- Question: To evaluate the correlations between blood loss (measured using a gravimetric method, visual estimation by an anesthesiologist, visual estimation by an obstetrician, and the Triton System) and hemoglobin values after elective cesarean delivery.
- Findings: We observed a statistically significant but weak correlation between blood loss measured by the Triton System and postcesarean hemoglobin (r = −0.33; P = .01), and no statistically significant differences in the magnitude of the correlations across the 4 measurement modalities.
- Meaning: There is limited clinical utility in estimating postcesarean hemoglobin levels from blood loss values measured across the 4 studied measurement modalities.
Postpartum anemia is an underappreciated yet important etiological factor associated with adverse maternal outcomes such as depression,1,2 maternal fatigue,3 and cognitive dysfunction.4 Preoperative anemia and postpartum hemorrhage (PPH) are strong independent risk factors for postpartum anemia after cesarean delivery (CD).5 However, little is known about the relationship between overall blood loss and post-CD hemoglobin (Hb) levels.
Assessing this association among patients undergoing CD has clinical relevance for several reasons. First, there is no consensus on the indications for postpartum Hb testing after CD. Accurate measurement of intraoperative blood loss may inform a decision to perform post-CD Hb testing. Second, in the absence of antenatal anemia or PPH, the likelihood of postpartum anemia is low, thus laboratory testing for postpartum anemia may be unnecessary. Conversely, if high-volume blood loss is underestimated, then screening for postpartum anemia may not occur. Third, examining the presence and strength of the relations between blood loss and post-CD Hb could be valuable for developing an algorithm for predicting Hb post-CD. Embedding such algorithms in electronic health record systems could prompt clinicians to screen for postpartum anemia. Therefore, assessing and comparing the correlations between blood loss with post-CD Hb across different blood loss measurement techniques have clinical relevance.
Visual estimation, gravimetric, and volumetric approaches are commonly used techniques to quantify blood loss during CD. However, volumetric blood loss measurement using noncalibrated collection bags can be inaccurate.6–8 Visual estimation can underestimate actual blood loss by 30%–50% in high blood loss situations, and conversely tend to overestimate in situations where little blood loss has occurred.7–9 Protocolized gravimetric measurement of blood loss can result in good correlation between blood loss and an adjusted fall in Hb (predelivery versus postdelivery) among women experiencing a major PPH (>1500 mL blood loss),10 but gravimetric measurement can be labor intensive and nonpractical for real-time analysis during a major bleeding episode.
A new device, the Triton System (Gauss Surgical Inc, Los Altos, CA) has attracted attention as a means of measuring intraoperative blood loss. The device photometrically measures blood loss using “Feature Extraction Technology” (FET) from iPad-derived images of blood-containing surgical material.11 Previous studies suggest that this technology provides an accurate measure of Hb mass on surgical sponges11,12 and in reconstituted blood contained in suction canisters.13 Gravimetric and volumetric methods are unable to differentiate amniotic fluid from the blood, and the volume of amniotic fluid collected in the suction canister may be highly variable. The Triton System may prevent overestimates of blood loss in cases where amniotic fluid represents a high percentage of the total fluid volume in a suction canister. Despite these advantages, limited evidence exists regarding which blood loss measurement most closely correlates with Hb after CD.
The aims of the study were to evaluate the correlations between blood loss (measured using a gravimetric method, visual estimation, and the Triton System) and post-CD Hb values, and to compare correlation coefficients among the methods used for measuring blood loss.
We performed a prospective observational study investigating the association between blood loss and postoperative Hb levels in women undergoing CD with neuraxial anesthesia. The study was approved by Stanford University Institutional Review Board (protocol no. 35648). Before patient enrollment, we registered the study at clinicaltrials.gov (NCT02667600; Principal Investigator: A.J.B.; date of registration: January 29, 2016). The study was conducted at Lucile Packard Children’s Hospital, a tertiary obstetric center, at Stanford University, CA, between April 2016 and November 2016. Written informed consent was obtained from study participants. Inclusion criteria were as follows: American Society of Anesthesiologists physical status class II or III, women between 18 and 50 years of age, ≥37 weeks’ gestational age, and elective CD or intrapartum CD after a trial of labor. Exclusion criteria included women requiring CD for urgent or emergent indications.
In the preoperative period, we measured a baseline venous Hb level. Spinal anesthesia was induced with intrathecal 0.75% hyperbaric bupivacaine 1.2–1.6 mL, fentanyl 15 µg, and morphine 100–150 µg. A 1000-mL coload of lactated Ringer’s solution was infused at the time of spinal injection. Patients were moved to the supine position accompanied by left lateral tilt. Surgery was commenced after verifying a bilateral T5 sensory block to pinprick. After infant delivery and umbilical cord clamping, all patients received a 1–2 U bolus dose of intravenous oxytocin. The obstetrician manually assessed uterine tone. As per institutional protocol, if uterine tone was deemed inadequate, additional oxytocin intravenous boluses (1–2 U) were administered. If a total of 6 U oxytocin was given and the uterine tone remained inadequate, a second-line uterotonic agent was used (methylergonovine maleate, carboprost, or misoprostol). The choice of drug administered was at the discretion of the anesthesia and obstetric teams. Surgical interventions for PPH management were at the discretion of the obstetrician.
Intraoperative blood loss was measured using 3 different modalities: a visual estimate, a gravimetric approach, and photometric analysis. Regarding visual estimates, the attending obstetrician and anesthesiologist were asked independently to estimate blood loss, hereafter referred to as the aBL (the anesthesiologist’s blood loss estimate) and oBL (the obstetrician’s blood loss estimate). Both the obstetrician and anesthesiologist were able to assess all blood-soaked materials and fields in the operating room, including laparotomy sponges, suction canister contents, and blood around the surgical field on surgical drapes. Both physicians were blinded to the gravimetric results and photometric analysis. Blood loss measurements using a gravimetric approach (hereafter referred to as gBL) were performed by 2 study investigators (K.F., K.M.S.) not involved in patient care. gBL was determined using a previously described technique and formula.14 In summary, gBL was quantified by summing the following measurements: weight of blood-soaked laparotomy sponges (measured using electronic scales, subtracting the dry weight of the sponge), an estimate of blood volume in the suction canister, and an estimate of blood loss in and around the surgical field after completion of surgery. After the hysterotomy incision, the surgical team suctioned amniotic fluid from the surgical field and a study investigator placed a mark corresponding with the level of amniotic fluid in the canister. To estimate the volume of blood in the canister, we deducted the amniotic fluid volume from the total volume of the fluid containing amniotic fluid and blood. The blood volume in the canister was also made available to the obstetrician and anesthesiologist who provided blood loss estimates for this study (oBL and aBL, respectively).
Photometric analysis was performed with the Triton System to calculate blood loss (hereafter referred to as tBL). Details of the Triton System have been described previously.11–13 In summary, the Triton System is a platform incorporating mobile computing with Gauss FET software (Gauss Surgical Inc, Los Altos, CA).11 FET uses a patented color density mathematical algorithm to calculate the Hb mass absorbed in surgical sponges. A built-in camera of an iPad (Apple Inc, Cupertino, CA) was used to take images of all surgical sponges used during CD. These images are encrypted and transferred wirelessly to a remote secure server. FET software on the server extracts geometric and pixel-level information from these photos and calculates Hb mass using proprietary classifiers and computational models. FET accounts for extraneous, nonbloody fluids absorbed by the sponge that could dilute the appearance of the sponge’s color, and compensates for the effect of variability in ambient light conditions.15 A cumulative Hb mass for all sponges is quantified, and a correlating Hb volume is reported. The Triton System can also assess the blood volume in a suction canister. A Triton suction canister (Medi-Vac Guardian Suction System, 3000 mL; Cardinal Health Inc, Dublin, OH), equipped with a color correction label and injection molded insert, is used as the sole reservoir for surgical fluid collection. On completion of each CD, photographs of individual suction canisters were taken using the iPad (Apple Inc, Cupertino, CA) and sent to the server. Features of blood in the images are extracted, and the color correction label is used as a standard from which pixels in the blood-containing image are normalized. Features captured from the images are inserted into a mathematical equation to correlate with predetermined Hb concentrations. To calculate Hb mass, the Hb concentration of fluid in the canister is multiplied by the final fluid volume. The Hb mass is divided by the baseline plasma Hb concentration to provide a volumetric estimate for the blood content in the canister.
Our primary outcome was the Hb level measured in the postanesthetic care unit (PACU) within 10 minutes of arrival after CD. We selected PACU Hb as our primary outcome for several reasons. First, we consistently measured the post-CD Hb at the same time point after CD for all study patients. Second, any examination of the relation between blood loss and a later Hb measurement (eg, on postoperative day [POD] 1]) would need to account for blood loss during and after CD. Such examinations may be challenging because post-CD blood loss is typically not measured during uterine fundal checks. Third, assessments of the intraoperative estimated blood loss-POD1 Hb relations may be limited by variability in postpartum intravenous fluid regimens or blood transfusions. Despite these limitations, we acknowledge that providers may prefer to measure the first postpartum Hb level on POD1 Hb because this time point may be close to the nadir Hb. Therefore, we selected POD1 Hb as a secondary outcome.
Other perioperative data collected were preoperative Hb level, body mass index, intraoperative intravenous fluid volumes, blood product use, total dose of oxytocin, and need for second-line uterotonics.
We hypothesized that the correlation between blood loss measurements and post-CD Hb would be highest using the Triton device. To test this hypothesis, we assessed the correlations between blood loss and post-CD Hb using 4 different blood loss measurement modalities (tBL, aBL, oBL, and gBL) and performed direct comparisons of these correlation coefficients. For our secondary analysis, we examined correlations between blood loss with the percentage change in preoperative versus PACU Hb and the Hb level measured on the first POD. We also directly compared the 4 measures of blood loss with a calculated estimated blood loss, the latter of which was based on an algorithm previously described by Stafford et al.16 Calculation of blood loss was derived by multiplying maternal blood volume by the percent of lost blood volume. Maternal blood volume was calculated as 0.75 × ([maternal height in inches × 50] + [maternal weight in pounds × 25]) and percent blood volume lost as ([predelivery Hb – postdelivery Hb]/predelivery Hb).16 Using this algorithm, we made 2 calculations for blood loss. For the postdelivery Hb value in the algorithm, we used PACU Hb in our first calculation of blood loss (hereafter referred to as cBL1) and the Hb on POD1 Hb in our second calculation of blood loss (hereafter referred to as cBL2).
Demographic, obstetric, and intraoperative data are presented as mean (standard deviation), median (interquartile range), or number (%). Continuous data were assessed for normality using normality plots and the Kolmogorov-Smirnov test. Pearson correlation coefficient (r) was used to assess the correlations among blood loss (primary exposure of interest) with postoperative PACU Hb (primary outcome), the percentage change in preoperative versus PACU Hb, and POD1 Hb level (secondary outcomes). For our primary analyses, we calculated 95% confidence intervals (CIs) for the correlation coefficients by performing 1000 bootstrap replications. We assessed whether the difference between each set of correlation coefficients was statistically significant using William t test17 and computed 95% CIs for the estimated difference18; we used the COCOR package in R to perform these tests.19 We also performed univariate linear regression analyses to assess the relations between blood loss estimates with the PACU Hb.
For our secondary analyses, we assessed the correlations between blood loss with the percentage change in Hb values (using pre-CD versus PACU Hb values and POD1 Hb values). We also examined the agreement between cBL1 and cBL2 with each of 4 blood loss measures; we performed Bland-Altman analyses.20 The bias (mean absolute difference between blood loss measured by each modality and cBL1 and cBL2) and limits of agreement (the range in which 95% of the differences are expected to lie) were calculated. All analyses were performed using STATA vsn 12 (StataCorp LLC, College Station, TX), with a conservative P value < .0125 considered as statistically significant (using a Bonferroni correction).
Using a 2-sided hypothesis, we estimated that 50 patients would be required to observe a moderate correlation (r = 0.6) between tBL and postoperative PACU Hb with 80% power and a Bonferroni corrected α = .0125. We also performed a separate power analysis for assessing correlation differences within the same population, with the null hypothesis being that 2 dependent Pearson correlations ρac and ρab are identical. For this power analysis, Xa = PACU Hb, Xb = gBL, and Xc = tBL were considered as continuous variables. Assuming a ρac = 0.6, ρab = 0.4, and ρbc = 0.9, the sample size of 44 patients would be required to detect a statistically significant correlation difference with 80% power and an α = .008 (0.05/6 [which accounts for 6 direct comparisons for all correlation coefficients]). To account for protocol violations or technical difficulties associated with blood loss measurement or blood sampling, we planned to enroll 61 patients for this study.
Sixty-one patients were enrolled between April 2016 and November 2016. Patient characteristics are presented in Table 1. The majority of patients (92%) underwent elective CD. Two patients received red blood cells (RBCs) intraoperatively. One patient who underwent primary CD had a triplet pregnancy complicated by a complete placenta previa with a history of several myomectomies. She experienced PPH (oBL, 2300 mL) due to uterine atony requiring transfusion with 2 units of RBCs. One patient had fibroids and underwent a classical hysterotomy incision related to refractory uterine atony; her oBL was 1500 mL, and she received 2 units of RBCs. Data from women who received intraoperative RBCs were not included in the correlation or regression analyses. No plasma, platelets, and cryoprecipitate were transfused to any patient. No patients received RBC transfusion within the first 24 hours after CD.
Hb levels and blood loss data (aBL, oBL, gBL, and tBL) are presented in Table 2. The median (interquartile range) difference between the preoperative Hb and the PACU Hb was 0.6 (0.3–1.25) g/dL. The median (interquartile range) difference between preoperative Hb and POD1 Hb was 1.9 (1–2.6) g/dL. The median percentage change in preoperative Hb and PACU Hb was −5.3% (−10% to −2.5%). The median percentage change in preoperative Hb and POD1 Hb was −15.6% (−20.7% to −9.2%). Correlations between blood loss measured with the 4 modalities and PACU Hb are presented in Table 3. We observed a statistically significant but weak correlation between tBL with PACU Hb (r = −0.33; 95% CI, −0.59 to −0.07; P = .01), and no statistically significant correlations among aBL, oBL, and gBL with PACU Hb, respectively. We observed no statistically significant differences between any pair of correlation coefficients (Supplemental Digital Content 1, Table 1, https://links.lww.com/AA/C383) among all 4 tested modalities. In the univariable regression analyses, we only observed a significant association between tBL and PACU Hb (β-coefficient, −0.1; 95% CI, −0.2 to −0.02) (Table 4). In each univariable model, there was a nonsignificant inverse relationship between blood loss with PACU Hb, with 100 mL blood loss associated with a 0.05–0.1 g/dL decrease in PACU Hb.
In our secondary analyses, we observed weak correlations among aBL, oBL, and gBL with percentage change in Hb that were statistically significant (Table 3). We also observed statistically significant but weak correlations for all 4 blood loss measures with POD1 Hb (Table 3); tBL had the strongest correlation with POD1 Hb (r = −0.41). Findings from the Bland-Altman analyses for aBL, oBL, gBL, and tBL against calculation of blood loss (based on PACU Hb values) suggest that each modality overestimated cBL1 with wide limits of agreement. Bias and limits of agreement were as follows: aBL versus cBL1 (bias, 451 mL; limits of agreement, −111 to 1014 mL) (Supplemental Digital Content 2, Figure 1, https://links.lww.com/AA/C384); oBL versus cBL1 (bias, 411 mL; limits of agreement, −143 to 968 mL) (Supplemental Digital Content 3, Figure 2, https://links.lww.com/AA/C385); gBL versus cBL1 (bias, 327 mL; limits of agreement, −697 to 1351 mL) (Supplemental Digital Content 4, Figure 3, https://links.lww.com/AA/C386); and tBL versus cBL1 (bias, 198 mL; limits of agreement, −440 to 838 mL) (Supplemental Digital Content 5, Figure 4, https://links.lww.com/AA/C387). Findings from the Bland-Altman analyses for each modality against cBL2 suggest that each modality underestimated cBL2 (based on POD1 Hb values) with wide limits of agreement. Bias and limits of agreement were as follows: aBL versus cBL2 (bias, −65 mL; limits of agreement, −914 to 784 mL) (Supplemental Digital Content 6, Figure 5, https://links.lww.com/AA/C388); oBL versus cBL2 (bias, −125 mL; limits of agreement, −984 to 734 mL) (Supplemental Digital Content 7, Figure 6, https://links.lww.com/AA/C389); gBL versus cBL2 (bias, −191 mL; limits of agreement, −1322 to 940 mL) (Supplemental Digital Content 8, Figure 7, https://links.lww.com/AA/C390); and tBL versus cBL2 (bias, −332 mL; limits of agreement, −1151 to 486 mL) (Supplemental Digital Content 9, Figure 8, https://links.lww.com/AA/C391).
In this prospective study of women undergoing uncomplicated CD, we assessed the correlations between 4 commonly used modalities for measuring blood loss against PACU Hb levels. Among the 4 modalities, only tBL had a statistically significant, albeit weak, correlation with PACU Hb. Moreover, for all 4 modalities, we observed weak correlations with wide CIs and no statistically significant differences between any pair of correlation coefficients. Similarly, statistically significant, albeit weak, correlations were also observed between each modality and POD1 Hb. These findings suggest that, in the setting of uncomplicated elective CD, there may be limited clinical utility in estimating PACU Hb or POD1 Hb with aBL, oBL, gBL, or tBL values.
In a prior observational study examining 70,939 women undergoing CD, severe postpartum anemia (Hb, <8 g/dL) affected 7.3% of the post-CD population.5 These findings, together with the lack of recommendations for anemia screening after CD by any professional obstetric society, suggest that postpartum anemia after CD is an underappreciated phenomenon. Therefore, identifying approaches that alert physicians to screen women for postpartum anemia may be important in limiting or preventing anemia-related morbidity after hospital discharge.21 Studying the relations between blood loss and post-CD Hb in a typical cohort of healthy women undergoing elective CD could provide important baseline data for developing an algorithm for predicting anemia post-CD. For our study, we selected 2 common approaches for measuring blood loss: visual estimation and gravimetric measurement. Gravimetric approaches are recommended in obstetric hemorrhage bundles published by national obstetric agencies, including the Council for Maternal Safety22 and California Maternal Quality Care Collaborative,23 and are firmly rooted in obstetric practices. However, noncalibrated collection bags do not accurately measure the volume of postpartum blood loss.6–8 Furthermore, some experts have questioned whether gravimetric or volumetric measurements can positively impact the timing of PPH diagnosis or the progression of blood loss to severe PPH.6 Gravimetric assessment can be also labor intensive and logistically difficult to perform real time during rapid blood loss.
The Triton System uses an objective method of blood loss measurement and can assess blood content in different containers or materials, including surgical laps and suction canisters. Studies of obstetric and nonobstetric populations have validated the ability of the device to measure Hb content of blood in surgical sponges and suction canisters.11,15,24 Similar to our findings, in a retrospective cohort study comparing gBL versus tBL estimates during CD among 2025 women, the mean tBL value was lower than the gBL (556 vs 662 mL).25 Interestingly, the incidence of PPH (blood loss >1000 mL) was higher in the tBL group than the gBL group (14% vs 3.5%), which provides further evidence that large blood losses may be underestimated using gravimetric estimation. However, no assessments were made to determine whether tBL more accurately predicts post-CD Hb compared to gBL. In our study, we found evidence of a statistically significant correlation between tBL and PACU Hb (r = −0.33); however, the correlation was relatively weak and we observed no significant differences between this correlation and correlations for aBL, oBL, and gBL. These findings are in line with those of Lilley et al10 who also reported a weak correlation between measured blood loss and corrected fall in Hb (r = 0.07) among women with blood losses <1500 mL. We observed similar weak correlations between each modality and POD1 Hb values. These data suggest that, among women undergoing uncomplicated elective CD, intraoperative blood loss may have limited clinical utility in predicting PACU Hb and POD1 Hb. Cost-effectiveness analyses may help determine whether gBL or tBL is preferred to visual estimation.
In our exploratory Bland-Altman analyses assessing the accuracy of each blood loss modality against cBL1, we observed that all blood loss measurements (aBL, oBL, gBL, and tBL) overestimated cBL1 values, with wide limits of agreement. Although the reason is not clear, it is possible that each modality “overestimated” blood loss because measuring Hb in PACU does not allow sufficient time for physiological equilibration. However, based on additional Bland-Altman analyses comparing blood loss with cBL2, we observed that each modality underestimated cBL2 with persistently wide limits of agreement. These findings suggest that there are no substantial improvements in precision when the reference standard is quantified using POD1 Hb values. In some cases, observed blood loss differences were small or negative which raises further questions about the utility of comparing measured blood loss values against calculated blood loss values as a reference standard. Other studies have also reported small or negative values when comparing measured blood loss against a reference standard. Using a simulation model, Lilley et al10 reported small or negative differences when comparing gravimetric or volumetric blood loss against a confirmed volume of artificial blood.10 Similarly, Stafford et al16 reported that the mean (25th, 75th percentiles) for the differences between visual estimates and calculated blood loss after CD was 75 mL (−296 to 530 mL).16 An additional problem using calculated blood loss as the reference standard is that postpartum Hb values may not decrease, and in some cases may increase, if women experience minimal blood losses. In a study of 152 women who underwent vaginal delivery, 55% increased their hematocrit at 48 hours postdelivery.26 In a separate study, among 34 women undergoing CD, Ueland et al27 reported similar mean hematocrit values before delivery, 1 hour postpartum, and on POD1 (34.5%, 34.3%, and 34%, respectively).
We acknowledge that our study has several limitations. Our cohort comprised healthy patients undergoing uncomplicated CD; therefore, blood loss values measured by all modalities were moderate and no patients incurred severe anemia in PACU (ie, Hb <8 g/dL). Thus, future studies should examine the relations between blood loss and post-CD Hb among patients who experience PPH during CD. Of note, Lilley et al10 reported that, among women who experienced PPH, a strong correlation (r = 0.8) existed between gBL with a delta Hb (calculated as Hb on admission minus Hb at hospital discharge).10 To our knowledge, there are no practice surveys reporting when providers order postpartum Hb tests after CD, and few longitudinal data exist of postpartum Hb levels. In an observational study of 98 nonanemic healthy women, Richter et al28 reported that the Hb decreases 0.8 g/dL within 24 hours of delivery and then plateaus for 4 days before increasing by 0.2 g/dL more than the predelivery value on POD 14. Although POD1 Hb levels may approximate the nadir Hb, future studies are needed to identify the ideal time point for measuring Hb after CD. We did not standardize the attending obstetrician and anesthesiologist who supervised each case; therefore, it is possible that variation in blood loss estimation may exist within each set of providers. Last, we used cBL1 and cBL2 as the reference standards in our Bland-Altman analyses. Although this allowed us to assess the agreement between the 4 modalities with calculated blood loss, the latter is not considered the “gold standard” method for quantifying blood loss in current clinical obstetric practice. Therefore, the findings of our Bland-Altman analyses are at best exploratory and should be cautiously interpreted. We did not measure blood loss expressed from the uterus or vagina at the end of surgery in the gBL or tBL estimates. It is also unclear whether “expressed blood” in the operating room was accounted for by physicians in aBL or oBL estimates.
In conclusion, among women undergoing elective CD who do not incur major postpartum bleeding, we observed weak correlation among aBL, oBL, gBL, and tBL with PACU Hb and POD1 Hb values. These findings suggest that blood loss measurements may not be useful for predicting postpartum Hb among women having low-volume blood loss after elective CD.
Name: Kelly Fedoruk, MD, FRCPC.
Contribution: This author helped with the acquisition, analysis, and interpretation of the data; and helped draft the work and revise it critically, and approve the final version of the manuscript.
Conflicts of Interest: None.
Name: Katherine M. Seligman, MD.
Contribution: This author helped with the conception of the work, and acquisition, analysis, and interpretation of the data; helped draft the work and revise it critically, and approve the final manuscript.
Conflicts of Interest: None.
Name: Brendan Carvalho, MBBCh, FRCA, MDCH.
Contribution: This author helped with the conception of the work, and helped interpret the data, revise the draft critically, and approve the final manuscript.
Conflicts of Interest: None.
Name: Alexander J. Butwick, MBBS, FRCA, MS.
Contribution: This author helped with the conception of the work, and acquisition, analysis, and interpretation of the data; and helped draft the work and revise it critically, and approve the final manuscript.
Conflicts of Interest: A. J. Butwick has received an unrestricted gift for research from Gauss Surgical, Inc, Los Altos, CA; a portion of this funding was used to cover costs for hemoglobin tests in this study. He is also supported by an award from the Eunice Kennedy Shriver National Institute of Child Health and Development (K23HD070972).
This manuscript was handled by: Jill M. Mhyre, MD.
1. Gaynes BN, Gavin N, Meltzer-Brody S, et al.Perinatal depression: prevalence, screening accuracy, and screening outcomes. Evid Rep Technol Assess (Summ). 2005;119:1–8.
2. Corwin EJ, Murray-Kolb LE, Beard JLLow hemoglobin level is a risk factor for postpartum depression. J Nutr. 2003;133:4139–4142.
3. Lee KA, Zaffke MELongitudinal changes in fatigue and energy during pregnancy and the postpartum period. J Obstet Gynecol Neonatal Nurs. 1999;28:183–191.
4. Beard JL, Hendricks MK, Perez EMMaternal iron deficiency anemia affects postpartum emotions and cognition. J Nutr. 2005;135:267–272.
5. Butwick AJ, Walsh EM, Kuzniewicz M, Li SX, Escobar GJPatterns and predictors of severe postpartum anemia after cesarean section. Transfusion. 2017;57:36–44.
6. Hancock A, Weeks AD, Lavender DTIs accurate and reliable blood loss estimation the ‘crucial step’ in early detection of postpartum haemorrhage: an integrative review of the literature. BMC Pregnancy Childbirth. 2015;15:230.
7. Toledo P, McCarthy RJ, Hewlett BJ, Fitzgerald PC, Wong CAThe accuracy of blood loss estimation after simulated vaginal delivery. Anesth Analg. 2007;105:1736–1740.
8. Patel A, Goudar SS, Geller SEDrape estimation vs visual assessment for estimating postpartum hemorrhage. Int J Gynaecol Obstet. 2006;93:220–224.
9. Yoong W, Karavolos S, Damodaram MObserver accuracy and reproducibility of visual estimation of blood loss in obstetrics: how accurate and consistent are health-care professionals? Arch Gynecol Obstet. 2010;281:207–213.
10. Lilley G, Burkett-St-Laurent D, Precious EMeasurement of blood loss during postpartum haemorrhage. Int J Obstet Anesth. 2015;24:8–14.
11. Konig G, Holmes AA, Garcia RIn vitro evaluation of a novel system for monitoring surgical hemoglobin loss. Anesth Analg. 2014;119:595–600.
12. Holmes AA, Konig G, Ting VClinical evaluation of a novel system for monitoring surgical hemoglobin loss. Anesth Analg. 2014;119:588–594.
13. Konig G, Waters JH, Hsieh EIn vitro evaluation of a novel image processing device to estimate surgical blood loss in suction canisters. Anesth Analg. 2018;126:621–628.
14. Butwick A, Hilton G, Carvalho BNon-invasive haemoglobin measurement in patients undergoing elective caesarean section. Br J Anaesth. 2012;108:271–277.
15. Sharareh B, Woolwine S, Satish S, Abraham P, Schwarzkopf RReal time intraoperative monitoring of blood loss with a novel tablet application. Open Orthop J. 2015;9:422–426.
16. Stafford I, Dildy GA, Clark SL, Belfort MAVisually estimated and calculated blood loss in vaginal and cesarean delivery. Am J Obstet Gynecol. 2008;199:519.e1–519.e7.
17. Williams EJThe comparison of regression variables. J R Stat Soc Series B Methodol. 1959;21:396–399.
18. Zou GYToward using confidence intervals to compare correlations. Psychol Methods. 2007;12:399–413.
19. Diedenhofen B, Musch JCocor: a comprehensive solution for the statistical comparison of correlations. PLoS One. 2015;10:e0121945.
20. Bland JM, Altman DGStatistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1:307–310.
21. Prabhu M, Bateman BTPostpartum anemia: missed opportunities for prevention and recognition. Transfusion. 2017;57:3–5.
22. Main EK, Goffman D, Scavone BM, et alNational Partnership for Maternal Safety; Council on Patient Safety in Women’s Health Care. National Partnership for Maternal Safety: consensus bundle on obstetric hemorrhage. Obstet Gynecol. 2015;126:155–162.
23. California Maternal Quality Care Collaborative. Obstetric Hemorrhage Toolkit: Improving Health Care Response to Obstetric Hemorrhage. Available at: https://www.cmqcc.org/ob_hemorrhage/ob_hemorrhage_compendium_of_best_practices
. Accessed March 16, 2018.
24. Doctorvaladan SV, Jelks AT, Hsieh EW, Thurer RL, Zakowski MI, Lagrew DCAccuracy of blood loss measurement during cesarean delivery. AJP Rep. 2017;7:e93–e100.
25. Rubenstein AF, Zamudio S, Al-Khan A, et al.Clinical experience with the implementation of accurate measurement of blood loss during cesarean delivery: influences on hemorrhage recognition and allogeneic transfusion. Am J Perinatol. 2017 [Epub ahead of print].
26. Gharoro EP, Enabudoso EJRelationship between visually estimated blood loss at delivery and postpartum change in haematocrit. J Obstet Gynaecol. 2009;29:517–520.
27. Ueland KMaternal cardiovascular dynamics. VII. Intrapartum blood volume changes. Am J Obstet Gynecol. 1976;126:671–677.
28. Richter C, Huch A, Huch RErythropoiesis in the postpartum period. J Perinat Med. 1995;23:51–59.