Introduction
According to the Kangaroo Foundation (KF) database, 11,953 oxygen-dependent (OD) low birth weight infants (LBWI) were followed in the ambulatory Kangaroo Mother Care Program (KMCP) in Bogotá, Colombia between 1998 and the first semester of 2021. It was observed that the frequency of oxygen dependency decreased from 43% between 2003 and 2007 to 7% between 2018 and 2021. Patients <32 weeks gestational age (GA) were between 41-62%. In addition, hospital stay was variable according to the severity of perinatal compromise.
For the last 24 years, the KMCP has managed LBWI with ambulatory oxygen. A first study conducted in 2004 in 12 hospitals of Bogotá, showed that weight, GA at birth, mechanical ventilation, intrauterine growth restriction, and type of institution (i.e., low vs. intermediate-high mortality) were independently associated with bronchopulmonary dysplasia (BPD) of increasing severity or even death1. Of this cohort, 194 patients were followed up to 1-year corrected age (CA); breastfeeding at term was successful in 76%, growth at one year was appropriate, 74% of the cohort were still receiving home oxygen at 40 weeks and at 3 months 23%. Moreover, around 57 % of the cohort were readmitted to hospital at least once, 47% of them because of respiratory conditions2. A latter study was conducted between 2017 and 2018 with 445 patients ≦33 weeks GA and OD at admission. It was observed that 56% had a history of sepsis and 49% had received invasive ventilation. Multivariate analysis also found that oxygen dependency was associated with low hemoglobin levels at admission, the number of blood transfusions received, and lower GA. On the other hand, weight gain and exclusive breastfeeding until oxygen weaning were identified as protective factors3.
Different therapeutic approaches have been described to prevent and treat BPD, such as xanthines, diuretics, bronchodilators, vitamin A and D, corticosteroids, and probiotics, without achieving clear effectiveness4,5. For more than 40 years, xanthines have been used to manage premature patients6. Within this group Caffeine, Aminophylline, Doxapram, and Theophylline stand out, with the first one being the most well-known. These medications work by stimulating respiratory centers and promoting diaphragmatic contractility, improving pulmonary distensibility, and reducing airway resistance. These effects are of particular importance in apnea of prematurity and in chronically ventilated newborns who may develop muscular fatigue with skeletal and diaphragmatic muscular atrophy. By improving muscle contractility, they contribute to chest wall stabilization, increasing functional residual capacity and allowing successful extubation; this last fact has been associated with less days with oxygen7,8.
Theophylline is an orally administered xanthine used for the treatment of respiratory diseases such as asthma and chronic obstructive pulmonary disease9. It is extensively metabolized, and its main metabolic product is caffeine due to N-methylation. Apnea of prematurity and BPD treatment has been described as its principal uses10. Theophylline (1,3-dimethylxanthine) therapy in newborns has the inconvenience of greater toxicity due to its erratic absorption and elimination, with a prolonged half-life of approximately 30 hours, nine times longer than in adults, due to the immaturity of the cytochrome monooxygenase P450 enzymes11. Several adverse effects associated with Theophylline use have been described, such as cardiac arrhythmias, nausea, vomiting, headache, diarrhea, irritability, and insomnia 12,13, without having a specific antidote.
From 1996 until 2017 oral Theophylline was systematically implemented in ambulatory KMCPs in Colombia as an adjuvant medication to reduce the number of days of ambulatory oxygen14. In 2017, in view of the available evidence of increased risk of toxicity with Theophylline, routine treatment was discontinued, and recommendations were modified in the new KMC technical guidelines of the Health Ministry15.
This study aims to evaluate the effectiveness of oral Theophylline to reduce the days of oxygen dependency and to assess the frequency of adverse events related to this drug.
Materials and Methods
Quasi-experiment before and after withdrawal of systematic Theophylline given to LBWI with ambulatory oxygen in two KMCPs. We compared two cohorts of LBWI OD at admission in the outpatient KMCPs of two teaching hospitals in Bogotá, Colombia (Hospital Universitario San Ignacio and Hospital Infantil San José).
The first cohort received oral Theophylline up to 40 weeks’ GA between July 25, 2017 and May 31, 2018. The second cohort included OD LBWI admitted to the KMCP without Theophylline treatment, between June 01, 2018 and April 20, 2019. Patients with a history of seizures, congenital heart disease or cardiac arrhythmia were excluded.
We compared days of oxygen requirement, hospital readmissions, number of KMCP consultations, feeding pattern, anthropometric measures, incidence of tachycardia, gastroesophageal reflux (GER), colic, seizures and other side effects that could be attributed to oral Theophylline treatment at 40 weeks GA and at oxygen weaning.

Sample size

It was calculated using the software STATA 14. The main outcome was days of oxygen dependency during follow-up in the ambulatory KMCPs. According to the KMCP database, OD patients received on average 70 days of oxygen (SD 64 days). To obtain a 15 days reduction in oxygen requirement, an alpha of 0.05 and a power of 80% with two-tailed estimation, and follow-up losses of 20%, the sample size per group was 344 patients.

Intervention

Statistical Analysis

For the analysis Stata 14 was used. Quantitative data was reported as medians with their interquartile range given their non-normal distribution. Qualitative data was reported as absolute and relative frequencies. Bivariate analysis of quantitative data was done with the Mann Whitney nonparametric test and for qualitative data we used Chi-square or Fisher’s exact test.
Propensity Score Matching was done for unbalanced baseline variables between the groups. Then, multivariate analysis was performed with multiple linear regression having as dependent variable days of oxygen dependence, and as independent variables those that according to the evidence, are associated with oxygen dependence.