Keywords cesarean delivery - blood loss quantification - postpartum hemorrhage - quality improvement
Quantification of blood loss (QBL) is recommended over visual estimation of blood
loss for women having cesarean delivery.[1 ]
[2 ] QBL is thought to enhance the accuracy of blood loss measurements, improve hemorrhage
recognition, and allow for appropriate activation of treatment stage–based protocols
which may consequentially improve patient outcomes.[3 ]
A common way of measuring QBL is the gravimetric method that involves weighing of
soiled sponges and measurement of fluid in calibrated suction canisters.[4 ] After adjusting for the dry weight of soiled materials, the estimated amount of
amniotic fluid and the amount of surgical irrigation are deducted from the total weight,
presumably resulting in an accurate measure of blood loss. While intuitively sensible,
technical problems obtaining accurate measurements, difficulty in estimating amniotic
fluid, and retention of irrigation fluid in the abdomen can lead to inaccurate results.[5 ]
[6 ] Moreover, some blood loss may not be captured on surgical sponges and in suction
canisters leading to additional error.
In a previous publication that compared visually estimated blood loss, gravimetric
QBL, and a colorimetric method of estimating blood loss to a reference hemoglobin
assay measurement of the amount of blood on surgical sponges and in suction canisters
during scheduled cesarean delivery, we found that gravimetric QBL measurements were
poorly correlated to the more accurate reference hemoglobin assay results.[7 ] The aim of this secondary analysis was to determine the sources of error and limitations
of the gravimetric QBL method in greater detail, using both correlations and Bland–Altman
measures of agreement.
Materials and Methods
The original protocol was approved by the Santa Clara Valley Medical Center Institutional
Review Board (San Jose, CA; reference no.: 12–003; August 12, 2013). The details of
the study methodology have been previously described.[7 ] Briefly, all sponges and calibrated surgical canisters from 50 patients having scheduled
cesarean deliveries were evaluated, and relevant patient and procedural information
was collected. The original study compared visually estimated blood loss, gravimetric
QBL, and a colorimetric blood loss estimation to a reference hemoglobin extraction
assay during scheduled cesarean delivery. This secondary analysis is focused on the
relationship between gravimetric QBL and the reference hemoglobin assay.
Quantitative gravimetric measurement methods were adopted from published guidelines.[4 ] At the time of the uterine incision, the surgical technician or circulating nurse
recorded the canister volume using the graduated markings. After aspiration of amniotic
fluid, a second measurement was made, and the difference was recorded as the estimated
amniotic fluid volume. At the conclusion of the surgery, the technician recorded the
total amount of any irrigation fluid used. Additionally, any blood loss expressed
by uterine massage was collected and transferred to one or more sponges. Immediately
following the case, all sponges and suction canisters were individually weighed using
a calibrated digital scale, and dry sponge weights and canister weights were subtracted.
To determine the total QBL estimate, all individual sponge and canister measurements
were tallied, and the measured amount of amniotic fluid and irrigation fluid used
was subtracted from the total weight.
For the reference hemoglobin extraction assay method, all soiled laparotomy sponges
and calibrated suction canisters were transferred to an on-site benchtop facility
for hemoglobin measurement at the completion of each procedure.[7 ] Sponges were individually soaked in 400 mL of normal saline, compressed by hand
for 60 seconds to a mean weight of 50 g. This process was repeated four times and
the hemoglobin concentration of the final extraction fluid was measured using the
plasma/low spectrophotometer (Hemo-Cue AB, Ängelholm, Sweden). The yield of the extraction
process was independently characterized by depositing banked blood on sponges in known
quantities and performing the same mechanical extraction. A linear regression analysis
revealed mean hemoglobin recovery rates of 89.5% (95% confidence interval [CI]: 86.8–92.1%)
for individual sponges (n = 116) and this value along with the patient's preoperative hemoglobin was used to
convert the extracted amount of hemoglobin (g) to the amount of blood in the sponges
(mL).
The hemoglobin concentration in the canisters was separately assayed by using either
a whole blood or low-concentration hemoglobin analyzer and converted to a canister
blood volume based on the patient's preoperative hemoglobin concentration (in g/dL).[7 ] The blood loss in the canisters was then combined with the blood loss from the sponges
to give a total assayed blood loss. The blood and fluid that remained on the surgical
drapes, gowns, towels, and other materials was not accounted for by either the quantitative
gravimetric assessment or the hemoglobin assay.
Statistical Analysis
The sample size was based on the previous study. The distributions of the blood loss
measurements and other analysis quantities were assessed using descriptive univariate
statistics: the mean and standard deviation ( ± ), as well as the median and the
first and third quartiles. Since the primary outcome of this paper is the accuracy
of the gravimetric method compared with the hemoglobin assay method (i.e., considered
to be the accurate reference value), the absolute difference between the hemoglobin
assay and gravimetric methods was assessed instead of the signed difference between
the measures.
Estimates of sensitivity and specificity used a binomial model. Since only four cases
were observed with assayed blood loss greater than 1,000 mL, CIs for sensitivity were
computed using the exact small-sample formula. CIs for specificity used the standard
large sample formula for the binomial distribution.
The strength of the relationship between the gravimetric and assay methods was assessed
by correlation coefficients; Fisher's z -transformation was used to compute CIs for correlation coefficients. A quantitative
relationship between the gravimetric and assay methods was derived using the nonparametric
regression method of Passing and Bablok.[8 ] The Passing–Bablok regression line provides a nonparametric estimate of the linear
relationship between two measures that assumes both are subject to error, without
making further assumptions about the statistical distributions of the measures. The
strength of correlations was assessed as follows: r < 0.5 (low); r = 0.5 to < 0.7 (moderate), r = 0.7 to < 0.9 (high); and r ≥ 0.9 (very high). For comparative analysis, a p -value of 0.05 was considered significant.
In addition, agreement between the extraction assay and gravimetric QBL was evaluated
using the Bland–Altman method.[9 ] The Bland–Altman bias (mean the difference between the two measures) and upper and
lower limits of agreement (mean ± 1.96 × SD) with the respective 95% CIs was computed.
All statistical computations were computed using SAS, Version 9.4 (SAS Institute,
Cary, NC).
Results
Data were successfully collected from all 50 patients.[7 ] The mean preoperative hemoglobin level was 12.2 ± 1.0 g/dL. Additional patient characteristics
were described in the previous publication.[7 ]
The gravimetric QBL method was moderately correlated to the actual (hemoglobin assayed)
blood loss (r = 0.564; mean absolute difference = 422 ± 330 mL; median absolute difference = 365;
[204–538] mL; p < 0.001).[7 ] The 95% CI for the correlation coefficient was [0.334, 0.725]. As previously reported,[7 ] the assessment of agreement between gravimetric QBL and the extraction assay revealed
a bias (95% CI) of 353 ± 405 mL. The upper limit of agreement (95% CI) was 1,145 mL
and the lower limit of agreement (95% CI) was −441 mL. The Bland–Altman plot is represented
in [Fig. 1 ].
Fig. 1 Assessment of agreement between gravimetric QBL and the extraction assay method according
to the Bland–Altman method. QBL, quantification of blood loss.
Compared with the hemoglobin assayed amount, gravimetric QBL overestimated blood loss
for 44 of the 50 patients (88%). In 32 patients (64%) the overestimate was greater
than 250 mL and for 14 patients (28%), the overestimate was greater than 500 mL. In
two instances, there was an underestimation of more than 500 mL when compared with
the assay (528 and 784 mL, respectively). Collectively, gravimetric QBL was only within
250 mL of the actual amount in 16 (32%) patients and all but three of these patients
had an assayed blood loss of less than 500 mL (mean = 378 ± 179 mL). The Passing–Bablock
regression line ([Fig. 2 ]) demonstrates that gravimetric QBL is typically about twice the blood loss determined
by the hemoglobin assay method.
Fig. 2 Assayed blood loss versus gravimetric QBL with Passing–Bablock regression line. QBL,
quantification of blood loss.
Assayed blood loss was greater than 1,000 mL in four patients. For three of the four
patients gravimetric QBL recognized that the blood loss was greater than 1,000 mL
(sensitivity = 75%; 95% CI: 0.194–0.932) but overestimated it by 617, 533 and 255 mL,
respectively. For the fourth patient the assayed blood loss was 1,041 mL but the gravimetric
QBL was only 513 mL. Seventeen patients (34%) had an assayed blood loss of greater
than 500 mL. For those 17 patients, the mean absolute difference between the QBL measurement
and the assay was 549 ± 420 mL (median = 528; [275–696] mL). Gravimetric QBL overestimated
the assayed blood loss in 13 of the 17 patients (mean overestimation = 597 ± 445 mL)
and underestimated it in 4 (mean underestimation = 394 ± 326 mL). Only 8 of the 17
(47%) values were within 500 mL and quantitative QBL was within 250 mL of the assayed
result in only 3 (18%) cases.
Gravimetric QBL was greater than 1,000 mL (the American College of Obstetricians and
Gynecologists [ACOG] criteria for hemorrhage recognition[10 ]) in 12 patients. While three of these had an assayed blood loss of greater than
1,000 mL, 9 of the 46 patients (20%) with blood losses of less than 1,000 mL by the
hemoglobin assay were incorrectly identified as having postpartum hemorrhage (false
positives). The specificity of gravimetric QBL was 80.4% (95% CI: 0.690–0.919). In
37 cases (74%), there was concordance among both methods that the blood loss was less
than 1,000 mL. These relationships are illustrated in [Table 1 ]. Interestingly, if the criteria for hemorrhage recognition was 900 mL rather than
1,000 mL, gravimetric QBL would have recognized hemorrhage in 16 patients, only 3
of whom had an assay of greater than 900 mL. Thus, 13 of the 46 patients (28%) with
blood losses of less than 900 mL would have been incorrectly identified as having
postpartum hemorrhage. Conversely, if the criteria for hemorrhage recognition were
increased to 1,100 mL, the false positive rate would be 8 of 47 patients (17%).
Table 1
Distribution of blood loss by quantitative QBL and the reference assay (mL)
Assay ≤ 1,000 mL
Assay > 1,000 mL
QBL ≤ 1,000 mL
37
1
QBL > 1,000 mL
9
3
Abbreviation: QBL, quantification of blood loss.
Corrections for amniotic fluid and irrigation were collectively applied to the sponge
and calibrated canister quantified amounts. The measured amount of amniotic fluid
varied from 100 to 2,100 mL (mean = 632 ± 507 mL; median = 500 mL; [250–850]). The
measured amount of irrigation varied from 50 to 2500 mL (mean = 760 ± 437 mL; median = 775 mL;
[400–1,000]).
An average of 15 surgical sponges were used per case (mean = 15.1 ± 4.9; median = 14.5;
[12–17]). The gravimetric QBL on the sponges was substantially higher than the hemoglobin
assayed amount of blood on the sponges in every case (mean absolute difference = 480 ± 182 mL).
The number of sponges used was highly correlated with the total assayed blood loss
on the sponges (r = 0.748; 95% CI: [0.588, 0.847]; p < 0.0001; i.e., the use of more sponges was associated with a higher assayed blood
loss).
A single-calibrated suction canister was utilized in 49 cases; the remaining case
required use of two canisters. The assayed amount of blood in the canisters was substantially
less than the total volume of fluid collected in the canisters (mean absolute difference = 1,264 ± 627 mL).
The correlation between the total amount of fluid in the canisters and the actual
amount of blood in the canister was low to moderate (R
2 = 0.466; 95% CI: [0.211, 0.656]). Irrigation and amniotic fluids affected both the
sponge and canister values. Assuming that all of the amniotic fluid was aspirated
into the canisters, 88% of the irrigation fluid went into the canisters and only 12%
was absorbed by the sponges ([Fig. 3 ]).
Fig. 3 Difference in canister volume (post–pre) versus irrigation volume (mL) with Thiel
regression line.
Discussion
Our study found that despite the use of trained technicians to carefully perform gravimetric
QBL using recommended guidelines, the results deviated substantially from the actual
(hemoglobin assayed) amounts of blood on the soiled sponges and in the surgical canisters
by greater than 250 mL in 68% of cases and by greater than 500 mL in 32% of cases.
While overestimation was more common than underestimation, both occurred. Detection
of hemorrhage (blood loss greater than 1,000 mL)[9 ] was aligned in 40 patients (80%); however, in 10 cases (20%), the results were not
aligned. If quantitative QBL were used to implement a hemorrhage protocol,[4 ] the protocol would have been inappropriately applied in the unaligned group.
Postpartum hemorrhage is the leading preventable cause of maternal mortality both
in the United States and worldwide.[11 ]
[12 ] Since treatment is often complicated by delayed or imprecise recognition, accurate
and timely measurement of the actual amount of bleeding is recommended to identify
hemorrhage early and institute protocol-based treatment plans that have led to improved
outcomes.[4 ]
[13 ]
During vaginal delivery, blood loss is captured by calibrated under-buttocks drapes,
surgical sponges and other substrates such as towels and pads that can be accurately
weighed. Estimation of the volume of amniotic fluid is relatively simple since the
collection of amniotic fluid precedes most of postpartum bleeding. During cesarean
delivery, however, there is mixing of amniotic and irrigation fluid with maternal
blood loss, further complicating the measurement of maternal blood loss.
Strengths and Limitations
Strengths and Limitations
The strength of our analysis is that the actual amount of blood loss was diligently
assayed creating a blood loss “gold standard” for comparison. The major limitation
was that only four of the patients studied (8%) had a hemoglobin assayed blood loss
of greater than 1,000 mL, fulfilling the criteria for postpartum hemorrhage.[13 ] The large differences in patients whose hemoglobin assayed blood loss was greater
than 500 mL and greater than 1,000 mL demonstrate that the inaccuracy of quantitative
QBL is substantial at higher amounts of blood loss.
Visual estimation of blood loss is known to be inaccurate.[14 ]
[15 ] Therefore objective measures, such as gravimetric QBL, are recommended and promoted.[1 ]
[2 ] This analysis demonstrates significant inaccuracies in the gravimetric QBL method,
suggesting alternative objective measures to accurately measure blood loss during
cesarean delivery are needed. Several studies have shown that the colorimetric method
is a practical and reliable way of accurately estimating maternal blood loss in the
operating room.[7 ]
[16 ] Using the change in hemoglobin from the preoperative level to the value following
delivery, Saoud et al confirmed the accuracy of the colorimetric method.[17 ] Katz et al recently demonstrated that quantifying blood loss using the colorimetric
method resulted in improved identification of postpartum hemorrhage, changes in patient
management and cost savings.[18 ] Further studies regarding the utility of accurate QBL using the colorimetric method
and its effect on reduction of maternal morbidity and mortality are needed.
Conclusion
In summary, this study demonstrates that quantitative QBL is only moderately correlated
with actual blood loss and is potentially inaccurate when bleeding is substantial.
These inaccuracies were apparent despite a dedicated assessor of gravimetric QBL and
optimal methodology within a study. Further studies evaluating the quantitative method
in patients with high-risk deliveries and in emergency cases with higher amounts of
blood loss, as well as comparisons with alternative methodologies and effects on clinical
outcomes, are needed.
