Table 2: Key components of each study Table 2: Key components of each study Table 2: Key components of each study Table 2: Key components of each study Table 2: Key components of each study Table 2: Key components of each study Table 2: Key components of each study Table 2: Key components of each study Table 2: Key components of each study Table 2: Key components of each study Table 2: Key components of each study Table 2: Key components of each study Table 2: Key components of each study Table 2: Key components of each study Table 2: Key components of each study
Study Author Study Design Study Population Exclusion Criteria Outcome with Definition and Method of Measurement Candidate Predictors, n Participants, n (events) Events per Variable Modelling Method Predictors in Final Model, n Calibration Discrimination Validation Presentation of final prediction model or tool Primary author nterpretation of presented model
Ahmadzia et al., 201814
Retrospective Cohort
Patients who underwent caesarean section delivery
Vaginal delivery and incomplete data for intrapartum and postpartum blood transfusion
Intraoperative and postoperative blood transfusion
Model 1: 16 Model 2: 23
56, 967 (1488)
Model 1: 93 Model 2: 64.7
Logistic
14
Association between predicted probability of transfusion and observed incidence in deciles of the risk score distribution’ Calibration graph provided.
Model 1: AUC 0.77 (95% CI 0.75-0.78) Model 2: AUC 0.83 (95% CI 0.81-0.84)
Internal validation by random split of data
Online Risk Calculator
Confirmatory
Baba et al., 2014 15
Retrospective Cohort
All women with placenta praevia who underwent a caesarean section
None
Blood Transfusion: haemoglobin <6.0g/dL or systolic BP <70mmHg or estimated blood loss reached >2500ml
9
205 (20)
2.22
Logistic
5
None
AUC 0.844 (95% CI 0.731-0.958)
None
Equation
Exploratory
Biguzzi et al., 2011 16
Unclear
All women who underwent vaginal delivery
Age <18 years, Caesarean section, Delivery before the 37th week of gestation, Twin pregnancy, No comprehension of the Italian language, Refusal to sign a written informed consent. Deliveries that occurred on Friday afternoon
Postpartum haemorrhage (blood loss of ≥500ml) measured by visual estimation using graduated basin
16
6011 (1435)
89.69
Logistic
5
Graph plotting predicted probabilities against observed outcomes – inadequate information on how curve was developed
AUC 0.70
Internal Validation with bootstrap sampling
Nomogram
Exploratory
Chi et al., 2016 17 Retrospective Case-control
Parturient patients with complete prenatal examination data
None
Postpartum haemorrhage
21
923 (447)
21.29
Logistic
15
None
None
None
Score assigned
Exploratory
Dunkerton et al., 2018 18
Retrospective cohort
All caesareans at a single university hospital trust in U. K
None
Postpartum haemorrhage (estimated blood loss of ≥ 1000ml, annual staff training with pictorial reference cards)
17
24,230 (2997)
176.29
Hothorn et al’s non-parametric recursive partitioning algorithm
9
None
None
Internal validation by data splitting
Decision tree
Confirmatory
Kim et al., 2017 19
Registry data
Patients with placenta praevia who underwent caesarean section
Coagulopathie-s and incomplete data
Massive transfusion (transfusion of ≥8 units pRBC within 24h after delivery)
12
238 (31)
2.58
Logistic
5
Hosmer-Lemeshow test (p=0.44)
Internal validationAUC (0.84) External validation AUC (0.88)
Internal validation by bootstrapping and External Validation of score model
Scoring system
Confirmatory
Lee et al., 2018 20
Retrospective Cohort
Pregnant women with placenta praevia
Multiple pregnancies, Maternal obstetrical or medical problems other than placenta praevia and Fetal anomalies
Massive haemorrhage defined as blood loss ≥ 2000ml during surgery, postpartum transfusion of four or more pints of packed red blood cells, caesarean hysterectomy or uterine arterial embolization triggered by post-partum bleeding.
16
560 (73)
4.56
Logistic
7
None
AUC (0.856)
None
Sum score
Confirmatory
Prata et al., 2011 21
Prospective Cohort
Women anticipating a singleton vaginal delivery of gestational age >36 weeks
Delivery by caesarean section, missing information on delivery type
Postpartum haemorrhage defined as blood loss greater than 500ml during the first 4 hours after delivery (measured by calibrated drapes)
20
2510 (93)
4.65
Logistic
8
None
None
None
Cumulative scores: number of risk factors vs risk
Exploratory
Rubio-Alvarez et al., 2018 22
Retrospective cohort
Women with singleton pregnancies who had vaginal birth
Antenatal fetal demise, Multiple pregnancies and Gestations <35 weeks
Excessive postpartum blood loss defined as a reduction in haemoglobin (Hb)levels greater than 3.5g/dL and a final Hb <8g/dL between onset of birth and 24 hours after it
14
2336 (43)
3.07
Logistic
6
Hosmer-Lemeshow test (no result reported)
AUC 0.9 (0.85-0.93)
Internal Validation by random-split of data External validation (temporal) AUC of 0.83 (0.74-0.92)
Risk Tool for Excel
Confirmatory
Tsu 1994 23
Case-control
Women with singleton vertex deliveries and spontaneous onset of labour
None
Postpartum blood loss of at least 600ml after a normal, unassisted vaginal delivery
5
653 (151)
30.2
Logistic
5
None
only ROC no AUC
None
Equation
Exploratory
Yoon et al., 2014 24
Prospective Cohort
Singleton pregnancy women with placenta praevia delivered by caesarean section
None
Occurrence of peripartum complications: Perioperative blood transfusion (determined by attending anaesthesiologist during CS when clinical evidence of inadequate oxygen-carrying capacity or ongoing profuse blood loss) or uterine artery embolization (to control moderate uterine bleeding immediately after CS when vital status of the patient was stable) or caesarean hysterectomy (vital status of patient was unstable despite vigorous massive transfusion and haemostatic manoeuvres)
12
110 (38)
2.89
Logistic
6
None
None
None
Scoring model
Exploratory