Delivery by caesarean section and childhood cancer: a nationwide follow-up study in three countries


  • Linked article: The article has journal club questions by Eichelberger KY, at the end of the article.



To investigate the association between delivery by caesarean section and risk of childhood cancer.


A population-based, follow-up study using register data from three countries.


Denmark, Sweden and Finland.


Children born in Denmark (1973–2007), Sweden (1973–2006) and Finland (randomly selected sample of 90%, 1987–2007; n = 7 029 843).


Exposure was delivery by caesarean section and the outcome was childhood cancer diagnosis. Follow-up started from birth and ended at the first of the following dates: cancer diagnosis, death, emigration, day before 15th birthday or end of follow-up. Cox regression was used to obtain hazard ratios.

Main outcome measures

Childhood cancer diagnosis.


A total of 882 907 (12.6%) children were delivered by caesarean section. Of these, 30.3% were elective (n = 267 603), 35.9% unplanned (n = 316 536) and 33.8% had no information on planning (n = 298 768). Altogether, 11 181 children received a cancer diagnosis. No evidence of an increased risk of childhood cancer was found for children born by caesarean section (hazard ratio, 1.05; 95% confidence interval, 0.99, 1.11). No association was found for any major type of childhood cancer, or when split by the type of caesarean section (elective/unplanned).


The evidence does not suggest that caesarean section is a risk factor for the overall risk of childhood cancer and possibly not for subtypes of childhood cancer either.


Delivery by caesarean section (CS) has been suggested to influence development and functions of both innate and acquired immunity, and subsequent risk of disease.[1] Caesarean delivery has been associated with an increased risk of diseases such as asthma[2-4] and type 1 diabetes mellitus.[5-7] Research findings regarding the association between CS and childhood cancer have, however, been conflicting.[8-19]

Three potential mechanisms have been identified by which the mode of delivery may affect the immune system, and subsequent cancer risk. First, there is an impact on the microbiome of the offspring: bacterial colonisation of the gut varies between children delivered vaginally and by CS, because of a lack of exposure to maternal vaginal and intestinal microbiota following CS.[20] There is an absence of Bifidobacterium species,[21] and gut flora may be altered for months[22] or even years[23] after birth. Second, labour elicits a ‘stress response’. The subsequent cortisol release and hypothalamic–pituitary–adrenal axis activation[24] have been suggested to aid gut and immune system maturation.[1] Offspring born by planned CS, carried out prior to the onset of labour, will not undergo the same birth stress. Third, differences in the epigenetic regulation of genetic expression have been suggested to play a role; leucocytes in cord blood of children delivered by planned CS demonstrate higher levels of DNA methylation.[25] The different possible mechanisms mean that the risks for childhood cancers may differ by type of CS.

Some workers have examined the association between delivery by CS and specific types of childhood cancer, namely neuroblastoma[8-10, 13, 17, 19] and leukaemia,[11, 12, 14-16, 18] mainly as part of more general studies on birth characteristics. However, the findings have been diverging, which may reflect differences in populations and chance as a result of smaller samples.

In the present cohort study, we used nationwide data from Denmark, Finland and Sweden to investigate the association between delivery by CS and risk of childhood cancer. We hypothesised that there would be a modest increased risk of childhood cancer among those born by CS, with the highest rate among those born by elective CS. We examined the risk of childhood cancer in general, and the risks of specific childhood cancers.

Materials and methods

Study participants and follow-up

This population-based, follow-up study used nationwide data in Denmark, Finland and Sweden, where all liveborn children are assigned a unique civil register number at birth or immigration. This number was used to link information from different national registers,[26] as performed in previous studies of childhood cancer risk by the authors.[27, 28] We included all children with a recorded mode of delivery born in Denmark from 1973 to 2007 and Sweden from 1973 to 2006, and, as a result of national regulation, a random sample of 90% of children born in Finland from 1987 to 2007. We excluded children diagnosed with cancer within 6 months of birth to remove cancers that may have been present at birth (n = 830, 6.9% of those with a childhood cancer diagnosis) and those who were diagnosed with cancer but did not have a date of cancer diagnosis (n = 104, 0.9% of those with a childhood cancer diagnosis). The remaining study population included 7 029 843 children in total: 2 364 160 from Denmark, 3 394 981 from Sweden and 1 270 702 from Finland.


Information on mode of delivery was obtained from the Medical Birth Registers for the entire period for Finland and Sweden. For births in Denmark before 1994, information was obtained from the Medical Birth Register; for births from 1994 onwards, information was obtained from the Danish National Hospitalisation Register, using the International Classification of Diseases, 10th revision (ICD-10, codes O82.0–O82.9 and O84.2 for caesarean delivery). In Sweden, information on the type of CS (planned versus unplanned) was available after 1990. Elective caesarean delivery was defined as CS before the onset of labour and unplanned CS was defined as CS where labour started spontaneously or by induction. In Finland, an indicator for elective CS was introduced in October 1990 and, in January 2004, a division for urgent and emergency CS was introduced. In the Danish data, it was possible to distinguish between elective and unplanned CS between 1982 and 1990, and then from 1994 onwards, as some ICD codes for CS indicate the type (elective and emergency CS ICD-10 codes O82.0 and O82.1, respectively). Where there was no information on planning, type was classified as unknown. The follow-up started at birth, and ended at the first of the following events: relevant cancer diagnosis, death, emigration, the day before the 15th birthday or end of follow-up (31 December 2006 in Sweden, 31 December 2007 in Denmark and 31 December 2010 in Finland).


Data on cancer diagnoses were extracted from the Danish, Swedish and Finnish Cancer Registers; the registration and coding of cancers have been described previously.[29-32] The main outcomes of interest were all incident cancers (ICD-7 codes 104–205; ICD-10 codes C00–97) diagnosed before the children's 15th birthday. In addition, more specific childhood cancer diagnoses were considered, including leukaemias (ICD-7 204; ICD-10 C91–95; ICD-O C42; and morphology codes 9800–9948), Hodgkin's lymphoma (ICD-7 201; ICD-10 C81; ICD-O C77; and morphology codes 9650–9667), non-Hodgkin's lymphoma (ICD-7 200 and 202; ICD-10 82–85; ICD-O C77; and morphology codes 9590–9599 and 9670–9728), central nervous system and sympathetic nervous system tumours (ICD-7 193; ICD-10 C47 and C70–72), renal tumours (ICD-7 180; ICD-10 C64), hepatic tumors (ICD-7 155; ICD-10 C22) and testis cancer (ICD-7 178; ICD-10 C62).

Statistical analysis

Hazard ratios (HRs) with 95% confidence intervals (CIs) were estimated using Cox regression for children whose mode of delivery was known. The proportional hazards assumption was evaluated using the estat phtest function, which tests the assumption on the basis of Schoenfeld residuals. The analyses were stratified by the type of CS (elective, unplanned, unknown). Covariates were selected a priori: potential confounders at baseline (birth) included birthweight (<2500, 2500–3999, ≥4000 g), gestational age (<37, ≥37 weeks), multiplicity (single or multiple birth), maternal age (≤26, 27–30, ≥31 years), and parity (1, 2, ≥3). In subanalyses, we also adjusted for maternal status in the labour market (available from 1981 onwards in Denmark, 1973 onwards in Sweden and 1990 onwards in Finland), maternal smoking status during pregnancy (from 1991 in Denmark, 1983 in Sweden and 1987 in Finland), congenital anomalies (in Denmark from 1977), paternal age group (in Denmark and Sweden), 5-minute Apgar score (from 1978 in Denmark and 1973 in Sweden), 1-minute Apgar score (in Sweden and Finland) and country of birth's CS rate by birth year. Analyses were repeated, splitting the follow-up time into two periods (0–4 and 5–14 years), to determine whether the risks were greater in the period closer to exposure and because the assumption of proportional hazards was not met for some of the models. Follow-up for cancer of the testis was extended beyond 15 years of age because the incidence peaks after childhood (between 25 and 34 years).[33] Data on cancers diagnosed from 15 years of age were available in Denmark and Sweden only. Statistical analyses were carried out in Stata 11 (StataCorp LP, College Station, TX, USA).

The study was approved by the Danish Data Protection Agency (j nr. 2008-41-2680), Scientific Ethics Committee of Central Jutland Region (VEK, sagnr. M-20100252), the Research Ethics Committee (EPN) at the Karolinska Institute (Ref. no. 2008/4:6), STAKES (National Research and Development Centre for Welfare and Health, currently THL National Institute for Health and Welfare) (Ref. no. 1646/604/2006) and Statistics Finland (Ref. no. TK53-1266-08).


Caesarean section was the mode of delivery for 882 907 children (12.6%); of these, at least 30.3% were delivered by an elective CS (n = 267 603) and 35.9% by an unplanned CS (n = 316 536). The type of CS was unknown for the remaining 33.8% (n = 298 768). The proportion born by CS was 11.2% in Denmark, 12.3% in Sweden and 16.0% in Finland. In the cohort, 11 181 children were diagnosed with cancer (incidence rate of 13.77 per 100 000 person years), the most common being leukaemias (n = 3559), cancers of the central and sympathetic nervous systems (n = 2779), renal tumours (n = 726) and non-Hodgkin's lymphoma (n = 636). During the first 5 years of follow-up, 5624 children were diagnosed with cancer and, between 5 and 14 years of age, 5557 were diagnosed with cancer. The total follow-up time was 79 571 514 person years.

The characteristics of the study population at birth are presented in Table 1. Statistically significant differences were seen for the characteristics between the subgroups; in particular, CS was more frequent in those with low birthweight, born before 37 weeks, born in multiple births, born to older mothers and of low birth order.

Table 1. Descriptive statistics of the study population by mode of delivery
VariableTotal (n = 7 029 843)Caesarean section (CS) (n = 882 907)Not CS (n = 6 024 346)Unknown (n = 122 590)
n n (%)n (%)n (%)
  1. a

    Available from 1979 in Denmark, 1973 in Sweden and 1987 in Finland.

  2. b

    Available from 1981 in Denmark, 1973 in Sweden and 1990 in Finland.

  3. c

    Available from 1991 in Denmark, 1983 in Sweden and 1987 in Finland.

Birthweight (g) a
<2500285 785115 316 (40)170 275 (60)194 (<1)
2500–39994 992 980595 061 (12)4 397 093 (88)826 (<1)
>40001 158 280141 025 (12)1 017 072 (88)183 (<1)
Missing163 3854074 (2)64 560 (40)94 751 (58)
Gestational age (weeks)
<37391 482132 989 (34)258 284 (66)209 (<1)
≥376 391 935741 758 (12)5 649 225 (88)952 (<1)
Missing246 4268160 (3)116 837 (47)121 429 (49)
Male3 606 248466 774 (13)3 077 123 (85)62 351 (2)
Female3 423 505416 120 (12)2 947 151 (86)60 234 (2)
Missing9013 (14)72 (80)5 (6)
Yes6 685 411751 696 (11)5 929 661 (89)4054 (<1)
No184 01267 686 (37)113 921 (62)2405 (1)
Missing160 420112 (<1)44 177 (28)116 131 (72)
Maternal age (years)
≤262 593 646234 090 (9)2 289 087 (88)57 835 (2)
27–301 968 464229 106 (12)1 712 535 (87)26 823 (1)
≥312 457 502407 032 (17)2 022 274 (82)28 196 (1)
Missing10 23145 (<1)450 (4)9736 (95)
12 988 508427 416 (14)2 508 310 (84)52 782 (2)
22 519 001279 918 (11)2 208 951 (88)30 132 (1)
≥31 493 329174 606 (12)1 298 215 (87)20 508 (1)
Missing29 005967 (3)8870 (31)19 168 (66)
Maternal employment status b
Not in labour market996 320117 296 (12)876 841 (88)2183 (<1)
Blue collar worker1 245 028158 119 (13)1 086 475 (87)434 (<1)
White collar worker1 887 972263 477 (14)1 623 131 (86)1364 (<1)
High position1 063 159142 172 (13)918 557 (86)2430 (<1)
Missing1 094 582135 074 (12)879 846 (80)79 662 (7)
Maternal smoking during pregnancy c
Yes880 162119 763 (14)760 251 (86)148 (<1)
No3 687 054520 219 (14)3 166 007 (86)828 (<1)
Missing310 28338 100 (12)248 271 (80)23 912 (8)
Denmark2 364 160263 662 (11)1 982 152 (83)118 346 (5)
Sweden3 394 981416 466 (12)2 978 525 (88)0 (0)
Finland1 270 702202 789 (16)1 063 669 (84)4244 (<1)

Table 2 displays the associations between delivery by CS and childhood cancer. An adjusted HR of 1.05 (95% CI, 0.99, 1.11) was seen for all cancers over the entire follow-up period for children delivered by CS compared with those delivered vaginally. The adjusted HRs were 1.03 (95% CI, 0.95, 1.11) for cancer diagnoses between the ages of 0 and 4 years, and 1.07 (95% CI, 0.98, 1.17) between 5 and 14 years. In subanalyses, when we also adjusted for maternal status in the labour market, maternal smoking during pregnancy, congenital anomalies, paternal age at birth, country's CS rate in year of birth, 1-minute Apgar score and 5-minute Apgar score, results were essentially unchanged (data not shown). No substantial differences were seen when stratifying results by type of CS.

Table 2. Hazard ratios (HR) for all childhood cancers according to mode of delivery
Mode of deliveryEntire follow-up period0–4 years5–14 years
Cancer cases (rate per 100 000 person years)Crude HR (95% CI)Adjusted HR (95% CI)a,bCancer cases (rate per 100 000 person years)Crude HR (95% CI)Adjusted HR (95% CI)a,cCancer cases (rate per 100 000 person years)Crude HR (95% CI)Adjusted HR (95% CI)a,d
  1. a

    Adjusted for birth year, country, gestational age, multiple birth, birth weight group and maternal characteristics at birth (maternal age and parity).

  2. b

    Reduced to 10 297 failures.

  3. c

    Reduced to 5295 failures.

  4. d

    Reduced to 5002 failures.

Unexposed9657 (13.74)1.0 (ref)1.0 (ref)4854 (17.31)1.0 (ref)1.0 (ref)4803 (11.38)1.0 (ref)1.0 (ref)
Caesarean section1408 (15.13)1.09 (1.03, 1.15)1.05 (0.99, 1.11)732 (18.57)1.08 (1.00, 1.16)1.03 (0.95, 1.11)676 (12.61)1.11 (1.02, 1.20)1.07 (0.98, 1.17)
Type of Caesarean section
Elective343 (14.73)1.04 (0.93, 1.16)1.02 (0.92, 1.14)211 (18.54)1.08 (0.94, 1.24)1.02 (0.89, 1.18)132 (11.09)0.98 (0.82, 1.16)1.03 (0.87, 1.23)
Emergency430 (15.53)1.09 (0.99, 1.21)1.06 (0.96, 1.17)263 (19.42)1.13 (1.00, 1.28)1.05 (0.92, 1.20)167 (11.81)1.04 (0.89, 1.21)1.07 (0.91, 1.25)
Unknown635 (15.09)1.12 (1.03, 1.21)1.05 (0.97, 1.15)258 (17.79)1.02 (0.90, 1.16)1.00 (0.88, 1.15)377 (13.67)1.20 (1.08, 1.33)1.09 (0.98, 1.23)

The associations between mode of delivery and specific childhood cancers are shown for the entire follow-up period in Table 3. For some types of cancer, numbers of cases were small. The strongest associations were seen for cancer of the testis (HR = 1.64), hepatic tumours (HR = 1.27), renal tumours (HR = 1.25), non-Hodgkin's lymphoma (HR = 1.21) and bone cancers (HR = 1.18), but results were statistically insignificant. When splitting the follow-up period into 0–4 and 5–14 years, the strongest associations were seen for cancer of the testis during both the 0–4-year (HR = 1.51; 95% CI, 0.81, 2.82) and 5–14-year (HR = 2.02; 95% CI, 0.71, 5.77) age periods, and for hepatic cancer at 5–14 years of age (HR = 1.75; 95% CI, 0.77, 4.01). Stratifying the analyses did not suggest significant effect measure modification by country, sex or multiple births. Considering the association between CS and the risk of all childhood cancers, dividing by country gave the following results: HR = 1.05 (95% CI, 0.93, 1.19) for Denmark, HR = 1.04 (95% CI, 0.96, 1.13) for Sweden and HR = 1.04 (95% CI, 0.92, 1.18) for Finland. Splitting by sex resulted in HR of 1.03 (95% CI, 0.95, 1.12) for boys and HR of 1.06 (95% CI, 0.97, 1.16) for girls. Stratifying by multiplicity gave a higher HR for those born in multiple births (HR = 1.26; 95% CI, 0.97, 1.62) than for singletons (HR = 1.04; 95% CI, 0.98, 1.10).

Table 3. Hazard ratios (HR) for specific childhood cancers according to mode of delivery
Cancer Entire follow-up period
Cases (rate per 100 000 person years)Crude HR (95% CI)Adjusted HR (95% CI)a
  1. CNS, central nervous system; SNS, sympathetic nervous system.

  2. a

    Adjusted for birth year, country, gestational age, maternal characteristics at birth (maternal age and parity) and multiple birth.

LeukaemiasUnexposed3067 (4.36)1.0 (ref)1.0 (ref)
Caesarean section450 (4.84)1.09 (0.98, 1.20)1.02 (0.92, 1.13)
Hodgkin's lymphomaUnexposed212 (0.30)1.0 (ref)1.0 (ref)
Caesarean section30 (0.32)1.13 (0.77, 1.65)0.96 (0.64, 1.45)
Non-Hodgkin's lymphomaUnexposed539 (0.77)1.0 (ref)1.0 (ref)
Caesarean section96 (1.03)1.35 (1.08, 1.67)1.21 (0.96, 1.52)
CNS and SNSUnexposed2433 (3.46)1.0 (ref)1.0 (ref)
Caesarean section334 (3.59)1.03 (0.92, 1.16)1.03 (0.91, 1.16)
EyeUnexposed366 (0.52)1.0 (ref)1.0 (ref)
Caesarean section50 (0.54)0.98 (0.73, 1.32)0.86 (0.63, 1.19)
RenalUnexposed612 (0.87)1.0 (ref)1.0 (ref)
Caesarean section105 (1.13)1.24 (1.01, 1.53)1.25 (1.00, 1.53)
HepaticUnexposed113 (0.16)1.0 (ref)1.0 (ref)
Caesarean section23 (0.25)1.48 (0.95, 2.32)1.27 (0.79, 2.05)
BoneUnexposed339 (0.48)1.0 (ref)1.0 (ref)
Caesarean section47 (0.51)1.09 (0.80, 1.48)1.18 (0.85, 1.62)
TestisUnexposed85 (0.12)1.0 (ref)1.0 (ref)
Caesarean section19 (0.20)1.62 (0.98, 2.66)1.64 (0.96, 2.80)
Endocrine glandsUnexposed366 (0.52)1.0 (ref)1.0 (ref)
Caesarean section62 (0.67)1.26 (0.97, 1.65)1.11 (0.83, 1.47)

When follow-up was extended beyond 15 years of age for cancer of the testis, the HR was 1.13 (95% CI, 0.93, 1.37) when considering the entire available follow-up period. For the period of 15 years and over, there were 1682 testicular cancer diagnoses (rate, 5.34) and the HR was 1.07 (95% CI, 0.87, 1.31).


Main findings

In this large cohort study, we found no convincing evidence that CS is associated with a marked increased risk of childhood cancer. As CS delivery is common and childhood cancer is a serious event, even a modest increase in risk may be of clinical interest. A slightly increased, but statistically insignificant, risk of childhood cancer was seen among children born by CS in both periods of follow-up. The results obtained when splitting the exposure by CS type suggested that no particular type of CS led to an increased childhood cancer risk. Although elevated risks were seen for some cancer subtypes, statistical significance was not present.

Strengths and limitations

The methodological strengths of this study are its large size: the number of cancer cases, including cases of types of cancer, exceeded all previous studies in this field. We examined the association between CS and both the overall risk of childhood cancer and of specific cancers. However, larger numbers for specific cancers would be useful to provide more precise estimates. We also used prospectively collected nationwide data from three countries on both the exposure and outcome. The cancer registries have high levels of completeness in Denmark, Sweden and Finland.[29-31] To our knowledge, this is the first study of this association which has separated the delivery method by type of CS. A review on the subject of mode of delivery and immune development highlighted the need for studies on this topic to take into account the type of CS to assess the increased risk that the absence of birth stress, experienced when labour has commenced, may confer on children born by elective CS.[1] The type of CS not only reflects different potential exposures, but also different reasons for CS; stratification by type of CS is therefore important when examining confounding by indication. Our results, however, did not demonstrate differences in risks between children delivered by elective and emergency CS.

We did not have complete data on CS status, with missing data for the mode of delivery for 1.5% of the study population; moreover, information on the type of CS was available for only 55.2% of children born by CS, which reflects different registration systems over time. There are some potential confounders that we have been unable to control for. For example, in infants delivered by CS, breast feeding is often delayed (because of delayed lactation),[34] of shorter duration or absent.[35] Breast milk is a key contributor to the colonisation of the gut, providing immunoglobulins and cytokines.[36] However, as this can be a consequence of delivery by CS, it could be on the causal pathway, and therefore should not be adjusted for. The type of anaesthesia has also been suggested as a confounder for the association,[13] which appears not to be a problem unless anaesthesia is a preventive factor. The use of other drugs, such as labour-inducing drugs, may also be different between the exposed and unexposed children. In addition, as a result of the current lack of knowledge regarding the aetiology of childhood cancer, it is likely that there are unknown confounders, for example, genetic factors. These would have to be causally linked to both CS and childhood cancers, which may be plausible, given that some childhood cancers have been suggested to have a prenatal aetiology.[37, 38]


Some previous studies have found statistically significant positive associations between CS and the risks of neuroblastoma[13, 19] and myeloid leukaemia.[11] Our findings, however, did not show statistically significant associations between CS and overall childhood cancer risk or risks of major types of childhood cancer. These findings lend support to other studies, which found no association between CS and the risks of neuroblastoma[8-10, 17] or leukaemia.[14-16, 18] However, despite statistically insignificant results, some HRs were elevated, for example the HR for testis cancer. During childhood, this is a relatively rare cancer type and the lack of statistical significance may be a consequence of a lack of power. One previous study on testicular cancer in adults found a statistically significantly increased risk among those born by CS.[39] The study also found an increased risk in children who were in the breech position at birth; the study discussed that breech presentation was associated with cryptorchism and that CS was a common mode of delivery for babies presenting in the breech position. However, when we added congenital malformations to the model, we obtained similar results. Other studies of adult testicular cancers have reported null associations.[40, 41]

Caesarean section rates in Nordic countries are fairly low compared with rates in the rest of the world, which was highlighted in the recent PERISTAT report. Rates in Europe in 2010 were highest in Cyprus (52%), Italy (38%), Romania (37%) and Portugal (36%). Outside of Europe, The Organization for Economic Co-operation and Development reported 2010 rates to be 47% in Brazil, 46% in China and 32% in the USA. This study found differences in the rates of delivery by CS between Denmark, Sweden and Finland; research studies have not investigated the reasons for the differences between the three countries; however, differences seen between European countries in the PERISTAT report were considered to be a result of differences in practices, but also of maternal factors and reporting differences.[42]


Caesarean section rates are high and increasing in most European countries,[42] and even a very small increase in childhood cancer risk following CS would have public health impact. However, the results from this large, multinational study suggest no or, at most, a very weak association between delivery by CS and childhood cancer in general, and for the most common types of childhood cancer. For less common types of childhood cancer, the numbers of cases were too small to provide any strong conclusions. It seems unlikely that delivery by CS is an important contributor to individual risk of childhood cancer, despite the suggested major effects of CS on immune function.

Disclosure of interests

No conflicts of interest declared.

Contribution to authorship

JO and JL conceived the research. JO, JL and NCM planned and designed the study. NCM analysed the data and wrote the first draft of the manuscript. NCM, JO, SC, MG and JL contributed to data analysis, interpretation of the results and revision of the manuscript.

Details of ethics approval

The study was approved by the Danish Data Protection Agency (j nr. 2008-41-2680), Scientific Ethics Committee of Central Jutland Region (VEK, sagnr. M-20100252), the Research Ethics Committee (EPN) at the Karolinska Institute (Ref. no. 2008/4:6), STAKES (National Research and Development Centre for Welfare and Health, currently THL National Institute for Health and Welfare) (Ref. no. 1646/604/2006) and Statistics Finland (Ref. no. TK53-1266-08).


This work was supported by a grant from the European Research Council [ERC-2010-StG no. 260242, the European Commission's Seventh Framework Programme (FP7)], the Nordic Cancer Union (2012) and NordForsk (070331), the Danish Medical Research Council (project nos. 09-060229, 09-063494 and 09-072986) and the Swedish Council for Working Life and Social Research (Grant no. 2010-0092).

Journal Club

Paper for discussion

NC Momen, J Olsen, M Gissler, S Cnattingius, J Li. Delivery by caesarean section and childhood cancer: a nationwide follow-up study in three countries. BJOG 2014; 121: 1343–50


A paediatric oncologist at your institute observes that a substantial proportion of her cancer patients were born by caesarean section. She asks if you know of any data suggesting an association between caesarean birth and childhood cancers.

Description of research

Participants Children born in Denmark (1973–2007), Sweden (1973–2008) and Finland (1987–2007) with a recorded mode of birth
Intervention Birth by caesarean section (elective, unplanned or unknown)
Comparison Children not born by caesarean section
Outcomes All childhood cancers. Childhood cancer, stratified by subtype
Study design Retrospective cohort

Discussion points

  • Do the hypotheses with regard to caesarean section and childhood cancer favour an association or causation? How does one prove causation?
  • What are the advantages of a retrospective cohort design? What other study type might have been used?
  • What potential confounding factors were not addressed in the adjusted models?
  • How does confounding by indication relate to the association between caesarean section and testicular cancer?
  • If a significant association between caesarean section and testicular cancer had been found, would that affect your clinical practice?
  • Based on this work, how would you answer the paediatric oncologist in the scenario above? (Data S1)
  • KY Eichelberger

  • Greenville Health System/University of South Carolina, School of Medicine, Greenville, SC, USA

Join us at #BlueJC: Follow @BJOGTweets to stay updated on #BlueJC sessions or email to host a journal club on Twitter. Find out more on our journal club page by visiting