Home-based versus hospital-based postnatal care: a randomised trial
Dr M. Boulvain, Department of Obstetrics and Gynaecology, Maternité-HUG, CH-1211 Geneva 14, Switzerland.
Objective To compare a shortened hospital stay with midwife visits at home to usual hospital care after delivery.
Design Randomised controlled trial.
Setting Maternity unit of a Swiss teaching hospital.
Population Four hundred and fifty-nine women with a single uncomplicated pregnancy at low risk of caesarean section.
Methods Women were randomised to either home-based (n= 228) or hospital-based postnatal care (n= 231). Home-based postnatal care consisted of early discharge from hospital (24 to 48 hours after delivery) and home visits by a midwife; women in the hospital-based care group were hospitalised for four to five days.
Main outcome measures Breastfeeding 28 days postpartum, women's views of their care and readmission to hospital.
Results Women in the home-based care group had shorter hospital stays (65 vs 106 hours, P < 0.001) and more midwife visits (4.8 vs 1.7, P < 0.001) than women in the hospital-based care group. Prevalence of breastfeeding at 28 days was similar between the groups (90%vs 87%, P= 0.30), but women in the home-based care group reported fewer problems with breastfeeding and greater satisfaction with the help received. There were no differences in satisfaction with care, women's hospital readmissions, postnatal depression scores and health status scores. A higher percentage of neonates in the home-based care group were readmitted to hospital during the first six months (12%vs 4.8%, P= 0.004).
Conclusions In low risk pregnancies, early discharge from hospital and midwife visits at home after delivery is an acceptable alternative to a longer duration of care in hospital. Mothers' preferences and economic considerations should be taken into account when choosing a policy of postnatal care.
Duration of hospital stay following a normal delivery is declining. In countries where early postnatal discharge has been implemented this policy remains controversial.1,2 Early discharge accompanied by provision of health care at home has been insufficiently evaluated and the safety of this policy has been questioned.3,4 Several randomised controlled trials compared early discharge, usually combined with some form of home follow up, with usual hospital care in the postnatal period.5–12 There was substantial variation between the trials in the definition of ‘early discharge’ and in the form and amount of follow up care provided. Low compliance with the intervention and large numbers of postrandomisation exclusions are common limitations of these studies. The authors of the largest study were unclear in their reporting of the methods of randomisation and of the results.7 These limitations prevent us from drawing firm conclusions about the benefits and risks associated with a policy of home care after an uncomplicated delivery.13
Despite the lack of evidence of benefits, reduced length of stay in hospital combined with home-based care are increasingly popular in Switzerland and in other countries. This reduction is largely driven by economic constraints and by the number of hospital beds available. In Geneva, the usual duration of stay in hospital declined from 12 days in 1950 to 4 days in 1990. Presently, the usual length of stay is three to four days after a vaginal delivery and five to six days after an uncomplicated caesarean section. Postnatal care at the public hospital and up to 10 post-discharge home visits by a midwife are fully reimbursed by the insurance companies. A group of well-organised independent midwives provide good quality home care to the mothers and their newborns. Despite the availability of this service, paediatricians and other hospital staff are reluctant to further reduce the length of hospital stay, fearing delays in detecting and treating complications. This is counterbalanced by parental demands to return home soon after the birth of their baby.
The objective of our randomised controlled trial was to compare a reduced hospital stay supplemented with home visits by a midwife and hospital-based care following an uneventful delivery. Our aim was to evaluate the impact on breastfeeding, maternal and infant morbidity and women's views on their care. An additional objective was to conduct an economic evaluation of these policies of postnatal care, which is reported separately.14
We recruited low risk women in the randomised controlled trial during an antenatal visit at or beyond 37 weeks of gestation. Nulliparous and multiparous women were eligible if they were at low risk for caesarean delivery and for postnatal complications. Exclusion criteria included placenta praevia, pre-eclampsia, diabetes treated with insulin, medical complication of pregnancy requiring postnatal hospital surveillance, past history of postnatal complication (e.g. postnatal depression), difficult socio-economic situation, multiple pregnancy, suspicion of intrauterine growth retardation (estimated fetal weight below the 5th centile) or estimated fetal weight above the 95th centile for gestational age, fetal malformation or genetic disease. We asked eligible women about their preferences regarding the duration of stay in hospital after an uneventful delivery. Women with strong preferences, either for a short or a long hospital stay, were not included in the trial.
Consenting women were allocated to home or hospital care postpartum by opening a consecutively numbered, opaque, sealed envelope. The envelopes were opened by a third party, usually the trial co-ordinator, during a telephone call with the research midwife. The allocation sequence was prepared with a computer program generating a random sequence of numbers in blocks of varying size (four, six and eight participants), arranged in random order.
After vaginal delivery, women allocated to the home-based care group were hospitalised for 24 to 48 hours, after which postnatal care was provided by a midwife at home. The number of midwifery visits and the interval between visits were determined by the needs of the family. Women in the hospital-based care group were hospitalised for four to five days. In both groups, length of stay was increased by two days in the case of caesarean section. When complications were detected, discharge from hospital was delayed but, in the home care group, participants were discharged as soon as midwifery follow up at home was considered to be reasonable.
The outcomes measures of this trial were: the proportion of women continuing breastfeeding beyond 28 days postpartum and total duration of breastfeeding; women's views of the quality of postnatal care; and the morbidity of both mothers and their infants. Maternal morbidity included the following indicators: mother's readmission to hospital, postnatal depression defined as a score higher than 12 on the Edinburgh Postpartum Depression Scale (EPDS)15 and mean scores of physical and mental health as measured by the Short-Form-12 Health Survey (SF-12).16 Neonatal morbidity indicators included: jaundice, defined as a bilirubinaemia above the cutoff for deciding phototherapy (e.g. 300 μmol/L on the third day) or phototherapy itself; rehospitalisation for more than 24 hours; missed screening test for metabolic and endocrine congenital diseases; and missed administration of vitamin K.
Baseline characteristics, socio-demographic, obstetric and perinatal variables, were recorded before recruitment and during the hospital stay. Most of the outcome measures were derived from responses to questionnaires contained in two small booklets. These booklets were given to the mother before discharge from hospital and she was asked to complete the questionnaires 7 and 28 days after delivery. Participants were contacted by telephone at these times to remind them to complete and to return the questionnaires by post. During the same telephone call, the trial co-ordinator answered any queries. A short postal questionnaire that assessed duration of breastfeeding, readmission to hospital and optimal duration of postnatal stay in hospital was completed by the participants at six months postpartum.
Means of continuous variables were reported with their standard deviation (SD) and the difference between groups was tested using the Student's t test. Medians were reported with their interquartile ranges (25th and 75th centile, IQ) and Mann–Whitney U tests. The effect of the intervention was estimated by the relative risk (RR) and 95% confidence interval (CI), and proportions were compared between groups using the χ2 test or the Fisher's exact test. Mean duration of breastfeeding was estimated from life tables and survival curves were compared using the log-rank test. Data management and analysis were performed using EpiInfo (CDC, Atlanta, Georgia) and SPSS (SPSS, Chicago, Illinois) software.
A sample size of 540 women was required to have 80% power (with a risk of type I error of 5%, two-tailed) to show a difference between 85% and 75% in the proportion of women breastfeeding at 28 days postpartum. This sample size was large enough to show a difference of one-fourth of the standard deviation in continuous variables. Because of a low incidence in the main outcome measures, recruitment slower than expected and financial constraints, we terminated the trial after two years, with a lower number of participants than anticipated.
Women were recruited into the trial over a two-year period: November 1998 to October 2000. A total of 6464 deliveries took place during that period. The trial profile is presented in Fig. 1. The main reason for not being contacted by a research midwife was delivery before having had any antenatal visit at the hospital. One woman was excluded from the analysis because of intrauterine death before labour (home-based care group).
There was a slightly higher proportion of nulliparas and of smokers in the home-based care group (Table 1). In the hospital care group, there were more women planning to work following their postnatal leave, more instrumental deliveries, postpartum haemorrhages and anal sphincter tears (Table 2). All other socio-demographic characteristics, preferences about duration of hospital care, obstetric interventions and neonatal outcomes were similarly distributed between the groups.
Table 1. Baseline characteristics according to group. Results are reported as n (%) or mean [SD]. Percentages may vary because of missing values.
|Age||29 [4.8]||29 [5.5]|
|Nulliparity||137 (60)||131 (57)|
|Current smoker||54 (25)||38 (17)|
|Speaking French at home||175 (77)||173 (75)|
|Swiss origin||71 (31)||70 (30)|
|Married||185 (83)||187 (82)|
|Education ≤13 years||115 (52)||113 (51)|
|Living in an apartment||207 (92)||209 (91)|
|<50,000 CHF||62 (27)||56 (24)|
|≥50,000 CHF||131 (57)||132 (57)|
|Missing||35 (15)||43 (19)|
|Planning to work after the postnatal leave||150 (67)||172 (76)|
Table 2. Obstetric interventions and neonatal indicators according to group. Results reported as n (%) or mean [SD]. Percentages may vary because of missing values.
|Induction of labour||65 (29)||59 (26)|
|Augmentation of labour||55 (24)||73 (32)|
|Epidural analgesia||136 (61)||134 (59)|
|Mode of delivery|
| Spontaneous||164 (72)||149 (65)|
| Instrumental||40 (18)||55 (24)|
| Caesarean section||24 (11)||27 (12)|
|Any perineal trauma||145 (65)||144 (64)|
|Anal sphincter tear||8 (3.6)||13 (5.8)|
|Blood loss >500 mL||12 (5.3)||27 (12)|
|Blood loss >1000 mL||3 (1.3)||4 (1.7)|
|Manual removal of the placenta||15 (7.0)||21 (9.5)|
|Male sex||117 (51)||110 (48)|
|Apgar score ≤7 at 5 minutes||2 (0.8)||4 (1.7)|
|Birthweight||3420 ||3480 |
|Admitted to neonatal care unit||10 (4.4)||14 (6.1)|
Women who were randomised to home-based care had a mean duration of hospital stay of 65 hours, compared with 106 hours in the hospital-based care group (P < 0.001). Compliance with the duration of hospital stay specified in the study protocol was 50% (n= 115) in the home-based care group and 72% (n= 167) in the hospital-based care group. The mean number of midwife home visits was 4.8 (SD = 2.3) in the home care group compared with 1.7 (SD = 2.2) in the hospital-based care group (P < 0.001).
The proportion of women who stopped breastfeeding during the first week and during the first month was low and similar in the two groups (Table 3). During the first week, fewer newborns in the home-based care group received complementary feeding and fewer women reported having not received help or being dissatisfied with the help received. A high percentage of women in both groups reported having had cracked nipples or breast engorgement.
Table 3. Breastfeeding according to group. Results reported as n (%) or mean [SD]. Percentages may vary because of missing values.
|Started breastfeeding||224 (99)||223 (97)||0.50 (0.13–2.0)|
|Breastfeeding at day 7 (n= 224 and 223)||221 (99)||218 (98)||0.60 (0.14–2.5)|
|Breastfeeding at day 28 (n= 224 and 223)||202 (90)||194 (87)||0.76 (0.45–1.3)|
|Breastfeeding at 6 months (n= 220 and 215)||78 (36)||78 (36)||0.98 (0.76–1.3)|
|First week postpartum||217||226|| |
|Satisfaction with breastfeeding*||87 ||84 ||P= 0.19|
|Complementary feeding||29 (14)||58 (28)||0.50 (0.33–0.74)|
|Reported problems with breastfeeding||138 (64)||139 (63)||1.0 (0.88–1.2)|
|Did not receive help or dissatisfied with help||17 (7.8)||40 (19)||0.42 (0.25–0.72)|
|Reported cracked nipples||97 (45)||109 (48)||0.93 (0.76–1.1)|
|Reported breast engorgement||87 (40)||73 (32)||1.2 (0.97–1.6)|
|Fourth week postpartum|
|Satisfaction with breastfeeding*||86 ||86 ||P= 0.91|
|Complementary feeding||33 (17)||42 (22)||0.77 (0.51–1.2)|
|Reported problems with breastfeeding||47 (22)||76 (35)||0.64 (0.47–0.87)|
|Did not receive help or dissatisfied with help||21 (9.7)||40 (18)||0.55 (0.33–0.90)|
|Reported cracked nipples||28 (13)||36 (16)||0.81 (0.51–1.3)|
|Reported breast engorgement||22 (10)||42 (19)||0.55 (0.34–0.88)|
At four weeks postpartum, a smaller proportion of women in the home-based care group reported problems with breastfeeding, not receiving help or being dissatisfied with the help received. Total duration of breastfeeding was similar in the two groups (means: 127 and 121 days in the home- and hospital-based care group, respectively; P= 0.42).
Satisfaction with care received was similar in the two groups (Table 4). Very few women reported not being satisfied with any of the aspects of care we explored. A substantial proportion of women in both groups reported fatigue and having received contradictory advice from health care providers.
Table 4. Women's views on their care and their infant's care, according to group. Results reported as n (%). Percentages may vary because of missing values.
|Dissatisfied with the help received||5 (2.3)||12 (5.4)||0.14|
|Dissatisfied with the care received||6 (2.8)||7 (3.1)||1.0|
|Would not like to receive the same care||15 (6.9)||11 (5.0)||0.42|
|Felt that the support from health care providers was insufficient||7 (3.2)||11 (5.0)||0.47|
|Felt that preferences were overlooked by health care providers||9 (4.2)||14 (6.3)||0.39|
|Felt that health care providers were not available when needed||11 (5.1)||10 (4.5)||0.83|
|Felt that health care providers did not respond well to queries||7 (3.3)||5 (2.3)||0.57|
|Dissatisfied with the level of surveillance by health care providers||3 (1.4)||8 (3.6)||0.22|
|Any of the above||31 (14)||31 (14)||1.0|
|Felt that the return home was not easy||20 (9.4)||20 (9.0)||1.0|
|Felt incapable of caring for baby||3 (1.4)||2 (0.9)||0.68|
|Could not usually explain why baby cried||15 (6.9)||22 (9.9)||0.30|
|Felt that postnatal rest was insufficient||68 (31)||71 (32)||0.92|
|Felt that health care providers gave contradictory advice||83 (39)||93 (43)||0.50|
Few women were readmitted to hospital during the first six months postpartum (Table 5). One mother died, from unknown cause, before the sixth month. Scores of depression (EPDS) and physical and mental health (SF-12) were similar between groups.
Table 5. Indicators of maternal morbidity: hospitalisation, contacts with health care providers after discharge from hospital, depressed mood (EPDS*), and physical and mental health scores (SF-12). Results reported as n (%) or mean [SD]. Percentages may vary because of missing values.
|Readmitted during the first month postpartum||4 (1.8)||2 (0.9)||2.0 (0.37–11)|
|Readmitted to hospital 2–6 months postpartum||4 (1.8)||3 (1.3)||1.4 (0.31–6.0)|
|Total readmitted during the first 6 months postpartum||8 (3.5)||5 (2.2)||1.6 (0.54–4.9)|
|One or more visits to a gynaecologist during the first month||33 (16)||48 (22)||0.70 (0.47–1.1)|
|EPDS* >12 at 7 days||27 (12)||27 (12)||1.0 (0.62–1.7)|
|EPDS* >12 at 28 days||16 (7.4)||21 (9.4)||0.79 (0.42–1.5)|
|SF-12* physical summary score||46 [7.9]||45 [8.6]||P= 0.42|
|SF-12* mental summary score||47 ||48 [9.6]||P= 0.44|
A larger number of infants in the home-based care group were readmitted to hospital during the first six months, but the difference was not statistically significant during the first month (Table 6). Five infants in the home-based care group were readmitted to hospital for the treatment of hyperbilirubinaemia, compared with two infants in the hospital-based care group.
Table 6. Indicators of infant morbidity: hospitalisation, contacts with a paediatrician after discharge from hospital, hyperbilirubinaemia, missed administration of vitamin K and missed screening for congenital endocrine and metabolic screening. Results reported as n (%). Percentages may vary because of missing values.
|Readmitted during the first month postpartum||12 (5.3)||5 (2.2)||2.4 (0.87–6.8)|
|Readmitted to hospital 2 to 6 months postpartum||18 (7.9)||6 (2.6)||3.0 (1.2–7.5)|
|Readmitted during the first 6 months postpartum (total)||28 (12)||11 (4.8)||2.6 (1.3–5.1)|
|Two or more visits to a paediatrician during the first month||38 (18)||37 (17)||1.1 (0.70–1.6)|
|Hyperbilirubinaemia (see definition in the Methods section)||14 (6.1)||14 (6.1)||1.0 (0.49–2.1)|
|Phototherapy||9 (3.9)||10 (4.3)||0.91 (0.38–2.2)|
|Jaundice (hyperbilirubinaemia or phototherapy)||16 (7.0)||15 (6.5)||1.1 (0.55–2.1)|
|Missed administration of vitamin K||18 (7.9)||6 (2.6)||3.0 (1.2–7.5)|
|Missed metabolic and endocrine screening||0 (0.0)||0 (0.0)||–|
We found no significant difference between the groups in the main outcome measures. The proportion of women who continued breastfeeding beyond one month postpartum and the duration of breastfeeding were similar. Differences were found, however, in several aspects of breastfeeding. Women in the home-based care group reported having used less complementary feeding, having had fewer problems with breastfeeding at one month and feeling less helpless or dissatisfied with the help they received. The results of the other trials reporting on duration of breastfeeding are conflicting.5–8,10,12 Most trials conducted in a setting where a high proportion of women continued breastfeeding beyond one month suggested that early discharge was associated with earlier stopping.8,10,12 On the other hand, the two trials where the majority of women stopped breastfeeding before one month showed an advantage for the early discharge group.6,7 The pooled estimate of the results of the six trials does not suggest a higher risk of stopping breastfeeding associated with early discharge (pooled RR: 0.96; 95% CI: 0.78–1.2).13
Women's views of their care and their infant's care were similar in the two groups. We found no differences between the groups in postnatal depression, evaluated by the EPDS. Postnatal depression may be detected later (e.g. three months postpartum).17 We decided, however, to administer the questionnaires early after delivery so as to obtain women's views of more recent events, to reduce loss to follow up and to minimise the number of contacts. There were also no differences between the groups in terms of the mothers' health status at 28 days.
We observed a trend towards more readmissions of the infant in the home-based care group during the first month postpartum, which persisted and became statistically significant at six months of age. This may be a chance finding given that this was not our pre-specified primary outcome and given the number of comparisons we have made. That the difference increased from month one to month six, well after the intervention took place, also suggests that this finding may be coincidental. However, the other trials also showed a trend towards more readmissions in the early discharge groups (pooled RR: 2.3; 95% CI: 0.70–7.8).13 Some observational studies showed an increased risk of readmission associated with early discharge,18 while others were unable to find this association.19
Despite this trial being the largest on the topic, its sample size was not sufficient to detect small differences in the most important outcomes. Readmissions, early weaning and dissatisfaction with care were all infrequent, thus limiting the power of the trial to detect differences between alternative forms of postnatal care. Furthermore, participation in the trial among eligible women was lower than expected. We therefore caution against extrapolating our findings to women who have strong opinions about the appropriate duration of in-hospital postnatal care.
Compliance with the allocation among randomised women was closely monitored and hospital staff were encouraged by the research midwives to comply with the planned intervention. Despite that, some women or their newborns had postpartum complications preventing early discharge, while other women preferred to leave the hospital before or after the requisite number of days. Early discharge (three to four days) after caesarean section was particularly unacceptable for both women and health care providers, resulting in a very low compliance in this subgroup (8%). Some women in the hospital-based care group received midwife visits, while others in the home-based care group received none. Lower compliance with the allocated intervention has been reported by other trials, encouraging some of the authors to exclude a high proportion of women from their analyses.5,6,8–12 A pilot study conducted in Australia suggested that a trial would be impossible to conduct, because of non-compliance with the allocated intervention.20 We have conducted all the analyses based on the ‘intention to treat’, in order to avoid selection bias.
We conducted this trial using the opportunity to provide home care by an existing and well-organised network of independent midwives. The results of the trial must be interpreted keeping in mind this favourable situation. It is possible that results may differ if early discharge from hospital is not combined with intensive home support.
In low risk pregnancies, early discharge from hospital supplemented with midwife visits at home after delivery is an acceptable alternative to a longer duration of care in hospital. Therefore other considerations, such as mothers' preferences or economic considerations, should guide the choice of treatment modality.
The authors would like to thank Prof Ellen Hodnett and Prof Judith Lumley for their advice in the development of the study questionnaires.
The authors would also like to thank the hospital personnel and the independent midwives of the ‘Arcade des Sages-femmes’ for their help in collecting the information.
Trial co-ordinator: Mrs Joëlle Vieille.
Safety Monitoring Committee: Dr Sophie Alexander (obstetrician, Brussels, Belgium) and Prof Jacques Rigo (paediatrician, Liège, Belgium).
Research assistants: Mrs Anny Crisafulli and Mrs Catherine Villard.
Mrs Martine Fuhrer was responsible for co-ordinating the study at the ‘Arcade des Sages-femmes’ and for liaison with the study team. She participated in the design of the study questionnaires.
The study received financial support from the Swiss National Science Foundation (grant #32-52954.97) and from the Quality of Care Program, Geneva University Hospitals.