiganon P. Accuracy of the blood loss estimation in the third stage of labor. International Journal of Gynecology and Obstetrics 2000;71(1):69-70. [DOI: 10.1016/S0020-7292%2800%2900294-0] [DOI] [Google Scholar]
  • 25.Razvi K, Chua S, Arulkumaran S, Ratnam SS. A comparison between visual estimation and laboratory determination of blood loss during the third stage of labour. Australian & New Zealand Journal of Obstetrics and Gynaecology 1996;36(2):152-4. [DOI: 10.1111/j.1479-828X.1996.tb03273.x] [DOI] [Google Scholar]
  • 26.Rubenstein AF, Zamudio S, Douglas C, Sledge S, Thurer RL. Automated quantification of blood loss versus visual estimation in 274 vaginal deliveries. American Journal of Perinatology 2021;38(10):1031-5. [DOI: 10.1055/s-0040-1701507] [DOI] [Google Scholar]
  • 27.Ushida T, Kotani T, Imai K, Nakano-Kobayashi T, Nakamura N, Moriyama Y, et al. Shock index and postpartum hemorrhage in vaginal deliveries: a multicenter retrospective study. Shock 2021;55(3):332-7. [DOI: 10.1097/SHK.0000000000001634] [DOI] [Google Scholar]
  • 28.Pacagnella RC, Borovac-Pinheiro A, Silveira C, Siani Morais S, Argenton JL, Souza JP, et al. The golden hour for postpartum hemorrhage: results from a prospective cohort study. International Journal of Gynecology and Obstetrics 2022;156(3):450-58. [DOI: 10.1002/ijgo.13823] [DOI] [Google Scholar]
  • 29.Gallos I, Widmer M, Gulmezoglu M, Piaggio G, Qureshi Z, Bello N, et al. Data (as supplied 12 June 2024). Data on file.
  • 30.Le Bas A, Chandraharan E, Addei A, Arulkumaran S. Use of the "obstetric shock index" as an adjunct in identifying significant blood loss in patients with massive postpartum hemorrhage. International Journal of Gynecology and Obstetrics 2014;124(3):253-5. [DOI: 10.1016/j.ijgo.2013.08.020] [DOI] [Google Scholar]
  • 31.Schorn MN. Measurement of blood loss: review of the literature. Journal of Midwifery and Women's Health 2010;55(1):20-7. [DOI: 10.1016/j.jmwh.2009.02.014] [DOI] [Google Scholar]
  • 32.Ruiz MT, Azevedo NF, Resende CV, Rodrigues WF, Meneguci J, Contim D, et al. Quantification of blood loss for the diagnosis of postpartum hemorrhage: a systematic review and meta-analysis. Revista Brasileira de Enfermagem 2023;76(6):e20230070. [DOI: 10.1590/0034-7167-2023-0070] [DOI] [Google Scholar]
  • 33.Makino Y, Miyake K, Okada A, Ikeda Y, Okada Y. Predictive accuracy of the shock index for severe postpartum hemorrhage in high-income countries: a systematic review and meta-analysis. Journal of Obstetrics and Gynaecology Research 2022;48(8):2027-37. [DOI: 10.1111/jog.15292] [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary material 1 Search strategies

CD016134-SUP-01-searchStrategy.html (36KB, html)

Supplementary material 2 Characteristics of included studies

CD016134-SUP-02-characteristicsOfIncludedStudies.html (152.9KB, html)

Supplementary material 3 Characteristics of excluded studies

CD016134-SUP-03-characteristicsOfExcludedStudies.html (224.3KB, html)

Supplementary material 4 Characteristics of studies awaiting classification

CD016134-SUP-04-characteristicsOfAwaitingStudies.html (37.6KB, html)

Supplementary material 5 Characteristics of ongoing studies

CD016134-SUP-05-characteristicsOfOngoingStudies.html (32.1KB, html)

Supplementary material 6 Analyses

CD016134-SUP-06-analyses.html (290.9KB, html)

Supplementary material 7 Data package

CD016134-SUP-07-dataPackage.zip (50KB, zip)

Supplementary material 8 QUADAS‐2 template with signalling questions

CD016134-SUP-08-other.html (40.3KB, html)

Supplementary material 9 QUADAS‐2 assessments for included studies

CD016134-SUP-09-other.html (26.4KB, html)

Data Availability Statement

Supplementary material 6; Supplementary material 7

Tests for diagnosis of postpartum haemorrhage at vaginal birth (2025)
Top Articles
Latest Posts
Recommended Articles
Article information

Author: Dan Stracke

Last Updated:

Views: 6517

Rating: 4.2 / 5 (43 voted)

Reviews: 82% of readers found this page helpful

Author information

Name: Dan Stracke

Birthday: 1992-08-25

Address: 2253 Brown Springs, East Alla, OH 38634-0309

Phone: +398735162064

Job: Investor Government Associate

Hobby: Shopping, LARPing, Scrapbooking, Surfing, Slacklining, Dance, Glassblowing

Introduction: My name is Dan Stracke, I am a homely, gleaming, glamorous, inquisitive, homely, gorgeous, light person who loves writing and wants to share my knowledge and understanding with you.