Volume 85, Issue 12 p. 1448-1452
Free Access

Estimation of blood loss after cesarean section and vaginal delivery has low validity with a tendency to exaggeration

CHRISTINA LARSSON MD

Corresponding Author

CHRISTINA LARSSON MD

Division of Obstetrics and Gynecology, Karolinska Institutet, Danderyd Hospital, SE-182 88, Stockholm, Sweden

: Christina Larsson, Division of Obstetrics and Gynecology, Karolinska Institutet, Danderyd Hospital, SE-182 88, Stockholm, Sweden [email protected]Search for more papers by this author
SISSEL SALTVEDT

SISSEL SALTVEDT

Stockholm South General Hospital, Sweden

Search for more papers by this author
INGELA WIKLUND

INGELA WIKLUND

Division of Obstetrics and Gynecology, Karolinska Institutet, Danderyd Hospital, SE-182 88, Stockholm, Sweden

Search for more papers by this author
SARA PAHLEN

SARA PAHLEN

University of Linkoping, Sweden

Search for more papers by this author
ELLIKA ANDOLF

ELLIKA ANDOLF

Division of Obstetrics and Gynecology, Karolinska Institutet, Danderyd Hospital, SE-182 88, Stockholm, Sweden

Search for more papers by this author
First published: 31 December 2010
Citations: 110

Abstract

Background. Excessive bleeding is one of the major threats to women at childbirth. The aim of this study was to validate estimation of blood loss during delivery. Methods. Bleeding was estimated after 29 elective cesarean sections and 26 vaginal deliveries and compared to blood loss measured by extraction of hemoglobin using the alkaline hematin method, according to Newton. Results. Inter-individual agreement of estimation showed good results. Estimated loss in comparison with measured loss resulted in an over-estimation. In vaginally delivered women, there was no correlation between estimated and measured blood loss (r2 = 0.13), and in women delivered by elective cesarean section, the correlation was moderate (r2 = 0.55). Agreement, according to Bland and Altman, indicated that measured blood loss could vary from 570 ml less to 342 ml more than estimated blood loss. Conclusions. The standard procedure of estimation of obstetric bleeding was found to be unreliable. In this study, blood loss was over-estimated in cesareans. In vaginal deliveries, there seemed to be no correlation. Estimated blood loss as a quality indicator or as a variable in studies comparing complications must be used with caution. For clinical purposes, estimation of blood loss and measurement of post partum hemoglobin is of low value and may lead to the wrong conclusions.

Introduction

Excessive bleeding after delivery is one of the top five causes of maternal mortality in developed and developing countries. Although pregnancy leads to an increase of the blood volume by roughly 1l, control of hemorrhage is necessary. Bleeding from lacerations or an atonic uterus may be profuse and amount to a few liters within minutes. This can lead to life-threatening situations, such as disseminated intravascular coagulation.

Anemia increases the risk of infection and interferes with recovery after delivery (1). It may, therefore, impair early infant contact during this important period (2).

Blood loss may be estimated in different ways. The content of drainage bottles can be measured. Swabs, pads and diapers can be weighed. In most deliveries, both vaginal and caesarean sections, blood loss is also estimated visually.

Estimation of blood loss in surgery corresponds reasonably well to the actual amount according to some studies (3), but less well according to other (4, 5).

After vaginal delivery, visual estimation is the predominant method. In obstetrics, the estimation is impaired by the amount of amniotic fluid and blood from the placenta.

Few recent studies have evaluated the accuracy of estimation of blood loss in obstetrics. Estimation correlated relatively well with small bleedings but not with large (6). Bleeding was usually underestimated after vaginal delivery (7–9).

Massive hemorrhage was only noticed if blood pressure and pulse were affected (10).

Studies of the estimation of hemorrhage after cesarean section are few (11, 12), and not relevant due to the change in anesthesiologist and surgical techniques during recent years (13). No study has compared the validity of estimation of blood loss in different modes of delivery, which is of interest when comparing complications.

The aim of this study was to validate estimation of blood loss after vaginal delivery and elective cesarean section, and to investigate if post partum hemoglobin reflects blood loss.

Material and methods

Women undergoing elective or semi-acute cesarean section or vaginal delivery were included. As the obstetric procedure was not to be studied, but rather the assessment of blood loss, both primiparas and multiparas were included. All cesareans were performed under spinal anesthesia. After cesarean delivery, blood loss was estimated by the obstetrician and the anesthetic nurse together. The content of drainage bottles was measured and added. After vaginal delivery, the blood loss was visually estimated by the midwife according to the routine of the hospital. When there were difficulties in estimating the blood loss or when the loss visually exceeded 500 ml, pads, swabs and diapers were weighed. These estimations represent standard procedure.

Initially, to compare inter-individual variation of estimation of blood loss, two skilled midwives estimated the blood loss in 10 vaginal deliveries independent of one another. Blood loss was also estimated in 16 cesarean sections simultaneously by 2 persons, independent of one another (the obstetrician and anesthetic nurse on duty).

To study the validity of estimation, estimated blood loss was compared to the measured amount. For this purpose, results from another 29 women delivered by cesarean section (s) and 26 women delivered vaginally (v) were analysed. Estimation of blood was performed as described above. Measurement of blood loss was performed using the alkaline hematin method (14). All the blood-stained pads, diapers and swabs and the content in the drainage bottle were collected, put in a plastic bag and blended with 5% NaOH solution. The plastic bag was then transferred to the Stomacher Lab Blender (Model 3500, Seward Laboratories, London, UK) and processed for a few minutes. In this way, the hemoglobin was extracted. A portion of the fluid was collected and diluted with 5% NaOH solution. The concentration of alkaline hematin was obtained by assay in a spectrophotometer at 546 nm with the appropriate NaOH as a blank. The intra-assay coefficient for analysing the concentration is 1%. The blood loss was then calculated using the patient's hemoglobin at admittance as a reference. On admittance, as well as 3–7 days postpartum, the level of hemoglobin in the blood was analysed in all participants.

The study was approved by the local Medical Ethics Committee and the subjects gave their informed consent.

Statistical methods

Intra-class correlation coefficient [ICC] was calculated with the SSPS program to assess the inter-observer repeatability (15).

All other statistical analyses of the data were performed using JMP statistical package (SAS Institute Inc., JMP Sales, SAS Campus Drive, Cary, NC 27513, USA).

Blood loss was not normally distributed and, therefore, the Wilcoxon two-sample test was used. The co-variation between variables was assessed by bivariate linear regression. Agreement between methods was analysed according to Bland and Altman (16).

Results

When the blood loss was estimated simultaneously by two persons, the ICC was 0.92 (95% CI 0.70–0.98) in vaginal deliveries and 0.97 (95% CI 0.91–0.99) in cesarean sections. Both correlations are significant with p-values <0.001.

As for the comparison of estimated and measured blood loss, results from another 29 women delivered by cesarean section (s) and 26 women delivered vaginally (v) were analysed. In all, 20 midwives were involved in the vaginal deliveries and 7 obstetricians in estimating the blood loss of the cesarean sections. Women in this latter part of the study delivered by cesarean section were older, 34 years (CI 95% 32.3–35.1) versus 30 years (CI 95% 28.9–31.8), and gestational age was shorter, 270 days (CI 95% 269–275) versus 281 days (CI 95% 278–284). As expected, the gestational age was skewed to the left in women who delivered by elective cesarean section as opposed to those who delivered vaginally. Parity did not differ: there were 12 primiparas out of 29 (s) versus 17 out of 26 (v).

The median blood loss in the vaginal group was 325 ml (200–1300 ml) according to estimation and 254 ml (102–715) according to the hemoglobin extraction method (p = 0.07). The median blood loss in the cesarean group was 500 ml (200–1500 ml) according to estimation and 440 ml (135–1000) according to the hemoglobin extraction method (p = 0.1). Estimated blood loss for the whole group (n = 55) was 400 ml compared to the measured loss of 370 ml (p = 0.05) (Figure 1). Regression showed a moderate correlation (r2 = 0.55) between estimated and measured blood loss in the (s) group (Figure 2). In the (v) group (r2 = 0.13), there was no correlation (Figure 3). Plotting the difference between estimated and measured blood loss against measured blood loss, there was no correlation (r2 = 0.001) (Figure 4).

Details are in the caption following the image

Estimated and measured blood loss in women delivered by cesarean section and vaginal delivery. Values are given as median and percentiles with the ends of the box as the 25th and 75th percentiles and the 10th and 90th percentiles indicated.

Details are in the caption following the image

Estimated and measured blood loss in women delivered by cesarean section. Estimated blood loss is plotted as the dependent variable and measured as the independent.

Details are in the caption following the image

Estimated and measured blood loss in women delivered vaginally.

Details are in the caption following the image

There was no correlation between the difference of estimated and measured blood loss and measured.

Agreement between the methods, according to Bland and Altman (16), showed that estimation tended to over-estimate the bleeding by a mean of 114 ml (SD 228). This indicates that the true result of measured blood loss could be between −570 (−2 SD) and +342 (+2 SD) ml from the estimated Figure 5. For vaginally delivered only, mean would be −101 (SD 240) indicating a range from −581 to +379. For women delivered by cesarean section, a mean of −26 (SD 220) indicating a range of −566 to +314.

Details are in the caption following the image

Mean of difference between measured and estimated blood loss for all women is −114 (SD 228) indicating that the difference could vary between −570 (−2 SD) and +342 (+2 SD) ml.

The decline of hemoglobin (difference between hemoglobin analysed at admittance to the delivery ward and hemoglobin 3–7 days post partum) did not correlate to the result of blood loss (r2 = 0.05). Only 17 patients were available for this analysis.

Discussion

The alkaline hematin method has been used in other studies (3). The Stomacher Lab-Blender simplifies the procedure and measurements of known volumes of blood in pads have been excellent (14). The measured amount in our study is, therefore, considered to be ‘the true amount’. Errors can occur if all pads are not collected. Since estimation was performed on the same occasion as measurement, this was less likely to occur.

There was good agreement between estimations performed by different persons in the same hospital. Since estimation of blood loss is routine in our hospital, there are guidelines to assist the staff even if no written instructions are at hand.

On the other hand, estimation in comparison with measured blood loss resulted in an over-estimation of blood loss in our study. This indicates that clinical routine has to be checked with objective methods. In women delivered by cesarean section, measured blood loss correlated better with estimated loss than in women delivered by vaginal delivery, where there was no correlation. In the first case, blood loss was estimated by the obstetrician and the anesthetist nurse, and in the latter, by the midwife. Anesthetic nurses, who are used to various surgical procedures, may be more precise in their estimation, while midwives may be biased, expecting the blood loss to vary less in normal deliveries.

The interpretation of correlation coefficients when comparing two methods is difficult. Therefore, a method recommended by Bland and Altman was also used, showing estimation to be inaccurate both in the cesarean and the vaginal group. Other studies have shown an over-estimation in small bleedings and an under-estimation in larger bleedings (10), or an under-estimation when compared to measured (8, 12). The conclusion of these three studies is that estimation of blood is grossly inaccurate. This indicates that our findings are generally applicable. Routines on how to estimate blood loss may also differ from hospital to hospital and even between different professions. This indicates that how estimation is carried out may vary in different hands and locations. Using estimated blood loss as a variable in quality assurance may therefore be problematic. Likewise, comparisons of studies should be performed with caution.

Bleedings over 1000 ml were few in this small study. The alkaline hematin method is also less suitable for excessive bleedings. Previous studies have shown that large bleedings are detected only if blood pressure and pulse are affected (6, 10).

No correlation could be seen between measured blood loss and decline of hemoglobin. Only a few patients were available for the analysis, but this agrees with a previous study, where post partum hemoglobin of almost 700 patients were compared with estimated blood loss during delivery (17). After delivery, changes in body fluids occur which may partly explain this lack of correlation. Post partum hemoglobin, therefore, seems to be of limited value.

In summary, estimation of blood loss in association with either vaginal delivery or cesarean section is imprecise, and attention should be paid to this fact in the clinical situation. Vigilance in monitoring blood loss after delivery is, however, mandatory. Estimation is the only simple method available, but its inaccuracy must be kept in mind.

Acknowledgement

The study was supported by Praktikertjänst AB, Stockholm, Sweden.