Impact of family-centred postnatal training on maternal and neonatal health and care practices in district hospitals in two states in India: a pre–post study

Program description

In July 2017, 12 district hospitals, 6 in each state, were selected by the governments of Punjab and Karnataka to launch the Care Companion Programme (CCP) as a pilot programme. These hospitals have monthly delivery volumes of 100–800 deliveries and are the primary government hospitals in their respective districts. The hospitals were selected based on need for the programme and willingness to implement the programme. Noora Health inspired CCP was implemented as a public-private partnership undertaking. The role of the private partner was to design the model including creating and deploying the teaching tools, training the hospital staff, working with administrators and other hospital staff to ensure integration of the training into their daily workflow. The formulation and approval of the curriculum, early testing of material and localisation and approvals from various stakeholders took place in the first half of 2017. The hospitals were to provide the infrastructure support, staff time and administrative support to help run the programme.

The programme uses an evidence-based curriculum for postnatal counselling, covering multiple behaviours for improving neonatal and maternal health.9 Skills taught include key healthy behaviours like exclusive breastfeeding, dry cord care, skin to skin care, mother’s diet, early recognition of danger signs and complications for mother and baby and appropriate healthcare seeking by the families. Behaviour change specialists created tools and methods to teach this curriculum using a human-centred design process, including needs finding interviews with families and discussions with local and regional health experts. Visuals and materials were tailored to the cultural practices of local populations using field testing, key informant interviews and iterative design. The Department of Health in the respective states, as well as District Surgeons and Medical Officers of each hospital, provided final approvals for materials included in the training curriculum. These materials included videos played on a television monitor installed in the postnatal wards, flipcharts, dolls for role play and hand-outs. Examples of the training materials can be accessed at https://drive.google.com/drive/folders/1e6Kl75IGLIxdxXqkDDet2h2Lsb9hcHiP

The initial training of trainers was conducted by a master trainer who trained nurses and counsellors from the 11 district hospitals in a workshop over two 8-hour days. Learning was assessed by a pre–post test. Nurses and counsellors were introduced to the tools, adult learning principles, health communication skills to engage their audiences and the specific health topics and skills that they needed to teach families. A hospital-specific rotational roster was created in order to maximise the number of deliveries covered, acknowledging that some families may discharge before being offered the class due to length of stay and resource constraints. Typically, 30–45 min sessions were held 1–2 sessions per week in the postnatal wards, including intensive care and caesarean delivery wards. In the hospitals, mothers and families were taught these skills by the nurses or counsellors in group sessions which facilitates behaviour adoption.10 On an average 1–2 members of a family attended these sessions.

Classes were held at times outside of physician rounds and food services to ensure that nurses were available for teaching. Security guards often helped gather patients and families in the wards and made attempts to include maximum family members. The nurse or counsellor covered a predefined list of topics, using flip charts and demonstrations followed by a video. Interaction was encouraged throughout the class and a question and answer session took place at the end of the session. The nurses were also encouraged to bring in their expertise to adapt to the needs of the group and local context. Families were encouraged to practise some of the skills while still in the hospitals. Programme managers visited each hospital once a week to monitor the fidelity of the programme, gather learnings for future refinement and provide feedback to improve the sessions.

Training of the trainers occurred after the preintervention data collection phase in June 2017. The programme launched in July 2017 and by August 2017, all hospitals were running the CCP.

The primary SMART (Specific, Measurable, Applicable, Realistic, and Timely), aim of the CCP is to improve patients’ postdischarge outcomes, reduce complications and increase families’ adherence to recommended newborn and maternal care practices. Our objective in the study was to assess the effect of CCP on family’s reported adoption of newborn care practices and newborn outcomes in the neonatal period.

Study design

We conducted a quasi-experimental study where a preintervention group received standard of care (SoC) and served as an historical control and the postintervention group receiving the CCP served as the intervention arm. All outcomes were self-reported by the mother or a primary caregiver who lived with the mother and were collected at 2-week postdelivery via telephone survey. We were interested in the overall effect of the programme at the systems level. Eleven sites in total were included in the study out of the 12 pilot sites; data collectors in one district hospital of Punjab could not be appointed in time before launch.

Sample

Field investigators created a list of deliveries each day referencing the hospital’s delivery registers. Using survey software, we selected a random sample of women who delivered a newborn in the hospital during the study timeframe. Field investigators checked for inclusion and exclusion criteria by talking to the families and using individual patients’ chart data and developed a final list. Families in the preintervention group were recruited between May and June 2017. For the postintervention group, recruitment occurred between August and October 2017. We calculated sample size to detect a 19% difference in postdischarge prevalence of exclusive breast feeding between the intervention and control groups. Assuming a design effect of 1.2 due to differences in hospital-level sampling and a 2:1 ratio of samples in intervention versus control groups, the required sample size to detect the difference with 80% power and 95% CI was 1710 in the control group and 3420 in the intervention group.

Inclusion and exclusion criteria

All women who delivered a live newborn in the selected district hospitals during the study recruitment were eligible and included in the study sampling frame. Families were excluded from participating if baby or mother died during hospital stay or if by the time of the survey call, mother was younger than 18 years, no one living near the mother had access to a telephone, no one living with the mother spoke one of the multiple languages the surveyors spoke, mother or baby were transferred to another hospital during their stay in the hospital, medicolegal cases and if the mother left the hospital before data collectors could collect telephone numbers.

Patient and public involvement

Study questionnaires were piloted with the target population prior to the start of study data collection and based on the pilot feedback. The instrument was fine tuned for the comfort and understanding of the patient population.

Once published, participants will be informed of the results through a dedicated section in the website (www.noorahealth.org), each hospital involved will be sent the results and the Government stakeholders at the state level will be informed.

Measurements

Population-based surveys like the Demographic and Health Survey11 and National Family Health Survey12 collect limited data about newborn care practices and neonatal and postpartum maternal health, only measuring breastfeeding practice and the occurrence of excessive vaginal bleeding or fever in the postpartum period. The study team developed and tested a phone survey to additionally measure other newborn care practices shown to reduce newborn illness and death: skin to skin care, exclusive breast feeding, infection prevention including handwashing and clean umbilical cord care and care seeking for newborn illness. In addition, we chose to measure receipt of postdelivery instructions and challenges like hospitalisation and readmissions.

A phone survey was developed in English and translated into Kannada, Hindi and Punjabi by certified translators. Where there was overlap in measurement, questions about exclusive breast feeding (EBF) and postpartum complications were adapted from the National Family Health Survey (NFHS) and District Health Survey (DHS) surveys. The translated surveys were evaluated for face and content validity by survey managers fluent in the language and familiar with the study aims. Subsequently, the survey pilot was tested with 20 families for understandability and changes were made to question ordering, phrasing and option choices. Finally, the survey with modifications after testing was back translated by the study team that was bilingual and assessed for adherence to the original question concepts. Given all measures were self-reported, several techniques were employed in the design of the survey to minimise desirability bias and recall bias: open-ended questions, small recall periods and specific question ordering. An example of the use of neutral interviewing style and open-ended questioning to mitigate desirability bias was asking ‘How did you clean the umbilical cord?’ without immediately prompting-specific items to elicit information about whether the mother reported ever cleaning the umbilical cord.

The survey was electronically programmed into SurveyCTO, a tool for digital data collection. Required responses, response validity checks and skip patterns, were used to ensure completeness of data; thus, only missing data occurred when a survey respondent abandoned the survey. In addition, survey phone calls were recorded with permission of families and one survey per interviewer was audited daily by survey managers for survey response accuracy and quality. Interviewers were women, from the same or nearby district, spoke the local language and were either trained as social workers or nurses. Families were asked to provide two phone numbers and preferred time of calling. Phone calls were made at 2 weeks postdelivery for 5–7 consecutive days before classifying families as not contactable. The primary outcome measures are described in table 1. Confounders measured were age of the mother, sex of the baby, whether the baby was premature or not, education of the mother, birth weight less than 2500 g, parity, delivery type and whether the baby was sent to Special Newborn Care Unit (SNCU).

Data analysis

Frequencies and percentages were used to describe maternal and child sociodemographic characteristics and confounding variables such as birth weight, gender of baby, whether the delivery was premature, whether it was first pregnancy and whether delivery was by C-section. χ² test for independence of characteristics between the two intervention phases was used to determine significance at 5% level. Logistic regression models with bootstrap SE were used to assess the effect of intervention on odds of outcome. In the first step, a null model was fit with no predictors. Then models were fit separately for each outcome specified in table 1 with the intervention group as the primary independent variable and adjusted for confounders. Intention to treat analysis was applied, so that families were included in the intervention group regardless of their level of exposure to postnatal education. All models were executed for the full sample. Model information efficiency was compared using Akaike’s information criterion. Effect estimates are reported in terms of risk ratios (RRs) which were calculated from ORs and significance was reported at 95% CI. Absolute risk differences are also reported in table 3.

Health system engagement instructions

Participants in the CCP group were 10% more likely to have received baby care instructions (RR=1.10, 95% CI 1.01 to 1.19). Participants in the CCP group were more likely to report receiving instructions on baby warning signs, but this difference was not statistically significant.

Healthy behaviours

Furthermore, we examined healthy behaviours and found skin-to-skin/kangaroo care and dry cord care practices to be significantly improved in the CCP group. Participants in the CCP group were 78% (RR=1.78, 95% CI 1.37 to 2.27) more likely to adopt kangaroo care practices and 4% (RR=1.04, 95% CI 1.02 to 1.06) more likely to practice dry cord care. Exclusive breast feeding and mothers’ following an unrestricted diet were increased in the CCP group, but this was not statistically significant. There was a statistically non-significant decrease in reported hand hygiene in the CCP group.

Healthcare seeking

Health-seeking behaviours were significantly improved in the CCP group compared with the SoC group. Outpatient visits were 27% (RR=1.27, 95% CI 1.10 to 1.46) more likely in the CCP group. Readmissions in the newborn, adjusted for confounders, were 56% (RR=0.44, 95% CI 0.31 to 0.61) lower in the CCP group.

Complications

The overall risk of mother or newborn complications reduced significantly in the CCP group as compared with the SoC group. Risks of newborn complications were reduced by 16% (RR=0.84, 95% CI 0.76 to 0.91) and mother complications by 12% (RR=0.88, 95% CI 0.79 to 0.97).

Summary

This study explored the feasibility and effectiveness of a hospital-run, family-focused training programme that provides education on multiple evidence-based newborn care practices. Our findings indicate that participation in the in-hospital programme was associated with an increase in the uptake of multiple key newborn care practices and health-seeking behaviours.

The CCP incorporates postnatal care principles to facilitate the transfer of healthy practices during the critical window of newborn care. Our study showed positive behaviour uptake in the right direction for all behaviours except handwashing. However, only behavioural changes associated with skin-to-skin care and dry cord care were statistically significant.

Interpretation

The prevalence of dry cord care in several studies from community settings in India varied from 49% to 72%.4 Our study showed a much higher prevalence at 88%, even before the intervention; this may be because it consisted of only hospital-delivered populations. Our skin to skin care was very low in the SoC group at about 5% as compared with other studies in rural India reporting 15%.13 The reason for this difference from other studies was that we measured this practice only after discharge and did not include in-hospital behaviours. We also saw a high uptake of skin-to-skin care as compared with other behaviours. One reason could be that in most hospitals, skin to skin care is promoted only for premature and low-birth weight babies, whereas our training included this for all babies. Breastfeeding levels were high at baseline and did not show change with CCP. This may be due to a ceiling effect as most women were already breast feeding. However, we did not observe this ceiling effect with dry cord care.

Of note, our study found greater changes in complications and readmissions than changes in newborn and maternal care practices. The apparent difference in effect may be because different care practices have varying influence on complication risk. Probably illness recognitionand timely care seeking, improved as reported by increased outpatient visits in the CCP group. Thismay have had an effect on reducing complications and readmissions. On the other hand, it is also possible that our care process indicators did not capture the behaviours driving these reductions or that institutional changes around peri-partum and newborn care improved between the preintervention and postintervention period.

Within India, newborn practices vary and often are related to cultural and religious beliefs. Involvement of entire families allows for easier adoption as additional family members influence uptake.13 Yet, a comprehensive review of 77 studies in LMICs, of educational strategies for postnatal care, reveals no interventions have considered the involvement of family members beyond the parents.4 The family-centred care model is useful in discharge planning by equipping families with essential newborn health skills and improving newborn outcomes.14

The CCP sessions are held with groups of families and the programme emphasises the ability of family members to learn together and practice multiple high-impact health skills. The CCP taps into the collectivist mindset, increasing willingness to learn during life events.15 Hospital staff also are able to address any concerns and answer questions during training sessions, which allows for increased engagement and group discussion of difficult care practices.

Within overburdened care settings, educational interventions face implementation challenges due to limited availability of healthcare workers. The streamlined CCP design with rotation of staff conducting sessions coupled with group classes alleviates the burden of individual and repetitive counselling. In hospital, postnatal education using existing resources can be cost-effective compared with community outreach. The in-hospital setting is often the only form of newborn education families receive and prevents missed opportunities to deliver comprehensive information on newborn care.5 The multitopic curriculum can be delivered using multiple modalities to target caregiver activation during hospital stays and can improve neonatal outcomes.

Strengths and limitations

The primary strengths of our study are its large sample size, measurement of multiple health behaviours and representation of public health settings spanning 11 hospitals in two states. Prior studies evaluating hospital-based postnatal programmes consisted of single sites with small sample sizes or only assessed impact of education on single behaviours like breast feeding. Additionally, this study took place in district hospitals which account for a large proportion of deliveries occurring in government run facilities.

Despite these strengths, there are important limitations to this study. The observed outcomes are self-reported and subject to desirability bias. It is also possible that other programmatic changes could have influenced observed outcomes during the study period. In particular, the LaQshya programme, which aimed at improving labour room and operation theatre quality, was launched soon after the postintervention period in December 2017, and one hospital successfully gained quality improvement accreditation. These factors may contribute to an overoptimistic estimate of the effects.

Additionally, it is uncertain and possibly context dependent whether any trained healthcare staff can train families, what is the ‘correct’ format and intensity of training and how much customisation versus standardisation of training content is appropriate. The current model is suited for facilities with dedicated manpower like nursing or counselling staff, groups of patients present simultaneously with similar conditions and buy-in from stakeholders about the importance of patient and family engagement. Until, CCP has been implemented in 152 district hospitals and 10 medical colleges—facilities with the number of beds ranging from 75 to 500—and 41 subdistrict hospitals which have 31 to 100 beds. Replicability of this CCP model to smaller facilities and health centres may be limited. To adapt to these care settings and scale-up CCP effectively, programme modifications like shorter training sessions, mobile-based postdischarge patient and family engagement are being tested and evaluated. In terms of sustainability, CCP is implemented through a joint partnership between the state and the development partner, but the latter does monitoring and support for the programme. The transition of these responsibilities to the implementing states is also being explored.