Introduction

Vitamin A or retinol is essential for normal embryogenesis, immune response, visual functioning, genetic expression, and hematopoiesis [17]. It also regulates cellular growth and differentiation in the lungs, maintains integrity of respiratory epithelium, and helps in surfactant synthesis [3, 11]. Vitamin A deficiency (VAD) may lead to recurrent infections and an increased risk of bronchopulmonary dysplasia (BPD) [13]. In laboratory animals, necrotizing tracheobronchiolitis and squamous metaplasia pulmonary epithelium caused by VAD could be reversed after restoration of adequate vitamin A status [9]. Similar changes observed in ventilated infants with BPD lead to the speculation that early vitamin A supplementation (VAS) might be beneficial in high-risk infants [9].

Vitamin A is primarily transported to the fetus during third trimester of pregnancy leading to limited hepatic vitamin A reserves in preterm infants [36]. High prevalence of preterm births as well as maternal VAD in low and middle income countries (LMIC) compound the problem further [6, 38]. Various studies have documented an association of postnatal intramuscular (IM)-VAS with reduced mortality and/or oxygen requirement at 1 month and lower incidences of long-term neurodevelopmental disability in preterm very low birth weight (VLBW) infants [11, 16, 21, 22, 33, 34]. The possibility that IM-VAS may ameliorate other complications of prematurity, such as retinopathy of prematurity (ROP), intraventricular hemorrhage (IVH), sepsis, hemodynamically significant patent ductus arteriosus (hs-PDA), and necrotizing enterocolitis (NEC) has also been suggested [11].

Most of the studies have used multiple IM injections of vitamin A with doses ranging from 5000 to 10,000 IU/dose [16, 21, 22, 33, 34]. Although no serious adverse events were reported with the use of this dose, repeated IM injections are painful, difficult to administer in poor muscle mass, associated with potential risk of secondary infection, and often unacceptable to the parents. Moreover, high cost and limited availability of injectable vitamin A preparations further preclude the use of IM vitamin A injections [10]. Intravenous (IV) supplementation is not suitable for its invasive nature and risk of infection [30].

Very few studies have used oral vitamin A as a preventive measure for mortality or BPD, and the results are not conclusive [8, 36]. There is no consensus regarding the oral dose of vitamin A to be used. Recommended supplementation of vitamin A for VLBW infants ranges from 1000 to 1500 IU/Kg/day, irrespective of the route of administration [7], though higher doses have been recommended for prevention of morbidity and mortality [24]. Previous authors have used an oral dose of 5000 IU/day of vitamin A without any clinical or biochemical evidence of vitamin A toxicity [36].

The present study was conducted to investigate the effect of early postnatal oral VAS in VLBW infants with respiratory distress.

Methods

This randomized double-blind placebo-controlled trial was conducted at a tertiary care teaching hospital of India, over 20 months (January 2016 to August 2017) after obtaining approval from Institute Ethics Committee. Written informed consents in local language were taken from all parents before inclusion. The trial was registered with Clinical Trial Registry of India (Registration No: CTRI/2017/03/008131).

Study population

Inborn, VLBW (birth weight (BW) < 1500 g) neonates admitted in NICU and requiring respiratory support in the form of oxygen inhalation through nasal prongs or head box, continuous positive airway pressure (CPAP), high flow nasal cannula (HFNC), or mechanical ventilation (MV) at the age of 24 h, were included. Neonates with major congenital malformation, any life-threatening condition where immediate oral feeding was contraindicated such as reversal of umbilical artery end-diastolic blood flow on antenatal Doppler, perinatal asphyxia with moderate to severe hypoxic ischemic encephalopathy, shock with escalating doses of vasopressors, recurrent seizures, and suspected inborn errors of metabolism were excluded.

Randomization and allocation

Randomization into vitamin A or placebo group was done using random permuted blocks of 4, 6, and 8, prepared by an independent statistician not involved in the study. Allocation into vitamin A or placebo group was done using serially numbered opaque and sealed envelopes by on-duty residents who were appropriately trained for the process beforehand. Allocation concealment was maintained throughout the study.

Method of blinding/masking

Vitamin A and placebo oral solutions were supplied in identical bottles of 20 mL with dropper marked at 1 mL. Vitamin A bottle contained 10,000 IU of retinol/mL in aqueous base, whereas placebo contained only base solution without any drug. The color and smell of the solution contained in the bottles were identical. Treating physicians, nursing staffs, and the parents were unaware about the composition of the bottles. Oral administration was done by nursing staff during hospital stay and later by parents at home, if discharged. Both the groups were trained beforehand for the procedure.

Method of administration of the intervention

Neonates in the vitamin A and placebo groups received 1 mL of syrup vitamin A or placebo on alternate day for 28 days, starting at 24 h of life (total 14 doses). In neonates on parenteral and/or orogastic tube (OGT) feeding, the solution was administered through OGT followed by a chasing with 1 mL sterile water. OGT was not aspirated later unless the neonate developed abdominal distension (abdominal girth increasing by 2 cm from the previous measurement). In infants on cup/breastfeed, the solution was put directly into the neonate’s mouth and feeding was continued for a few minutes. If the neonate vomited within 5 min, the dose was repeated.

Outcome variables

Primary outcome variable was composite incidence of all-cause mortality and/or oxygen requirement for 28 days, measured at day 28 of life. Secondary outcome variables were safety and tolerability of high-dose VAS, serum retinol concentration at recruitment and day 28, total duration of oxygen requirement, and respiratory support by CPAP/HFNC/MV and incidences of complications such as sepsis, echocardiography-confirmed hs-PDA, NEC (Bell stage II and beyond), IVH (grade II and beyond), periventricular leukomalacia (PVL), and ROP. All-cause mortality was measured again at post-menstrual age (PMA) of 36 weeks along with BPD.

Clinical work up

Maternal and neonatal details were recorded. Neonates were examined thoroughly after birth and anthropometric details were recorded. Intra-uterine growth categorization was done as per INTERGROWTH 21ST standards [35].

Study neonates were managed according to our NICU protocol. Initial nutritional support was provided by IV fluids and parenteral nutrition. Enteral feeding with expressed breast milk was started through OGT as soon as the infant became hemodynamically stable. Antibiotics were started in presence of risk factors for sepsis and modified/stopped as per clinical condition, sepsis screen, and blood culture reports. The nature of respiratory support was guided by Downe score [12], chest X-ray, and arterial blood gas parameters under strict monitoring with pulse oximeter for a saturation target of 90–94%. Caffeine was started if the infant was on CPAP/HFNC/MV. Surfactant replacement therapy (Curosurf 200 mg/kg initially, repeated on deterioration) was given if the neonate showed clinical and radiological evidence of respiratory distress syndrome (RDS). None of the neonates was given any steroid (injectable or inhalational), diuretics, or azithromycin to prevent BPD.

During hospital stay, infants were observed for the development of sepsis, hs-PDA, acute kidney injury (AKI), neonatal hyperbilirubinemia (NNH), IVH, NEC, PVL, BPD [20], and ROP [19]. Hs-PDA was treated with IV paracetamol 15 mg/kg/dose 6 hourly for 72 h. Cranial ultrasound was done at recruitment and at days 3, 7, 28, and at clinical suspicion for detection of IVH and PVL. Germinal matrix-intraventricular hemorrhage was graded as per Papile et al. [28]. ROP screening was done at 4 weeks of postnatal age and repeated as per the advice of the ophthalmologist.

Infants were observed for the features of raised intracranial pressure or mucocutaneous lesions suggestive of hypervitaminosis A. If the infant developed NEC or frank gastro-intestinal hemorrhage at any time, oral vitamin A/placebo supplementation was stopped, and the infant was managed appropriately. Progress during the hospital stay and outcome were noted. If discharged earlier, parents were contacted telephonically to remind for vitamin A administration and to note occurrence of side effects, if any. Infants were called after completion of 28 days for estimation of serum retinol. Study neonates are currently being followed up for growth, development, and morbidity.

Estimation of serum retinol

Peripheral venous blood samples were taken at recruitment and on day 28 of life for estimation of serum retinol. Sampling was clustered with other investigations. Serum was separated immediately by centrifugation and stored at − 60 °C until further analysis. Retinol concentration was measured by spectrophotometric method of Bessey et al. [5].

Sample size calculation

As per the record of our NICU, the combined incidence of death and BPD (defined as oxygen requirement for ≥ 28 days) in VLBW neonates requiring respiratory support at 24 h of age in the previous year was 64%. Assuming a similar incidence and expecting a relative risk reduction of 20% in vitamin A group compared with placebo, with a confidence level of 95% and power of 80%, a minimum total sample size calculated was 178 using online power/sample size calculator (http://www.stat.ubc.ca/~rollin/stats/ssize/b2.html). Expecting a 10% attrition rate, the final total sample size estimated was 196, with 98 in each group.

Statistical analysis

The statistical program SPSS version 16.0 (SPSS Inc., Chicago, IL) was used for data entry and analysis. Independent samples T test, Mann–Whitney U test, chi-square test and Fisher exact test were used to compare continuous and categorical variables between groups. Relative risk (RR) with 95% confidence interval (CI), and number needed to treat for benefit (NNTB) were calculated for outcome variables using MEDCALC® (www.medcalc.org/calc/relative_risk.php). Survival analysis of the study neonates was done using Kaplan Meier survival plot analysis. A p value of < 0.05 was considered statistically significant.

Results

Flow of participants

One hundred and ninety-six neonates were randomized into vitamin A (n = 98) and placebo (n = 98) groups. Allocated intervention was started in all. Eighty-five (85/98) infants in the vitamin A group and 76 (76/98) infants in placebo group completed the intervention. The statistical analysis was done on intention-to-treat basis (Fig. 1).

Fig. 1
figure 1

Flow of participants. *Developed necrotizing enterocolitis stage II. REDF—Reversed end-diastolic flow; LAMA—Left against medical advice for financial constraints

Full size image

Profile of the study population

Both the groups were comparable with respect to maternal and neonatal characteristics (Table 1). Mean BW of vitamin A and placebo groups were 1185 ± 194 and 1163 ± 181 g, respectively; and mean gestational ages (GA) were 30.9 ± 2.9 and 30.7 ± 2.7 weeks, respectively. There was no difference in distribution of gender, intrauterine growth patterns, Apgar scores, Downe score, surfactant, and SpO2 at recruitment between the groups. However, requirement of paracetamol therapy for hs-PDA was significantly less in vitamin A group (7/98 in vitamin A vs. 20/98 in placebo; p < 0.05). The most common cause of respiratory distress was RDS (57/98 in vitamin A and 59/98 in placebo), followed by early-onset sepsis (EOS) with intrauterine pneumonia (30/98 in vitamin A and 27/98 in placebo). Transient tachypnea of newborn and meconium aspiration syndrome were responsible in a minor percentage of cases (3/98 and 1/98 vs. 4/98 and 1/98 in vitamin A and placebo groups, respectively). Both the groups had similar other sources of vitamin A from parenteral nutrition and milk feeds.

Table 1 Maternal and neonatal characteristics
Full size table

Outcome variables

Outcome variables are summarized in Table 2. Composite incidence of all-cause mortality and oxygen requirement for 28 days were significantly lower in vitamin A group, compared with placebo (RR (95% CI), 0.440 (0.229–0.844); p < 0.05, NNTB 7). Among secondary outcome variables, although there was no difference in EOS between the groups, the incidence of late-onset sepsis (LOS) was significantly lower in vitamin A group (RR (95% CI), 0.564 (0.346–0.918); p < 0.05, NNTB 7). Klebsiella pneumoniae was the organism most commonly grown both for EOS and LOS in either group. Incidence of hs-PDA was significantly higher in placebo than vitamin A group (RR (95% CI), 0.350 (0.155–0.789); p < 0.05, NNTB 8). Though incidence of BPD at 36 weeks’ PMA was less in vitamin A than placebo (2/98 vs. 9/98), the difference was not statistically significant. Other complications and total number of all-cause mortality at 36 weeks PMA were similar between the groups.

Table 2 Comparison of outcome variables
Full size table

Respiratory support

Compared to placebo, requirement, and duration of oxygen supplementation and non-invasive respiratory support by CPAP/HFNC were significantly less in vitamin A group. Though number of infants requiring MV was less in vitamin A compared with placebo (17/98 vs. 25/98), the difference was not statistically significant. The duration of MV was also similar (Table 3).

Table 3 Nature of respiratory support
Full size table

Adverse effects of intervention and outcome

No major adverse effect was observed in either group. Transient vomiting was observed in three neonates in vitamin A and five neonates in placebo group; p > 0.05. None of the infants in either group had diarrhea, bulging fontanel, or mucocutaneous lesions. Interventions were stopped in one and four infants in vitamin A and placebo groups, respectively, for the development of NEC stage II. All of these neonates recovered without any sequelae.

In vitamin A group, 9/98 infants expired and 86/98 were discharged. In placebo group, 16/98 infants expired and 80/98 were discharged (p > 0.05). All the deaths were because of sepsis and its associated complications. However, the median duration of hospital stay was longer in the placebo group compared with vitamin A (median (IQR), 14 (9–22) vs. 12 (9–15) days; p < 0.05). Three infants in vitamin A group and 2 in placebo group left against medical advice for financial/personal reasons and could not be followed up.

Serum retinol concentration in study neonates

Number of neonates with low serum retinol (< 20 μg/dL) were high (over 60/98) in both the groups at baseline. A significant increase in mean serum retinol concentration was observed in vitamin A group at 28 days. Thirty-three (33/98) neonates had low serum retinol concentrations in placebo compared with none in vitamin A group (RR (95% CI), 0.014 (0.001–0.240); p < 0.01, NNTB 2) (Table 4). None of the infants who received vitamin A had a high serum retinol concentration (> 100 μg/dL) on day 28 of life.

Table 4 Serum retinol concentration
Full size table

Kaplan Meier survival analysis

Mean survival of vitamin A and placebo groups was 26.1 and 24.7 days, respectively. Log rank test failed to detect any statistically significant difference between the groups (χ2 = 2.277; p = 0.131) (Fig. 2).

Fig. 2
figure 2

Kaplan Meier survival analysis curve

Full size image

Discussion

In the present study, a statistically significant reduction in composite incidence of all-cause mortality and oxygen requirement for 28 days was observed after oral VAS. Among secondary variables, duration of oxygen supplementation, non-invasive respiratory support, hospital stay, and incidences of LOS and hs-PDA also favored VAS. No major adverse effect of VAS was documented.

VAS in the neonatal period was initially proposed as a means to increase the body’s vitamin A stores [18], and more recently as a strategy to improve infant survival [37]. Three trials, conducted in Indonesia, India, and Bangladesh, have shown a reduction in mortality during infancy after VAS during neonatal period [18, 23, 29].

Though a recent meta-analysis [11] demonstrated a marginal benefit of IM-VAS in VLBW in reducing mortality and oxygen requirement at 1 month (NNTB 20), and the risk of BPD at 36 weeks (NNTB 11) and benefits of enteral VAS in VLBW are not well explored yet. One randomized controlled trial (RCT) using a daily oral vitamin A (5000 IU/day) documented serum retinol concentrations similar to IM supplementation [25], no decrease in the incidence of BPD or death was documented [36]. The sample size of this study was less and many of the infants received postnatal steroids. Another RCT of oral vitamin A prophylaxis (30,000 IU/kg for 6 weeks starting within first 48 h) in infants of GA ≤ 32 weeks and BW ≤ 1250 g on oxygen support, did not find any significant difference in the incidences of RDS, LOS, PDA, pneumothorax, severe intracranial hemorrhage, ROP, BPD, and mortality [8].

One of the reasons for the lack of beneficial effects of oral VAS may be due poor bioavailability of vitamin A from oil-based solution. Immature gut in VLBW may lead to decreased hydrolysis of retinyl esters, poor availability of vitamin A carrier-proteins in enterocyte, and inadequate bile salts for micelle formation [15]. Absorption of vitamin A may be better from aqueous solution, as used in our study, as smaller particle size might have had better diffusion and better bioavailability [27].

Significantly lower incidence of LOS in vitamin A group may be due to better immune function after vitamin A supplementation [2]. Meta-analysis of three published trials showed a non-significant trend towards a reduction in culture-proven nosocomial sepsis [4, 22, 33]. Vitamin A is required in early gestation for normal cardiopulmonary development [26], and postnatally, it accelerates the development of oxygen induced contraction of the ductus arteriosus in the rat model [32], which might have contributed to the significantly lower incidence of hs-PDA found in this study. Another study did not find any difference in the spontaneous closure rate of hs-PDA by day 14 in ventilator-dependent preterm VLBW infants after IM VAS (2000–3000 IU/kg IM thrice weekly for 4 weeks) [31].

There is no “standard” regimen of oral VAS. Ambalavanan et al. [1] compared the “standard” IM regimen of VAS (5000 IU for 3 doses/week for 4 weeks) [33], with a higher IM dose (10,000 IU 3 doses/week for 4 weeks) and a once weekly IM dose (15,000 IU/week for 4 weeks). Adverse effects were seen in less than 5%. Higher dose did not increase retinol or improve outcome. Once weekly regimen led to lower serum retinol levels, but outcome was similar.

Adequate concentration of serum retinol in VLBW infants is not known. Concentrations below 20 μg/dL (0.70 μmol/L) have been considered “deficiency” in premature infants, and concentrations below 10 μg/dL (0.35 μmol/L) indicate severe deficiency and depleted liver stores [11, 33]. In the present study, approximately 60% of study infants in both the groups had deficiency at baseline, but after VAS for 28 days, 33.7% neonates in placebo group had deficiency compared to none in vitamin A group. A recent Indian study reported a very high rate of VAD (over 90%) at birth in neonates of BW < 1250 g [14]. None of our study infants in vitamin A group had a high (> 100 μg/dL) serum retinol concentration which could be considered as toxic level.

A reasonably large sample size, well-adhered study protocol and use of aqueous solution of vitamin A were the strengths of the present study. Major limitations were inclusion of relatively larger and more mature neonates, low rate of antenatal steroid coverage, lower requirement of MV, and lack of long term follow-up.

To conclude, early postnatal oral VAS was associated with better composite outcome of all-cause mortality and oxygen requirement for 28 days in VLBW neonates with respiratory distress. In LMIC, where the burden of preterm/VLBW is high, oral VAS may be implemented as a cost-effective strategy to improve the clinical outcome in VLBW neonates with respiratory distress. However, long term follow-up is necessary to document the effect of high-dose VAS on respiratory, growth, and neurodevelopmental outcome.