PDF icon Download PDF
Open Access
CC BY-NC-ND 4.0 · Geburtshilfe Frauenheilkd
DOI: 10.1055/a-2561-6555
GebFra Science
Original Article

Detection of Nuchal Cord Based on Specific CTG Patterns Intrapartum – A Myth?

Article in several languages: English | deutsch
Lars Brodowski 
1   Klinik für Gynäkologie und Geburtshilfe, Medizinische Hochschule Hannover, Hannover, Germany (Ringgold ID: RIN9177)
,
Roza Röchte-Christoforatou
2   Zentrum für Pränatalmedizin, Hannover, Germany
,
Ismini Staboulidou
2   Zentrum für Pränatalmedizin, Hannover, Germany
,
Constantin Sylvius von Kaisenberg
1   Klinik für Gynäkologie und Geburtshilfe, Medizinische Hochschule Hannover, Hannover, Germany (Ringgold ID: RIN9177)
,
Elna Kühnle
1   Klinik für Gynäkologie und Geburtshilfe, Medizinische Hochschule Hannover, Hannover, Germany (Ringgold ID: RIN9177)
› Author Affiliations
› Further Information
Also available ateRefHebammen
 
 PDF Download Permissions and Reprints

Abstract

Background

Nuchal cord, the situation when the umbilical cord is wrapped around the neck of the fetus, is a common occurrence during pregnancy and intrapartum. Abnormal cardiotocography patterns (CTG patterns) during nuchal cord are usually not associated with increased perinatal morbidity. This retrospective data analysis aimed to investigate the occurrence of specific CTG patterns with nuchal cord.

Methods

150 CTGs with and 150 CTGs without nuchal cord at 60 and 30 minutes prior to delivery were randomly selected from the obstetric database of Hanover Medical School out of a total cohort of 7573 births (spontaneous delivery, vaginal-operative delivery, secondary caesarean section) between 2014 and 2017. After anonymization in accordance with the 2015 FIGO criteria, CTG patterns were interpreted by three physicians with varying levels of professional experience. The physicians were also asked to assess whether nuchal cord was present or not. The interrater variability between investigators regarding the interpretation of the CTG patterns was also investigated using proportion of agreement and kappa statistics.

Results

Nuchal cord was present in 11% of the total cohort. The study cohort and the total cohort were equivalent in terms of patient characteristics. No significant differences were found in the assessments of the three physicians with regard to CTGs with and without nuchal cord. Likewise, logistic regression analysis was unable to identify a specific CTG pattern in cases with nuchal cord intrapartum. High interpretation variability between physicians was found regarding CTG interpretations (PoA > 0.5).

Conclusion

CTG interpretation just before delivery of the infant is not useful to detect nuchal cord. Moreover, despite the existence of defined criteria, the variability between the CTG interpretations of the three physicians was high.


 

Keywords

nuchal cord - CTG interpretation - quality indicators

Introduction

Nuchal cord (NC), when the umbilical cord is wrapped around the neck of the fetus, is quite common in clinical practice during delivery. The clinical relevance of NC intrapartum is still controversially discussed in the literature. Some studies report an increased rate of birth complications such as secondary caesarean section and vaginal-operative delivery, fetal stress and neonatal adaptation disorders [1] [2] [3]. Other studies have come to the conclusion that no serious short or long-term damage to the neonate is expected in cases with NC during delivery [4] [5] [6] [7].

Monitoring the fetus intrapartum is essential to ensure the birth of a healthy neonate in an optimal condition. For decades, cardiotocography (CTG) has been the method of choice to monitor the fetus throughout the entire birth process [8]. However, this method is also prone to interpretation mistakes as making a visual diagnosis of CTG patterns is subjective and there is a risk of variability of interpretations among different clinicians [9]. The precise impact of NC on fetal heart rate during the birth and on the perinatal outcome is still not known. Studies on this topic have come to different results [2]. On the CTG, variable decelerations are often observed if NC is present [2] [6]. The majority of studies on this topic have reported an increased occurrence of variable or late decelerations intrapartum if NC is present [2] [3] [10].

In the literature, the birth mode for cases with NC is still disputed. A retrospective analysis of 11748 planned vaginal births found no increased rate of vaginal-operative births or deliveries by secondary caesarean section in the group with NC, irrespective of the number of NC. This finding is in contrast with the results presented by Schäffer et al., who found abnormal changes in fetal heart rate patterns more often during delivery in the group with NC [10]. Larson et al. reported a higher rate of vaginal-operative births when multiple NC was present. However, they found no increase in the rate of necessary caesarean sections [11]. The study by Mastrobattista et al. reported a significantly reduced rate of caesarean sections in the group with multiple NC [12]. Another study group, however, found a higher rate of caesarean sections if NC was present once or multiple times [13].

There are currently no studies which can provide evidence of specific changes to CTG patterns caused by NC. The current study situation is heterogeneous and the results are still controversially discussed. The majority of variable decelerations reported on CTGs are not specific to NC as they can also occur without NC being present. A randomized blinded study of CTG assessments which aims to identify NC based on CTG changes has not been previously published. The aim of this study is therefore to use randomized blinded CTG interpretations to examine whether a CTG pattern can be identified in cases with NC intrapartum which specifically and reliably indicates the presence of NC. This study aims to investigate whether it is possible to infer the presence of NC during delivery based on the CTG pattern.


 

Methods

Study cohort

The entire dataset was obtained from the obstetric database of the gynecological department of Hanover Medical School (MHH). The dataset consisted of all births between 2014 and 2017 in the gynecological department which ended in spontaneous delivery, vaginal-operative delivery, or secondary caesarean section after 37 weeks of pregnancy. Primary caesarean sections, preterm births, multiple births, births of neonates with intrauterine growth restriction and pathological Doppler findings, congenital malformations or umbilical pathologies were excluded. Acute obstetric emergencies such as shoulder dystocia which occurred during delivery of the infant were also excluded.


 

CTG interpretation

For the retrospective CTG interpretations, 150 births with NC and 150 births without NC were randomly selected from the total cohort and the CTG tracings for the period from 60 minutes prior to delivery and from 30 minutes prior to delivery were made available for evaluation. Differentiating between the presence or absence of NC was done by reviewing the partographs. All cases were singleton pregnancies with spontaneous delivery in cephalic presentation and the full electronic CTG tracings were available for the above-listed measurement times. The NC group consisted of 120 spontaneous births with simple NC and 30 births with double NC. Cases with more complex NC were not included in the study cohort.

The CTG tracings were interpreted by a medical specialist with more than three and less than five years’ professional experience (test person A) and two medical specialists with more than five years’ professional experience (test person B and test person C). The test persons doing the interpretations did not receive any information about the outcome parameters of the respective neonates. The interpretations of the CTG tracings were based on the 2015 FIGO criteria [14]. An assessment questionnaire consisting of five multiple-choice questions (Fig. S1) (supplementary material, online) was compiled and provided to each investigator. The duration of each CTG was 30 minutes (scenario 1: from 60 minutes prior to delivery to 30 minutes prior to delivery; scenario 2: from 30 minutes prior to delivery to the birth). The CTG interpretations were carried out by the investigators independently from one another over a period of four months.


 

Statistical analysis

Equivalence tests (two-sided t-tests [TOST]) were done to test for statistical differences in the response behaviors of test persons for each individual investigator and for all the investigators together. Logistic regression analysis was additionally carried out to examine whether selected response options were associated with the occurrence of NC intrapartum. To assess the level of agreement between the answers of the three investigators, proportion of agreement (PoA) with 95% confidence intervals was calculated for each question and for each pair of test persons. To ensure robust and high-quality testing of the level of agreement between the three test persons with respect to their CTG interpretations, kappa statistics were used to calculate the extent of interobserver agreement which cannot be explained by coincidence [15]. Microsoft Excel version 16.61 was used for the descriptive statistics of the demographic characteristics and the birth characteristics of the total cohort. R software version 4.1.0 was used for the statistical analysis of the CTG interpretation data. Statistical significance was assumed for all p values less than 0.05.


 

Ethical approval

The study was approved by the local ethics commission of Hanover Medical School and was carried out in accordance with the principles of the Declaration of Helsinki (approval number: 7768_BO-K_2018).


 

 

Results

Description of the study cohort

There was no nuchal cord in 89.03% of the total cohort (N = 6742). Simple NC was present in 9.3% (N = 704) of cases, double NC in 1.45% (N = 110) and multiple NC in 0.22% (N = 22) of cases. The descriptive characteristics of the study cohort are summarized in [Table 1]. There were no significant differences in investigated demographic and birth-related criteria between the total cohort (N = 7573) and the study cohort (N = 300).

Table 1 Incidence of nuchal cord in the total cohort and the study cohort.

Nuchal cord and characteristics

Total cohort (N = 7573)

Study cohort (N = 300)

Maternal age (years)

  • No NC

31.5

31.1

  • Simple NC

31.3

31.7

  • Double NC

31.2

31.1

BMI (kg/m2)

  • No NC

29.9

29.5

  • Simple NC

30.1

30.2

  • Double NC

30.1

33.2

Gestational age at birth (in GW)

  • No NC

39.73

39.88

  • Simple NC

39.89

39.92

  • Double NC

39.94

39.84

Birth weight (in grams)

  • No NC

3447.11

3592.90

  • Simple NC

3429.91

3502.11

  • Double NC

3390.41

3436.00

Sex of the child

Male

  • No NC

50.61%

50.67%

  • NC

53.43%

52.00%

Female

  • No NC

49.39%

49.33%

  • NC

46.57%

48.00%

 

Interpretation of the CTG as a function of nuchal cord diagnosis

All test persons showed a similar response behavior both for the births with NC and the births without NC. This applied both to the period from 60 to 30 minutes prior to delivery and the last 30 minutes prior to delivery. Equivalence tests were able to dismiss the hypothesis that answers for CTGs with NC differed from those without NC in every case and at every level and the rejection of the hypothesis was statistically significant. This means that no difference could be observed in response behaviors to CTGs with and those without NC.

Logistic regression analysis was additionally carried out to investigate whether NC could be identified based on the type of deceleration on the CTG. The results of logistic regression analysis of NC depending on the manifestation of the decelerations are shown in supplementary Tab. S1 (supplementary material, online). Statistical significance was only found for the response option “no decelerations” of test person B for the period 30–60 minutes prior to the birth.

When the association between suspicious or pathological CTGs and presence of NC during delivery was analyzed with logistic regression, a relation between abnormal CTGs and the presence of NC was found in several cases ([Table 2]). But to summarize the results, no specific CTG pattern was found for NC during delivery.

Table 2 Logistic regression coefficients with NC as the dependent variable and suspicious or pathological CTGs as explanatory variables for each test person A, B or C.

Period

Intercept

Test person A: suspicious CTG

Test person A: pathological CTG

Test person B: suspicious CTG

Test person B: pathological CTG

Test person C: suspicious CTG

Test person C: pathological CTG

*: statistically significant at a 10% level

**: statistically significant at a 5% level

***: statistically significant at every level

60–30 min prior to delivery

1.454 (1.06)

−0.1397 (0.3530)

0.3769 (0.6047)

0.5252 (0.3264)

1.297*** (0.4552)

−0.2714 (0.3223)

−2.321* (1.267)

30 min prior to delivery

1.5191 (1.0722)

0.1834 (0.3011)

0.6302* (0.3469)

0.5784* (0.3468)

0.4507 (0.4073)

−0.2995 (0.3285)

−1.1394 (0.7075)

60–30 min prior to delivery

1.352 (1.192)

−0.0041 (0.3654)

0.238 (0.6307)

0.2994 (0.3415)

1.083** (0.4657)

−0.1603 (0.3334)

−2.003 (1.271)

30 min prior to delivery

1.3912 (1.213)

0.1871 (0.3101)

0.5459 (0.3599)

0.4285 (0.3538)

0.3278 (0.4146)

−0.3499 (0.334)

1.1861 (0.7766)

 

Interobserver variability in the CTG assessments

As the extent of the interobserver agreement among the three test persons could have had a major impact on the results of the study, the interobserver agreement with regard to CTG interpretations was assessed. Agreement values and PoA values were calculated for every investigated parameter and for each pair of test persons. Kappa statistics were used to ensure robust and high-quality testing of the level of agreement between the three test persons with respect to their CTG interpretations [16].


 

Interobserver variability: baseline

The level of agreement between each pair of test persons was very high when baseline was normal. When the baseline was abnormal, the extent of agreement was very low. Test person A tended to suspect bradycardia significantly more often than test person B and test person C, especially in the period 30 minutes prior to delivery (31 cases out of 300). In 22 and 26 of these 31 cases, test person B and test person C reported normal baseline. Test person C tended to suspect tachycardia significantly more often, both in the period from 30 to 60 minutes prior to delivery and in the last 30 minutes prior to delivery. In the 30 minutes prior to delivery, test person C reported tachycardia in 14 out of 300 cases while test person A and test person B only noted tachycardia in four and six out of 300 cases, respectively.


 

Interobserver variability: oscillation

The level of agreement between each pair of test persons was very high when oscillation was normal. When oscillation was abnormal, the extent of agreement was significantly lower. Test persons B and C tended to report reduced oscillation significantly more often than test person A, who classified oscillation as normal in most cases. In the period 30–60 minutes prior to delivery test person A observed reduced oscillation in 15.33%, test person B in 20.33% and test person C in 28.67% of all cases.


 

Interobserver variability: decelerations

The level of agreement for abnormal findings was very low. Test person A tended to report normal findings significantly more often, while the other two test persons, especially test person B, usually found early decelerations. Test person A usually classified the interpreted decelerations as typical variations and less often as late decelerations. In contrast, test person B most often reported early decelerations. Because of this difference, test person B found the most CTGs with decelerations and test person B was defined the fewest CTGs without classified decelerations. Test person C showed a more uniform response behavior in so far as no cluster of a preferentially classified type of decelerations was noted.

This means that the level of agreement between all three pairs of test persons was insufficient with regard to all deceleration types. The highest level of agreement was for the AC pair of test persons regarding typical variable decelerations at 30 minutes prior to delivery and for the BC pair of test persons with regard to early decelerations at 30–60 minutes prior to delivery. The lower 95% CI limits were 0.2079 and 0.2813, respectively. These results demonstrate insufficient agreement regarding the interpretation of decelerations for all three investigators.


 

Interobserver variability: CTG classification

As with the decelerations described by test person A, test person A also tended to report normal CTG according to the FIGO criteria significantly more often than the other two test persons, both 30–60 minutes and 30 minutes before the birth. Test person B generally interpreted more CTGs as suspicious and pathological. The extent of agreement between test persons was insufficient in most cases. The lower limit of the 95% confidence intervals of the PoA was less than 0.5 for all three pairs of test persons only when the CTG was categorized as normal according to FIGO criteria and only for the period 30–60 minutes prior to delivery. The extent of agreement between test person B and test person C was slightly better for all response options compared to test person pairs AB and AC.

[Table 3] summarizes the agreement figures for the various pairs of test persons using the lower limits of the 95% confidence intervals of the PoA for normal and pathological findings.

Table 3 Lower limits of the 95% confidence intervals of the PoA for normal findings and pathologies; a = 60–30 minutes prior to delivery; b = 30 minutes prior to delivery.

Test persons

ABa

Abb

Aca

ACb

BCa

BCb

Normal baseline

0.917

0.797

0.872

0.786

0.898

0.836

Normal oscillation

0.760

0.652

0.725

0.599

0.724

0.663

No decelerations

0.685

0.448

0.800

0.611

0.729

0.491

Normal CTG

0.584

0.366

0.620

0.415

0.629

0.498

Fetal bradycardia

0

0

0

0

0

0

Reduced oscillation

0.281

0.217

0.305

0.188

0.401

0.378

Late decelerations

0

0.024

0.043

0.017

0.027

0.039

Pathological CTG

0.114

0.201

0.019

0.018

0

0.032

 

Assessment of presence of nuchal cord

Statements by the test persons on whether nuchal cord was present based on interpretations of the CTG tracings were heterogeneous. Test person B interpreted CTGs as indicating NC more often, whereas test person A did not suspect NC in the majority of cases. For the period 30–60 minutes prior to delivery, test person B reported expecting NC in 20% of all cases based on the existing CTGs whereas test person A and test person C only reported expected NC in 9% of all cases. [Table 4] shows the extent of agreement between the different pairs of test persons using the lower limit of the 95% confidence intervals of the PoA for both response options “no NC” and “NC”. Overall, only response b (no NC) showed a good agreement between all pairs of test persons, with both the PoA values and the lower limits of the 95% CI of the PoA values significantly higher than 0.5 (range 0.6545–0.7476) at both 30–60 minutes and 30 minutes prior to delivery.

Table 4 Lower limits of the 95% confidence intervals of the PoA for the question whether NC could be present; a = 60–30 minutes prior to delivery; b = 30 minutes prior to delivery.

Aba

ABb

Aca

Acb

Bca

BCb

No NC

0.736

0.655

0.733

0.687

0.748

0.657

NC

0.157

0.236

0.125

0.143

0.142

0.205

With regard to the actual presence or non-presence of NC, it is clear that all three test persons did not reliably recognize the presence of NC using only CTG tracings.

Test person B recognized 37 out of 150 and 36 out of 150 NCs at 30–60 minutes and in the last 30 minutes prior to delivery. The rates of test person A and test person C for recognizing NC based on CTG tracings were even lower and completely insufficient. The sensitivity and specificity of the individual test persons is shown in [Table 5]. The mean sensitivity of the method for all three test persons was 14.7% for the period 30 to 60 minutes prior to delivery and 18.9% for the last 30 minutes prior to delivery. The specificity was 85.1% and 77.8%, respectively.

Table 5 Sensitivity and specificity of the CTG method for each test person.

Test person

30–60 minutes prior to delivery

30 minutes prior to delivery

Sensitivity

A

9.3%

16.0%

B

24.7%

24.0%

C

10.0%

16.7%

Specificity

A

82.7%

78.0%

B

80.7%

70.0%

C

92.0%

85.3%

 

 

Discussion

Speculations on whether NC is present based on an abnormal CTG during delivery are common. These speculations can unsettle both the staff providing the care and the pregnant women herself and ultimately affect the management of the birth, for example by increasing the rate of vaginal-operative deliveries [2] [6] [17].

Recent relevant studies show that a risk to the fetus based on the presence of simple NC alone is unlikely. Most studies found no significant differences in neonatal outcomes between births with and those without NC. It is very probable that NC does not increase long-term morbidity [5] [6] [7] [17] [18].

Our study was unable to detect a specific CTG pattern indicating that NC was present. All the test persons showed similar response behaviors when evaluating CTG tracings, both when NC was actually present and when it was not. Regression analysis of NC depending on the type of deceleration also found no difference for the last 30 minutes prior to delivery. For the period 30 to 60 minutes prior to delivery, the only statistically significant positive correlation was between “no NC” and “no decelerations”. This is plausible as, according to the relevant literature, the presence of NC can lead to decelerations during delivery [2]. It was not possible to precisely differentiate between deceleration types to detect NC.

While an increase in variable decelerations on CTG tracings intrapartum when NC is present has been reported many times in the literature, our study was unable to confirm this. One reason for this could be our small sample size of 300 births. The working group of Carter et al. was only able to show an association between repeat late or variable decelerations in the last 30 minutes before delivery and NC in a cohort of more than 8500 births [2]. The working group of Tagliaferri et al. also reported a statistically significant increase in the number of variable decelerations during delivery in a NC group after week 37 of gestation from a midsized cohort of 808 births, but they used computerized analysis and their finding was not based on visual analysis of CTGs [19].

The results of our study were able to show a correlation between NC and suspicious or pathological CTGs. But this correlation was not detected by all three test persons at the same time. This result reinforces what is already known from many studies, namely, that suspicious or pathological CTGs, summarized as “non-reassuring” CTGs, are present significantly more often when NC is present [2] [5] [13] [20] [21]. Again, the recorded CTG abnormalities were unspecific and could take the form of repeat late or variable decelerations or fetal bradycardia or tachycardia or another inexplicable loss of oscillation. In their retrospective study of a total of 690 births, Miser et al. were able to demonstrate a doubling of pathological CTGs including fetal bradycardia and variable decelerations during delivery in a group with NC [21].

Even though CTG abnormalities are more significantly common when NC is present, the CTG method is unable to identify NC as the cause of observed abnormalities in fetal heart rates. No NC-specific changes in fetal heart rates can be identified. There are several reasons why a CTG may show pathological patterns, including fetal acidosis, taking medication, or an infection [14] [22]. This means that the causes of CTG abnormalities are usually multifactorial. This should make it clear that the FIGO interpretation criteria for CTGs cannot be used to specify the causes of different pathological tracings.

It is very probable that a pathological CTG indicates a fetal risk. Ekengård et al. have shown that the FIGO criteria have a sensitivity of 71% to 97% for fetal acidosis during delivery [23] [24]. However, the specificity of the CTG method is low and a pathological CTG is not always indicative of compromised fetal well-being [25]. In addition, there is also the issue of high interobserver variability of CTG interpretations despite the existence of defined criteria, which was reflected in the results of our study and has also been investigated in detail in a number of studies [9] [15] [26].

One of the limitations of our study is its retrospective nature. The percentage of raw information which may have been recorded incorrectly, for example, relating to the actual presence of NC during delivery, cannot be precisely estimated retrospectively. The test persons were not informed about the 1:1 distribution of the presence or absence of NC in the cohort; nevertheless, the prevalence of NC in our study cohort was higher than would be expected in reality and could have led to selection bias. Due to the subjective nature of CTG interpretations it is impossible to exclude the risk of bias in the study results, even though the internationally defined, sufficiently validated 2015 FIGO criteria were used when interpreting the CTGs. Visual CTG analysis does not always provide repeatable and reliable results. This is especially the case with regard to the decreased agreement on the interpretation of suspicious and pathological CTGs, a finding which is in line with the results of numerous previous studies [9] [15] [26]. As the fetal heart rate is affected by multiple factors during delivery, it is impossible to exclude all influence parameters or to statistically remove them. But this appears to be an important precondition for identifying an apparent criterion which could be present more often with NC. Although all births with known risk factors which could potentially have an impact on fetal heart rate patterns during delivery were consistently excluded from the total cohort, possible unknown influence factors remained. Other limitations affecting interpretation were caused by the intrinsic limitations of the CTG diagnostic method. CTG tracings obtained in the last minutes prior to delivery often show complex, sometimes pathological, changes. Interpreting such CTGs during delivery is particularly challenging and is often associated with diagnostic discrepancies between different medical colleagues [27]. Multifactorial effects on fetal heart rate patterns are one of the main reasons why CTGs and their classification criteria cannot reliably assign fetal heart rate abnormalities to a specific underlying cause.

This study provides an important contribution to the ongoing medicolegal discussion. It is not uncommon for midwives and doctors to be involved in legal disputes about misinterpretations of CTG tracings which resulted in appropriate treatment. The results of our study, which accord with those of many other published studies, can be used to show that CTG alone cannot provide a reliable assessment of the fetal situation and obstetric outcome [9] [15] [26]. At the same time, these data can be used to counter often inflated expectations regarding the meaning and significance of CTGs with which expectant parents confront obstetricians. Providing honest, scientifically based information to parents is the basis for trust-based cooperation and ultimately also for safe and positive birth experiences.

If the CTG method were able to differentiate CTG abnormalities caused by an apparently harmless NC from abnormalities caused by fetal acidosis, the obstetric team could avoid unnecessary interventions such as induction of labor, vaginal-operative deliveries, or delivery by caesarean section. Currently, as the results of this study show, it is not possible to differentiate using CTG alone. Future studies will be needed to find a suitable instrument and criterion to differentiate pathologies identified by CTG more precisely, detect new CTG-specific criteria, or establish other forms of examination which would be capable of reliably detecting NC and could thereby avoid possibly unnecessary interventions.


 

Supplementary Material

  • Supplementary Table S1: Logistic regression coefficients with NC as the dependent variable and the extent of deceleration as the explanatory variable.

  • Supplementary Fig. S1: Questionnaire for the CTG assessments of 300 births with and without nuchal cord.


 

 

Conflict of Interest

The authors declare that they have no conflict of interest.

Supplementary Material


Correspondence

Dr. Lars Brodowski
Klinik für Gynäkologie und Geburtshilfe, Medizinische Hochschule Hannover
Carl-Neuberg-Straße 1
30625 Hannover
Germany   

Publication History

Received: 05 January 2025

Accepted after revision: 15 March 2025

Article published online:
22 May 2025

© 2025. The Author(s). This is an open access article published by Thieme under the terms of the Creative Commons Attribution-NonDerivative-NonCommercial-License, permitting copying and reproduction so long as the original work is given appropriate credit. Contents may not be used for commercial purposes, or adapted, remixed, transformed or built upon. (https://creativecommons.org/licenses/by-nc-nd/4.0/).

Georg Thieme Verlag KG
Oswald-Hesse-Straße 50, 70469 Stuttgart, Germany


  Permissions and Reprints