Shoulder dystocia is an obstetric emergency with serious potential risks for both fetus and mother. It is an uncommon and highly unpredictable event,1,2 with no cost-effective means of prevention for the large majority of women at higher risk. Despite identification of best practice in the 1990s, a review of obstetric brachial plexus injuries in the British Isles concluded that the incidence of obstetric brachial plexus injury had not changed over the last 20 years.3 It is thought that poor fetal outcome is commonly a result of inappropriate management,4–6 excessive traction in particular being implicated in the development of brachial plexus injury. Training might therefore be the most effective means of reducing morbidity and mortality related to shoulder dystocia.
Specific training in force perception has been proposed as a means of decreasing the probability and severity of brachial plexus injuries.5 Low-fidelity doll-and-pelvis mannequins have been available for many years, but these do not simulate the impaction of the shoulder on the symphysis pubis, and they cannot provide feedback on the traction applied. Advent of more life-like mannequins provides an opportunity for training in force control. The aim of this study was to compare effectiveness of training with low- and high-fidelity mannequins for the management of shoulder dystocia.
MATERIALS AND METHODS
This study was part of a large randomized controlled trial (Simulation and Fire-drill Evaluation, SaFE Study), commissioned by the Department of Health of England and Wales. The Simulation and Fire-drill Evaluation Study used a 2×2 factorial design, to evaluate simulation center and local in-hospital training, with and without additional teamwork training (www.safestudy.org.uk). Three simulated obstetric scenarios—eclampsia, postpartum hemorrhage, and shoulder dystocia—were used to investigate the benefits of different training methods; shoulder dystocia was used to measure individual skills and behavior. This paper reports the evaluation of the shoulder dystocia simulation.
At the time the Simulation and Fire-drill Evaluation Study was designed (2003), there were no data of the effectiveness of training upon which to base power calculations. Therefore, the study was powered to detect differences in posttraining knowledge of obstetric emergency management (evaluated by a 185-question multiple-choice examination), rather than performance during simulated emergencies. A sample size of 36 per intervention was estimated to give 89% power to detect a difference of 20 in mean score (α=0.05, estimated within-group standard deviation=35). Therefore, the target recruitment was 144 participants. The present report focuses on a comparison between high- and low-fidelity training; 72 participants in each group should, therefore, give 85% power (α=0.05) to detect a difference in procedure rate of 70% versus 90%.
Participants were recruited from six hospitals in the southwest of England, with delivery rates ranging from 2,500 to 4,600 per year. All midwives (hospital- and community-based), and all doctors working in the maternity departments were considered but were excluded if they had participated at a nationally accredited obstetric emergencies course within the previous 12 months, had participated in the pilot phase of the present study, were trainers for the study, or were on maternity or long-term sick leave. Recruitment took place from September to November 2004 in three hospitals and from December 2004 to February 2005 in the remaining three hospitals. The aim was to recruit 24 staff (eight junior and eight senior midwives (with 5 years or less of experience and with more than 5 years of experience, respectively) and four junior and four senior obstetricians (with three years or less of experience and with more than three years of experience, respectively) from each hospital. Participating hospitals provided lists of eligible staff, stratified by staff group and years of experience. A local midwife approached and consented staff within each hospital. After recruitment, subjects were randomly assigned, using a computer-generated number sequence, to one of four training arms: 1) 1-day local hospital obstetric emergency clinical training course, 2) 2-day local hospital obstetric emergency clinical and teamwork training course, 3) 1-day simulation center obstetric emergency clinical training course, or 4) 2-day simulation center obstetric emergency clinical and teamwork training course. If staff withdrew after giving consent, further members of staff were recruited in an attempt to attain the target number.
Each participant underwent a baseline assessment in which they managed a standardized simulated shoulder dystocia scenario. Up to 3 weeks later, participants attended training, which included a 40-minute practical workshop on the management of shoulder dystocia. Each participant was reevaluated managing a standardized simulated shoulder dystocia up to 3 weeks after training (Fig. 1). All training, regardless of the location or mannequin used, commenced with a discussion of shoulder dystocia management and a demonstration of delivery maneuvers. Every participant then practiced each of the delivery maneuvers during training. Training at the central simulation center used a high-fidelity shoulder-dystocia training mannequin (PROMPT Birthing Trainer, prototype IV, Limbs and Things Ltd, Bristol, UK). The fetal mannequin had anatomical articulated limbs and housed a 20 Hz electronic strain gauge that measured force applied across the neck with a range of 0–250 Newtons (N). Printed feedback of applied force was given to each participant during training (Fig. 2). In contrast, training conducted peripherally used locally available mannequins (simple doll-and-pelvis or S500 Childbirth Simulator, Gaumard Scientific Company, Miami, FL). These models had no means of measuring forces applied during simulation (Fig. 1). Trainers were senior obstetricians and midwives from the local centers. To standardize training, all the trainers attended a Training-the-Trainers day course.
Timing of the pre- and posttraining assessments was determined by factors such as school holidays and availability of the training facilities. Participants were aware they would manage a simulated obstetric emergency but were not aware of its nature. Participants were told they could call for help and that they were allowed to stop the simulation. Each participant was taken individually into a delivery room, given a standardized description of the scenario, and asked to complete the delivery. The participant discovered that there was shoulder dystocia as they assisted the delivery. The prototype PROMPT Birthing Trainer was integrated with a patient-actor on a delivery bed for assessment simulations (Fig. 3), the patient-actor followed a standardized script during scenarios. If help was requested, an assistant attended but would not perform any tasks unless specifically directed and would not take over or complete the delivery. The simulation was continued until delivery of the posterior arm or until the participant chose to stop. The simulation was also halted if the delivery was not completed at 5 minutes, a time chosen because 47% of babies who died after shoulder dystocia in the UK Confidential Enquiry into Stillbirths and Deaths in Infancy had a head-to-body interval of less than 5 minutes.6 Force data and key events (start of simulation, key delivery maneuvers, end of drill) were contemporaneously logged to a specifically designed computer program during each scenario. Simulations were recorded using four ceiling-mounted cameras and a microphone. Images were transferred to a digital recorder.
Immediately after each simulation, the actor subjectively assessed the quality of communication during delivery using a five-point Likert scale: “I felt well informed due to good communication” (strongly disagree=1, disagree=2, neither agree nor disagree=3, agree=4, strongly agree=5). Two trained assessors (senior midwife and obstetrician) independently reviewed each video recording using a checklist of appropriate and inappropriate actions derived from the published literature. The reviewers were blind to the training intervention and timing of the simulation (pre- or posttraining). The reviewers viewed the videoed simulations in different orders randomly generated by computer. The shoulder dystocia simulation was assessed by using the following outcome criteria: 1) achievement or failure of delivery, 2) the head-to-body delivery time, 3) performance of appropriate and inappropriate actions, 4) force applied, and 5) communication.
McNemar tests for change were used to look at overall success of training in terms of increased proportions of successful deliveries and completion of all basic actions. Logistic regressions were used to look at posttraining differences between high- and low-fidelity groups in the proportions of participants who carried out specific actions. Adjustments were made for staff group and interactions between high-fidelity training and staff group were explored. Team training formed part of the original study design, but it was thought unlikely to influence the outcomes studied here. Indeed team training was not found statistically significant in any model and has been excluded from the results below. Where frequencies were small and logistic regression was not possible, the high- and low-fidelity groups were compared by using two-tailed Fisher exact tests. Proportions of participants showing good communication were compared by using a χ2 test. Posttraining delivery times for high-fidelity versus low-fidelity groups were compared by using a Mann-Whitney U test. The effect of high-fidelity training on mean peak force and mean total force were explored by using “censored” normal regression and analysis of variance, respectively. Both variables required prior logarithmic transformation to remove skewness. A 5% level of significance was used throughout. The statistical software used was Stata 8 (StataCorp, College Station, TX).
Ethical approval was granted by the Regional Research Ethics Committee and five Local Research Ethics Committees granted site-specific approval. Research and Development approval was granted by each Healthcare Trust.
There were 975 members of staff working in the six participating hospitals. Although the target had been to recruit 144 staff, this was not achieved; three doctors were unable to attend due to clinical commitments, and a midwife withdrew on the morning of the assessment due to illness (Fig. 4). One hundred and forty staff entered the study, 45 doctors and 95 midwives, 132 of whom completed the posttraining assessment, a dropout rate of 5.7%, all due to illness (Fig. 5).
Six pretraining scenarios were not recorded due to a fault in recording equipment, and three and a half of the posttraining scenario video recordings were overwritten before being saved; 134 of 140 (95.7%) of pretraining videos and 128.5 of 132 (97.3%) of posttraining videos were therefore available for analysis. Some data from the nonvideoed drills were available (call for help, McRoberts' position, suprapubic pressure, evaluation for episiotomy, delivery of the posterior arm, and delivery success or failure, and delivery time) from the computer log of key actions and forces contemporaneously collected during each simulation. Force data from 27 of 140 (19.3%) preevaluation scenarios were unavailable for analysis. No data were recorded for three participants, and a fault occurred with the strain gauge at one hospital. Force data were not available for 3 of 132 (2.3%) posttraining drills because of a malfunction in the force data collection system. In a further eight simulations, the peak force was greater than the upper limit (250 N) of the strain gauge (five pretraining, three posttraining). The peak force in the eight cases was only known to be greater than this, and so the total force applied during the scenario could not be calculated for these participants.
Overall, there was a significant increase in the proportion of successful deliveries after training: 60 of 140 (42.9%) pretraining compared with 110 of 132 (83.3%) posttraining (P<.001, McNemar test for change). There was also an increase in the performance of all basic actions (call for help, McRoberts' position, and suprapubic pressure) after training, from 114 of 140 (81.4%) to 125 of 132 (94.7%) (P=.002).
Training with the high-fidelity mannequin was associated with a greater likelihood of delivery than low-fidelity training, 94% participants compared with 72% (P=.002), and with a shorter head-to-body delivery interval, median delivery intervals 135 seconds versus 161 seconds (P=.004, Mann Whitney U test in those who delivered). The majority of participants who did not deliver the baby appeared unable to gain access to the maternal pelvis and successfully deliver the posterior arm. The high fidelity effect did not differ significantly between different staff groups. Similarly we were unable to show significant interactions for the other variables in Tables 1 and 2.
Training on the high-fidelity mannequin was also associated with a significantly higher chance of delivering the posterior arm after training (P=.001; Table 2). However, those trained on the high-fidelity mannequin were significantly less likely to have called for pediatric support (P=.003; Table 1). They also tended to be less likely to have performed all basic actions and to have evaluated for episiotomy, but these findings were not statistically significant (P=.099 and P=.118, respectively). Fundal pressure (a potentially harmful intervention) was mentioned in 14 of 133 (11%) pretraining simulations and 12 of 129 (9%) after training and was actually performed on five occasions (4%) before training and twice (2%) after training (Table 3). Participants who had force training on the high-fidelity mannequin applied a lower mean peak force than those who did not have force training, but the mean difference was just 10 N, and it was not statistically significant (Table 2). The total force applied during simulations (area-under-the-curve) was significantly lower for those who underwent force training (Table 4). The significance levels shown adjust for staff group differences; no significant interactions were found between high fidelity and staff group.
Communication scores were awarded in all but one of the 140 pretraining scenarios and in all 132 posttraining scenarios. Overall, the number of participants with good communication (score of 4 or greater) increased after training, from 79 of 139 (56.8%) to 109 of 132 (82.6%) (P<.001, McNemar test for change), but there was no difference in the posttraining communication according to the type of mannequin used (P=.697, χ2 test; Table 5).
In addition to the comparison of high- and low-fidelity training, this study provides a guide to the current standard of management of shoulder dystocia across a large health region in England. Baseline data demonstrated that one in 27 used an inappropriate (harmful) maneuver (fundal pressure), almost two thirds failed to call for pediatric support, and 57% were unable to deliver the baby. Two UK Governmental surveys produced similar findings: the National Health Service Litigation Authority (NHSLA) reported that fundal pressure was performed in 7% of inappropriately managed deliveries complicated by brachial plexus injury4 and the fifth Confidential Enquiry Into Stillbirths and Deaths in Infancy (CESDI) documented that avoidable factors were identifiable in 66% of deaths after shoulder dystocia, and different management would reasonably have been expected to have improved the outcome.7 Our data add to the findings of previous surveys and suggest that little has changed in the last few years despite recommendations for skills training.
We compared the effect of training with high- and low-fidelity mannequins on ability to effectively manage shoulder dystocia, the amount of force applied, and communication with a patient-actor during delivery. The high-fidelity mannequin was used both to train and to assess; therefore, the apparent additional benefits of training with the high-fidelity mannequin might be due to additional exposure during training. However, the high-fidelity mannequin was the only mannequin able to measure applied force and was consequently required during all assessments. Some might argue that our study is weakened by two thirds of participants being midwives managing shoulder dystocia in isolation. However, analysis of fatal cases of shoulder dystocia6 found that the lead professional at delivery was a midwife in 45% of cases. Clearly, training must be targeted at midwives as well as obstetricians, and both must be able to act independently if necessary.
Currently there is no evidence that training improves maternal or neonatal outcome of shoulder dystocia. Two studies have previously demonstrated an improvement in the management of simulated shoulder dystocia with training,8,9 but both studies were small. Our study has shown that all simulation training with mannequins improved the management of shoulder dystocia, including completion of basic actions, increased delivery rate, and communication. Training with a high-fidelity mannequin appeared to offer some additional training benefits, a higher rate of delivery, shorter head-to-body delivery time, and a reduction in total applied force. However, training with the high-fidelity mannequin offered no additional benefit in the perceived quality of communication with the patient-actor.
The aim of training is to enable accoucheurs to both maximize efficiency (to limit hypoxia) and minimize force (to limit trauma). These findings suggest that high-fidelity simulation works toward the first goal, but the effect of force training is less clear, with no significant difference in the maximum applied force. The reduction in total force may simply be a reflection of the shorter head-to-body interval. The clinical significance of a reduction in total force is unknown. However, total force has been measured during two actual cases of shoulder dystocia with a greater total force applied in the delivery complicated by a neonatal obstetric brachial plexus injury and fractured clavicle.5
Calling for pediatric support is important during shoulder dystocia, an emergency in which fetal hypoxia-acidosis is common. The Confidential Enquiry Into Stillbirths and Deaths in Infancy found that, in 19% of fatal cases, no pediatric staff had been present 3 minutes after delivery.6 It is of concern that those trained on the high-fidelity mannequin were significantly less likely to call for pediatric support. It could be that too much emphasis was placed on manual skills and force training and that this detracted from the wider management of the emergency. An alternative explanation is that training in a local setting encourages a better approach to general management.
This study confirms both the requirement for and benefit of simulation training for shoulder dystocia. Before training, only 43% of midwives and obstetricians achieved delivery. Mannequin-based simulation training improved the management of simulated shoulder dystocia. Training on a high-fidelity mannequin offered additional training benefits. Force perception training appeared promising but requires further development. The results of the trial have highlighted that practical skills, wider management of the emergency, and patient communication should be included in shoulder dystocia training programs. The ultimate challenge will be to determine whether simulation improves the management, and therefore the clinical outcome, of shoulder dystocia.
1. Gross TL, Sokol RJ, Williams T, Thompson K. Shoulder dystocia: a fetal-physician risk. Am J Obstet Gynecol 1987;156:1408–18.
2. Ouzounian JG, Gherman RB. Shoulder dystocia: are historic risk factors reliable predictors? Am J Obstet Gynecol 2005;192:1933–5.
3. Evans-Jones G, Kay SP, Weindling AM, Cranny G, Ward A, Bradshaw A, et al. Congenital brachial palsy: incidence, causes, and outcome in the United Kingdom and Republic of Ireland. Arch Dis Child Fetal Neonatal Ed 2003;88:F185–9.
4. National Health Service Litigation Authority. Summary of substandard care in cases of brachial plexus injury. NHSLA Journal 2003;2 suppl:ix–xi.
5. Allen R, Sorab J, Gonik B. Risk factors for shoulder dystocia: an engineering study of clinician-applied forces. Obstet Gynecol 1991;77:352–5.
6. Hope P, Breslin S, Lamont L, Lucas A, Martin D, Moore I, et al. Fatal shoulder dystocia: a review of 56 cases reported to the Confidential Enquiry into Stillbirths and Deaths in Infancy. Br J Obstet Gynaecol 1998;105:1256–61.
7. Focus group: shoulder dystocia. In: Confidential Enquiry Into Stillbirths and Deaths in Infancy (CESDI) 5th Annual Report. London, UK: Maternal and Child Health Research Consortium; 1996. p. 73–9.
8. Deering S, Poggi S, Macedonia C, Gherman R, Satin AJ. Improving resident competency in the management of shoulder dystocia with simulation training. Obstet Gynecol 2004;103:1224–8.
9. Crofts JF, Attilakos G, Read M, Sibanda T, Draycott TJ. Shoulder dystocia training using a new birth training mannequin. BJOG 2005;112:997–9.