Midwifery and nurse staffing of inpatient maternity services – a systematic scoping review of associations with outcomes and quality of care

Design

A scoping literature review methodology was selected in order to summarize the breadth of the evidence from a range of sources (Levac et al., 2010). Unlike a systematic review, a scoping review allows researchers to identify all the relevant literature regardless of study design. A protocol was not registered in advance as this scoping review developed iteratively to discover the nature of the literature available.

Search strategy

Searches were completed in Medline (Ovid), Embase (Ovid), CINAHL (EBCSCO), Cochrane Library, TRIP, Web of Science and Scopus on 6 th April 2020. Search terms were entered as key words and subject headings, to identify primary research relating to staffing and maternity care (See Appendix 1 for full search strategy). No limitations were placed on the date of publication.

The reference lists of eligible studies were scanned to identify further references. All eligible studies were entered into the Cited Reference Search in Web of Science to identify citations and potential new primary studies in the same field.

Study selection

Studies were eligible for inclusion if they investigated the quantitative association between a measure of midwifery staffing levels and/or skill mix and outcome for mother baby, or staff members, costs or quality of care. All quantitative designs were included including controlled trials, time series, cross-sectional, cohort studies and case controlled studies. Studies on the effects of implementing changes to staffing levels or mix were included, as were studies on the effects of implementing a mandatory minimum staffing policy or a tool to measure demand and guide staffing decisions.

To be eligible for inclusion, staffing levels had to be quantified in measures such as staff per bed, staff to mother ratio, or hours per patient day. An assumption was made that continuous support from a midwife in labour was similar to a staffing ratio of 1:1, and therefore papers reporting staffing in this way were eligible for inclusion. Staff groups include registered midwives, nurse midwives or equivalent, and assistant staff working under the supervision of registered professionals. Studies reporting a quantitative measure of subjective staffing adequacy were included but purely qualitative studies were excluded.

Staffing in any or all inpatient settings were considered including ante-natal, labour/delivery and post-natal care. Studies which were based in neonatal units and midwifery community settings were excluded.

All references arising from the search were imported into Endnote X9™ reference management software where duplicates were removed. Studies were screened and excluded if titles were unrelated to the subject area. The abstracts of 266 studies were read and studies excluded if it was clear that the inclusion criteria were not met by reading the abstract alone. Forty-six full text articles were screened against the inclusion criteria. All included papers were checked, and the decision verified by at least two reviewers. Of the excluded papers, double rating of a sample suggested a high level of agreement. Data charting was performed by one investigator.

Statistical meta-analysis was not attempted but all results were tabulated to show both the direction and statistical significance of the observed effects. From this a description of the overall pattern of results was derived. Sources of bias were identified by considering the study design, measurement of exposure and outcomes, and risk adjustment.

Summary of included studies

The online searches yielded a total of 3280 records. The PRISMA flow diagram is shown in Figure 1 below. Twenty one separate studies were identified which were published from 1988 −2020. These studies are tabulated in detail in Appendix 2. Data were extracted from 23 papers as two studies were reported separately. One study was available as an abstract only (Mercer, 2016). There were three randomised controlled trials, eleven cohort studies, one case control study and six cross-sectional studies.

• Download figure
• Open in new tab

Figure 1:

Outcome of search strategy

Nine studies were conducted in the UK, and the remaining studies were conducted in USA, Canada, France, Italy, Indonesia, Korea, Thailand and Iran. Six studies included only participants at low risk of complications. Three studies included only complex cases such as women having postpartum haemorrhage (Prapawichar et al., 2020), those having oxytocin in labour (Clark et al., 2014) or caesarean section (Kim et al., 2016). The majority of studies (14/21) reported only outcomes relating to labour and birth. No studies of antenatal inpatient care were found, and there were four studies of postnatal care outcomes, including those studying readmission rates.

There were 17 multicentre studies and many were large. Nine studies had over 30,000 participants and five studies had over 400,000 participants. In terms of measurement of staffing, 16 studies used the term ‘midwife’ while others looked at staffing by ‘nurses’ or ‘nurse-midwives’ in a labour setting. Three studies also included the impact of health care assistant/support worker staffing, and eight studies also examined medical staffing in terms of obstetricians, anaesthetists or neonatal doctors.

Quality of the evidence

Three randomised controlled trials (Gagnon et al., 1997; Hodnett et al., 2002; Kashanian et al., 2010) compared patients all receiving one to one care in labour with usual staffing levels, although all had some limitations. Hodnett et al. (2002) excluded patients where one-to-one care was deemed medically necessary. Kashanian et al. (2010) included only 100 women and the usual labour care involved a lack of privacy, no birth companion and women were not permitted to eat and drink. The third RCT (Gagnon et al., 1997) was relatively small and incorporated other therapeutic measures along with the one-to-one care which limits the ability to assess the effects of the staffing ratio alone.

Of the eleven cohort studies, only the Tucker et al. (2003) study provided data on objective patient outcomes while also adjusting for baseline risk and other confounders. Other cohort studies considered care processes such as time to theatre transfer for caesarean section, quality of record keeping, mode of birth or labour interventions (Cerbinskaite et al., 2011; Knape et al., 2014; Rowe et al., 2014; Bailey et al., 2015; Zbiri et al., 2018). These outcomes may not translate directly into benefits for patients. The study by Clark et al. (2014) was conducted in a select patient group receiving oxytocin, limiting the generalisability of findings. The measurement of staffing was based on opinion, and the background risk was not adjusted for. The Dani et al. (2020) study did not measure staffing exposure directly and was at risk of bias due to differences in settings and patient acuity between the two groups. Cohort studies by Kim et al. (2016) and Stilwell et al. (1988) were deemed to be at high risk of bias in the assessment of staffing exposure and had limited risk adjustment. Mercer (2016) was published only as an abstract and therefore the methodology could not be scrutinised.

Of the six cross-sectional studies, four were large scale studies which used routine data to assess exposure to staffing and patient-centred outcomes such as perineal damage, maternal mortality, readmission rates, still birth and neonatal mortality (Joyce et al., 2004; Gerova et al., 2010; Sandall et al., 2014; Makhfudli et al., 2020). Other cross sectional studies focused on the outcome of mode of birth (Joyce et al., 2002; Gerova, 2014) or had a narrow focus on epidural use (Kpéa et al., 2015). All of these studies controlled for risk in terms of maternal age, deprivation, and some measures of clinical risk. These cross-sectional studies considered aggregate measures of staffing such as the number of midwives employed at institutional level or the number of midwives in relation to patients or births. This represents a major difficulty in determining that staffing exposure is causally linked to outcomes for patients, as the time period and fluctuating staffing exposure may not match patient stay. It also does not account for deployment of midwives within the service as some may have non-clinical roles.

Maternal outcomes in relation to staffing

Nine studies examined the outcomes for mothers after birth (Table 1). On the whole, most of these suggest improved outcomes where more staff were present. The outcomes studied included severe maternal outcome (death or near miss), perineal trauma, post-partum haemorrhage, maternal readmission, satisfaction, and maternal infection.

Delivery with bodily integrity and intact perineum were more common when more midwives were employed (Sandall et al., 2014). This finding of reduced perineal trauma was supported by studies by Gagnon et al. (1997) and Hodnett et al. (2002) although significance was not reached. In the case control study by Prapawichar et al. (2020), hospitals which had below the standard nurse midwife to patient ratio had significantly increased odds of postpartum haemorrhage OR 2.3 (95% CI 1.08 to 4.92, p=0.03). Two studies found that maternal readmission was lower when more midwives or nurses were employed in the organisation (Gerova 2010, Kim 2015).

In contrast to this, the study by Clark et al. (2014) found opposite effects for rates of complications in their population of high risk women receiving oxytocin. The lack of risk adjustment in this study could not eliminate confounding by indication, that is higher risk women had higher staffing levels because of the increased risk. Makhfudli et al. (2020) found that the odds of a severe maternal outcome, as defined by World Health Organization (2019) was lower when women were admitted to units with higher nursing staffing (OR 0.48, 95% CI 0.31 to 0.74) but rates were increased in units where midwifery staffing was higher (OR 1.81, 95% (CI 1.07 to 3.06).

Neonatal outcomes in relation to staffing

Ten studies examined the outcomes for neonates (Table 2). Outcomes studied included Apgar scores, birth asphyxia, need for neonatal resuscitation, breastfeeding, admission to the neonatal unit, stillbirth, neonatal death and a composite measure entitled healthy baby. Other potentially important outcomes for babies including neonatal readmission, neonatal hypoglycaemia, sustained breastfeeding, jaundice, and weight loss were not studied.

Three studies report significantly improved outcomes which favour more staff, and one study shows results in the opposite direction. Dani et al. (2020) found higher breastfeeding rates with increased staffing (88% vs 78%, p=0.048), although comparisons took place in two different settings. They also report lower Neonatal Unit admission (2% vs 9%), and this is supported by further studies by Hodnett et al. (2002) and Tucker et al. (2003), although these findings did not reach significance. Gagnon (1997) provides evidence to the contrary, with rates of neonatal unit admission of 7.2% vs 4.9%, RR1.46 (95% CI 0.67, 3.18), thereby presenting a mixed picture for this outcome. Considering the overall pattern, 11 studies have point estimates in favour of more staff while four show results favouring less staff.

Of the higher quality studies (Tucker et al., 2003; Sandall et al., 2014), these suggest that higher staffing was associated with improved neonatal outcomes. Tucker et al. (2003) reported that fewer babies needed neonatal resuscitation using advanced measures (OR 0.97, 95% CI 0.94, 0.99). This was also noted by Hodnett et al. (2002) although no risk adjustment was undertaken in this study.

Events during labour

Ten studies examined events during labour in relation to staffing (Table 3). Outcomes studied included the quality of record keeping, continuous fetal monitoring in low risk women, fetal distress, augmentation of labour, epidural use, speed of theatre transfer for caesarean section, and length of labour. These care process measures are difficult to interpret as they may not translate into differences in patient outcomes. Many of the findings favour more staff, with seven comparisons reaching statistical significance in that direction. Ten further comparisons show non-significant results in favour of more staff. Three comparisons favour having less staff, although some of these result from subgroup analyses.

Fetal distress was lower in facilities that offered 1:1 care more frequently (Clark et al., 2014) and the completeness of the partogram improved (Bailey et al., 2015). Kpéa et al. (2015) found that if the midwifery workload was high, 58.3% of women had an epidural or spinal for pain relief, compared to 49.7% if the workload was not high (OR 1.1, 95% CI 1.0-1.2). This finding was also supported by other studies, although non-significant effects were seen (Gagnon et al., 1997; Joyce et al., 2002; Rowe et al., 2014). Lower staffing was associated with higher augmentation rates, and this reached significance for multiparous women (Rowe et al., 2014). These findings suggest higher intervention rates when staffing levels fall, possibly representing a lack of support for women to manage pain or to facilitate progress of labour.

Cerbinskaite et al. (2011)studied the time taken to enter theatre for emergency caesarean section, and found this to be reduced when more midwives were present. For example, transfer time to theatre for grade 1 caesarean section was achieved within 15 mins for 81/82 (99%) cases where staffing was 1:1 or better, compared to 34/40 (85%) when the ratio fell below this target.

Mode of birth in relation to staffing

Ten studies examined mode of birth as an outcome measure, examining rates of emergency caesarean section, instrumental birth and spontaneous vaginal birth (Table 4). The results were mixed, and no patterns emerged favouring more or less staff.

Measures of birth without assistance were defined differently in the studies, using the terminology ‘normal birth’ and ‘spontaneous vaginal birth’ at times. Within this theme, only Gerova (2014) found a significant association between increased staffing and more normal birth, while studies by Sandall (2014), Hodnett (2002) and Rowe (2014) offered inconclusive findings. An extension of this outcome ‘straightforward birth’ was used by Rowe (2014) to include unassisted birth with no serious perineal trauma or blood transfusion.

In terms of caesarean section rates, only two studies (Kashanian et al., 2010; Zbiri et al., 2018) found a positive association between more staff and reduced caesarean section rate. Rowe et al. (2014) found the opposite, in that understaffing was significantly associated with reduced caesarean section rates, and this was significant for nulliparous women. The majority of other studies examining this outcome found no significant differences (Gagnon et al., 1997; Hodnett et al., 2002; Joyce et al., 2002; Clark et al., 2014; Gerova, 2014; Sandall et al., 2014; Kim et al., 2016). All studies examining the effect of staffing on instrumental birth had non-significant findings and the directions of effect were not consistent (Joyce 2002, Gagnon 1997, Gerova 2014, Hodnett 2002, Rowe 2014).

Effect of midwifery support worker staffing

Three studies (Gerova 2014, Sandall 2014, Kim 2016) reported on the addition of health care support workers and relationship with outcomes. Gerova (2014) found that increases in health care assistants were not significantly related to the probability of emergency section (OR=0.99, 95%CI 0.96-1.03), instrumental birth (OR=1.003, 95%CI 0.96-1.05) or normal birth (OR=0.99, 95%CI 0.95-1.03). Kim (2016) evaluated the impact of increasing the total number of nurses, both licenced and unlicensed. As the total workforce increased, this was not significantly associated with the risk of readmission within 30 days (RR1.01, 95% CI 1.0,1.02).

Sandall (2014) found no significant differences in outcomes for increasing support worker staff in the adjusted overall analysis. Sensitivity analyses were performed in different risk groups and parity. Increasing support workers was associated with an increase in birth with bodily integrity for lower-risk women (OR 1.04) but not for higher-risk women (OR 0.96). The chances of the healthy mother outcome being met were reduced when the number of support workers increased, irrespective of parity (ORs range from 0.87 to 0.93). Support worker staffing levels were associated with a reduced healthy baby outcome (ORs range from 0.90 to 1.00 for women of different parity). When considered together, the above findings do not highlight substantial benefits or detriments for increasing support worker numbers in the workforce.

Effects on staff delivering care

There were no published studies which reported a numeric association between staffing levels and measures of staff wellbeing in the maternity services. No studies were found relating staff retention, job satisfaction or sickness absence to staffing levels.

Economic analyses

Economic analyses were included in primary studies by Clark (2014) and Sandall (2014). Clark (2014) noted that considerable investment would be required to implement one-to-one care for patients undergoing oxytocin induction or augmentation within the USA. They found insufficient evidence of benefit in their trial to justify the additional costs.

Sandall (2014) modelled staffing in relation to cost per birth and found that higher midwifery staffing was associated with increased delivery costs. The relationship was not strong, and this variable plus Trust size and case mix accounted for only 17% of cost variation between Trusts. Cookson et al. (2014) provided an economic impact assessment based on the Sandall (2014) data above. In their calculations, an increase in 1 Full Time Equivalent midwife per 100 births provided an incremental cost effectiveness ratio of £85,560 per additional healthy mother and £193,426 per mother with bodily integrity.

Strengths and limitations

In this scoping review, literature searching was completed in a systematic way, however, there may be undetected studies in the grey literature or in press that have not been accessed. The eligibility screening was not performed independently for all the papers, so it remains possible that some excluded papers might have been included by another reviewer. The high levels of agreement obtained on samples means that it is unlikely that this would make a substantial change to the overall number of included studies or the conclusions about the body of literature as a whole. Although major methodological issues have been discussed, the quality of the evidence has not been rigorously evaluated, which is consistent with the scoping review methodology. This means that poorer quality studies have been included, and these findings are more prone to bias.

Recommendations for further research

Further evidence is needed so that policy makers can make informed decisions about staffing levels and configurations, and the likely impact on outcomes. High quality research is needed to clarify the direction and strength of effects. Studies should examine a range of outcomes in addition to those on labour ward. These could include maternal mental health, neonatal weight loss, jaundice, sustained breastfeeding, and neonatal readmission following discharge home. Support worker contribution and the impact on workforce wellbeing also requires further research. Improved attempts should be made to measure staffing at a ward level or individual patient level if possible. The impact of different workforce configurations and staff groups should be considered. It is important that future studies adjust for underlying risk as well as other predictive factors such as parity, gestational age, pre-existing conditions, and socioeconomic status. Economic studies could model health care costs in terms of staffing numbers, but also potential cost-savings related to intervention rates in labour, readmissions and the cost of advanced neonatal care or maternal morbidity.