Volume 87, Issue 7 p. 751-759
Free Access

Outcome of planned home births compared to hospital births in Sweden between 1992 and 2004. A population-based register study

HELENA E. LINDGREN

Corresponding Author

HELENA E. LINDGREN

School of Health, Care and Social Welfare, Mälardalen University, Eskilstuna, Sweden

: Helena E. Lindgren, Department of Caring and Public Health Sciences, Mälardalen University, Box 325, S-63105, Eskilstuna, Sweden. E-mail: [email protected]Search for more papers by this author
INGELA J. RÅDESTAD

INGELA J. RÅDESTAD

School of Health, Care and Social Welfare, Mälardalen University, Eskilstuna, Sweden

Search for more papers by this author
KYLLIKE CHRISTENSSON

KYLLIKE CHRISTENSSON

Division for Reproductive and Perinatal Health, Department of Woman and Child Health, Karolinska Institutet, Stockholm, Sweden

Search for more papers by this author
INGEGERD M. HILDINGSSON

INGEGERD M. HILDINGSSON

Department of Health Sciences, Mid Sweden University, Sundsvall, Sweden

Search for more papers by this author
First published: 31 December 2010
Citations: 71

Abstract

Objective. The aim of this population-based study was to measure the risk of adverse outcome for mother and child in planned home births in a Swedish population irrespective of where the birth actually occurred, at home or in hospital after transfer. Design. A population-based study using data from the Swedish Medical Birth Register. Setting. Sweden 1992–2004. Participants. A total of 897 planned home births were compared with a randomly selected group of 11,341 planned hospital births. Main outcome measures. Prevalence of mortality and morbidity among mothers and children, emergency conditions, instrumental and operative delivery and perineal lacerations were compared. Results. During this period in Sweden the neonatal mortality rate was 2.2 per thousand in the home birth group and 0.7 in the hospital group (RR 3.6, 95% CI 0.2–14.7). No cases of emergency complications were found in the home birth group. The risk of having a sphincter rupture was lower in the planned home birth group (RR 0.2, 95% CI 0.0–0.7). The risk of having a cesarean section (RR 0.4, 95% CI 0.2–0.7) or instrumental delivery (RR 0.3, 95% CI 0.2–0.5) was significantly lower in the planned home birth group. Conclusion. In Sweden, between 1992 and 2004, the intrapartum and neonatal mortality in planned home births was 2.2 per thousand. The proportion is higher compared to hospital births but no statistically significant difference was found. Women in the home birth group more often experienced a spontaneous birth without medical intervention and were less likely to sustain pelvic floor injuries.

Introduction

For some women a home birth is considered to be the optimal choice when planning where to give birth, but uncertainty regarding potential risk factors might make it difficult for the woman and her partner to make evidence-based decision regarding place of birth for their child.

The observed perinatal and neonatal mortality rates were similar in home and hospital settings for women with normal pregnancies and an expected normal delivery in studies from Holland, Canada and the USA 1–5. Bastian et al. (6) found a three-fold risk of perinatal death among babies born at home in Australia. The studied population included women with diabetes, pregnancies exceeding 42 weeks, breech and twin births. After excluding high-risk pregnancies there was no difference in perinatal mortality. Pang et al. (7) reported an increased risk of perinatal death among babies born at home in the US. In their study planned home births could not be distinguished with certainty from unintended, potentially high-risk out-of-hospital births.

Olsen (4) concluded after a meta-analysis comprising six studies of planned home births that women with low-risk pregnancies and expected normal deliveries who planned to give birth at home had fewer operative or instrumental deliveries, fewer lacerations and a lower frequency of Apgar scores below seven at five minutes after delivery than women who gave birth in hospital. Lower rates of cesarean sections and episiotomies among planned home births compared to planned hospital births have also been reported by Johnson (1) and Jansen (2). Anderson (8) reported the outcome of 11 788 intended home births in the United States. Emergency care was needed in 0.09% of the planned home births: 0.05% due to prolapsed cord, 0.02% to placenta praevia and 0.02% due to placental abruption. Mori and co-authors (9) reported an increased risk for intrapartum-related mortality was reported for unintended and transferred planned home births in England and Wales between 1990 and 2003.

In Sweden, the health authorities do not recommend, and seldom fund, home births. A woman who wishes to give birth at home has to find a licensed midwife willing to assist her and she must also pay for the service herself. Only licensed nurse-midwives have the right to call themselves midwives in Sweden, and they are obliged to hold records and refer to an obstetrician in case of complications irrespective of where the birth takes place. In the event of transfer during a planned home birth the hospital midwife or an obstetrician usually assumes responsibility.

No studies on the outcome of planned home births have been carried out in any of the Scandinavian countries. The aim of this population-based study was to measure the risk of adverse outcome for mother and child in planned home births in a Swedish population between 1992 and 2004.

Material and methods

We attempted to document outcome of all births in Sweden between the 1st of January 1992 and the 31st of December 2004. Register data are incomplete concerning planned home births. To classify women as having, or not having, a planned home birth we utilized additional sources than the Swedish Medical Birth register which contains data on 97–99% of all births (10). Moreover, to increase study efficiency, we took a random sample of the hospital births.

We used two methods to recruit women with experience of an intended home birth. Indirect recruitment was done by contacting all home-birth midwives in Sweden. Women (n = 315) were also recruited by advertisements in newspapers and on the Internet (direct recruitment). The home-birth midwives (n = 43) attempted to contact all women (n = 448 were recruited this way) they had assisted in a planned home birth between 1st January 1992 and 31st December 2004 and asked the women if they wanted information about the present study to be sent to them. All the women and midwives we came into contact with were asked whether they knew of other planned home births, and by doing this we were able to identify ‘new’ home births. Few women were found by this ‘snowball method’, and we consider the data collection nearly complete.

All women who agreed to take part in the study stated their personal identification number on an application form which was linked to the medical birth register. A total of 757 women replied, and they had experienced a total of 1038 planned home births. Of these women, 657 were found in the birth register and these had a total of 1602 registered births. 100 women with a total of 141 self-reported planned home births were not found in the register. According to the returned application forms, 100 births were intentionally unassisted by a professional birth assistant. These births are not included in the present study since no birth records were kept and no register data are therefore available.

After excluding births that occurred outside the study period (n = 551) 1051 births remained. Data from the application forms were compared with the register data. All women (n = 147) for which there was any uncertainty regarding intended place of birth were contacted by phone and asked about the circumstances surrounding the actual birth. Births that were not planned home births according to the women in the telephone interviews (n = 154) were excluded, and 897 planned home births remained after the telephone interviews (Figure 1). In the Medical Birth Register 47% (n = 418) of these births were coded as out-of-hospital births and 53% (n = 479) were coded as hospital births.

Details are in the caption following the image

Flowchart of inclusion in study population.

In the present study a birth is considered a planned home birth if the woman had decided to give birth at home and the birth started at home with contractions or rupture of membranes. A birth that was planned to take place at home according to the woman and started at home with contractions or rupture of membranes is considered a planned home birth irrespective of whether the baby was actually born at home or the woman was transferred to hospital during labor or immediately after birth.

Twins (16 babies), pre-term (11 babies) and post-term (79 babies) births are included in the planned home birth group but not in the planned hospital group. Breech births are included in both groups. Analyses were conducted separately with and without inclusion of complicated pregnancies. Neonatal mortality defines death intrapartum or during the 28 first days of age.

A randomly selected control group containing 11 341 (1:10) hospital births was obtained from the Medical Birth Register. Criteria for inclusion in the control group were a spontaneous, full-term, singleton birth (gestational week 37 to 42) during the study period. The control group was geographically matched since home birth is more common in certain areas of Sweden.

Data analysis

All births that were planned to take place at home and started at home are included in the home birth group irrespective of where the baby actually was born, at home or after transfer in the hospital. Differences in characteristics of women with planned home and planned hospital births were calculated as risk ratios with a 95% confidence interval using the Mantel and Haenszel method. The relative risks of adverse outcome were calculated in the same way. In a logistic regression model the relative risks were adjusted for potential confounders. The SPSS 12.0 software package was employed for the data analysis.

The Regional Ethical Review Board at Karolinska Institutet, Stockholm approved of the study.

Results

A total of 12,238 births (897 planned home births and 11,341 planned hospital births) in Sweden between 1st January 1992 and 31st December 2004 are included in the study. The prevalence of planned home births was 0.85 per thousand births.

Women in the planned home birth group were more often: older than 35 years (RR 1.8, CI 1.5–2.1); born in a European country other than Sweden (RR 1.8, CI 1.5–2.2); employed in fields where an educational qualification is needed (RR 1.5, CI 1.2–1.8) and less likely to be unemployed (RR 0.3, CI 0.1–0.5) than women in the planned hospital group. Women in the planned home birth group were four times more likely to be multiparas and they were less likely to be smokers (RR 0.3, CI 0.2–0.4) or have a high Body Mass Index (RR 0.7, CI 0.6–0.9) compared with women in planned hospital group. There was no difference in marital status (RR 1.0, CI 0.7–1.3) or the prevalence of pre-pregnancy diseases (RR 0.9, CI 0.8–1.0) (Table I).

Table I. Characteristics of women with planned home births compared with women with hospital births.
Home births n = 897 n (%) Hospital births n = 11,341 n (%) Relative risk of home birth 95% confidence interval P-values for relative risks
Mother <25 years 88/897 (10) 2,173/11,341 (19) 0.4 0.4–0.7 <0.001
Mother 25–34 years 579/897 (64) 7,578/11,341 (67) 1.0 Reference
Mother 35 years or more 230/897 (26) 1,590/11,341 (14) 1.8 1.5–2.1 <0.001
Missing 0 (0) 0 (0)
Educationally unqualified 227/434 (53) 6,770/10,830 1.0 Reference
Educationally qualified 143/434 (31) 2,930/10,830 (27) 1.5 1.2–1.8 0.001
Housewife 52/434 (12) 610/10,830 (6) 2.5 1.8–3.4 <0.001
Unemployed 12/434 (3) 520/10,830 (5) 0.3 0.1–0.5 <0.001
Missing 463/897 (52) 511/11,341 (5)
Married/coha bitating# 655/698 (94) 9,758/10,400 (94) 1.0 Reference 0.9
Single/other 43/698 (6) 642/10,400 (6) 1.0 0.7–1.3
Missing 199/897 (22) 941/11,341 (8)
Born in Sweden 799/897 (90) 9,589/11,341 (85) 1.0 Reference
Born in European country other than Sweden 85/897 (9) 533/11,341 (5) 1.8 1.5–2.2 <0.001
Born outside Europe 13/897 (1) 1,219/11,341 (10) 0.1 0.1–0.2 <0.001
Missing 0 (0) 0 (0)
First child 229/897 (26) 7,039/11,341 (62) 1.0 Reference <0.001
2nd–3rd child 513/897 (57) 3,895/11,341 (34) 3.7 3.2–4.3 <0.001
4th child or more 155/897 (17) 407/11,341 (4) 8.8 7.3–10.5
Missing 0(0) 0(0)
No pre-pregnancy diseases## 763/897 (85) 9,362/11,341 (83) 1.0 Reference
Pre-pregnancy diseases## 134/897 (15) 1,979/11,341 (17) 0.9 0.8–1.0 0.06
Mother not smoking 679/709 (96) 9,100/10,507 (87) 1.0 Reference
Mother smoking 30/709 (4) 1,407/10,507 (13) 0.3 0.2–0.4 <0.001
Missing 188/897 (21) 834/11,341 (7)
BMI <20 80/587 (14) 1,076/9,250 (12) 1.0 0.8–1.4 0.5
BMI 21–25 382/587 (65) 5,567/9,250 (60) 1.0 reference
BMI >25 125/587 (21) 2,607/9,250 (28) 0.7 0.6–0.9 0.007
Missing 313/897 (35) 2,081/11,341 (18)
  • #Cohabitating couples are not distinguished from married couples in the register.
  • ##Diagnoses included in the register are repeated UTIs, renal disease, diabetes, epilepsy, asthma, ulcerative colitis, SLE and hypertension.

The majority of women in both groups had a normal vaginal delivery, 95% in the planned home birth group versus 84% in the planned hospital group. Seven planned home births were breech births, of which all were vaginally delivered. The hospital group included 146 breech births, 63 (43%) were vaginally delivered and 83 (57%) were delivered by cesarean section.

There were no maternal deaths in either the planned home birth group or the planned hospital group. No cases of uterus rupture, cord prolapse, placental abruption or placenta previa were identified in the planned home birth group. There were a total of 38 (0.3%) emergency cases among women in the planned hospital group: eleven (0.1%) women had a uterus rupture; thirteen (0.1%) women had a prolapsed cord, nine (0.07%) had a placental abruption and five (0.04%) had a placenta previa. Women in the planned home birth group were less likely to experience hemorrhage after birth (RR 0.4, CI 0.2–0.8), and after adjusting for parity, BMI, smoking and nationality the relative risk was 0.5 and the 95% confidence interval was 0.2–1.0.

All neonatal deaths (death before 28 days) in the planned home birth group and the planned hospital group are presented in Table II. Altogether nine children, two in the home birth group and seven in the hospital group, died before the age of 28 days. There was no statistically significant difference between the two groups with regard to neonatal mortality (RR 3.6, CI 0.8–17.2). One child (0.1%) in the planned home birth group and six children (0.06%) in the hospital group died intrapartum or before the age of six days (RR 1.8, CI 0.2–14.7). One child in the planned home birth group and one child in the planned hospital group died between seven and 28 days of age, both due to congenital anomalies. Two babies, one hospital birth and one planned home birth, died after labor in the birthing pool. The birthing pool was used for pain relief in both cases, in the home birth the baby was born in the water; in the hospital case cesarean section was performed due to placental abruption. In both cases the babies were alive at birth and transferred to the neonatal intensive care unit (Table II).

Table II. Perinatal mortality <28 days among 897 home births and 11,341 hospital births.
Place of birth, characteristics of mother Day of death Mode of delivery Pain relief Weeks of gestation Parity Apgar at 5 minutes Cause of death
Home birth, mother 26 years, married, working full-time. 1 Vaginal waterbirth Bath 40 1 4 Girl, weight missing, asphyxia, transferred to intensive care immediately after birth, died in intensive care unit.
Home birth, mother 29 years, married, working part-time (secretary) 19 Vaginal None 37 2 5 Boy, 3,706 g, born with neuroblastoma, transferred to intensive care immediately after birth and treated there for 19 days.
Hospital birth, mother 32 years, married, working full-time (economist) 0 Vacuum extraction Enthonox, acupuncture, epidural blockade 37 1 2 Boy, 4,010 g, Asphyxia, shoulder dystocia
Hospital birth, mother 21 years, single, obese, unemployed. 0 Vacuum extraction Epidural blockade 41 1 5 Boy, weight missing, asphyxia, epicranial hemorrhage caused by birth injury, other non-specified hemorrhages from delivery
Hospital birth, mother 28 years, married, working full-time (nurse assistant) 0 Vacuum extraction Enthonox, epidural blockade 40 1 1 Girl, 5,770 g, dystocia, posterior presentation, asphyxia
Hospital birth, mother 22 years, single, unemployed. 2 Vacuum extraction Pethidine, enthonox, pudendus blockade, epidural blockade 40 1 5 Boy, 3,706 g, neonatal distress, attacks of cramp after delivery.
Hospital birth, mother 40 years, diabetic, married, working part-time (lab assistant) 2 Vaginal None 40 5 10 Girl, 3,445g, neonatal heart defects, malformations of aorta.
Hospital birth, mother 27 years, married, working full-time (receptionist) 2 Cesarean section Birthing pool, general anesthesia 39 1 0 Girl, weight missing, asphyxia due to placenta abruption.
Hospital birth, mother 28 years, married, not working outside home (nurse assistant) 9 Vaginal Pethidine 41 2 9 Boy, 3,610 g, microcephali.

Compared with a national sample of all full-term births during the study period, the relative risk of mortality (0–28 days after birth) among babies in the planned home birth group was 0.9 (CI 0.1–6.5).

There were no differences in frequency of low Apgar scores between babies born to women who intended to give birth at home and babies born to women in the planned hospital group. After adjusting for parity, smoking, BMI, epidural blockade and use of oxytocin, the relative risk of low Apgar scores was 1.0 with a confidence interval of 0.4–2.2.

Birth injuries and medical interventions following planned home births and planned hospital births are presented in Table III. The risk of sphincter or rectal rupture was one fifth among women in the planned home birth group compared to women in the planned hospital group. The relative risk of having a vaginal tear was 0.7 (CI 0.6–0.9) in the planned home birth group. There was no difference in the rate of perineal tears among the women in the two groups (Table III).

Table III. Multivariate analysis of birth injuries and medical interventions among planned home births (including births transferred to hospital) and hospital births.
Planned home births n = 897 n (%) Hospital births n = 11,341 n (%) Relative risk in planned home births 95% CI Adjusted relative risk# 95% CI of adjusted risk P-values for adjusted risks
Vaginal tears 161 (18) 3,577 (31) 0.5 0.4–0.6 0.7 0.6–0.9 0.001
Perineal tears 178 (20) 2,587 (23) 0.8 0.7–1.0 1.0 0.8–1.3 0.65
Sphincter/rectal rupture 3 (0.3) 311 (2.7) 0.1 0.0–0.4 0.2 0.0–0.7 0.01
Episiotomy 8 (1) 820 (7) 0.1 0.0–0.2 0.1 0.0–0.2 <0.001
Cesarean section 22 (2) 776 (7) 0.4 0.3–0.5 0.4 0.2–0.7 0.002
Vacuum extraction 20 (2) 1,089 (10) 0.2 0.1–0.4 0.3 0.2–0.5 <0.001
  • #All variables adjusted for parity, BMI, smoking and nationality. Tears in vagina, perineum or sphincter/rectum also adjusted for epidural blockade and use of oxytocin.

The cesarean section rate for planned home births was 2.4% compared to 6.8% in the planned hospital group, and the rate of vacuum extraction was 2.2% among intended home births compared to 9.6% in the planned hospital group. In the case of episiotomies the relative risk in the planned home birth group was 0.1 with a 95% confidence interval 0.0–0.2. The episiotomy rate among planned home births was 1% compared with 7.2% in the planned hospital group (Table III). Adjusting for vacuum extractions did not change the relative risk of episiotomy in the planned home group (RR 0.1, CI 0.0–0.3).

In all, having an episiotomy increased the risk of anal sphincter tear (RR 1.9, CI 1.4–2.6), but there were no sphincter ruptures in the home birth group after episiotomy, whereas in the hospital group 14% of the sphincter ruptures occurred after the intervention of an episiotomy. In this study nulliparity was associated with a threefold risk of having a sphincter rupture. After adjusting for parity the relative risk was 0.2 with a 95% confidence interval of 0.1–0.6.

All analyses were done with and without inclusion of complicated births. The exclusion of those cases did not change the results with respect to adverse outcome.

Discussion

In this nationwide population based study of planned home births a four fold risk for the baby's life was found when compared with planned hospital births. In the hospital group it was five times more common to have perineal trauma such as sphincter or rectal ruptures and the risk of having an episiotomy was ten times higher in the hospital group. The rates of cesarean sections and vacuum extractions were significantly lower in the planned home birth group irrespective of whether the birth occurred at home or the woman was transferred to hospital during labor.

Two babies died in the home birth group, the mortality rate for the newborn babies (0–28 days of age) was 2.2 per thousand. This is more than the hospital group (0.6 per thousand) but no statistically significant difference was found. To reach a statistically significant difference from 0.6 to 1.2 per thousand cases of neonatal mortality a sample of 47 361 planned home births is required (p-value 0.05, 80% power). Inductions of labor and elective cesarean sections were excluded from the planned hospital group, as were twins and deliveries after 42 weeks of pregnancy. This was not the case for the planned home birth group where 16 babies were twins and 79 babies had a gestational age exceeding 42 weeks. The mortality rate in the national sample of all singleton births during the study period was 2.5 per thousand.

A total of nine children (two in the planned home birth group and seven in the planned hospital group) in this study died intrapartum or before 28 days of age. In two cases the death occurred after labor in the birthing pool. The home birth case with death after birthing-pool labor has been subject to legal proceedings with three experts reviewing the case. Their conclusion is that the water birth, was the main reason for the death of the baby. Swedish national guidelines discouraging water births were consequently outlined (12).

The potential risk of complications which require immediate medical attention is an argument against home as a place of confinement from the point of view of safety. Emergencies such as placental abruption, cord prolapse or severe hemorrhage can occur quickly and without warning during labor or immediately after birth. There were no emergency cases in the planned home birth group according to the register. Serious and life-threatening complications are uncommon in low-risk populations, and the absence of emergencies might reflect some kind of selection of women giving birth at home even though the home birth group included complicated births. However, in this study there is a possibility that complications have been underreported by the midwife. It might also be that the records have been reported but not registered. In fact, one case of cord prolapse in a planned home birth has come to our knowledge but was not found in the register. The baby was delivered at home eleven minutes after the waters broke and had an Apgar score of eight at five minutes. The ambulance was called and arrived after the baby was born. No transfer was needed. According to the assisting midwife this information was recorded and sent to the Medical Birth Register but was not found and this case is therefore not included in this study. It can not be excluded that complicated cases are more likely to be underreported in the registers.

Planning a home birth reduced the risk of anal sphincter rupture. This is in line with previous results from research on planned home births (5, 13, 14). The increasing rate of sphincter ruptures among birthing women has recently been an issue in the literature. In Sweden an increase from 0.5 to 3% in a 20-year period has been observed (15). The most important risk factors for a third- or fourth-degree laceration are forceps delivery, vacuum extraction and having an episiotomy 15–19. Following a Finnish study the authors suggested that upright birth positions and midwives’ lack of control over the perineum were the main reasons for the increased rate of sphincter rupture (19). A randomised control trial in Sweden showed that women using a kneeling position during second stage had a statistically significant higher frequency of having an intact perineum compared with those who used a semi-sitting position (51% vs. 73%). The authors also reported an increase in episiotomies among those who gave birth in a semi-sitting (10%) position compared with those who were kneeling (2%) (20).

In the case of cesarean sections and vacuum extractions we found that planning to give birth at home significantly reduced the risk. The cesarean section rate in our study was 2.4%. Johnson (2) found that among 5418 women who intended to give birth at home in North America the cesarean section rate was 3.7%, significantly lower than for planned hospital births. Similar results have also been reported from the USA (9) and Australia (21) where 2.9 and 5.5% respectively of the births ended with a cesarean section. An increasing cesarean section rate is being seen in Sweden as well as in other high-resource countries. Higher maternal age, more women with a BMI over 30, previous cesarean sections and a changed policy regarding breech presentations are some explanatory factors (22).

Women who choose to give birth at home may differ from women who choose to give birth in hospital in other ways than what we can discover from register data. Wiegers (6) and Neuhaus (23) propose that women who intend to give birth at home view labor and birth as physiological processes which should not be disturbed by routine interventions and that they prepare for a vaginal delivery supported by a midwife, well aware that pharmacological pain relief is not available at home. It is possible that enhanced belief in one's own ability to give birth physiologically is an explanation for the lower rates of interventions and pelvic floor injuries in this study (24). The women who planned for a home birth were probably highly motivated; they pay for the service themselves and organize for the birth independent of the routine care. Furthermore, the birthing woman may have an advantage in knowing the midwife who will assist her during birth, which is usually the case in Sweden. In contrast to the situation in a delivery unit, the home birth midwife can be focused on the woman in labor and she is not disturbed by sharing her attention with other women in labor which might effect the outcome (25). The midwives who assist home births might also differ in experience and skills compared to midwives working only in delivery wards.

A limitation of our study is the data missing in the medical birth register. A total of 141 births that were reported by the women in the recruitment process as planned home births are missing from this study. No information on these births was found in the Medical Birth Register. We discovered from application forms that 100 births were intentionally unassisted; hence, no records were sent to the birth register. This might explain two thirds of the missing births. For the remaining unregistered births one explanation might be that the midwife did not send the birth records to the register or that these data for some reason have not been recorded. It is possible that information regarding complicated cases or infant mortality has been shielded from public attention. However, we do not find it likely that women would return application forms if information concerning the birth was controversial. We also checked the answers concerning physical problems after birth reported by the women in questionnaires sent to all women who consented to participation by returning the application forms. Misclassification in the register occurred in the case of more than half (53%) of the planned home births: the code for place of confinement indicated that the birth had taken place in hospital. This is probably due to three factors: home-birth midwives use records from delivery units; the code for out-of-hospital birth has not been observed during the registration process; and in the case of transfer the birth will be coded as an intended hospital birth at the delivery unit. Another limitation is the possibility that midwives have forgotten or not kept records for all births they had assisted during this period. Since documentation is one of the duties in health care we find it unlikely that this is a major problem.

The strength of this study is that we probably have reached by and large all women in Sweden who have planned to give birth at home during a 13-year period. In all 97.3% of all women the midwives contacted returned the application forms. According to the midwives who had assisted the births there were no cases of adverse outcome among those who did not reply. The level of internal missing data in the register on outcome variables that we measured in this study is low. In research concerning the safety of the chosen place of birth it is essential that the intended place of birth is the factor deciding who is to be included in the study group (5, 11). In the present study all births that were reported by the women as planned home births and started at home at term of pregnancy are included, irrespective of whether the baby was actually born at home or in hospital after a transfer which was the case in about 12% of the planned home births (26). By using self-reported information regarding planned place of birth we have been able to learn more about planned home births in Sweden than would have been possible without this information.