Associations of gestational weight loss with birth-related outcome: a retrospective cohort study


Dr A Beyerlein, Ludwig-Maximilians University of Munich, Institute for Social Paediatrics and Adolescent Medicine, Division of Epidemiology, Heiglhofstr. 63, 81377 Munich, Germany. Email


Please cite this paper as: Beyerlein A, Schiessl B, Lack N, von Kries R. Associations of gestational weight loss with birth-related outcome: a retrospective cohort study. BJOG 2011;118:55–61.

Objective  Although the prevention of gestational weight loss (GWL) has become a priority for clinicians in the past few decades, recent work has suggested that GWL may be beneficial for obese mothers. We aimed to identify the potential beneficial or adverse associations of GWL with pregnancy outcome stratified by maternal body mass index (BMI) category.

Design  Retrospective cohort study.

Setting and population  Data on 709 575 singleton deliveries in Bavarian obstetric units from 2000–2007 were extracted from a standard dataset for which data are regularly collected for the national benchmarking of obstetric units.

Methods  We calculated the odds ratios (ORs) for adverse pregnancy outcome by GWL (explanatory variable) compared with nonexcessive weight gain with adjustment for confounders and stratification by BMI category (underweight, BMI < 18.5 kg/m2; normal weight, BMI = 18.5–24.9 kg/m2; overweight, BMI = 25–29.9 kg/m2; obese class I, BMI = 30–34.9 kg/m2; obese class II, BMI = 35–39.9 kg/m2; obese class III, BMI ≥ 40 kg/m2).

Main outcome measures  Pre-eclampsia, nonelective caesarean section, preterm delivery, small or large for gestational age (SGA/LGA) birth and perinatal mortality.

Results  GWL was associated with a decreased risk of pregnancy complications, such as pre-eclampsia and nonelective caesarean section, in overweight and obese women [e.g. OR = 0.65 (95% confidence interval: 0.51, 0.83) for nonelective caesarean section in obese class I women]. The risks of preterm delivery and SGA births, by contrast, were significantly higher in overweight and obese class I/II mothers [e.g. OR = 1.68 (95% confidence interval: 1.37, 2.06) for SGA in obese class I women]. In obese class III women, no significantly increased risks of poor outcomes for infants were observed.

Conclusions  The association of GWL with a decreased risk of pregnancy complications appears to be outweighed by increased risks of prematurity and SGA in all but obese class III mothers.


Interest has focused recently on gestational weight gain (GWG) in relation to short- and long-term pregnancy outcomes, such as birth weight and offspring obesity.1–8 The prevention of low GWG or even gestational weight loss (GWL) has become a priority for clinicians in recent decades.9 The Institute of Medicine currently recommends GWG of at least 5, 7 and 11.5 kg for obese, overweight and normal-weight mothers, respectively.10 Although some studies have suggested that obese women should gain even less weight to reduce the risk of adverse pregnancy outcomes,1,4 Kiel et al.3 have shown that, for obese class III women, even GWL may be beneficial.

In a very recent study based on a small number of observations (n = 2012), it was hypothesised that GWL might benefit obese mothers with respect to maternal postpartum weight retention, offspring obesity, preterm delivery and small (SGA) or large (LGA) for gestational age births.5 Previous analyses carried out by our study group have suggested that modest GWL does not lead to adverse birth weight outcome for term deliveries in overweight and obese mothers.11

We used data for 709 575 singleton deliveries in Bavarian obstetric units between 2000 and 2007 to assess the prevalence of GWL stratified by maternal body mass index (BMI) category and to identify potential associations of GWL with a variety of pregnancy-related maternal and neonatal health outcomes.


Maternal and neonatal data on 819 905 singleton deliveries were available in Bavarian obstetric units from 2000–2007.

The data were extracted from a standard dataset for which data are regularly collected electronically for the national benchmarking of obstetric units in terms of clinical performance. These data are recorded by medical and paramedical hospital staff. Standardised definitions for these data, as well as corresponding performance indicators, are published by the Institut für angewandte Qualitätsförderung und Forschung im Gesundheitswesen [the Institute for Applied Quality Promotion and Research in Public Health Care (AQUA Institute); (homepage in German)]. Random data checks are conducted to assess data quality. To date, no systematic biases have been detected. The Bayerische Arbeitsgemeinschaft für Qualitätssicherung (BAQ; Bavarian Working Group on Clinical Quality Assessment) conducts corresponding regional evaluations for Bavaria. Data are transferred electronically from obstetric units after information that may identify individuals has been removed and replaced by an anonymous unique reference number.

Data on maternal weight and gestational age at booking, as well as on smoking, are extracted by midwives and nurses from the mother’s pregnancy booklet. Weight measurements are routinely taken by the medical staff in the outpatient clinics or by midwives at the time of appointment and recorded in the pregnancy booklet. There is a general agreement that the patient is weighed without shoes and heavy clothes.

Gestational age was calculated from the expected date of delivery based on the date of the last menstrual period and corrected by first-trimester ultrasound measurement of the crown–rump length in cases in which there was a substantial difference (usually >14 days) between the two estimates. Neonatal anthropometric data were abstracted from the hospital records. We calculated GWG as the maternal weight at last measurement prior to delivery minus the maternal weight at booking (both recorded in kilograms), and defined GWL as negative GWG.

Women were grouped by BMI (kg/m2) at booking as underweight (BMI < 18.5), normal weight (18.5–24.9), overweight (25–29.9), obese class I (30–34.9), obese class II (35–39.9) and obese class III (≥40).3,12 We classified the following GWG values as ‘excessive’ according to the Institute of Medicine (IOM) criteria:10 >18.0 kg for underweight women, >16.0 kg for normal-weight women, >11.5 kg for overweight women and >9.0 kg for obese women (including all degrees of obesity).

We excluded a total of 81 371 mothers with missing values for weight at booking (n = 57 485) or delivery (n = 15 210), or for height (n = 9840), or with implausible values (n = 456) for maternal height (<100 or >220 cm) or weight at first antenatal care visit or prior to delivery (<30 or >300 kg). We excluded a further 28 959 mothers with booking at <4 weeks before delivery or unknown date of booking, leaving 709 575 deliveries remaining for our analyses. For the main analysis, in which we compared mothers with GWL with those with non-negative GWG, we also excluded women with excessive GWG, yielding an eventual sample size of 445 323. We excluded these women because women with excessive GWG constitute a group with elevated risk for several adverse pregnancy outcomes, such as pre-eclampsia (data not shown), and their inclusion in the non-negative GWG group might have led to an underestimation of the effects of GWL.

The following conditions were considered to be adverse pregnancy or birth outcomes: pre-eclampsia, nonelective caesarean section, preterm delivery (prior to 37 + 0 completed weeks of gestation), the infant being born SGA or LGA, and perinatal mortality (stillbirths and deaths within the first 7 days of life). SGA and LGA birth weights were defined as the lower and upper 10%, respectively, of the 1992 German sex- and gestational age-specific reference percentiles for birth weight.13 Pre-eclampsia was diagnosed on the basis of pregnancy-induced high blood pressure (>140/90 mmHg later than 20 weeks of gestation) and proteinuria (≥300 mg protein/day), according to the guidelines of the International Society for the Study of Hypertension in Pregnancy (ISSHP). The mode of delivery and the condition of the newborn (alive/not alive) were immediately recorded at the time of delivery, and deaths within the first few days of life are routinely reported to BAQ. We were not able to analyse long-term outcomes, such as maternal weight retention or childhood obesity.

We calculated the proportions of these outcomes for each BMI class. In addition, we computed odds ratios (ORs) for GWL versus nonexcessive GWG, adjusting for gestational and pre-gestational diabetes, smoking during pregnancy, gender, parity (primiparous versus multiparous) and maternal age, as all these variables showed significant differences (P < 0.05) between women with and without GWL. In order to assess whether the risks of perinatal mortality were mainly attributable to prematurity, we also adjusted for preterm delivery in the respective logistic regression models.

To explore the clinical relevance of preterm delivery, we calculated the prevalences of very early (before 30 completed weeks of gestation) early (30–33 weeks) and moderate (34–36 weeks) preterm deliveries, stratified by BMI and nonexcessive GWG versus GWL. To analyse the potentially different effects of low, medium and high GWL, we assessed the adjusted ORs for adverse pregnancy outcomes in logistic regression models with GWL as a dummy-coded explanatory variable (low, 1–5 kg; moderate, 6–10 kg; high, >10 kg; with nonexcessive GWG as reference).


The median booking date was at eight completed weeks of gestation. <1% of the mothers for whom records were available booked before 5 weeks of gestation. The distributions of pregnant women with GWG and GWL with respect to date of booking were similar: the proportion of mothers booking before 15 weeks of gestation was 95% amongst those with GWG and 92% amongst those with GWL (24 weeks: 99% and 98%, respectively). The prevalences of each outcome under study differed by <15% between the available dataset as a whole and the data used for the final analyses (including women with excessive GWG; data not shown).

The proportion of pregnant women with GWL increased with increasing maternal BMI (Figure 1); most women with GWL showed low GWL values. All factors, except maternal nationality and age, and the sex of the newborn, showed significant (P < 0.01) associations with GWL. Among the 3521 mothers with GWL, there were 343 mothers (9.74%) with a history of underlying severe disease and 50 mothers (1.42%) with recorded hyperemesis. These proportions were slightly lower in the group of mothers with no GWL (Table 1).

Figure 1.

 Proportions of women with different amounts of gestational weight loss (GWL) by body mass index (BMI) category: underweight (UW), normal weight (NW), overweight (OW), obese class I (OB1), obese class II (OB2) and obese class III (OB3).

Table 1.   Maternal and pregnancy-related variables stratified by gestational weight change category: gestational weight loss (GWL), nonexcessive gestational weight gain (NEGWG; reference) and excessive gestational weight gain (EGWG)
 GWL (n = 3521)NEGWG (n = 441 802)EGWG (n = 264 252)
  1. SD, standard deviation.

  2. *Difference between GWL and NEGWG groups: P < 0.01.

Diabetes mellitus [n (%)]*51 (1.45)1714 (0.39)1338 (0.51)
Gestational diabetes [n (%)]*196 (5.57)6648 (1.50)4169 (1.58)
Hyperemesis [n (%)]*50 (1.42)1748 (0.40)975 (0.37)
Severe disease in the past (e.g. heart, lung, liver, kidney, central nervous system or psychiatric) [n (%)]*343 (9.74)34 956 (7.91)21 773 (8.24)
Smoking in pregnancy [n (%)]*563 (15.99)35 491 (8.03)24 195 (9.16)
German mother [n (%)]2874 (81.62)359 916 (81.47)215 927 (81.71)
Male offspring [n (%)]1665 (47.29)222 255 (50.32)126 048 (47.71)
Primiparous [n (%)]*1146 (32.55)199 175 (45.08)144 187 (54.56)
Underweight [n (%)]*19 (0.54)26 173 (5.92)4328 (1.64)
Normal weight [n (%)]*480 (13.63)331 439 (75.02)128 673 (48.69)
Overweight [n (%)]*808 (22.95)54 512 (12.34)90 015 (34.06)
Obese class I [n (%)]*889 (25.25)18 308 (4.14)30 043 (11.37)
Obese class II [n (%)]*688 (19.54)7756 (1.76)8229 (3.11)
Obese class III [n (%)]*637 (18.09)3614 (0.82)2964 (1.12)
Maternal body mass index (kg/m2) [mean (SD)]*33.5 (9.7)23.1 (4.2)25.4 (4.8)
Maternal age at delivery (years) [mean (SD)]30.2 (5.5)30.6 (5.4)29.7 (5.2)
Birth weight (g) [mean (SD)]*3207 (666)3275 (541)3497 (502)
Gestational age (weeks) [mean (SD)]*38.4 (2.7)38.8 (2.0)39.2 (1.6)

As there were only 19 underweight mothers with GWL, the results for this group were not meaningful and results for underweight women are not presented in Table 2. No underweight mother showed medium or high GWL.

Table 2.   Proportions and crude/adjusted odds ratios (ORs) for adverse pregnancy outcomes in mothers with nonexcessive gestational weight gain (reference) versus gestational weight loss (GWL)
  Normal weightOverweightObese class IObese class IIObese class III
Reference (n = 331 439)GWL (n = 480)Reference (n = 54512)GWL (n = 808)Reference (n = 18 308)GWL (n = 889)Reference (n = 7756)GWL (n = 688)Reference (n = 3614)GWL (n = 637)
  1. Adjustments were made for gestational and pre-gestational diabetes, smoking in pregnancy, the offspring’s sex, parity and maternal age, and preterm delivery, if appropriate. ORs in bold are significant.

  2. CI, confidence interval; LGA, large for gestational age; SGA, small for gestational age.

  3. *Additionally adjusted for preterm delivery.

Crude OR (95% CI)0.51 (0.13, 2.03)1.10 (0.65, 1.87)0.67 (0.42, 1.07)0.56 (0.37, 0.86)0.63 (0.46, 0.87)
Adjusted OR (95% CI)0.55 (0.14, 2.19)1.03 (0.59, 1.79)0.70 (0.44, 1.11)0.59 (0.39, 0.91)0.64 (0.46, 0.89)
Nonelective caesarean section%10.0010.6311.169.0312.948.6615.2411.6316.9113.19
Crude OR (95% CI)1.07 (0.80, 1.43)0.79 (0.62, 1.01)0.64 (0.50, 0.81)0.73 (0.58, 0.93)0.75 (0.58, 0.95)
Adjusted OR (95% CI)1.13 (0.84, 1.52)0.74 (0.58, 0.95)0.65 (0.51, 0.83)0.80 (0.62, 1.02)0.76 (0.60, 0.98)
Preterm delivery%7.8218.969.1412.509.2210.469.6310.6110.2110.68
Crude OR (95% CI)2.76 (2.20, 3.47)1.42 (1.15, 1.75)1.15 (0.92, 1.43)1.11 (0.86, 1.44)1.05 (0.80, 1.38)
Adjusted OR (95% CI)2.73 (2.17, 3.43)1.38 (1.12, 1.71)1.17 (0.94, 1.46)1.14 (0.88, 1.47)1.05 (0.80, 1.38)
Crude OR (95% CI)1.63 (1.28, 2.08)1.63 (1.33, 1.98)1.73 (1.41, 2.11)1.40 (1.08, 1.81)1.26 (0.93, 1.70)
Adjusted OR (95% CI)1.50 (1.17, 1.92)1.49 (1.22, 1.82)1.68 (1.37, 2.06)1.45 (1.12, 1.89)1.31 (0.96, 1.79)
Crude OR (95% CI)0.60 (0.37, 0.96)0.70 (0.53, 0.92)0.92 (0.74, 1.14)0.60 (0.46, 0.78)0.63 (0.49, 0.80)
Adjusted OR (95% CI)0.63 (0.39, 1.01)0.77 (0.58, 1.02)0.94 (0.76, 1.17)0.58 (0.45, 0.76)0.60 (0.47, 0.77)
Perinatal mortality%0.471.460.681.110.901.240.981.600.720.63
Crude OR (95% CI)3.12 (1.48, 6.59)1.64 (0.84, 3.19)1.38 (0.75, 2.55)1.64 (0.87, 3.11)0.87 (0.30, 2.51)
Adjusted OR (95% CI)3.08 (1.46, 6.51)1.61 (0.83, 3.13)1.41 (0.76, 2.61)1.72 (0.91, 3.26)0.92 (0.32, 2.65)
Adjusted OR (95% CI)*1.58 (0.73, 3.40)1.31 (0.66, 2.60)1.24 (0.66, 2.34)1.65 (0.85, 3.23)0.88 (0.30, 2.55)

In obese women, GWL was associated with a lower risk of pre-eclampsia [e.g. adjusted OR = 0.59 (95% CI: 0.39, 0.91) in obese class II women; Table 2]. In addition, a lower risk for nonelective caesarean section was found for overweight and obese women [e.g. adjusted OR = 0.65 (95% CI: 0.51, 0.83) in obese class I women]. GWL was not associated with beneficial effects on these outcomes for underweight and normal-weight women.

In all maternal BMI groups, except for underweight women, GWL was associated with an increase in SGA and a decrease in LGA births [e.g. adjusted ORs for SGA and LGA births in obese class I women: 1.68 (95% CI: 1.37, 2.06) and 0.94 (95% CI: 0.76, 1.17), respectively]. The ORs for preterm delivery in women with GWL were increased significantly for normal-weight and overweight mothers and decreased gradually with increasing maternal BMI category. Similarly, lower risks for perinatal mortality were observed in higher BMI groups, but no significant association was observed in any BMI group after adjustment for preterm delivery. In obese class III women, no significantly increased risk of any poor neonatal outcome was observed.

In mothers with GWL, the prevalence of any kind of premature delivery (at <30, 30–33 and 34–36 weeks) was considerably higher than in mothers with nonexcessive GWG, irrespective of maternal BMI category (data not shown).

Reduced risks of pre-eclampsia and nonelective caesarean section in overweight and obese women were also detected for low GWL (≤5 kg) alone. Similar associations with low GWL were also found with respect to adverse effects on the child’s health, such as preterm delivery. No dose effects for moderate and high GWG were observed (data not shown).

Discussion and conclusions

In this analysis of representative data on more than 700 000 deliveries in Bavarian hospitals, GWL was found to occur predominantly in obese women and was very rarely observed in normal-weight or overweight mothers. For overweight and obese mothers, GWL was associated with a reduction in the risk of pregnancy complications, such as pre-eclampsia and nonelective caesarean section. However, GWL was associated with an increased risk of adverse outcomes in the offspring, such as preterm delivery and SGA birth weight in normal-weight, overweight and obese class I or II mothers. In obese class III mothers, the potential benefits with respect to pregnancy complications may not be outweighed by an increased neonatal risk. Virtually all associations of GWL with pregnancy outcomes were detectable in mothers who lost ≤5 kg during pregnancy and therefore do not solely reflect an impact of extreme GWL.

As a consequence of its cross-sectional design, our study does not formally allow conclusions to be drawn about the causal relationships between GWL and pregnancy outcomes. Under the hypothetical assumption that our findings indicate causality, the observed effect sizes may be compared as follows. Among 1000 overweight women who lose weight during pregnancy, 15 cases of pre-eclampsia and 26 cases of LGA birth would be prevented according to the results shown in Table 2, and 50 additional cases of SGA birth and 34 additional preterm deliveries would occur. Among 1000 obese class II women, GWL would prevent 24 cases of pre-eclampsia and 55 cases of LGA birth, but lead to an additional 28 cases of SGA birth and ten additional preterm deliveries. In 1000 obese class III women, GWL would lead to 36 fewer cases of pre-eclampsia and 62 fewer cases of LGA birth, and lead to only 16 additional cases of SGA birth and five additional preterm deliveries.

Perinatal mortality represents the most severe outcome of those considered in our study, but was not found to be associated significantly with GWL in overweight and obese mothers. In normal-weight mothers, significant associations between GWL and mortality disappeared after adjustment for preterm delivery. These findings indicate that GWL is not likely to affect perinatal mortality substantially in overweight and obese mothers, but the opposite might be true for normal-weight mothers, for whom preterm delivery appears to be one of the causes of perinatal mortality.

Our results are in accordance with those reported by Kiel et al.,3 who showed no potential beneficial effects of GWL in obese class I or II women, but found that obese class III women may benefit from GWL. In our study, the considerably increased risks of preterm delivery and SGA birth emphasise the dangers of GWL, especially in overweight women. Assuming causality, our data may therefore provide an argument against GWL in overweight or obese class I women, for whom Oken et al.5 suggested potential beneficial effects based on analyses of a comparatively small population with few cases of GWL. Although an increased risk of prematurity was reported in the study by Oken et al.,5 the absence of valid data on rare adverse fetal outcomes, such as late fetal and early neonatal deaths, as a result of the small sample size, precluded the assessment of these important outcome parameters.

Although there were independent associations of GWL with perinatal mortality, preterm delivery was a major confounder for this important outcome. The risk of preterm delivery associated with GWL decreased markedly with increasing maternal BMI category, suggesting that potential adverse effects of GWL might be smaller in heavier women. Our previous analyses showing potentially beneficial effects of GWL for overweight and obese women with respect to avoidance of adverse birth weight outcome were based on full-term deliveries only, and therefore allowed no conclusions to be drawn regarding the effect of GWL on preterm delivery.11

The numbers of underweight mothers, even in our large study, were too small to allow valid conclusions to be drawn, given the very low prevalence of GWL in underweight mothers. However, there is apparently no reason why GWL should be recommended for underweight mothers.

Unfortunately, our data do not allow us to distinguish between intentional and unintentional GWL. If GWL were related to underlying morbidity, the observed effects may reflect underlying disease rather than independent effects of GWL. However, as the prevalences of hyperemesis and a history of underlying severe disease were similarly low in mothers with and without GWL, it is unlikely that the observed associations are caused by other illness. It seems likely that most mothers who lost weight did so intentionally, but not on the physician’s advice, as it is no longer considered to be good practice in Germany to encourage women to lose weight during pregnancy. Therefore, and because GWL occurred only rarely, the BAQ survey did not attempt to ascertain the reasons for GWL. Moreover, a prospective study assessing the differential impacts on perinatal outcomes by reason for GWL would require data on even larger numbers of women than included here, although the findings of such a study would be of interest. The objective of our study was to assess the potential impact of GWL irrespective of the underlying cause.

A major strength of the present study is the large number of pregnancies and neonates available for analysis, allowing for stratified analyses of the impact of GWL on rare adverse perinatal events. The data were collected for purposes not related to the study hypothesis. Data quality is high, as completeness of the data is monitored annually across obstetric units as an integral part of benchmarking of healthcare provision. BAQ, which provided the data for our study, is part of an established national programme for the generation of annual statistics to show an individual obstetric unit’s deviation from national targets. This programme has been run within all federal states over the last three decades. Data are collected in similar programmes in the UK and USA, and have been analysed in other studies on determinants of pregnancy outcome.14–18

Some items are more stringently checked than others, depending on their relevance to the process of quality assurance and their use for administrative purposes within the hospital. Therefore, there were no missing values with respect to the relevant pregnancy and perinatal complications considered in this study. Height and weight at booking, although essential for the assessment of GWG and pre-pregnancy BMI, are not essential for either quality assurance or administrative purposes, and therefore are not as rigorously checked as other variables, accounting for missing or implausible values in 10% of the cases recorded, which were excluded from our study. Implausible values of height and weight accounted for <1% of these exclusions. Both variables are routinely measured by the family physician/obstetrician well ahead of the first occurrence of the pregnancy or perinatal complications considered in our survey. Therefore, excluded values of height or weight at booking are unlikely to be related to the pregnancy outcomes considered, and information on weight prior to delivery was available in the majority of cases recorded.

To prevent the misclassification of mothers in the GWL or nonexcessive GWG groups as a result of short-term fluctuations in body weight, we decided not to consider women who had booked within 4 weeks of delivery or who had an unknown date of booking. In most cases, however, booking occurred before 15 weeks in mothers with either GWG or GWL. Therefore, a spurious inflation of the number of premature births in mothers with GWL appears to be unlikely, as even mothers who delivered, for example, after 24 weeks were likely to have been under observation for at least 9 weeks.

In addition, it must be kept in mind that even GWG of 0–3 kg might effectively mean a weight loss for the mother during pregnancy after deducting the child’s birth weight. However, we did not deduct the birth weight from the maternal weight at delivery in order to make the results consistent with those of other studies in this field.1–8 Our results may not be directly comparable with those of studies that use self-reported pre-pregnancy weight to determine GWG.2,5–8 However, the validity of self-reported weight is not clear.

In conclusion, although GWL in overweight and obese class I women was associated with lower risks of pregnancy complications, these potential benefits were clearly outweighed by the increased risks of prematurity and SGA birth. In obese class III mothers, the significant and marked reduction in the rate of pregnancy complications may outweigh the risks of GWL.

Disclosure of interest

The funding body had no involvement in the design or conduct of the study, the collection, management, analysis or interpretation of the data, or the preparation, review or approval of the manuscript. None of the authors has a conflict of interest to declare.

Contribution to authorship

AB had full access to the study data and takes responsibility for the integrity of the data and the accuracy of the data analysis. He contributed to the first and final drafts of the manuscript. BS was responsible for data acquisition and contributed to the final draft of the manuscript. NL was responsible for making the data available and for preparation of the data, and contributed to the penultimate and final drafts of the manuscript. RvK contributed to the conception and design of the study, and to the first and final drafts of the manuscript.

Details of ethics approval

Not applicable (observational study).


This work was supported by the project Perinatal Prevention of Obesity Development (PEPO) and the Deutsche Forschungsgemeinschaft (DFG; grant KR 1926/3-1). PEPO is part of the German obesity network funded by the German Federal Ministry of Education and Research (BMBF).