Study population

The study was carried out in south east England in a district general hospital with about 3000 deliveries a year. Recruitment took place between May 1999 and August 2000. Mothers in the postnatal ward between the hours of 0800 and 1600 on at least one of the minimum five days of recruitment a week were potential study participants.

Intervention

The randomised interventions were routine examination of the newborn by a paediatric SHO or a community midwife. All community midwives conducting the examination had completed the English National Board (ENB) N96 post-registration course in examination of the newborn. SHO training for the newborn examination was more informal and had been carried out by registrars or a consultant paediatrician.

Babies included in the study were those that would have normally been eligible for a midwife examination, according to criteria formulated by senior paediatric and midwifery staff at the hospital. Exclusion criteria were: (a) maternal problems, including chronic disease, infection, drug dependency, drugs with known side effects taken during pregnancy, family history of genetic or inherited diseases; (b) infant problems, including emergency or elective caesarean section under general anaesthetic, instrumental or operative delivery with perinatal complications, Apgar score < 5 at one minute and/or resuscitation required at birth, gestation under 37 weeks, birth weight below 2500 g or above 4500 g, abnormalities detected antenatally or at birth needing follow up, jaundice, problems since birth requiring medical investigation, abnormal neurological responses, dysmorphic features, admissions to the special care baby unit, and history of symptomatic meconium stained liquor. All other births were potential candidates for inclusion in the study. We estimated that about a half of all births would be eligible for the trial. Eligibility for the trial was assessed by the midwife managing the postnatal ward as part of the usual procedure for admitting the healthy newborn.

Newborn examinations were usually carried out 6–24 hours after birth, which is considered an optimal period by paediatric and midwifery management at the hospital, and is also the time recommended by Hall.¹ Examinations by midwives were carried out either in hospital (84%) or at the mother's home (16%) depending on the duration of her postnatal stay in hospital. All examinations by SHOs were carried out in hospital. Referrals as a result of the examination were recorded by the examiner on the specially designed newborn examination form. Referrals to a registrar, consultant, or outpatient department were followed subsequently through the medical notes for outcome. The study protocol was reviewed by the NHS Trust ethics committee.

Randomisation

A research midwife working on the postnatal ward provided written information and explained the study to eligible women. Women who agreed to participate were asked to provide signed consent. Each consenting mother and baby pair were individually assigned to an intervention group using random numbers generated by a coordinator of the study using a statistical software package. An identification sticker was placed on the front cover of the infant's notes to ensure that babies were examined by the allocated examiner. Examiners were therefore not blinded to trial participation, and it was not possible to conceal the identity of the examiner from the mothers.

Outcome measure

The primary outcome measure was maternal satisfaction with the newborn examination. As no tool existed to measure maternal satisfaction with the routine examination of the newborn, a self completion questionnaire was developed by the multidisciplinary research team (comprising psychologists, a health economist, paediatric and midwifery staff, health services researchers, and a statistician), which was piloted on 60 mothers. As well as measuring overall satisfaction with the examination, the questionnaire was designed to determine the degree of satisfaction with specific aspects, including the manner and communication style of the examiner, information given, and handling of the baby, using a seven point Likert-type scale ranging from “very dissatisfied” (−3) to “very satisfied” (+3) (fig 1⇓). Satisfaction with care during labour and since delivery, maternal mood, and bonding with the baby were also assessed.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Items on the questionnaire identified as relating to maternal satisfaction.

The mother was handed the questionnaire and a prepaid envelope after the newborn examination had been carried out. Participants had the option of either leaving their completed questionnaire in a specially placed box on the ward reception or completing it at home. Non-respondents were sent a written reminder, followed by a telephone call. Respondents and non-respondents²⁸ were compared with respect to sociodemographic, obstetric, and medical data, which were collected from mother and infant hospital records.

Principal components analysis with varimax rotation on all respondents with a complete data set (n = 645) was used to indicate which questions assessed satisfaction with aspects of the newborn examination. The analysis identified seven of the questions, loading on a first factor, that assessed satisfaction with the newborn examination. These items in the scale loaded relatively equally (loading: 0.82–0.90), so mean scores were calculated by summing the actual scores for the respective questions and dividing by seven. A Cronbach's α coefficient of 0.94 indicated that the mean satisfaction scale had high internal reliability. Two other factors were identified; these represented the mood of the mother and bonding with the baby (factor 2), and satisfaction with care during and since labour and delivery (factor 3). Factors 2 and 3 had only weak correlations with the newborn satisfaction scale (Spearman's rank correlation (r) = 0.15, p < 0.001; and r = −0.09, p = 0.02 for factors 2 and 3 respectively), indicating that the satisfaction scale had high discriminatory validity.

Eligibility

A retrospective census was conducted on 100 consecutive births at the study hospital, to determine the potential number of babies eligible for midwife examination according to the exclusion criteria.

Statistical analysis

Analysis of a pilot study indicated that around 1000 subjects could be recruited within 15 months at the study site. We estimated that this sample size would be adequate to detect about a 9% difference in satisfaction levels between the two study groups at 80% power and 5% significance.

Results from the pilot study indicated that satisfaction scores were skewed, with a high proportion of women reporting that they were satisfied or very satisfied. A decision was made a priori to dichotomise the distribution into two categories: low satisfaction scores (33% lowest scoring of sample) and medium to high satisfaction scores. Analysis was by intention to treat using a maximum likelihood logistic regression model to predict low satisfaction. Sociodemographic, obstetric, and health variables (table 1⇓) were tested as potential confounders, and examination specific variables (table 3⇓: healthcare issues discussed, problems detected, length of examination, newborn examined by midwife who clerked first antenatal appointment, and place of examination) were tested as potential factors explaining intervention group differences in satisfaction. Modelling was aided using likelihood ratio tests. The final model included predictors significantly and independently associated with maternal satisfaction. This analysis was repeated on an actual treated by SHOs v midwives basis (explanatory analysis). Statistical analysis was performed using Stata 6.0 software, and two sided α values are reported.

Differences by intention to treat

Table 2⇓ shows the relative distribution of low and medium/high satisfaction by intention to treat. Although overall satisfaction was high, women in the midwife group were significantly less likely to report a low satisfaction score than were women in the SHO group (midwives 27% v SHOs 40%: χ² (df 1) = 11.3; OR 0.54, 95% CI 0.39 to 0.75, n = 645).

Confounders and examination specific variables

Table 1⇑ shows that despite randomisation, the SHO group contained more women who were primiparous (χ² = 5.3, p = 0.02) and fewer women with spontaneous vaginal delivery (χ² = 12.4, p < 0.001).

Four out of five examination specific variables differed between midwives and SHOs (table 3⇓). Almost twice as many midwives as SHOs (61% v 33%) were reported to have discussed healthcare issues during the newborn examination, with feeding, skin care, jaundice, sleeping, and stools and nappy care being the most commonly discussed issues. The routine newborn examination was longer in the midwife group than in the SHO group (median 15 v 10 minutes, Wilcoxon rank sum z = −3.91, p < 0.001). The number of hospital referrals (to a registrar, consultant, or outpatient department) made by midwives and SHOs during the newborn examination for problems requiring further investigation did not differ (4% for both midwives and SHOs). Continuity of care was defined as the same midwife clerking the mother at the initial antenatal visit and conducting the neonatal examination. On an intention to treat basis, 95/318 midwives provided some continuity of care, and none of the SHOs had had contact with the mother or baby before the routine examination. Some babies were examined at home by a midwife rather than in hospital. Of the babies examined by a midwife, 101 were examined at home, 96 of which had been randomised to examination by a midwife and five to SHO examination.

Adjusted model

Neither parity (crude OR 1.11, 95% CI 0.80 to 1.55, n = 645) nor type of delivery (crude odds ratio: 1.04, 95% CI 0.70 to 1.53, n = 645) were related to maternal satisfaction with the examination. Of 16 factors (table 1⇑), only history of miscarriage was significantly associated with satisfaction rating, with experience of a miscarriage being associated with lower satisfaction with the newborn examination (crude OR 1.68, 95% CI 1.14 to 2.48, n = 645; table 4⇓).

Women who reported that healthcare issues had been discussed during the examination were less than half as likely to report low satisfaction with the examination (crude OR 0.43, 95% CI 0.30 to 0.60, n = 645). Neither length of the examination (crude OR = 0.96, 95% CI 0.92 to 1.00, n = 470) nor examination at home (crude OR = 0.89, 95% CI 0.56 to 1.42, n = 644) were significantly related to maternal ratings of satisfaction. If the midwife who clerked the first antenatal visit also performed the routine newborn examination, women were 67% less likely to report low satisfaction than if a different midwife or an SHO carried out the examination (crude OR = 0.33, 95% CI 0.18 to 0.58, n = 645).

In table 4A⇑, the adjusted odds ratios for predicting satisfaction with the newborn examination on an intention to treat basis are shown. After adjustment for history of miscarriage, discussion of healthcare issues, and continuity of care, intention to treat (midwife v SHO) no longer independently predicted satisfaction (adjusted OR = 0.82, 95% CI 0.57 to 1.20, n = 645). In contrast, even after adjustment for the other variables, the discussion of healthcare issues during the examination (adjusted OR = 0.49, 95% CI 0.34 to 0.70, n = 645), a history of miscarriage (adjusted OR = 1.61, 95% CI 1.08 to 2.40, n = 645), and continuity of care (adjusted OR = 0.43, 95% CI 0.23 to 0.81, n = 645) remained significantly associated with satisfaction with the newborn examination. Whether the baby was examined at hospital or at home was not related to satisfaction.

To ascertain that the discussion of healthcare issues, continuity of care, and experience of previous miscarriage are explanatory factors, the analysis was repeated according to actual treatment received by SHOs v midwives. The results (table 4B⇑) were virtually the same as for the intention to treat analysis.

Eligibility census

In the eligibility census, the records for 9% of babies were unobtainable. Of the 91 records that were found, it was determined that 48 (53%) babies were eligible for midwife assessment according to the criteria formulated by midwives and paediatricians at the trial hospital (see Methods). Fifteen of 48 (31%) eligible cases were recruited into the trial.

Methodological issues

The major outcome measure was a seven item scale of satisfaction with the newborn examination which was specifically developed for this study. The internal reliability of the scale was high (0.94), exceeding that reported for patient satisfaction scales in investigations of maternity care³⁰ or general practice.³⁷ Principal component analysis and the low and non-significant correlations with other scales relating to satisfaction with labour and postnatal care (r = −0.09) or maternal mood and bonding to her newborn (r = 0.15) indicate high construct and discriminant validity. That is, the major outcome measure assessed satisfaction specifically with the newborn examination independent of other care experiences or maternal wellbeing, and findings cannot be attributed to systematic bias.³⁸ The scale developed and tested here may be highly suitable for assessing satisfaction with the newborn examination in clinical practice for audit purposes, or research.

The randomised control trial was analysed by intention to treat. However, to test whether other and examination specific factors explain maternal satisfaction ratings, analysis according to those who were actually treated by a midwife or SHO should be conducted. This analysis by actual treatment for the final maximum likelihood logistic regression model is thus presented in addition (table 4B⇑). The results were virtually the same whether analysed by intention to treat or actual treatment and support the interpretation of the findings.

Conclusions

The results support the conclusion that, from the mother's perspective, the quality of midwife examination is at least as satisfactory as that of SHOs, when adequate training and paediatric support have been provided. Midwives more often provided information on healthcare issues that are important at this time for mothers, such as infant feeding, skin care, jaundice, sleeping, and stool and nappy care, thus enhancing the quality of care. Furthermore, midwives are able to provide continuity of care, which was found to increase maternal satisfaction with the newborn examination. It was apparent that more reassurance is required to increase satisfaction in new mothers who have had previous miscarriages.

On a cautious note, maternal satisfaction is important, but is only one of different indicators of quality of care. Full recommendations will be made when the study, including a current longitudinal evaluation on the safety, observed quality, and cost effectiveness of midwife versus SHO examination, as well as a national survey of current practice and qualitative analysis of stakeholders' views of the newborn examination and organisational implications, is completed.