Introduction

Access to a skilled birth attendant is a key strategy to reduce maternal and neonatal deaths in low/middle-income countries. India’s high maternal mortality of 130 per 100 000 live births contributes 19% of global maternal deaths.1 2 Institutional deliveries have been actively promoted by the government to increase access to skilled birth attendants; for example, since 2005 the government has offered cash incentives to pregnant women (and health workers at public facilities) if they deliver at a health facility. This has resulted in the institutional delivery rate doubling between 2005 and 2014 (from 39% to 79%).3 Notably, around 52% (54% in rural) of the deliveries nationwide occurred at a public sector facility; this highlights the prominent role of the public sector in providing women access to skilled birth attendants.3

The remarkable increases in institutional deliveries in India have also raised concern about the quality of obstetric care. Studies have pointed out that despite substantial increases in institutional deliveries there has not been a concomitant reduction in maternal mortality, and its effect on neonatal mortality is unclear.4–10 While the infrastructure and resources available at health facilities have been an ongoing challenge, more recently, attention has also focused on the performance of healthcare providers.11–14 A recent study of nurses’ management of obstetric complications at public facilities in India found that only 14% of nurses conducted initial assessments, 58% made a correct clinical diagnosis and 20% provided first-line care.14 A variety of factors constrain the ability of nurses to provide quality care—inadequate infrastructure, poor training, limited access to continuing education and professional support from experienced clinicians, lack of agency in the workplace and entrenched detrimental clinical practices.6 15–18

In health system contexts where trained clinicians are scarce, clinical functions are often shifted to health workers with less training and without appropriate support, which raises concerns about the quality of care.19 In the state of Bihar, where this study is situated, the lack of doctors and staff nurses at primary health centres (PHCs) has resulted in lesser trained auxiliary nurse-midwives (ANMs) providing obstetric care. Both staff nurses and ANMs are cadres within the public sector health system. Staff nurses have finished high school and have completed a diploma in General Nursing and Midwifery (GNM), a 3-year course in basic nursing.20 GNM nurses are trained to conduct deliveries and manage complications. ANMs have completed high school and possess a 2-year diploma that trains them to be multipurpose community health workers; their course work includes some training in managing deliveries.20 Discussions the authors had with ANMs in Bihar indicate that their diploma programmes include some course work but no practical training in managing deliveries. The curriculum of GNM and ANM programmes is set by the Nursing Council of India.20 ANMs were introduced into the health workforce in the 1950s to serve primarily as midwives at PHCs and subcentres.21 22 However, overtime their function changed from that of a midwife to a paramedical health worker who provides a range of services such as family planning, immunisations and some level of antenatal care.21 22 In Bihar, because staff nurses are scarce at PHCs, ANMs have taken on the responsibility of managing deliveries.

Efforts to improve clinical skills of in-service nurses (We use the term nurses to include both ANMs and staff nurses) have traditionally taken the form of short duration training sessions, often away from the health facility, which are not very effective for skill building. On-site clinical mentoring programmes, which involve longer term engagement and relationship building between mentor and mentee, have shown to be effective in improving nurse knowledge and skills around essential obstetric and neonatal care in other countries, and more recently in India.23–27 When mentors become part of the mentee’s environment, mentoring functions such as didactic instruction, demonstrations and support and encouragement, have the potential to improve mentee skills and self-efficacy to perform effectively and overcome challenges. On-site mentoring programmes for nurses at PHCs in Bihar report increased nurse knowledge and management of normal and complicated deliveries, increased adherence to quality protocols during delivery and increased confidence.26 27 In other contexts, mentoring was associated with a reduction in mortality for low birthweight babies.28

This paper reports on a large-scale mentoring programme to improve the quality of intrapartum care produced by ANMs (and staff nurses) at PHCs in the state of Bihar, India. Specifically, this study investigates if mentoring improved the ability of ANMs to manage normal and complicated deliveries; how mentored ANMs and staff nurses compared in ability, and the sustainability of the mentoring effect. Nurse ability was assessed in managing standardised cases of normal delivery, postpartum haemorrhage and neonatal resuscitation. These cases were selected because mentees were taught to manage these conditions; haemorrhage is a leading cause of maternal mortality, and birth asphyxia a leading cause of neonatal mortality in India.29 30

The state of Bihar (population 100 million) is among India’s resource poor states. PHCs provide outpatient and occasionally inpatient care, maternal health services, family planning, public health services and basic diagnostic and laboratory services. They cover between 20 000 and 30 000 people and are staffed with a doctor, nurses and other supporting staff. PHCs are central to facility-based delivery of obstetric services because they are the first level in the public system which has clinicians trained to provide obstetric care such as doctors and nurses. In 2015–2016, ~64% of all births were in health facilities and 36% at home; urban areas had 74% institutional births while rural areas had 63%.31 Institutional births in public facilities have increased from just 4% of all institutional births in 2005–2006 to almost 48% in 2015–2016.31

Methods

Results

Table 1 shows the characteristics of the mentored and comparison PHCs and nurses in the matched and full sample. The full sample (see Total sample) includes the 239 mentored and comparison PHCs and 474 nurses before matching was carried out. After matching, there were 134 mentored and 56 comparison PHCs, which yielded 256 mentored and 112 comparison nurses. The matched mentored and comparison PHCs are broadly similar in their characteristics indicating that matching was successful. However, mentored PHCs conducted more deliveries on average than the comparison ones. The PHCs and nurses in the full sample (ie, Total sample in table 1), are also broadly similar to the PHCs included in the matched sample in terms of PHC and nurse characteristics, though the matched PHCs have slightly higher values on all the PHC level indicators.

There are several notable points about the characteristics of nurses. First, the overwhelming majority of nurses were ANMs; staff nurses were a minority. The mentored group had a higher proportion of ANMs relative to the comparison group. Second, the majority of nurses were regular staff (as opposed to contractual staff); their proportion is higher in the comparison group. Finally, nurses across groups are similar in average age. Nurse age is also correlated with service duration; nurses in the mentored and comparison groups had similar practice durations.

Nurses in the AMANAT programme typically experienced six full months of mentoring. Overall 93% of nurses reported having at least 6 months of mentoring exposure (table 1). Duration of mentoring ranged from 2 to 9 months, with an average of 7 months of mentoring. These results suggest that there was high adherence to the AMANAT programme among those who participated.

Effect of mentoring

Mentored ANMs and staff nurses had significantly greater proportion, relative to the comparison group, of correct actions overall, and for the domains of history and examinations, case identification (except for normal delivery) and in case management (tables 2 and 3). Figure 1 illustrates the adjusted mentoring effect (ie, treatment effect on the treated) on ANMs and staff nurses. These are marginal effects from the regression models with other variables at their reference values. ANMs in the mentored group had 17.5% more correct actions for normal delivery relative to the comparison group, 25.9% more for postpartum haemorrhage and 28.4% more for neonatal resuscitation. Staff nurses in the mentored group also show similar levels of improved performance. Notably, in all cases the mentoring effect is statistically significant, that is, the 95% CI do not overlap zero.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Effect of mentoring on ability of ANMs and staff nurses. 

Both ANMs and staff nurses appear to have gained from mentoring, but ANMs have gained more. As seen in figure 1, the mentoring effect for each case is greater for ANMs than staff nurses. Finally, because no differences in the proportion of correct actions are seen between mentored ANMs and staff nurses (table 2), it appears that mentoring brought ANMs on par with staff nurses in the management of normal deliveries, postpartum haemorrhage and neonatal resuscitation.

Of interest is to know if the mentoring effect on nurse ability changes with time. Figure 2 illustrates this by plotting estimates from local area regressions within the matched sample of PHCs; it shows the probability of nurses demonstrating a correct action at different points in time after mentoring was completed. Note that the same nurses are not observed at each point in time. In all three cases, nurse ability declined with time. In the normal delivery case, just after mentoring nurses were able to provide 53% of the correct actions, while nurses observed 43 months (the observation farthest from completion of mentoring) after mentoring were able to provide 43% of correct actions, a 10% point drop. Similar trends are seen in the case of neonatal resuscitation and postpartum haemorrhage. In the neonatal resuscitation case, the slight increase seen at the tail end appears to be an artefact of the data. For this case, nurses just after mentoring were able to provide 55% of the correct actions, which declined to 44% after 43 months postmentoring. For postpartum haemorrhage, just after mentoring nurses were able to provide 50% of correct actions, which fell to 37% 43 months postmentoring, a drop of 13% points.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Trends in nurse ability after completion of mentoring (ANM and staff nurse combined, lowess estimates). 

Discussion

In India’s public sector health system, ANMs function as community-based paramedical health workers and provide a range of services such as family planning, immunisations and antenatal care.21 22 In human resource scarce states like Bihar, where there is a paucity of trained doctors and staff nurses, ANMs are the de facto providers of obstetric care at many primary care facilities even though they are not formally trained for this role. In our study, in the 239 PHCs surveyed, staff nurses comprised only 16% of the sampled nurses. In a policy environment that aggressively promotes institutional deliveries, the quality of care provided by ANMs is an important concern.

A principal finding from this study is that mentoring improved the average ability of ANMs and staff nurses at PHCs to manage normal and complicated deliveries. In all three cases—normal delivery, postpartum haemorrhage and neonatal resuscitation—mentored ANMs and staff nurses had better average ability compared with non-mentored ones. This finding is consistent with evaluations of mentoring programmes in other contexts and India.24–27 Mentored ANMs (staff nurses) were able to provide 17% (13%) more correct actions for managing normal deliveries, 26% (14%) more for postpartum haemorrhage and 28% (19%) more for neonatal resuscitation, compared with their non-mentored colleagues. The gains were greater for ANMs since they started from a lower level of ability relative to staff nurses. Among these three cases, for both ANMs and staff nurses the largest gain in ability was in neonatal resuscitation, followed by postpartum haemorrhage, and normal delivery. This is along expected lines since both ANMs and staff nurses are most familiar with the management of normal deliveries, so the incremental effect of mentoring would be relatively smaller. Further, the AMANAT mentoring programme emphasised management of complications such as postpartum haemorrhage and neonatal resuscitation, which explains the gains observed in this area.

ANMs that had completed mentoring had the same average ability as staff nurses to manage normal and complicated deliveries. Further, un-mentored ANMs had significantly poorer ability than un-mentored staff nurses. This mentored ANM achievement is particularly salient because ANMs typically operate as paramedical health workers providing services such as family planning, immunisations and some level of antenatal care. It is unlikely that improvements in ANM ability would be achievable outside a mentoring context. The AMANAT programme provided a combination of didactic teaching, supportive supervision and professional support over a long period of time in the context of a mentor–mentee relationship. To the best of our knowledge, there was no difference in the way staff nurses and ANMs were mentored, though the programme was focused on ANMs given that they primarily manage deliveries at PHCs.

Mentoring improved case identification among ANMs. The vast majority of mentored ANMs were able to correctly identify a normal delivery case, 84% correctly identified a postpartum haemorrhage case, and 83% the neonatal resuscitation case. While non-mentored ANMs and staff nurses performed similarly as mentored ANMs in correctly identifying normal deliveries, they did quite poorly in identifying the other two cases. For example, among non-mentored ANMs, only half of them were able to correctly identify the postpartum haemorrhage or neonatal resuscitation case. The biggest improvement in correctly identifying a case was among ANMs for postpartum haemorrhage and neonatal resuscitation. The improvements observed in recognising birth complications is important for identifying women for referral or complication management within the PHC.

Even as mentoring improved the average ability of nurses to provide quality care, there was substantial variation in nurse ability. Moreover, many mentored nurses were unable to provide a high level of correct actions. For example, 43% of the mentored nurses provided less than half the necessary actions in the normal delivery case. This suggests that participation in the mentoring programme did not raise the ability of all nurses to some minimum standard of ability. There are several reasons for this, such as, the differential capacity of individual nurses to learn, the persistence of habit or features in the work environment that prevents nurses from imbibing new learnings, or low baseline knowledge levels that impedes advanced learning. An important implication is that exposure to mentoring (or any training programme for that matter) does not automatically translate into improved ability to provide quality care for everyone. As such, it becomes important to assess how well mentored nurses provide quality care and only deploy those who meet some minimum quality standard.

There are several reasons why findings from this study might be generalisable to Bihar and to similar contexts. For one, the mentoring programme was large scale and covered most of the state. PHCs surveyed in the study covered the majority of the intervention PHCs and characteristics of the matched sample of PHCs and nurses aligns well with that of the larger pool.

It is important to be cautious about the extent to which quality of care can be improved by a single mentoring programme. Even after 6–9 months of mentoring through the AMANAT programme there remained considerable scope for improving quality of care. Mentored ANMs and staff nurses were able to provide a little more than half the correct actions on all three cases. This is also reflected in the ability domains of history and examinations and case management on which nurses were evaluated. For example, mentored ANMs (nurses) were able to provide only 51% (54%) of correct actions for case management in the normal delivery case, 37% (38%) in the postpartum haemorrhage case and 51% (56%) correct actions in the neonatal resuscitation case. Second, there will be variation in the ability of both ANMs and staff nurses post mentoring. Therefore, not all mentored ANMs and staff nurses will have the ability to properly manage normal or complicated deliveries. Moreover, the work environment can also affect nurse performance; the availability of adequate infrastructure, adequate staff, the management environment are all factors that can influence performance and modify the effect of mentoring.

Another reason for being cautious is that the mentoring effect on nurse ability appears to decay over time. In all three cases, we find that nurse ability was highest in the period immediately after mentoring and this declined with the passage of time. The decay was least for the normal delivery case and most for the neonatal resuscitation case. This has also been observed in other nurse mentoring studies in India, which have documented declines in nurse practice and knowledge.26 51 These patterns highlight the importance of continuing health worker education and make a strong case for mentoring or similar programmes which are embedded within the health system to be offered on a frequent basis to health workers.

A limitation of this study is that we did not observe the ability of nurses before they started mentoring. As such, our estimates of the mentoring effect will be biased if the matched comparison group nurses do not have the same average ability as those in the mentored group before they were mentored. If at baseline comparison nurses had lower ability relative to non-mentored nurse, estimates of the mentoring effect will be upwardly biased. There are several reasons why this may not be the case. One source of bias could arise from differences in nurse characteristics between mentored and comparison groups that are related to their ability. Both mentored and comparison group ANMs (or staff nurses) had the same standard preservice training and require the same qualifications to be recruited into government service. As such, it is unlikely that the extent or quality of nurse preservice training differed between groups. Further, in our regression, we control for difference in nurse and facility characteristics between groups. As such estimates of the mentoring effect are adjusted for differences in nurse characteristics such as age, years of service and employee type between mentored and non-mentored nurses.

Nurse ability can vary by the characteristics of the environment where they work. For example, studies have noted that the ability of clinical care providers varies with the economic condition of the community, that is, better-off areas have better quality providers.52 Similarly, nurses working in a facility located in economically better areas (eg, close to urban areas) might have better ability. As such, differences in the facility location between the mentored and comparison groups can bias estimates of the mentoring effect on nurse ability. If mentored PHCs were located in more favourable areas then estimates of the mentoring effect will be upwardly biased. To address this, we created the comparison group PHCs to be as similar as possible to the mentored PHCs in terms of location and physical environment. Comparison group PHCs were selected such that they were in close geographic proximity (from the same or adjacent administrative block) to the mentored PHCs. Further, we matched mentored and comparison PHCs on location, service readiness and PHC use criteria.

Improved nurse ability need not translate into better practice in managing deliveries. Factors such as motivation, the work environment and support from colleagues, among others, mediate the pathway between ability to practice. However, studies that have examined the association between ability and practice report that these measures are correlated, though there is a typically a gap between what clinicians know and what they do in practice.53 In general, more able clinicians also ‘do more’ in practice, and ability measures the upper bound of what is possible in practice.53

Mentoring programmes can improve the quality of care of lesser trained health workers like ANMs to levels comparable to those of staff nurses who have higher levels of training. In health systems which struggle to recruit highly trained health workers to underserved areas, such mentoring programmes offer a way of making task shifting strategies successful. However, other considerations like programme costs and the ability to sustain such programmes for long periods are also important for scale-up. To sustain the quality gains from mentoring, it is important to expose health workers to regular continuing education programmes.