BLOODLOSS AND TRANSFUSION NEED IN ORTHOGNATHIC SURGERY:
REVIEW OF LITERATURE
in the estimated blood loss, but the irrigation fluid and
the fluid collected in suction devices was [8,49-53].
In 5 papers/theses, the calculation was based on
a comparison of preoperative and postoperative
hematocrit level [6,42,54,55] or blood volume [56].
The problem with estimation of preoperative and
postoperative hemoglobin and/or hematocrit is that
the timing between the preoperative and postoperative
measurements differed between studies. It varied
between 6 hours postoperatively [38] and 1 week
postoperatively [46]. Ueki et al [46], however, did include
the perioperative blood loss estimation. Hemodilution
may be a problem when measuring the postoperative
hemoglobin level, and the rise in hemoglobin level will
differ according to individual body size.
The postoperative blood loss was not estimated, and the
blood lost in drapes and gowns was not measured. The
blood lost postoperatively in drains or swallowed by the
patient all contribute to the ultimate blood loss that may
affect the decision for blood transfusion. Also, blood lost
in sinuses or tissue spaces was not accounted for [44].
The intraoperative blood loss estimation is always less
than the calculated blood loss [6], but the estimation
accuracy worsens with the amount of blood lost. Böttger
calculated a linear regression where the estimated blood
loss=0.4115 × Calculated blood loss + 406.8. In this
calculation, a 1000 mL estimated blood loss appears
to correlate with 1462 mL calculated blood loss, and a
2000 mL estimated blood loss appeared to correlate
with 4250 mL calculated blood loss [6]. The deviation
between estimation and calculation increases as the
volume increases. The underestimation between the
estimated blood loss and the calculated blood loss also
was observed by Schaberg et al [45] in a radioisotope
study of red blood cells.
3.4. Measurement of duration of surgery
To know the duration of surgery, all papers that define
and measure the duration of surgery in the same manner
are valid candidates to be included if the duration can be
clearly attributed to a specified orthognathic procedure.
The operation time is defined as ‘missing’ if the starting
point of the measurement and the endpoint of the
measurement were not defined. In contrast to ‘age’,
‘weight’, and ‘length’, operation time allows variable
interpretations.
Out of the 51 papers retained for data acquisition, 44
did not mention at all how the duration of surgery was
measured; 2 stated that the duration was recorded or
documented; 2 clearly stated that the operating time
was calculated from the first incision to the last suture
[42,55]; one counted from the beginning of the BSSO
incisions to the last suture [29], and one started from
the injection of the local anesthetic to the last suture
[47]. Clearly, no uniform definition of operation time
exists. Using operation time as a predictive variable, in
the absence of the knowledge about which two points
were used to measure it, creates an important bias when
comparing studies.
3.5. Duration of surgery and blood loss
One would assume that the longer the operation lasts,
the more blood is lost. In bimaxillary surgery, Böttger [6]
indeed found a linear correlation between calculated
46
Table 2. Summary of the entries that were introduced in PubMed,
Scopus and LIMO.
Study Estimate Standard Error
Golia et al [15] -0.37310 0.3161034
Kasahara et al [17] 3.65315 0.7732564
Landes et al [20] 1.46156 0.6044895
Zelllin et al [48] 4.43946 1.1349279
Table 3. Relation bloodloss versus duration of surgery.
Evaluation between-study
heterogeneity
Random Effects Analysis
I 2 Q Df P-value Estimate
92.210 38.510 3 <.0001
2.151
SE p-value
1.124 0.056
I = percentage of variation in study estimates due to heterogeneity; Q = Cochrans Q
statistic. If Q is smaller than the number of degrees of freedom, then the estimated
heterogeneity equals zero (I 2 = 0%); Df = Degrees of Freedom for heterogeneity test;
Random Effects Analysis: DerSimonian and Laird method; SE = standard error
2
blood loss and operation time, but this correlation
was weak for 82 bimaxillary procedures: Spearman
correlation coefficient r = 0.325. Chen et al [31] found
a weak Spearman rank correlation between operation
time and blood loss in 30 mandibular surgery patients
(IVRO set-back and genioplasty). Rummasak et al [43], in
a retrospective review of 208 patients with bimaxillary
orthognathic surgery, reported the following correlation
between blood loss and operative time: blood loss = (2.64
* operative time) + 82.35. They reported a “significant”
relationship because the p value was < 0.001 for R2 =
0.15. However, we should be cautious: R2 = 0.15 signifies
R = 0.387; when attempting to predict one measure from
another, coefficients below 0.40 do not yield a guess
even 10% better than chance [57]. Further, the chart
seems to include all data points, indicating that the whole
population rather than a sample was reported. As a result,
p value reporting would not make sense.
A confusing presentation of blood loss versus operative
time is found in Ueki et al [46]. In regression analysis, y is
what we want to predict or to understand and is sometimes
called the dependent variable. X is called the independent
variable or a predictor. In the equation, we want to predict
blood loss on the basis of operative time. So, operative time
needs to be on the x-axis, as most authors have placed
it [22,29,31,43,55,58]. Because blood loss is a continuous
variable, linear regression was used [59].
Four papers [15,17,20,48] provide data of overall 163
individual patients concerning the peroperative bloodloss
and the duration of surgery (Table 2).
Based on these studies it was possible to estimate the
amount of bloodloss for every minute of surgery: per
minute of operation time the estimated bloodloss is 2.151
mL (= SE 1.124 mL; p = 0.056) (Table 3).
Several articles [6,22,29,31,43,46,58] do find a linear
correlation between blood loss and the duration of
surgery but these data could not be integrated due to
the lack of necessary information. Besides the linear
regres