Abstract
Purpose of Study
Assessment of accuracy, intra-observer and inter-observer and agreement for CTG interpretation utilizing NICE guidelines 2017.
Methods
This prospective observational study was completed at a tertiary-level hospital. 1650 CTG tracings were interpretations by 5 clinicians with various levels of clinical experience according to NICE 2017 CTG interpretation guidelines were analyzed. Inter-observer and intra-observer congruence was ascertained by employing Fleiss’ kappa and Cohen’s kappa using SPSS statistical software. Sensitivity, specificity, predictive values, likelihood ratios and overall accuracy of various CTG parameters for acidemia were computed.
Results
Different CTG parameters had varied inter and intra-observer congruence which improved among clinicians with more than 5 years of experience. The best agreement was seen for baseline fetal heart (kf = 0.678), and others like identification of variability pattern, type of decelerations and concerning features of variable decelerations had fair to only slight agreement. Again, various CTG parameters had very diverse predictive values for metabolic acidosis. Overall CTG category had a sensitivity, specificity and overall accuracy of 85.00%, 82.64% and 82.67%, respectively. As a single parameter variable deceleration with concerning characters surpassed all other CTG variables for the prediction of acidosis.
Conclusions
NICE 2017 guidelines for CTG interpretation were found to have fairly good predictive accuracy for acidemia. CTG had a high negative predictive value. A high level of non-correspondence in few CTG parameters as determining variability, identification of the deceleration type and concerning features probably point to the requirement for further objectivity while defining these parameters.
Keywords: Cardiotocography, Inter-observer agreement, Intra-observer agreement, NICE guidelines, CTG interpretation guidelines, Diagnostic accuracy
Introduction
Intrapartum hypoxia is considered as one of the important causes of perinatal asphyxia [1, 2]. Severe uterine contractions, reduced fetal capacity to compensate due to underlying placental or fetal pathology or a combination of both deprive the fetus of oxygen initiating to hypoxia-acidosis-ischemia cascade. In order to assess intrapartum fetal well-being, external electronic fetal monitoring (EFM) was introduced in the 1960s [3] and has become one of the most prevalent obstetrical procedures performed in modern-day obstetrics.
Subsequent to its introduction, various professional bodies provided different guidelines to interpret and standardize EFM tracings [4–7]. The first among these was the International Federation of Gynecology and Obstetrics (FIGO) recommendations for fetal heart rate (FHR) monitoring in 1987 [8], which gained international consensus and were revised in 2015 [9]. RCOG first issued CTG guidelines in 2001 [9], which were revised in 2007 and 2017 in collaboration with NICE. NICE 2017 CTG interpretation guidelines are the latest of these guidelines.
Unfortunately, the goal to standardize the interpretation of CTG has been met with a low degree of agreement between the observer and poor reproducibility [10]. There is a divided opinion amongst the published literature in regard to the ability of CTG to reduce overall perinatal mortality and cerebral palsy [3, 10]. The aim of the study was to evaluate the accuracy, intra-observer and inter-observer agreement of CTG analysis performed using NICE 2017 CTG interpretation guidelines.
Methods
Study Design and Study Participants
This observational study was undertaken prospectively at a tertiary-level teaching hospital in accordance with the human research ethical principles laid down by Declaration of Helsinki. The study was conceded approval by the institutional research and ethical committee (Certificate number: AIIMS/IEC/2019/1821). Consecutive laboring women at or beyond 37 weeks of gestation in active labor with a singleton non-anomalous fetus in a cephalic presentation were approached for enrollment in the study. A written informed consent was taken for their CTG tracings, intrapartum information and birth details to be used for research purposes in an anonymous way. Women experiencing acute intrapartum events inclusive of cord prolapse, maternal collapse, shoulder dystocia or anesthetic complications which can lead to acute fetal deprivation of oxygen and may have a confounding effect were excluded from the study. Similarly, cardiotocographs with more than 15% signal loss were excluded. A total of 210 women were recruited, however, in 45 patients umbilical cord pH was not accurately estimated due to delay in sampling or technical errors in sample processing. Hence, 165 cardiotocographs were included in the final analysis.
Methodology
Labor monitoring was performed according to institutional standard protocol besides intrapartum continuous external fetal monitoring for study participants. CTG traces of 60 min or more of eligible consenting women at a paper speed of 1 cm/min were cataloged along with the demographic, obstetric, medical, intrapartum and immediate postpartum details by investigators not involved in the interpretation of these cardiotocograph traces. Umbilical cord blood sampling was performed from the unclamped cord or clamped cord within 30 s using a pre-heparinized syringe. The samples were analyzed for blood gases and pH immediately. The digital copy of these 165 cardiotocograph traces was emailed to 3 consultants and 2 doctors in training with different levels of expertise in the specialty of obstetrics and gynecology. The clinicians were provided with the patient’s demographic details, obstetric history, medical history and intrapartum monitoring details inclusive of labor onset, the need for augmentation, the color of the liquor or any bleeding per vaginum during labor if present. Clinicians were blinded to the fetal outcome data and mode of delivery. An online form, with no patient identifying data, required the clinicians to interpret the CTG characteristics in accordance with NICE 2017 guidelines for CTG interpretation in a multiple-option format. After a locking period of 3 months, the order of the CTG traces was reshuffled and again sent for interpretation to the same assessors to assess intra-observer agreement since the human brain tends to forget 99.9% of information within 3 months [11].
Statistical Analysis
The results were compiled and analyzed using SPSS software version 26 to compute inter-observer and intra-observer agreement and a p-value < 0.05 was taken as significant. Intra-observer agreement was assessed by Cohen’s kappa statistics, while inter-observer agreement was evaluated using Fleiss’ kappa statistics. Kappa coefficient is preferred over simple percentage agreement in clinical research as it calculates agreement over and above the chance agreement [12]. A kappa coefficient value below 0.20 was taken as slight agreement, between 0.21 and 0.40 as fair agreement, while values between 0.41–0.60 and 0.61–0.80 indicated a moderate and substantial agreement, respectively. A kappa coefficient level of 0.80 or above was considered as almost perfect agreement [13]. For increased transparency of results, a 95% confidence interval was calculated.
Fetal heart rate was documented in the pro forma as a continuous variable, therefore, a receiver-operated characteristic (ROC) curve was developed for the diagnostic value of fetal heart rate for acidosis, and the area under the curve was estimated. Other CTG variables are included in the 2017 NICE CTG interpretation guidelines in a categorical fashion. Hence, sensitivity, specificity, diagnostic accuracy of these parameters was computed.
Results
The mean age of the pregnant women enrolled in the study was 25.36 years (range19-36 year) and 56.36% (N = 93) of women were primigravidas. One hundred and thirty-two women had a gestational age of under 40 weeks, while 33 women had a gestational age of more than 40 weeks. Nearly, three fifth of women had a vaginal delivery, 17 had instrumental delivery and 46 had a cesarean Sect. 165 CTG tracings were interpreted by 5 clinicians with various levels of clinical experience twice according to NICE 2017 CTG interpretation guidelines, making a total of 1650 evaluations.
Inter-observer agreement
Figure 1 depicts the inter-observer agreement of various observers on basic CTG parameters. The overall agreement was substantial for baseline heart rate (kf = 0.678), whereas it was moderate for overall category classification of the CTG tracing (kf = 0.423). Identification of reduced and increased variability, classification of patient baseline risk and type of deceleration was met with poor to no overall agreement. Inter-rater reliability was observed to be superior among clinicians with more experience in CTG interpretation (Table 1).
Fig. 1.
Inter-observer agreement for various CTG variables
Table 1.
Inter-observer agreement (Fleiss’ kappa) on various CTG parameters according to clinician experience level
| Experience in years | ||||
|---|---|---|---|---|
| CTG parameters | 5–10 years | < 5 years | ||
| Kf | 95% CI | Kf | 95% CI | |
| Baseline | 0.754 | 0.753–0.755 | 0.535 | 0.533–0.537 |
| Define risk | 0.305 | 0.302–0.308 | 0.109 | 0.104–0.114 |
| Variability (Reduced/Increased) | 0.208 | 0.205–0.210 | 0.169 | 0.165–0.173 |
| Acceleration | 0.415 | 0.412–0.417 | 0.225 | 0.220–0.229 |
| Deceleration | 0.632 | 0.629–635 | 0.61 | 0.605–0.615 |
| Concerning features | 0.192 | 0.186–0.197 | 0.116 | 0.106–0.125 |
| Abnormal features | 0.496 | 0.493–0.499 | 0.346 | 0.341–0.351 |
| Category | 0.418 | 0.416–0.420 | 0.386 | 0.383–0.390 |
Intra-observer Agreement
Among all CTG parameters, the intra-observer agreement was again substantial for defining baseline fetal heart rate and for the identification of decelerations (Table 2). Interpretation of the overall the CTG category had moderate agreement among the majority of interpreters. But few components like identification of abnormal variability and type of deceleration had only fair to slight intra-observer agreement. Furthermore, the senior-most consultant was observed to have better agreement compared to other consultants and residents (Table 2).
Table 2.
Intra-observer agreement (Cohen’s kappa) on various CTG parameters
| Consultant 1 | Consultant 2 | Consultant 3 | Resident 1 | Resident 2 | |
|---|---|---|---|---|---|
| Kappa (S.E., p) | Kappa (S.E., p) | Kappa (S.E., p) | Kappa (S.E., p) | Kappa (S.E., p) | |
| Baseline | 0.937 (0.02, < 0.0001) | 0.85 (0.031, < 0.0001) | 0.615 (0.05, < 0.0001) | 0.825(0.03, < 0.0001) | 0.367(0.05, < 0.0001) |
| Define risk | 0.706 (0.06, < 0.0001) | 0.800 (0.05, < 0.0001) | 0.585 (0.06, < 0.0001) | 0.366 (0.06, < 0.0001) | 0.229 (0.11, 0.002) |
| Variability | 0.470 (0.09, < 0.0001) | 0.227 (0.09, 0.001) | 0.241 (0.13, < 0.0001) | 0.23 (0.14, < 0.0001) | 0.122 (0.14, 0.096) |
| Acceleration | 0.593 (0.06, < 0.0001) | 0.584 (0.08, < 0.0001) | 0.504 (0.08, < 0.0001) | 0.481 (0.08, < 0.0001) | 0.277 (0.14, < 0.0001) |
| Deceleration | 0.717 (0.06, < 0.0001) | 0.702 (0.06, < 0.0001) | 0.558 (0.07, < 0.0001) | 0.697 (0.055, < 0.0001) | 0.622 (0.06, < 0.0001) |
| Concerning features | 0.377 (0.15, < 0.0001) | 0.464 (0.11, < 0.0001) | 0.206 (0.13, 0.132) | 0.421 (0.15, 0.008) | 0.335 (0.14, 0.021) |
| Type of deceleration | 0.212 (0.08, < 0.0001) | 0.372 (0.08, < 0.0001) | 0.33 (0.11, < 0.0001) | 0.533 (0.10, < 0.0001) | 0.086 (0.12, 0.34) |
| Type of variable deceleration | 0.45 (0.08, < 0.0001) | 0.373 (0.19, 0.033) | 0.086 (0.11, 0.433) | 0.601 (0.11, < 0.0001) | 0.356, (0.15, 0.014) |
| Abnormal features | 0.56 (0.08, < 0.0001) | 0.44 (0.08, < 0.0001) | 0.297 (0.10, < 0.0001) | 0.652 (0.09, < 0.0001) | 0.364 (0.1, < 0.0001) |
| Category | 0.552 (0.05, < 0.0001) | 0.398 (0.06, < 0.0001) | 0.389 (0.06, < 0.0001) | 0.555 (0.06, < 0.0001) | 0.473 (0.06, < 0.0001) |
CTG Prediction for Metabolic Acidosis
Of 1650 CTG evaluations, 62.96% of the traces were classified as normal, 18.78% as suspicious and 18.24% as pathological in accordance with NICE 2017 guidelines for CTG interpretation. One hundred and thirty-six neonates born had a normal umbilical cord pH (pH ≥ 7.20), while 6 and 2 newborns respectively were classed as having moderate (pH < 7.10 but more than 7.00) and severe acidosis (pH < 7.00). The remaining 21 neonates had pH values ranging from 7.10–7 to 19). Table 3 compares the birth umbilical cord pH values amongst all the CTG tracing categories.
Table 3.
Comparison of the birth umbilical cord pH values amongst all the CTG tracing categories
| Normal CTG category interpretations (N = 1040) | Suspicious CTG category interpretations (N = 310) | Pathological CTG category interpretations (N = 300) | |
|---|---|---|---|
| Normal pH | 91.06% (N= 947) | 79.03% (N = 245) | 59.67% (N = 179) |
| (pH ≥ 7.20) | |||
| Abnormal pH | 7.98% (N= 83) | 16.45% (N = 51) | 22.00% (N = 66) |
| (pH 7.11–7.19) | |||
| Moderate acidosis | 0.77% (N = 8) | 4.19% (N = 13) | 12.67% (N = 38) |
| (pH < 7.1) | |||
| Severe acidosis | 0.19% (N = 2) | 0.32% (N = 1) | 5.67% (N = 17) |
| (pH < 7.0) |
The area under the receiver-operated curve (ROC) of baseline fetal heart rate was 0.607 and 0.539 for moderate and severe acidosis, respectively. These values depict a poor discriminatory value of baseline fetal heart rate for fetal acidosis. Pathological CTG had a sensitivity of 85.00% and a specificity of 82.64% with an overall accuracy of 82.67% for severe acidosis at a prevalence of 1.21% for severe acidosis. The corresponding figures for the identification of moderate acidosis were 68.75%, 84.39% and 83.64% at a prevalence of moderate acidosis of 3.63%. Pathological CTG trace had a positive likelihood ratio of 4.9 and a negative likelihood ratio of 0.18 for severe acidosis. As a single parameter, presence of variable decelerations with concerning features over 50% of contractions for greater than 30 min showed the best positive likelihood ratio for moderate and severe acidosis with a global accuracy of 70.29% (Table 4). Reduced variability (< 5) showed reasonable sensitivity for acidosis with poor specificity and overall accuracy. On the other hand, excessive variability (> 25 for > 25 min) showed high specificity with low sensitivity.
Table 4.
Sensitivity, Specificity and accuracy of various CTG parameters for acidosis
| Severity of Acidosis | Sensitivity | Specificity | Positive likelihood ratio | Negative likelihood ratio | Positive predictive value | Negative predictive value | Accuracy | |
|---|---|---|---|---|---|---|---|---|
| Variability < 5 for 30–50 min | Abnormal pH (21.19%) | 62.50% | 42.02% | 1.08 | 0.89 | 22.47% | 80.65% | 46.36% |
| (43.69–78.90%) | (33.03–51.41%) | (0.79–1.47) | (0.54–1.46) | (17.55–28.30%) | (71.76–87.23%) | (38.22–54.65%) | ||
| Moderate acidosis (7.28%) | 72.73% | 42.14% | 1.26 | 0.65 | 8.99% | 95.16% | 44.37% | |
| (39.03–93.98%) | (33.85–50.77%) | 0.85–1.85 | 0.24–1.73 | (6.28–12.71%) | (88.02–98.14%) | (36.30–52.67%) | ||
| Severe acidosis (0.66%) | 100.00% | 41.33% | 1.70 | 0.00 | 1.12% | 100% | 41.72% | |
| 2.50–100% | 33.36–49.64% | 1.49–1.95 | 0.98–1.28% | (33.76–50.02%) | ||||
| Variability < 5 for > 50 min | Abnormal pH (21.19%) | 25.00% | 86.55% | 1.86 | 0.87 | 33.33% | 81.10% | 73.51% |
| (11.46–43.40%) | (79.09–92.12%) | (0.88–3.95) | (0.70–1.07) | (19.05–51.51%) | (77.63–84.14%) | (65.72–80.35%) | ||
| Moderate acidosis (7.28%) | 9.09% | 83.57% | 0.55 | 1.09 | 4.17% | 92.13% | 78.15% | |
| 0.23–41.28% | 76.38–89.29% | 0.08–3.72 | 0.89–1.33 | 0.64–22.62% | 90.54–93.46% | 70.70–84.45% | ||
| Severe acidosis (0.66%) | 0.00% | 84.00% | 0.00 | 1.19 | 0 | 99.21% | 83.44% | |
| (77.14–8.47%) | 1.11–1.28 | 99.16–99.27% | 76.54–88.99% | |||||
| Variability > 25 for 15–25 min | Abnormal pH (21.19%) | 6.25% | 80.67% | 0.32 | 1.16 | 8.00% | 76.19% | 64.90% |
| (0.77–20.81%) | (72.42–87.34%) | (0.08–1.30) | (1.03–1.32) | (2.12–25.90%) | (73.84–78.39%) | (56.72–72.48%) | ||
| Moderate acidosis (7.28%) | 9.09% | 82.86% | 0.53 | 1.10 | 4.00% | 92.06% | 77.48% | |
| 0.23–41.28% | 75.58–88.70% | 0.08–3.56 | 0.90–1.34 | 0.62–21.86% | (90.46–93.42%) | (69.98–83.87%) | ||
| Severe acidosis (0.66%) | 0.00 | 83.33% | 0.00 | 1.20 | 0 | 99.21% | 82.78% | |
| 76.39–88.91% | 1.12–1.9 | 99.15–99.26% | 75.80–88.43% | |||||
| Variability > 25 for > 25 min | Abnormal pH (21.19%) | 6.25% | 90.76% | 0.68 | 1.03 | 15.38% | 78.26% | 72.85% |
| (0.77–20.81%) | (84.06–95.29%) | (0.16–2.90) | (0.93–1.15) | (4.07–43.79%) | (76.40–80.01%) | (65.02–79.76%) | ||
| Moderate acidosis (7.28%) | 9.09% | 91.43% | 1.06 | 0.99 | 7.69% | 92.75% | 85.43% | |
| 0.23–41.28% | 85.51–95.49% | 0.15–7.42 | 0.82–1.21 | 1.18–36.83% | 91.34–93.95% | 78.78–90.64% | ||
| Severe acidosis (0.66%) | 0.00 | 91.33% | 0.00 | 1.09 | 0 | 99.28% | 90.73% | |
| 85.64–95.30% | 1.04–1.15 | 99.24–99.31% | 84.93–94.84% | |||||
| Variable decel with concerning features < 50% < 30 min | Abnormal pH (31.75%) | 25% | 63.12% | 0.68 | 1.19 | 23.97% | 64.41% | 51.02% |
| (18.07–33.02%) | (57.40–68.59%) | (0.49–0.94) | (1.04–1.35) | (18.59–30.33%) | (61.40–67.30%) | (46.25–55.78%) | ||
| Moderate acidosis (10.20%) | 20.00% | 65.80% | 0.58 | 1.22 | 6.16% | 87.80% | 60.77% | |
| 9.58–34.60% | 60.49–70.08% | 0.32–1.05 | 1.04–1.44 | 85.94–89.44% | 85.94–89.44% | 56.04–65.36% | ||
| Severe acidosis (3.40%) | 26.67% | 66.67% | 0.80 | 1.10 | 2.74% | 96.27% | 65.31% | |
| 7.79–55.10% | 61.97–71.13% | 0.34–1.87 | 0.80–1.50 | 1.19–6.18% | 94.97–97.24% | 60.66–69.74% | ||
| Variable decel with concerning features < 50% > 30 min | Abnormal pH (31.75%) | 21.43% | 72.09% | 0.77 | 1.09 | 26.32% | 66.36% | 56.01% |
| (14.95–29.16%) | (66.66–77.09%) | (0.53–1.11) | 0.97–1.22) | (19.86–33.98%) | (63.83–68.80%) | (51.24–60.70% | ||
| Moderate acidosis (10.20%) | 15.56% | 72.98% | 0.58 | 1.16 | 6.14% | 88.38% | 67.12% | |
| 6.49–29.46% | 68.32–77.29% | 0.29–1.16 | 1.01–1.33 | 3.15–11.64% | 86.87–89.73% | 62.52–71.49% | ||
| Severe acidosis (3.40%) | 6.67% | 73.47% | 0.25 | 1.27 | 0.88% | 95.72% | 71.20% | |
| 0.17–31.95% | 69.01–77.61% | 0.04–1.68 | 1.10–1.47 | 0.13–5.59% | 95.08–96.28% | 66.73–75.39% | ||
| Variable decel with concerning features > 50% < 30 min | Abnormal pH (31.75%) | 9.29% | 88.04% | 0.78 | 1.03 | 26.53% | 67.60% | 63.04% |
| (5.04–15.36%) | (83.83–91.48%) | (0.43–1.42) | (0.96–1.10) | (16.52–39.73%) | (66.11–69.06%) | (58.34–67.56%) | ||
| Moderate acidosis (10.20%) | 15.56% | 89.39% | 1.47 | 0.94 | 14.29% | 90.31% | 81.86% | |
| 6.49–29.46% | 85.93–92.25% | 0.70–3.07 | 0.83–1.08 | 7.38–25.86% | 89.11–91.39% | 77.94–85.35% | ||
| Severe acidosis (3.40%) | 13.33% | 88.97% | 1.21 | 0.97 | 4.08% | 96.68% | 86.39% | |
| 1.66–40.46% | 85.60–91.78% | 0.32–4.52 | 0.80–1.19 | 1.13–13.72% | 95.97–97.27% | 82.84–89.45% | ||
| Variable decel with concerning features > 50% > 30 min | Abnormal pH (31.75%) | 44.29% | 76.74% | 1.90 | 0.73 | 46.97% | 74.76% | 66.44% |
| (35.90–52.92%) | (71.55–81.40%) | (1.44–2.51) | (0.62–0.85) | (40.17–53.88%) | (71.62–77.66%) | (61.82–70.84%) | ||
| Moderate acidosis (10.20%) | 48.89% | 72.22% | 1.76 | 0.71 | 16.67% | 92.56% | 69.84% | |
| 33.70–64.23% | 67.53–76.58% | 1.25–2.47 | 0.53–0.95 | 12.48–21.91% | 90.28–94.34% | 65.32–74.09% | ||
| Severe acidosis (3.40%) | 53.33% | 70.89% | 1.83 | 0.66 | 6.06% | 97.73% | 70.29% | |
| 26.59–78.73% | 66.33–75.16% | 1.12–3.01 | 0.38–1.13 | 3.78–9.58% | 96.16–98.67% | 65.79–74.52% | ||
| Late decel < 30 min | Abnormal pH (33.33%) | 64.52% | 40.32% | 1.08 | 0.88 | 35.09% | 69.44% | 48.39% |
| (45.37–80.77%) | (28.05–80.77%) | (0.78–1.51) | (0.50–1.55) | (27.95–42.96%) | (56.41–79.96%) | (37.89–58.99%) | ||
| Moderate acidosis (12.90%) | 66.67% | 39.51% | 1.10 | 0.84 | 14.04% | 88.89% | 43.01% | |
| 34.89–90.08% | 28.81–50.99% | 0.71–1.71 | 0.36–1.96 | 9.54–20.18% | 77.47–94.90% | 32.78–53.69% | ||
| Severe acidosis (5.38%) | 40.00% | 37.50% | 0.64 | 1.60 | 3.51% | 91.67% | 37.63% | |
| 5.27–85.34% | 27.40–48.47% | 0.22–1.90 | 0.74–3.44 | 1.21–9.72% | 83.66–95.94% | 27.79–48.28% | ||
| Late decel > 30 min | Abnormal pH (33.33%) | 35.48% | 59.68% | 0.88 | 1.08 | 30.56% | 64.91% | 51.61% |
| (19.23–54.63%) | (46.45–71.95%) | (0.50–1.55) | (0.78–1.51) | (20.04–43.59%) | (57.04–72.05%) | (41.01–62.11%) | ||
| Moderate acidosis (12.90%) | 33.33% | 60.49% | 0.84 | 1.10 | 11.11% | 85.96% | 56.99% | |
| 9.92–65.11% | 49.01–71.19% | 0.36–1.96 | 0.71–1.71 | 5.10–22.53% | 79.82–90.46% | 46.31–67.22% | ||
| Severe acidosis (5.38%) | 60.00% | 62.50% | 1.60 | 0.64 | 8.33% | 96.49% | 62.37% | |
| 14.66–94.73% | 51.53–72.60% | 0.74–3.44 | 0.22–1.90 | 4.06–16.34% | 90.28–98.79% | 51.72–72.21% | ||
| Pathological category | Abnormal pH (16.97%) | 43.21% | 86.93% | 3.31 | 0.65 | 40.33% | 88.22% | 79.52% |
| 37.33–49.24% | 85.03–88.67% | 2.73–4.01 | 0.59–0.72 | 35.82–45.02% | 87.10–89.26% | 77.49–81.44% | ||
| Moderate acidosis (4.85%) | 68.75% | 84.39% | 4.41 | 0.37 | 18.33% | 98.15% | 83.64% | |
| 57.41–78.65% | 82.50–86.16% | 3.65–5.31 | 0.27–0.51 | 15.69–21.30% | 97.45–98.66% | 81.76–85.39% | ||
| Severe acidosis (1.21%) | 85.00% | 82.64% | 4.90 | 0.18 | 5.67% | 99.78% | 82.67% | |
| 62.11–96.79% | 80.71–84.45% | 3.96–6.05 | 0.06–1.87 | 4.63–6.91% | 99.37–99.92% | 80.75–84.46% |
After excluding other indications, 30 cesarean sections were performed for pathological CTG trace. Overall, 14 babies were born with a 1-min APGAR score of less than 7, and 7 neonates had a 5-min APGAR score of lower than 8. No case of stillbirth or hypoxic–ischemic encephalopathy was documented for the cohort.
Discussion
Electronic fetal monitoring for intrapartum fetal surveillance has gained acceptance among obstetric specialty in recent years with the expectation of decreasing perinatal mortality and morbidity. The theoretical rationale behind this expectation is the presumed early detection of intrapartum fetal hypoxia, by detecting the subtle fetal heart pattern changes which intermittent auscultation by stethoscope or by handheld Doppler can miss. Our study aimed to determine the variability among different observers while interpreting CTG and to determine the effectiveness of CTG for the prediction of fetal metabolic acidosis. Individual CTG parameters were analyzed for their accuracy to predict fetal acidemia and for inter-rater agreement.
The total number of evaluations in our study was 1650, which is comparable to most of the studies on the inter-observer and intra-observer agreement [14–18]. The majority of these studies interpretated CTG by means of ACOG, FIGO and NICE 2007 guidelines. Santo S and coworkers [19] compared the different guidelines (FIGO/ACOG/NICE2007) using 4077 evaluations. After thorough research of published literature, we were unable to find any previous study assessing inter and intra-observer agreement of CTG parameters based on NICE 2017 guidelines.
Inter-rater and intra-rater reproducibility were quite variable for different CTG parameters. A substantial inter-observer and intra-rater agreement was observed for baseline fetal heart rate rhythm and identification of fetal heart decelerations. A moderate agreement was seen for the overall CTG category, while a few CTG variables like identification of reduced variability, type of deceleration and defining baseline patient risk was met with fair to poor agreement. The low reproducibility for the type of decelerations may be partly explained by the relatively lower occurrence of late decelerations in the cohort as the kappa coefficient is dependent on the prevalence of the variable. Our results were comparable to a study conducted by Rie and colleagues [15] who estimated the inter-observer agreement for interpreting 151 CTG tracings by 6 clinicians according to FIGO 2015 guidelines. They used a proportion of agreement (Pa) and found a poor inter-observer agreement for reduced variability, saltatory pattern, absence of accelerations and absence of decelerations. Also, our results were similar to the S. Schiermeier et al [16], who observed the highest agreement for baseline and least for acceleration. On the other hand, Rhose and colleagues [14] studied inter-observer agreement of 79 non-reassuring CTGs among nine observers according to FIGO/STAN guidelines and estimated a poor agreement among most observers. Likewise, Devine [18] and Lalor found poor inter-observer agreement in their study.
Furthermore, both intra and inter-observer reproducibility was lower among clinicians with less than 5 years of experience as compared to more experienced observers. Our results were in agreement with others [20–22] who in a systematic review observed that CTG training led to improved CTG interpretation proficiency and superior inter-observer agreement. On the other hand, Rei M and colleagues [15] in 2016 and S. Schiermeier and group [16] had found no substantial difference in agreement with the increasing level of expertise.
3.63% of babies born to the study cohort had moderate acidosis, while 1.2% had severe acidosis. Pathological CTG was found to have a sensitivity of 68.75% and a specificity of 84.39% for moderate acidosis pH with a positive predictive value of 18.33 and a negative predictive value of 98.15%. This effectively means that if there were 100 pathological intrapartum CTGs, only 18 babies will be born with an umbilical artery pH between 7.00 and 7.10. This computes to a small but important increase in the probability of neonates having moderate acidosis that is, 18.33%, in women with pathological intrapartum CTG as compared to 3.63% baseline risk of moderate acidosis found in the cohort. Conversely, a normal intrapartum CTG trace assures the clinician of a 63% and 82% reduction of the probability of moderate acidosis and severe acidosis, respectively. Also, CTG was found to have a high negative predictive value with good sensitivity and specificity for a neonate to be born with severe acidosis. A low prevalence of event, that is acidosis, probably explains why CTG as a diagnostic modality, is better at ruling out the disease and worse at ruling in. By extrapolating this finding, it can be inferred that the positive predictive value of CTG can be higher in women at high risk of fetal acidosis. A normal CTG was able to rule out all cases of severe acidosis and nearly 83% of cases of moderate acidosis. Published literature quotes a spread-over range for CTG sensitivity (47%-79%) and specificity (79%-92%) [23–25].
The individual CTG parameters varied considerably in their ability to predict acidosis. Tachycardia, and abnormal fetal heart variability, reduced or exaggerated, showed high specificity at the cost of poor sensitivities. Next to the overall CTG category, variable decelerations with concerning characters especially if present in more than 50% of contractions for more than 30 min had a reasonable sensitivity, specificity and accuracy for acidosis. After thorough research of the published literature, we were unable to find any previous study assessing inter and intra-observer agreement of CTG parameters based on NICE 2017 guidelines.
An important strength of the study is a locking period of 3 months before evaluation of intra-observer agreement, to ensure independent interpretation on both occasions. The observers interpreting the CTG traces had varied levels of experience, which on one hand enabled us to study the influence of increasing experience on congruence but may have led to a slightly reduced level of overall agreement. The agreement on certain parameters like prolonged deceleration, sinusoidal pattern and late deceleration was impacted by the low prevalence. Moreover, the observers interpreted the CTG traces at leisure with ready accessibility to interpretation guidelines which takes away a few complexities of a real-life clinical scenario like time pressure, the stress of labor suite and memory recall of guidelines. Additionally, the predictive value of CTG may vary in a population with a dissimilar baseline risk of fetal acidemia.
Conclusions
NICE 2017 guidelines for CTG interpretation were found to have fairly good predictive accuracy for acidemia. All CTG parameters had a high negative predictive value and CTG had a good diagnostic modality for ruling out acidemia. Comprehensively, overall CTG category had the best sensitivity, specificity and accuracy for acidosis, giving support to the NICE 2017 guidelines for the interpretation of CTG. However, poor inter and intra-observer reproductivity of a few parameters like identifying the type of decelerations, ascertaining concerning features of variable decelerations and determining abnormal variability underlines the complexity of definitions. This possibly emphasizes the need for further clarity and objectivity in describing some of the CTG variables.
Human and Animal Participants
This prospective observational study was conducted in accordance with the human research ethical principles laid down by Declaration of Helsinki. The study was conceded approval by the institutional research and ethical committee for this (Certificate number: AIIMS/IEC/2019/1821).
Acknowledgements
The authors would like to acknowledge the nursing and paramedical staff for their cooperation throughout the study period.
Author Contributions
Dr. Navdeep Kaur Ghuman was involved in the conception of the study, guiding the study, interpretation of CTG traces statistical analysis and manuscript writing. Dr. Priya Singhmor was involved in the study conception, collection and filling of data, compiling of data and statistical analysis. Dr. Charu Sharma, Dr. Priyanka Kathuria, Dr. Priya Susan Mathew, Dr. Mayank Jain and Dr. Suma Shet were involved in the interpretation of the CTG traces. Dr. Pratibha Singh was involved in critical appraisal of the manuscript.
Funding
The study did not receive any monetary funding support. AIIMS Jodhpur provided the infrastructure, research approval and ethical approval for the conduction of the study.
Declarations
Conflict of interest
Authors have no conflict of interest to disclose.
Ethical approval
I, Dr Navdeep Kaur Ghuman as the corresponding author, on behalf of all the authors, declare that, This manuscript is not submitted to any other journal for simultaneous consideration. The submitted work is original and is not published elsewhere in any form or language except for the abstract of the study which was presented in the RCOG annual conference by Dr Priya Singhmor. There are no concurrent, secondary or partial publications of this manuscript. Results are presented clearly, honestly, and without fabrication, falsification or inappropriate data manipulation. Data Sharing Statement: This is not a clinical trial. The data of participants in the study can only be shared after approval from the institute-All India Institute of Medical Sciences, Jodhpur, India. Ethical approval was provided by the Institutional Research Committee & Institutional Ethics Committee approval (AIIMS/IEC/2019/1821).
Informed Consent
A written informed consent was taken for their CTG tracings, intrapartum information and birth details to be used for research purposes in an anonymous way.
Footnotes
Priya Singhmor is a Post Graduate Student, Navdeep Kaur Ghuman is a Additional professor, Priyanka Kathuria is a Associate professor, Charu Sharma is a Additional professor, Mayank Jain is a Post Graduate Student, Suma Shet is a Post Graduate Student, Priya Susan Mathew is a Post Graduate Student, Pratibha Singh is a Professor and Head.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Sendeku FW, Azeze GG, Fenta SL. Perinatal asphyxia and its associated factors in Ethiopia: a systematic review and meta-analysis. BMC Pediatr. 2020;20:1–1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Garabedian C, De Jonckheere J, Butruille L, et al. Understanding fetal physiology and second line monitoring during labor. J Gynecol Obstet Hum Reprod. 2017;46(2):113–7. [DOI] [PubMed] [Google Scholar]
- 3.Ayres-de-Campos D. Electronic fetal monitoring or cardiotocography, 50 years later: what’s in a name? Am J Obstet Gynecol. 2018;218(6):545–6. [DOI] [PubMed] [Google Scholar]
- 4.Ayres-de-Campos D, Spong CY, Chandraharan E. FIGO intrapartum fetal monitoring expert consensus panel FIGO consensus guidelines on intrapartum fetal monitoring: cardiotocography. Int J Gynecol Obstet. 2015;131(1):13–24. [DOI] [PubMed] [Google Scholar]
- 5.American Academy of Pediatrics. American College of Obstetricians and Gynecologists Fetal heart rate monitoring. Guidelines for perinatal care. 7th ed. Washington, DC: American College of Obstetricians and Gynecologists. 2012.
- 6.American College of Obstetricians and Gynecologists. Practice bulletin no. 116: Management of intrapartum fetal heart rate tracings. Obstet Gynecol. (2010), 116(5): 1232–40. 10.1097/AOG.0b013e3182004fa9 [DOI] [PubMed]
- 7.National Collaborating Centre for Women’s and Children’s Health commissioned by the National Institute for Health and Clinical Excellence. Intrapartum care. 2007. [DOI] [PubMed]
- 8.Ayres-de-Campos D, Bernardes J. Twenty-five years after the FIGO guidelines for the use of fetal monitoring: time for a simplified approach? Int J Gynecol Obstet. 2010;110(1):1–6. [DOI] [PubMed] [Google Scholar]
- 9.Royal College of Obstetricians and Gynaecologists. The use of electronic fetal monitoring. Evidence-based clinical guideline, number 8. London: RCOG Press; 2001.
- 10.Stout MJ, Cahill AG. Electronic fetal monitoring: past, present, and future. Clin Perinatol. 2011;38(1):127–42. [DOI] [PubMed] [Google Scholar]
- 11.Murre JMJ, Dros J. Replication and analysis of ebbinghaus’ forgetting curve. PLoS ONE. 2015;10(7): e0120644. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Bakeman, R., & Quera, V. (2011). Sequential analysis and observational methods for the behavioral sciences. Sequential Analysis and Observational Methods for the Behavioral Sciences.
- 13.McHugh ML. Interrater reliability: the kappa statistic. Biochemia medica. 2012;22(3):276–82. [PMC free article] [PubMed] [Google Scholar]
- 14.Rhöse S, Heinis AM, Vandenbussche F, et al. Inter- and intra-observer agreement of non-reassuring cardiotocography analysis and subsequent clinical management. Acta Obstet Gynecol Scand. 2014;93(6):596–602. 10.1111/aogs.12371. (Epub 2014 Apr 15 PMID: 24597920). [DOI] [PubMed] [Google Scholar]
- 15.Rei M, Tavares S, Pinto P, et al. Interobserver agreement in CTG interpretation using the 2015 FIGO guidelines for intrapartum fetal monitoring. Eur J Obstet Gynecol Reprod Biol. 2016;205:27–31. [DOI] [PubMed] [Google Scholar]
- 16.Schiermeier S, Westhof G, Leven A, et al. Intra- and interobserver variability of intrapartum cardiotocography: a multicenter study comparing the FIGO classification with computer analysis software. Gynecol Obstet Invest. 2011;72(3):169–73. [DOI] [PubMed] [Google Scholar]
- 17.Åmark H, Millde-Luthander C, Ajne G, et al. Single versus pairwise interpretation of cardiotochography, a comparative study from six Swedish delivery units. Sex Reprod Healthc. 2014;5(4):195–8. [DOI] [PubMed] [Google Scholar]
- 18.Devane D, Lalor J. Midwives’ visual interpretation of intrapartum cardiotocographs: intra-and inter-observer agreement. J Adv Nurs. 2005;52(2):133–41. [DOI] [PubMed] [Google Scholar]
- 19.Santo S, Ayres-de-Campos D, Costa-Santos C, et al. Agreement and accuracy using the FIGO, ACOG and NICE cardiotocography interpretation guidelines. Acta Obstet Gynecol Scand. 2017;96(2):166–75. [DOI] [PubMed] [Google Scholar]
- 20.Spilka J, Chudáček V, Janků P, et al. Analysis of obstetricians’ decision making on CTG recordings. J Biomed Inform. 2014;51:72–9. [DOI] [PubMed] [Google Scholar]
- 21.Gyllencreutz E, Hulthén Varli I, Lindqvist PG, et al. Reliability in cardiotocography interpretation - impact of extended on-site education in addition to web-based learning: an observational study. Acta Obstet Gynecol Scand. 2017;96(4):496–502. [DOI] [PubMed] [Google Scholar]
- 22.Pehrson C, Sorensen J, Amer-Wåhlin I. Evaluation and impact of cardiotocography training programmes: a systematic review. BJOG. 2011;118:926–35. [DOI] [PubMed] [Google Scholar]
- 23.Visser GH, Ayres-de-Campos D. FIGO intrapartum fetal monitoring expert consensus panel. FIGO consensus guidelines on intrapartum fetal monitoring: adjunctive technologies. Int J Gynaecol Obstet. 2015;131(1):25–9. [DOI] [PubMed] [Google Scholar]
- 24.Ekengård F, Cardell M, Herbst A. Low sensitivity of the new FIGO classification system for electronic fetal monitoring to identify fetal acidosis in the second stage of labor. European J Obst Gynecol Reproduct Biol: X. 2021;1(9): 100120. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Ekengård F, Cardell M, Herbst A. Impaired validity of the new FIGO and Swedish CTG classification templates to identify fetal acidosis in the first stage of labor. J Matern Fetal Neonatal Med. 2022;35(25):4853–60. [DOI] [PubMed] [Google Scholar]