Physician and nurse accuracy of estimating blood loss

Do estimates of blood loss differ between student midwives and midwives? A multicenter cross-sectional study

OBJECTIVE

the principal objective of this study was to assess the quality of blood loss estimates by midwives and student midwives. The secondary objectives were: to assess the intraobserver agreement of visual blood estimates and the rate of underestimation of blood loss by participants, and to estimate the sensitivity, specificity, and negative likelihood ratio of these estimates for clinically pertinent blood losses (≥ 500mL and ≥ 1000mL).

DESIGN

multicenter cross-sectional study.

SETTING

thirty-three French maternity units and 35 French midwifery schools participated in this study.

PARTICIPANTS

volunteer French midwifery students (n = 463) and practicing midwives (n = 578).

INTERVENTION

an online survey showed 16 randomly ordered photographs of 8 different simulated blood quantities (100, 150, 200, 300, 500, 850, 1000, and 1500mL) with a reference 50-mL image in each photo and asked participants to estimate the blood loss. The visual blood loss estimates were compared with Fisher's exact test. Intraobserver agreement for these estimates was assessed with a weighted kappa coefficient, and the negative predictive values (probability of no hemorrhage when visual estimate was negative) were calculated from prevalence rates in the literature.

FINDINGS

of the 16,656 estimates obtained, 34.1% were accurate, 37.2% underestimated the quantity presented, and 28.7% overestimated it. Analyses of the intraobserver reproducibility between the two estimates of the same photograph showed that agreement was highest (weighted kappa ≥ 0.8) for the highest values (1000mL, 1500mL). For each volume considered, students underestimated blood loss more frequently than midwives. In both groups, the negative predictive values regarding postpartum hemorrhage (PPH) diagnosis (severe or not) were greater than 98%.

KEY CONCLUSIONS AND IMPLICATIONS FOR PRACTICE

student midwives tended to underestimate the quantity of blood loss more frequently than the midwives. Postpartum hemorrhage (≥ 500mL) was always identified, but severe postpartum hemorrhage (≥ 1000mL) was identified in fewer than half the cases. These results should be taken into account in training both student midwives and practicing professionals.

Reliability of student midwives' visual estimate of blood loss in the immediate postpartum period: a cross-sectional study

BACKGROUND

In France, postpartum hemorrhage (blood loss≥500mL in the first 24h postpartum) is the leading direct obstetric cause of maternal mortality. In French practice, PPH is mainly diagnosed by a quantitative assessment of blood loss, performed by subjective methods such as visual estimates. Various studies have concluded that visual estimates are imprecise, tend to underestimate blood loss, and thus to delay diagnosis of PPH.

OBJECTIVES

The principal objective of this study was to assess the accuracy of visual estimates of blood loss by student midwives. The secondary objectives were to study intraobserver agreement of these assessments, to assess the accuracy of visual estimates for threshold values, and to look for a region effect.

DESIGN

A cross-sectional multicentre study.

SETTING

All French midwifery schools (n=35).

PARTICIPANTS

Volunteer French student midwives at their fifth (final) year (n=463).

METHODS

The online questionnaire contained 16 photographs (8 different, each presented twice) of simulated volumes of blood loss (100, 150, 200, 300, 500, 850, 1000, and 1500mL). A 50-mL reference standard for calibration accompanied each photograph. Only one answer could be selected among the 7 choices offered for each photograph. Comparisons used χ(2) and Kappa tests.

RESULTS

The participation rate was 48.43% (463/956), and 7.408 visual estimates were collected. Estimates were accurate for 35.34% of the responses. The reproducibility rate for the visual estimates (0.17≤к≤0.48) and for the accurate visual estimates (0.11≤к≤0.55) were moderate for 4 of the 8 volumes (100, 300, 1000, and 1500mL). The percentage of accurate responses was significantly higher for volumes≤300mL than for those ≥500mL (52.94% vs. 17.17%, p<0.0001) and those ≥1000mL (52.94% vs. 18.30%, p<0.0001). The percentage of accurate responses varied between the regions (p=0.042).

CONCLUSION

Despite the help of a visual aid, both the accuracy and reproducibility of the visual estimates were low.