Archives of Gynecology and Obstetrics

, Volume 288, Issue 1, pp 57–64

Pregnancy risk factors for very premature delivery: what role do hypertension, obesity and diabetes play?


    • Department of PediatricsUniversity of Lübeck
  • G. Stichtenoth
    • Department of PediatricsUniversity of Lübeck
  • J. Weichert
    • Department of Gynecology and ObstetricsUniversity of Lübeck
  • I. R. König
    • Institute of Medical Biometry and StatisticsUniversity of Lübeck
  • M. Schlaud
    • Robert-Koch-Institut
  • A. v.d. Wense
    • Department of NeonatologyAltonaer Kinderkrankenhaus
  • D. Olbertz
    • Department of NeonatologyKlinikum Südstadt
  • H. Gurth
    • Department of Neonatology and Pediatric Intensive CareCarl Gustav Carus University
  • J.-H. Schiffmann
    • Department of PediatricsKlinikum Nürnberg Süd
  • B. Bohnhorst
    • Department of Pediatrics IMedizinische Hochschule Hannover
  • L. Gortner
    • Department of PediatricsUniversity Hospital of Saarland
  • E. Herting
    • Department of PediatricsUniversity of Lübeck
  • W. Göpel
    • Department of PediatricsUniversity of Lübeck
  • The German Neonatal Network, GNN
Maternal-Fetal Medicine

DOI: 10.1007/s00404-013-2739-6

Cite this article as:
Spiegler, J., Stichtenoth, G., Weichert, J. et al. Arch Gynecol Obstet (2013) 288: 57. doi:10.1007/s00404-013-2739-6



Very premature delivery is a major cause of infant morbidity and mortality. Obesity, diabetes and pregnancy hypertension are known risk factors for pregnancy complications. The study aimed to scrutinize differences of pregnancy complications in a cohort of very premature deliveries compared to a national group.


In a multicenter study performed between January 2009 and December 2010 including 1,577 very low birth weight (VLBW) infants, we compared parental reported pregnancy problems of VLBW infants with a national cohort (KIGGS). We compared reported pregnancy complications to reasons for premature delivery and neonatal outcome within the group of VLBW infants.


While parents of the national cohort reported pregnancy-induced hypertension in 8 %, parents of VLBW infants reported this complication more frequently (27 %). Mothers of the national cohort were significantly younger (1 year), suffered less from obesity, anaemia, diabetes. Regression analysis showed that hypertension (OR = 5.11) and advanced maternal age (OR = 1.03) increased the risk for premature birth. Women with hypertension were likely to experience a clinically indicated premature delivery, had more VLBW infants with a moderate growth restriction, but less multiples and their infants had less intraventricular haemorrhages grade 3 or 4. Otherwise, neonatal outcome was correlated with gestational age but not with the pregnancy complications diabetes, hypertension or obesity.


Premature birth seems to be correlated to gestational hypertension and associated problems in about ¼ of VLBW infants. Further studies should focus on preventing and treating gestational hypertension to avoid premature delivery and associated neonatal morbidity.


VLBWHypertensionObesityDiabetesOutcomePremature delivery


Rates of premature births have been increasing over time. At the same time, the prevalence of obesity is rising throughout the world. Accordingly, pregnant women with elevated body mass index (BMI) are seen frequently. Pregnancy outcome of obese women has been the focus of numerous studies in the last few years. Most of them addressed problems seen in term infants, whereas there are few data regarding premature delivery in obese women. Bhattacharya et al. [1] showed that a BMI >35 kg/m2 (WHO grade II) had an odds ratio (OR) of 2.8 for a delivery before 33 weeks of gestation. Two groups [2, 3] reported that morbid obesity (WHO grade III; BMI >40 kg/m2) increased the risk of preterm delivery on medical decision due to pregnancy hypertension (OR 1.8 and 2.8). However, Aly et al. [4] showed that obesity increases the risk for other pregnancy complications, but were not able to show a direct correlation between obesity and premature delivery. Bakker et al. [5] demonstrated that hypertensive disorder during pregnancy negatively affected fetal growth and increased the risk for premature birth. In a Swedish cohort, gestational diabetes has been linked to preterm birth [6], and Weindling [7] described a fivefold increased risk of premature delivery for these women. Obesity, hypertension and gestational diabetes are frequent pregnancy complications; they have all been discussed with regard to premature delivery. However, it remains unclear as to which are the important determinants with respect to very premature delivery.

A representative cross-sectional German health survey (KIGGS) was conducted in the years 2003–2006 [8]. Our objective was to compare pregnancy problems as subjectively reported by parents in the KIGGS study with problems reported by mothers who gave birth to very low birth weight infants (<1,500 g). Furthermore, the purpose of the study was to determine whether specific pregnancy complications increased the risk for premature birth and neonatal complications.



Data from VLBW infants born between 22 + 0 and 36 + 6 weeks of gestation and a birth weight of less than 1,500 g were collected from 28 centres in Germany. Between 2009 and 2010, a total of 1,577 VLBW infants, who answered the KIGGS survey during their first year of life were included in our study. Proper assessment of clinical data was ensured by on-site-monitoring. Following data were taken from maternal or neonatal records and controlled during the on-site-monitoring: preeclampsia, HELLP, intrauterine growth restriction, premature labour, suspected amnion-infection syndrome, gender, birth weight, gestational age, multiples, surgery for PDA, surgery for NEC/FIP, surgery for ROP, IVH, PVL.

For the comparison group, we selected infants 6–12 months of age (n = 935) from a representative cross-sectional German health survey (KIGGS) conducted between 2003 and 2006. All parents received the KIGGS survey (1–2 years) within the first year of life. Both groups of parents received the questionnaire approximately at the same age with the VLBW group being older regarding chronological age, but younger regarding postterm age [VLBW chronological age (median (Q1–Q3)): 8.9 (8.1–9.9), postterm age: 6.4 (5.4–7.5), KIGGs: 7.5 (5.2–9.7)]. The comparison of the KIGGS and VLBW group has been described before [9]. From the questionnaire of both groups, following data were extracted: recent maternal BMI and maternal age. Mothers were asked whether the following diseases had been diagnosed during their pregnancy: anaemia, lower urinary tract infection, hypertension, oedema, Rh-incompatibility, gestational diabetes. Possible answers were: “yes”, “no” or “I do not know”.


According to the WHO classification, obesity was defined as BMI above 30 kg/m2. Bronchopulmonary dysplasia (BPD) was defined as need for supplemental oxygen at 36-week postmenstrual age. Intraventricular haemorrhage (IVH) was per definition any bleeding into the cerebral germinal matrix or the ventricles. Surgery for retinopathy of prematurity (ROP) was defined as any ROP treated with laser- or cryocoagulation. Surgery for necrotizing enterocolitis (NEC) or focal intestinal perforation (FIP) was defined as any surgery necessary for FIP or NEC, surgery for PDA was any surgical ligation for persistent ductus arteriosus.

Premature delivery was classified as clinically indicated if Preeclampsia, HELLP (haemolysis, elevated liver enzymes, low platelets) or intrauterine growth restriction were medically documented reasons for preterm delivery. It was classified as spontaneous if premature delivery was documented as due to suspected amnion infection syndrome or premature labour.

Statistical analyses

Data analysis was performed using SPSS 20.0 data analysis package (IBM) and R version 2.15.0 [10].

To compare the VLBW with the KIGGS group for the continuous variables maternal age and maternal BMI, quantiles were determined. For description, the groups were compared using a Mann–Whitney U test. For all binary variables, sex, obesity, diabetes, anemia, oedema, hypertension, urinary tract infection and rhesus incompatibility, VLBW and KIGGS were compared using Fisher’s exact test for description. Descriptive p values are reported.

In the next step, VLBW group was predicted using the previous variables in logistic regression models. First, a model was developed using all variables and considering all main effects. From this, backward selection was performed to reduce the model to contain only variables with a p less than 5 %. Second, a model was developed including all two-way interactions. Again, backward selection was performed reducing the global model.

To evaluate the stability of the resulting models, a tenfold cross-validation was performed. For this, from the entire data set, ten data sets were created by randomly excluding one tenth of the observations at a time. Within every data set, regression models using only main effects with backward elimination were developed.

For the further analyses, only data from VLBW children were considered. Here, we aimed to compare those with clinically indicated preterm delivery (reasons for delivery HELLP, preeclampsia, growth restriction) versus those with spontaneous preterm delivery (reasons for delivery suspected amnion infection, preterm labour). In detail, the interest was on the binary variables hypertension, diabetes and obesity. A logistic regression model predicting clinically indicated delivery was developed using main effects for hypertension, diabetes and obesity with backward elimination with a p value criterion of 5 %. A second logistic regression model was similarly developed considering all two-way interactions. Again, the stability of the resulting models was evaluated using a tenfold cross-validation. For this, from the entire data set, ten data sets were created by randomly excluding one tenth of the observations at a time. Within every data set, firstly regression models using only main effects with backward elimination were developed. Additionally, models with two-way interactions and backward elimination were developed.

We then aimed at comparing characteristics and complications in the VLBW children depending on risk factors in the mother. Therefore, only data from VLBW children were considered. For this, the following characteristics in the children were considered: birth weight, gestational age, multiples, birth weight <10th percentile, birth weight <3rd percentile, surgery for persistent ductus arteriosus, surgery for necrotizing enterocolitis or focal intestinal perforation, surgery for retinopathy of prematurity, bronchopulmonary dysplasia, intraventricular haemorrhage grade 3 or 4, and periventricular leukomalacia. As influencing maternal factors, we considered hypertension, diabetes, and obesity.

For an overview, descriptive statistics were evaluated for all dependent variables, separately for the risk factor groups. For a systematic comparison, logistic regression models were developed to predict the binary characteristics in the child from the risk factors in the mother taking all possible two-way interactions into account. As before, we applied a backward elimination with a p value criterion of 5 %. Analogously, linear regression models were developed to predict the continuous characteristics from the risk factors with all two-way interactions. As before, we applied a backward elimination with a p value criterion of 5 %.


The study was approved by the local committee on research in human subjects of the University of Luebeck, and the local committees of all participating centres.

Written informed parental consent was given for the research, and publication of the results of each infant included in the study.


Differences between VLBW and KIGGS groups are shown in Table 1. All data are subjective parental information inquired with the same questionnaire at approximately the same age postterm.
Table 1

Self-reported pregnancy-data


VLBW (n = 1,577)

KIGGS (n = 935)

p value


810 (51.4 %)

480 (51.3 %)


Birth weight (median, Q1/Q3)

1,110 (850/1,350)

3,400 (3,032/3,740)



534/1,577 (33.9 %)

31/935 (3.3 %)


Age mother (years)

32 (28/36)

31 (27/35)


BMI mother (median, Q1/Q3)

23.8 (21.3/27.7)

23.5 (21.0/26.6)



240/1,530 (15.7 %)

95/917 (10.3 %)



123/1,345 (9.1 %)

47/857 (5.5 %)



431/1,386 (31.1 %)

258/855 (30.2 %)



123/1,329 (9.3 %)

56/857 (6.5 %)


Lower urinary tract infection

138/1,320 (10.5 %)

72/861 (8.4 %)



391/1,418 (27.6 %)

71/862 (8.2 %)



31/1,276 (2.4 %)

28/837 (3.3 %)


Continuous variables are given as median (Q1/Q3), binary variables are given as absolute numbers and percentage

Based on logistic regression models, VLBW group was predicted from maternal risk factors. The final model showed predictive effects for maternal age (OR = 1.03, 95 % CI 1.01–1.05, p = 3.21e−04), oedema (OR = 0.58, 95 % CI 0.47–0.73, p = 1.24e−06), and hypertension (OR = 5.11, 95 % CI 3.77–6.91, p = 4.00e−26). Thus, whereas increasing maternal age and hypertension increase the chance for delivery of a VLBW infant, oedema decrease the risk statistically. Tenfold cross-validation showed that as in the original model, all of the ten models included the variables maternal age, oedema and hypertension with effects in the same direction as in the original model. The model from one data set additionally includes maternal BMI. This adds credibility to the original model as described above.

When two-way interactions were allowed, the same variables were included (maternal age: OR = 1.03, 95 % CI 1.01–1.05, p = 5.47e−04; oedema: OR = 7.82, 95 % CI 2.61–23.42, p = 2.36e−04; hypertension: OR = 34.93, 95 % CI 14.24; 85.69, p = 8.46e−15), and the model additionally included an interaction between oedema and hypertension (OR = 0.25, 95 % CI 0.14; 0.44, p = 2.75e−06). The interpretation of this model is facilitated considering the following numbers: Considering mothers without hypertension, 15 % in the VLBW group have oedema but 27 % in the KIGGS group. Conversely, within the mothers with hypertension, 73 % in the VLBW group have oedema but only 58 % in the KIGGS group.

The logistic regression model predicting clinically indicated delivery only contained hypertension as predictive factor (OR = 20.00, 95 % CI 14.29–33.33, p = 4.08e−56). In the model including two-way interactions, hypertension, obesity and their interaction survived backward elimination (hypertension: OR = 33.33, 95 % CI 16.67; 50.00, p = 8.60e−43; obesity: OR = 7.69, 95 % CI 1.61–33.33, p = 0.0101; obesity × hypertension: OR = 0.30, 95 % CI 0.12–0.72, p = 0.0070). In the tenfold cross-validation, in all of the ten data sets for main effects, hypertension was selected as only variable. For the models including two-way interactions, the interaction of hypertension with obesity was selected nine out of ten times, an interaction of hypertension with diabetes three times, and an interaction of diabetes with obesity four times. As for the previous analysis, this adds credibility to the original models.

When aiming at prediction of children’s characteristics based on maternal risk factors, no variable remained in the regression model for the variables birth weight <3rd percentile, surgery for persistent ductus arteriosus, surgery for necrotizing enterocolitis or focal intestinal perforation, surgery for retinopathy of prematurity, bronchopulmonary dysplasia, periventricular leukomalacia (results are summarized in Table 2). However, we found that multiples were less likely in mothers with hypertension (OR = 0.21; 95 % CI 0.15–0.30; p = 2e−16), and more likely in mothers with diabetes (OR 1.57; 95 % CI 1.02–2.40; p = 0.0399). Birth weight below the 10th percentile was likely in mothers with hypertension (OR = 2.33; 95 % CI 1.77–3.08; p = 2.13e−09). Higher grade IVH (grade 3 or 4) was less likely in infants of mothers with hypertension (OR = 0.35; 95 % CI 0.17–0.71; p = 0.0034). Birth weight was higher if mothers suffered from diabetes (estimate 62.12; 95 % CI 7.40; 116.84; p = 0.0263), and gestational age was higher if mothers suffered from either hypertension (estimate 1.08; 95 % CI 0.74; 1.41; p = 3.63e−10) or diabetes (estimate 0.58; 95 % CI 0.07; 1.10; p = 0.0254).
Table 2

Neonatal outcome (VLBW) stratified to maternal pregnancy complications


















Male gender (n, %)

174 (391, 44.5 %)

562 (1,027, 54.7 %)

58 (123, 47.2 %)

651 (1,222, 53.3 %)

123 (240, 51.3 %)

663 (1,290, 51.4 %)

Clinically indicated premature birth (n, %)

282a (319, 88.4 %)

204 (767, 26.6 %)

50 (101, 49.5 %)

376 (927, 40.6 %)

96b (172, 55.8 %)

415 (1,002, 41.4 %)

Birth weight (median, Q1/Q3)

1,100 (860/1,342)

1,120 (860/1,360)

1,190* (973/1,392)

1,110 (860/1,350

1,120 (860/1,360

1,105 (829/1,301)

Gestational age (median, Q1/Q3)

29.9** (28.0/31.6)

28.7 (26.6/30.6)

29.7* (27.4/31.1)

29.0 (26.9/30.9)

29.0 (27.0/30.9)

29.1 (27.0/30.5)

Multiples (n, %)

60** (391, 15.3 %)

427 (1,027, 41.6 %)

44 (123, 35.8 %)

427 (1,222, 34.9 %)

68 (240, 28.3 %)

458 (1,290, 35.5 %)

Birth weight < 10th percentile (n, %)

114** (391, 29.1 %)

154 (1,027, 15.0 %)

24 (123, 19.5 %)

219 (1,222, 17.9 %)

49 (240, 20.4 %)

241 (1,290, 18.7 %)

Birth weight < 3rd percentile (n, %)

27 (391, 6.9 %)

58 (1,025, 5.7 %)

6 (122, 4.9 %)

74 (1,220, 6.1 %)

11 (237, 4.6 %)

78 (1,288, 6.1 %)

Surgery PDA (n, %)

7 (388, 1.8 %)

35 (1,024, 3.4 %)

3 (122, 2.5 %)

38 (1,219, 3.1 %)

8 (240, 3.3 %)

38 (1,282, 3.0 %)

Surgery for NEC or FIP (n, %)

6 (388, 1.5 %)

34 (1,024, 3.3 %)

2 (122, 1.6 %)

38 (1,219, 3.1 %)

8 (240, 3.3 %)

38 (1,282, 3.0 %)

Surgery for ROP (n, %)

6 (386, 1.6 %)

20 (1,014, 2.0 %)

1 (120, 0.8 %)

22 (1,209, 1.8 %)

8 (236, 3.4 %)

22 (1,274, 1.7 %)

BPD (n, %)

48 (353, 13.6 %)

134 (922, 14.5 %)

14 (116, 12.1 %)

161 (1,092, 14.7 %)

33 (210, 15.7 %)

163 (1,169, 13.9 %)

IVH grade 3 or 4 (n, %)

9** (389, 2.3 %)

65 (1,020, 6.4 %)

7 (123, 5.7 %)

63 (1,214, 5.2 %)

11 (240, 4.6 %)

69 (1,280, 5.4 %)

PVL (n, %)

7 (383, 1.8 %)

32 (1,003, 3.1 %)

4 (116 3.4 %)

33 (1,195, 2.8 %)

6 (235, 2.6 %)

34 (1,259, 2.7 %)

* Significant differences p < 0.05; ** significant difference p < 0.01

aOR = 33.3

bOR = 7.7


If comparing pregnancy problems of VLBW mothers with controls from the KIGGS study, hypertensive disorders significantly increased the risk for preterm birth. Pregnancy-induced hypertension is a frequent problem worldwide with an increasing incidence. Whereas rates of 6–8 % of pregnancy-induced hypertension have been documented in the USA [11], there are even higher rates estimated in other populations [12]. In the KIGGS cohort, a rate of 8 % of pregnancy-induced hypertension meets the expected rate for Germany [13]. Mothers in our cohort who have had a very preterm birth had a threefold higher rate of hypertension. Logistic regression analyses showed hypertension as an important risk factor for premature birth in our cohort. Even though hypertension has been identified as a risk factor for premature birth, its impact has not been elaborated in larger studies of VLBW infants. However, our data can neither reflect the severity of hypertension nor targeted treatment strategies before birth or details of maternal surveillance. Interestingly, women with oedema but no hypertension were less likely to deliver prematurely. However, if suffering from oedema and hypertension, the risk for a premature delivery quadruples. Suffering from both problems putatively means that preeclampsia is the imminent problem causing a clinically indicated premature delivery. Gestational age might be an important bias for these data since oedema are often seen as an isolated harmless problem late during the pregnancy.

In our cohort, hypertension was the most striking pregnancy complication regarding preterm delivery. Even though advanced maternal age is a significant risk factor, clinically this difference is a not a discerning risk factor for the obstetrician. Interaction of pregnancy complications was not noted for other risk factors (advanced maternal age, obesity, diabetes, anemia), i.e. having combined pregnancy problems did not further increase the risk for premature delivery. This is surprising given the increasing risk for hypertension with age. Multiples were less likely seen in women with hypertension in our cohort; however, in current literature, higher rates for hypertension and preeclampsia have been described in twin gestations [14, 15]. Choosing birth weight and not gestational age causes a bias of our data towards more multiples due to the risk of growth restriction that is associated with hypertension. In an earlier cohort of our study, we could show that VLBW twins did not differ significantly regarding birth weight compared to singletons [16]. This finding is surprising and needs to be confirmed in other cohorts. However, our data only relate to VLBW infants and might not be true in advanced gestational ages. Similar to our findings, Geradin et al. [15] showed an increased risk for gestational diabetes in twin pregnancy.

Obesity is known to be one of the major, independent causes of gestational hypertension [17]. A recent meta-analysis [18] reported an increased risk of preterm birth in obese women and a subgroup of growth restricted infants. However, the analysis did not control for hypertension as a confounder. In our cohort, we were not able to show an association of obesity with premature delivery, even though women with hypertension had a higher BMI. Therefore, being obese increased the risk for having gestational hypertension but it did not seem to increase the risk for premature birth in the absence of hypertension in our cohort as previously published by Aly [4]. In our analysis, we used a maternal BMI inquired approximately 9 months after a premature delivery. There are insufficient data as to how a weight at this time correlates with maternal weight during pregnancy, especially after a premature delivery. However, we think that there will be only few women with a BMI <25 before pregnancy and a rapid weight gain/BMI >30 nine months later. Our data might be confounded by women who had a BMI between 25 and 30.

Generally, preterm delivery is classified as either clinically indicated or spontaneous. Spontaneous preterm births comprise about 2/3 of preterm deliveries and are grossly the consequence of preterm labour and/or preterm rupture of membranes. Acute or chronic infections have been identified as major risk factors in this context. Prevention of premature birth is the key issue and has mainly focused on identification and treatment of infection/inflammation over the last decades. Some of these findings have already been implemented in national guidelines [19]. In our cohort of VLBW infants, we could demonstrate that hypertension is associated with a considerably higher risk for a clinically indicated preterm delivery. Women without hypertension were more likely to experience spontaneous preterm birth due to premature labour, local infections and related problems. This group had the expected rate of spontaneous premature births. However, women with hypertensive disorders had a very high risk (OR 20.0) for a clinically indicated preterm delivery following HELLP, preeclampsia and/or intrauterine growth restriction. Infections and spontaneous preterm onset of labour were only rarely found in this group. Analysing combined pregnancy problems showed that hypertension in combination with obesity lowered the risk for a clinically indicated preterm birth, compared to hypertension alone. This might be explained with the fact that the combination of obesity and hypertension is treated more aggressively so that a clinically indicated preterm birth can be avoided and pregnancy be prolonged.

Following the analysis of the reasons for preterm delivery, we were interested in the outcome of VLBW infants with regard to reported pregnancy complications (hypertension, diabetes and obesity).

Birth weight of VLBW infants was higher in mothers with diabetes confirming the data seen in term infants of diabetic mothers. Moderate growth restriction was seen in mothers with hypertensive disorders, as shown previously [20, 21]. Infants of normotensive mothers were born more prematurely without being growth restricted. In this group, suspected infection and preterm labour were the most frequent causes of preterm delivery. Major neonatal outcome parameters of all groups were similar except for lower rates of intraventricular haemorrhage in the hypertensive group. As findings of the EPIPAGE study group [22] suggested, being exposed to intrauterine infections (premature labour/premature rupture of membranes) markedly increased the risk of intraventricular haemorrhage while infants born to hypertensive mothers are less at risk. Sarkar [23] confirmed via logistic regression analyses the impact of gestational age with the development of severe IVH. In our normotensive group infections were more frequent and children were born more prematurely. This explains in our opinion the lower rate of higher grade IVH in infants of mothers with hypertension.

One has to acknowledge that our analyses of reasons for a preterm birth as well as postnatal problems relate only to VLBW infants. This group is defined by birth weight and associated with a low gestational age. Data might be different for later gestational ages.

Is there a need to counsel mothers with impending premature birth according to their pregnancy complications and not merely gestational age? Probably yes. Gestational age still has the most crucial impact on postnatal outcome in our group of VLBW infants. However, there are slight differences in postnatal outcome. Moderate growth restriction is seen more often if mothers had gestational hypertension and IVH is seen more often after suspected intrauterine infections. These aspects can be taken into account when counselling parents with imminent preterm delivery.

Nevertheless, our study has several limitations. We present pregnancy complications as subjectively recorded by the parents. We had no data regarding the severity of pregnancy complications (diabetes, hypertension) and how these conditions have been diagnosed, treated and monitored.


Infections have been previously identified as a major risk factor for premature delivery and there is ongoing research how to prevent this. Based on our findings, we were able to show that premature birth is associated with gestational hypertension and associated problems in a high percentage of VLBW infants. Further studies should focus on preventing and effectively treating gestational hypertension to avoid premature delivery and associated neonatal morbidity. No significant association with premature delivery was seen in obesity and diabetes. However, hypertension and obesity were associated with higher rates of clinically indicated preterm births within the group of VLBW infants.

Regarding the neonatal outcome, maternal hypertension was associated with mild growth restriction but less higher grades of IVH, gestational diabetes was associated with higher birth weight, and gestational age of VLBW infants was higher in mothers with hypertension or diabetes.


This study is part of a larger project (German neonatal network, GNN) and is funded by the German Ministry for Education and Research (BMBF-Grant-no. 01ER0805).

Conflict of interest


Copyright information

© Springer-Verlag Berlin Heidelberg 2013