Trial design, settings and location
This was a two arm, hospital-based pilot study conducted in four urban, public, maternity hospitals in Nagpur, India from August 2010 – to June 2012. This pilot study was conducted to understand the effectiveness of weekly cell phone counselling and daily text messages meant for pregnant and lactating women attending antenatal care and infant immunization clinics at these hospitals. This pilot will be essential to design a larger cluster randomized control trial to be implemented in rural India.
Eligibility criteria
The participating hospitals (two in intervention and two in control) had to have annual deliveries of above 5000 and catered to women belonging to poor socio-economic background. Women in their third trimester (32–36 weeks), registered for antenatal clinics, planning to deliver at the same hospital and willing to give follow up till 6 months of infant age were considered eligible. An informed consent was obtained from all eligible women. Women with presence of complications in pregnancy that could affect exclusive breastfeeding such as severe anemia (Hb < 6 g/dL), at the risk of eclampsia or pre-eclampsia, consuming drugs contraindicated in pregnancy or HIV positivity were excluded. A record of all women screened, consented and attrition was maintained, including those ineligible and the reasons for not participating in the study.
Randomization
Standardized Baby Friendly Hospital Initiative re-training was imparted by certified instructors to healthcare providers at all the four hospitals using the ‘Breastfeeding Promotion Network of India’ curriculum. The hospitals were then randomized to intervention (cell phone counselling + Baby Friendly Hospital Initiative re-training) and control (Baby Friendly Hospital Initiative re-training only) by the toss of a fair coin.
Description of the intervention
Cell phone counselling was provided by certified lactation counsellors once a week, starting in the third trimester of pregnancy until a week after the infant was 6 months old. These counsellors were auxiliary nurse midwives with additional training for counselling over the phone. They provided advice on importance of antenatal care, iron-folic acid supplementation, maternal nutrition, appropriate infant and young child feeding practices, avoiding of pre-lacteal feeds (additional liquid supplements prior to initiation of breastfeeding), how to deal with problems regarding breastfeeding and infant immunizations. The counsellors also facilitated seeking of care at the hospitals if the mother or infant reported ill. Additionally, women received a text message daily, in the regional language to augment appropriate feeding practices. These women were also provided cell phones, seven free recharge vouchers and subsidized prepaid calling cards. Also, they could call the counsellors as and when needed, using a speed dial facility. During the study, if a mother lost her study cell phone, she was asked to use her personal or family cell phone.
The counsellors were trained to manage their counselling logs for scheduling their weekly calls and sending daily health promotional bulk text messages.
Implementation and data collection
Prior to randomization, the baseline exclusive breastfeeding rates at the participating hospitals were assessed at delivery; 6, 10, 14 weeks postpartum. The rates of exclusive breastfeeding 24 h post delivery were 71.8% in the intervention and 72.3% in the control; similarly, at 6 weeks the rates were 52.8% versus 64.3%, at 10 weeks 52% versus 65% and at 14 weeks they were 40.3% versus 57.6% respectively. Data were collected by independent, trained data collectors from the enrolled women at the study hospitals. These data collection visits coincided with the woman’s antenatal care and child immunization visits. Data were collected at the following time points – registration (visit 1), a week after registration (visit 2), within 24 h of delivery (visit 3) and at 6 weeks (visit 4), 10 weeks (visit 5), 14 weeks (visit 6) post delivery of a live birth. The last two visits were at 6 months (visit 7) and a week after 6 months (visit 8). At registration information on socio-demographic details and preliminary health status were collected. At visit 2, information was collected on maternal illness, whether routine breastfeeding advice has been received and if breast examination has been done. In visit 3, data regarding mode of delivery, birth outcome, place of delivery, infant anthropometry, breastfeeding initiation, pre-lacteal feeds given along with their reasons, maternal or infant illnesses that prolonged hospitalization were obtained. In post-natal visits (4, 5, 6, 7 and 8) data were collected on breastfeeding practices, infant immunization and initiation of complementary feeding. Maternal satisfaction was noted, in both arms, by using a pictorial Likert scale. Random unannounced home visits in 5% of the intervention sample were conducted by data collectors to assess exclusive breastfeeding and inquire from family members about presence of infant formula or bottle in the household.
Cost data collection
The health care costs incurred by the healthcare provider and patients were collected using micro–costing techniques. These costs were measured at enrollment, at delivery and on any subsequent hospitalization (maternal or infant). The costs of cell phones, caller plan subscription, text messages, dialed calls and recharge were recorded. The time and salaries of lactation counsellors, costs of health facility visits and hospitalizations were noted. The variable costs, i.e., direct medical (defined as cost of service, investigations and medication), direct non-medical (defined as cost of travel, food, living outstation etc.) and indirect costs (defined as wages lost during hospital visits) of the two study arms, were measured. The protocol driven costs were excluded from cost calculations. The mean differences in costs and the predictors of total cost were analyzed. The incremental cost-effectiveness of the two study arms was assessed as the incremental total cost of intervention per percentage increase in exclusive breastfeeding.
Outcomes
The primary outcome was exclusive breastfeeding rates at delivery, and postnatal 6, 8, 10, 14 weeks, 6 months and a week after 6 months. It was assessed using the standard World Health Organization’s 24-h recall questionnaire. An infant receiving only breastmilk and no supplemental liquids or solid foods other than vitamins, minerals supplements, medicines or Janamghuti (herbal supplement) in last 24 h was considered to be exclusively breastfed. Other outcomes assessed were timely initiation of breastfeeding (breastfeeding the infant within an hour of birth), pre-lacteal feeds (additional liquid supplements prior to initiation of breastfeeding), neonatal outcomes, bottle feeding rates (use of bottle with nipple / teat), timely initiation of complementary foods (initiation of semi-solid foods after completion of 6 months of infant’s age), infant hospitalizations (any hospitalization more than 24 h related to an illness), infant weight (unclothed weight to the nearest 10 g), maternal satisfaction and incremental cost-effectiveness.
Sample size
We anticipated a total of 1036 mothers-infant dyads (518 per group i.e. 259 per cluster) would participate in the trial, based on the cluster sample size calculation and analysis plan (PASS 2007 software) to achieve 80% power and 5% two-sided alpha to detect an absolute difference between the group proportions of 0.15 (46% exclusive breastfeeding in control group under null hypothesis and 61% under the alternative hypothesis). The test statistic used was the two-sided Z test (unpooled) and the intra cluster correlation coefficient was 0.008.
Data analysis
All the analyses were performed in STATA version 11.2, STATACorp, 4905, Lakeway drive, College Stations, Texas, United States of America. These analyses were conducted at the mother-infant dyad level, for both intervention and control arms (unclustered analyses). The primary analyses compared the prevalence of exclusive breastfeeding in children at 6 months using Pearson’s chi-square tests and 95% confidence intervals for the group differences. We used generalized linear mixed models for non-continuous outcomes (logistic mixed models for binary outcomes - percentage of exclusive breastfeeding). Modelling analyses examined the primary outcome variable taking into account the repeated measurements within children (time) as random effect and all co factors as fixed effects. Variables that may have had impact on the outcome based on a review of the literature were selected as covariates and adjusted for in the models.
Cost analysis was done by calculating the mean costs of cell phone use, counselling, the facility visits and inpatient stay if any. A robust boot-strap method was used to obtain the incremental cost effectiveness ratio. A re-sampling to 100,000 observations was done. Group differences in mean cost of the study arms were assessed using Student’s t-test after normalizing the data. For the incremental cost-effectiveness, the numerator was the difference in the predicted total costs and the denominator was the difference in effects such as the number of not exclusively breastfeeding avoided i.e. number of inappropriate practices that were avoided by an incremental cost of using cell phones.