The effect of learning curve on the outcome of caesarean section


Dr WY Fok, Department of Obstetrics and Gynaecology, Prince of Wales Hospital, The Chinese University of Hong Kong, Shatin, New Territories, Hong Kong. Email


Objective  To evaluate the operative outcomes when trainees first perform caesarean sections independently.

Design  A retrospective study in a tertiary obstetric unit.

Population  Five hundred caesarean sections, which represented the first 50 caesarean sections performed independently by each of ten trainees, were studied.

Methods  The effect of learning curve on outcome was analysed.

Main outcome measures  Total operative time, incision-to-delivery interval, operative blood loss, Apgar score, cord arterial pH, incidence of neonatal intensive care unit admission, postoperative complication rates and duration of hospitalisation.

Results  The mean operative time for the first five cases by trainees was 52.2 ± 11.4 minutes. It progressively decreased and reached 39.6 ± 8.4 minutes for the 46th to 50th cases. The operative time was significantly longer in the first 15 caesarean sections (P < 0.05). Moreover, the incision-to-delivery interval was also longer during the first five cases (P= 0.02). Besides the time of the operation, the trend for operative blood loss stabilised after the first ten caesarean sections (P < 0.05). Otherwise, there were no significant differences among other outcome variables.

Conclusion  This study shows that trainees need to perform 10–15 caesarean sections before their skills become more proficient. Senior obstetricians may need to provide guidance to the trainees during their first independent 15 caesarean sections.


In medicine, every procedure requires operator skill. These skills cannot be acquired solely from theoretical instruction. Direct supervision and practical experience are essential. Previous research has indicated that both the operative outcomes and complication rate may be directly related to experience of the surgeon.1,2 In any training scheme, it is important to ensure that trainees are competent in performing a procedure before they are allowed to do it independently. However, there is little quantitative information on the approximate number of particular operation a trainee should perform before they can be considered adequately skilled to perform the operation independently. Consequently, trainers need to make this decision based on observation, a qualitative decision.

In some procedures, such as external cephalic version, laparoscopic assisted vaginal hysterectomy, laparoscopic fundoplication, hand-assisted laparoscopic nephrectomy and stapler haemorrhoidectomy, the outcomes of the learning phase have been documented.1,3–6 It is therefore surprising that there is little information on caesarean section.7 Evaluating outcomes performed by trainees when they start performing caesarean section independently may allow us to determine whether teaching and supervision were adequate. The aim of the present study was to evaluate the outcomes of caesarean sections performed by trainees when they were first allowed to operate independently, in the absence of a supervisor in the operating theatre.


This is a retrospective study conducted in a tertiary obstetric unit. The training policy is described briefly. All specialty trainees in obstetrics and gynaecology must have been the first assistant before they can start performing the operation under direct supervision. These operations are supervised directly with specialist obstetricians as first assistant. After performing a variable number of operations under supervision, trainees are assessed to determine whether they are considered competent to perform the operation independently (assisted by an intern). There is always an on site specialist obstetrician available should the trainee require assistance.

We reviewed the medical records of the first 50 caesarean sections performed by a total of ten trainees from the point when they start performing caesarean sections without a supervisor being in the operating theatre. The variables we studied included the incision-to-delivery time interval, total operative time (skin incision to skin closure time interval), operative blood loss (which was estimated by anaesthetists through measuring the amount of blood collected in the suction bottle and weighing the used gauze swabs if necessary), Apgar score, umbilical cord blood pH, incidence of neonatal intensive care unit (NICU) admission, postoperative complication rates (febrile illness, wound infection, urinary tract infection) and the length of hospitalisation after caesarean section.

The 500 caesarean sections in the study were divided into ten groups according to the order in which they were performed by each trainee. The study was viewed as measuring the same response in the ten trainees at multiple time points. The variables in groups of five are thought of as responses to different time points. The ‘time’ effect of these repeated measures data (whether the responses of a measure stay equal over each consecutive group) was assessed by repeated measures analysis of variance. By repeating the analyses for groups 1–10, groups 2–10, groups 3–10, etc, we would identify a point where the variable of interest showed no significant difference for the remaining groups and hence the learning curve had ‘flattened’. In other words, we compared the variable of interest between group 1 and group 2–10, then compared group 2 and group 3–10 and so on until we identified a point when no significant difference was detected. Huynh–Feldt statistics with adjustment for possible violation of the sphericity assumption in the repeated measures analysis were reported. A P value of <0.05 was considered statistically significant.

Statistical analyses were performed using the Statistical Package for Social Sciences for Windows version 10.0 (SPSS Inc., Chicago, IL, USA). SAS 8.02 (SAS Institute, Cary, NC, USA) was used for repeated measures analysis.


The trainees assisted a mean of 29 caesarean sections (range 21–49) and on average performed 24 operations as principal surgeon (range 14–42) under direct supervision before they were considered competent to perform the procedure independently. A total of 500 cases of caesarean section (the first 50 cases from ten trainees) were included in the study. The demographic characteristics and delivery outcomes are shown in Table 1. The major indications for caesarean section are described in Table 1. There was no significant difference in the indication of the cases between trainees (P > 0.05). In 6 of 500 caesarean sections, senior obstetricians were ‘called in’ to assist the trainees mainly because of dense pelvic adhesion and bleeding. The two major intraoperative complications were a case of injury to the bladder serosa and a uterine tear. Both injuries were repaired with no adverse consequences.

Table 1.  Demographic characteristics and delivery outcomes of study subjects
  1. Data are presented as mean [SD], number (%) or median {interquartile range}.

Mean gestational week at delivery38.9 [1.6]
History of previous caesarean section121 (24.2)
Elective caesarean section143 (29.0)
Primary indication of caesarean section
Slow/no progress of labour132 (26.4)
Fetal distress91 (18.2)
Malpresentation82 (16.4)
Previous caesarean section72 (14.4)
Other123 (24.6)
Anaesthesia used for caesarean section
General34 (6.8)
Regional466 (93.2)
Mean operative time (minutes)43.8 [11.0]
Mean operative blood loss (ml)450.8 [270.4]
Birthweight (g)3247 [540]
Apgar score in first minute (<7)17 (3.4)
Median cord arterial pH7.29 {7.26, 7.32}
NICU admission1 (0.2)
Mean length of hospital stay (days)6.1 [2.0]
Postoperative complications
Fever21 (4.2)
Wound infection11 (2.2)
Wound dehiscence5 (1.0)
Urinary tract infection64 (12.8)

Figure 1 shows the changes in total operative time from groups 1 to 10. The mean operative time for the first five cases by trainees was 52.2 ± 11.4 minutes. It progressively decreased with increasing experience and reached 39.6 ± 8.4 minutes for the 46th to 50th cases. Repeated measures analysis of variance showed that the total operative time was significantly longer in groups 1, 2 and 3 (the first 15 cases). In other words, the operative time was significantly longer in group 1 cases compared with group 2–10 (P < 0.001), in group 2 compared with group 3–10 (P < 0.001) and in group 3 compared with group 4–10 (P= 0.027). From group 4 onwards (16th to 20th cases), there was no significant difference with the remainder of the cases.

Figure 1.

The learning curve of total operative time in the first 50 caesarean sections. The symbol ‘*’ represents P < 0.05. SE, standard error.

Incision-to-delivery time interval is shown in Figure 2. The curve also showed a steady reduction from groups 1 to 10, but the decrease was less obvious than that for operative time. When repeated measures analysis of variance was applied, only the first five cases (group 1) were significantly longer than the rest of the groups (P= 0.020). The time interval for group 2 was also shorter but did not reach statistical significance (P= 0.076).

Figure 2.

The learning curve of incision-to-delivery interval in the first 50 caesarean sections. The symbol ‘*’ represents P < 0.05. SE, standard error.

Figure 3 shows the operative blood loss. The reduction of blood loss with operative experience is only modest. After the first ten caesarean sections, the trend for operative blood loss is flattened (P < 0.05).

Figure 3.

The learning curve of operative blood loss in the first 50 caesarean sections. The symbol ‘*’ represents P < 0.05. SE, standard error.

There were no statistically significant differences in the other outcomes (Apgar scores, cord pH, NICU admission, postoperative complication rates and length of hospitalisation).

Among the 500 caesarean sections, 143 cases were elective and 357 were emergency. The ratio of elective to emergency sections was fairly similar among the ten time groups analysed. We then attempted to analyse elective and emergency caesarean sections separately. The ratio of elective to emergency caesarean sections for individual trainees (range from 10 [elective]:40 [emergency] to 20:30) and different time groups (range from 8 [elective]:42 [emergency] to 19:31) varied quite significantly. Due to this variation in distribution, we analysed the first 10 elective and first 30 emergency cases performed by each trainee. For elective caesarean, the first two cases performed by each trainee (total 20 cases) were put into the first time group and the third and fourth case into the second time group, etc. For emergency caesarean, the first three cases by each trainee were put into the first time group and the next three into the second time group, etc. We then had five time groups for elective and ten time groups for emergency section in the subgroup analysis. There was no significant difference among the operative outcomes in elective groups. However, the first six to nine emergency cases (i.e. the first one to three time groups) had significantly longer operative and delivery times (P < 0.05). For estimated blood loss, the difference was only marginally significant in the first six cases (first two time groups) (P= 0.05) (Table 2).

Table 2.  Change in operative outcomes among emergency caesarean sections
 Total operative time (minutes)Incision-to-delivery time interval (minutes)Total operative blood loss (ml)
  1. Values represent mean ± SD.

Group 154.0 ± 10.9 (P < 0.0001)11.9 ± 5.5 (P= 0.011)513.3 ± 147.6 (P= 0.057)
Group 248.7 ± 8.4 (P= 0.006)9.2 ± 2.2 (P= 0.042)460.0 ± 118.7 (P= 0.053)
Group 347.2 ± 5.6 (P= 0.035)8.8 ± 2.2623.6 ± 357.6
Group 442.5 ± 7.57.4 ± 2.1488.3 ± 150.9
Group 540.9 ± 6.87.0 ± 1.7425.0 ± 96.0
Group 640.0 ± 6.37.2 ± 1.8426.7 ± 170.7
Group 741.2 ± 8.67.6 ± 2.8405.0 ± 112.5
Group 841.0 ± 6.86.8 ± 1.3401.0 ± 135.0
Group 938.1 ± 5.37.3 ± 3.6340.0 ± 105.1
Group 1039.0 ± 6.86.5 ± 2.0426.7 ± 124.5


Teaching and passing on of operative skills to our successors is an important role in any specialty. However, there is a paucity of data in the literature on the learning curve of many operations, including caesarean section. Recommendations on the minimum number of procedures a trainee needs to perform before being allowed to operate independently is often based on little, if any, evidence. An example is laparoscopic sterilisation. In the guideline from the Royal College of Obstetricians and Gynaecologists, it is recommended that trainees should perform 25 laparoscopic sterilisations under supervision before they are allowed to do it independently.8 No reference was given for this recommendation in the guideline.

To our knowledge, there is only one study on the learning curve of residents performing the first 20 independent caesarean sections. Muller and Zimmermann7 evaluated the learning curves visually without statistical analysis and suggested that operative time and the induction to delivery time flattened after 20 cases. Our results showed that three variables demonstrated significant improvement with increase in trainee’s experience, namely total operative time, incision-to-delivery interval and operative blood loss. This supports our impression that operative time and blood loss are most sensitive to operators’ experience. This was particularly so for total operative time. The reduction from 52.2 to 39.6 minutes (a 24% reduction) is clinically relevant. During caesarean section, the vast majority of blood loss originates from the uterine wound and the placental implantation site. Rapid suturing of the uterine wound stops bleeding and also allows the uterus to contract. These are the keys to reduce blood loss. It is likely that beginners may take more time to close the uterine wound, resulting in greater blood loss.

Our results did not demonstrate a significant change in other outcome measures (such as wound infection rate). This is probably because the background incidence of these adverse events is low, and a much larger sample size would be required to demonstrate a difference. Nevertheless, our results showed that the total operative time and incision-to-delivery time were longer and operative blood loss was greater in the first 10–15 caesarean sections performed by trainees independently. Although lengthening of the operative time by approximately 12 minutes or an extra blood loss of 50 ml are unlikely to impose significant additional morbidity, these findings may suggest that the trainee’s skills had not been optimised in the early cases. However, as in all surgical training, there is a big psychological step between performing an operation under supervision and being the principal surgeon. There is also the issue of the surgical assistants, who are likely to be less experienced when trainees assume the role of principal surgeon. This is in effect a double step in the training process. From the risk management point of view, it may be preferable to offer additional supervision during these early cases. One option is to have a senior person to scrub as a second assistant for the first 10–15 cases although this will mean the trainee is still being supervised, thus delaying the transition to independence. Nevertheless, this may allow trainees to practice how to perform the procedure with the assistance of a more junior doctor. At the same time, the senior can provide guidance and assistance as an additional safeguard. A suggestion arising from these data is that trainees need to perform approximately 40 caesarean sections under different degrees of supervision before being independent.

Our results also demonstrated a different learning curve among elective and emergency caesarean sections. It is logical that elective cases are comparatively easier for the trainees. Therefore, they are able to demonstrate competence more rapidly than with emergency sections.

On the other hand, the caesarean sections performed by the trainees in our study were highly selected. They were all low-risk cases in order to help trainees to build up their skill and confidence to be prepared to do more difficult and advanced cases (including placenta praevia, preterm and second-stage caesarean sections). Based on our data, further studies need to be performed to evaluate the learning curve of the trainees in performing low-risk and high-risk caesarean sections.

In conclusion, it is gratifying to report that the skill necessary to perform a caesarean section, after adequate preparation, is rapidly acquired and the learning curve is relatively short. An excessive period of direct supervision is not necessarily better for maintenance of standards and may delay trainees reaching an adequate and appropriate level of confidence and maturity. Provided trainees receive a thorough preparation, first in assisting and then doing caesarean under supervision, their transition to independence should be relatively smooth. It was interesting to note that the number of operations the trainees did under supervision varied considerably. This may reflect varying natural surgical talent or levels of confidence of their supervisors in their surgical skills.