A total of 254 preterm infants with gestational age (GA) of ≤ 30 weeks (born in our hospital between June 2012 and June 2017) were enrolled in a previous retrospective study about the extra-uterine growth restriction and related neurodevelopmental outcomes, as previously published (10). All included infants regularly attended our follow-up program at least up to 24 months of corrected age (CA), as per our unit protocol (11), up to November 2019. Exclusion criteria were death after discharge, incomplete medical records, congenital malformations, genetic syndromes, large for gestational age (LGA) at birth, non-Italian speaker, lost to follow-up (not present at 2 sequential follow-up visits), and inclusion in any trial that could interfere with growth and outcomes (10). Data of patients were obtained from medical records.

In order to evaluate the effects of maternal intrauterine inflammation, we divided preterm infants into two groups: infants born to mothers with Ureaplasma/Mycoplasma colonization during pregnancy (cases) and unexposed infants (controls). We also considered unexposed infants born to mothers who were not screened, and pregnancy was uneventful until delivery (Figure 1).

All mothers with the threat of miscarriage or preterm labor were screened for Ureaplasma/Mycoplasma colonization; samples from the lower genital tract were obtained using vaginal swabs during physical examinations. Colonization was diagnosed via microbial culture or DNA extraction and PCR analysis according to diagnostic conditions employed at the microbiology unit of each laboratory where the mothers went.

All exposed preterm infants received clarithromycin in the first 3 days of life, after baseline respiratory specimens (nasal swabs if in spontaneous breathing or in non-invasive ventilation; bronchoalveolar lavage fluid if intubated) were collected. In our hospital, culture and identification were performed until 2015 using the Mycoplasma IST 2 kit (BioMérieux, Craponne, France), and we were able to identify U. urealyticum and M. hominis (12). Afterward, the method used has been changed into the new molecular Anyplex™ STI-7 Detection kit (Seegene Inc., Seoul, Korea), based on a multiplex real-time PCR method that can identify M. genitalium, M. hominis, U. parvum, and U. urealyticum (13).

Antibiotic treatment was stopped when samples resulted negative.

Data were collected in our follow-up facility care unit, including pregnancy and newborn characteristics.

The following maternal and neonatal characteristics were collected during the neonatal period: maternal age, antenatal corticosteroids, complete course of antenatal corticosteroids, multiple pregnancy, pathological umbilical artery Doppler parameters, cesarean section, preterm premature rupture of membranes, sex of the neonate, GA, birthweight, birth head circumference, Apgar at 5 min, number of days on invasive mechanical ventilation, incidence of bronchopulmonary dysplasia (BPD, defined by the need for supplemental oxygen at 36 w of PMA), postnatal steroids, presence of major brain lesions [defined as a grade ≥ III Intraventricular Hemorrhage (IVH) according to Papile or a Cystic Periventricular Leukomalacia (cPVL) according to De Vries (14, 15)], incidence of necrotizing enterocolitis (NEC, defined as a stage ≥ IIA according to Bell’s criteria), incidence of early-onset sepsis (EOS, defined as the presence of systemic signs suggestive of infection and positive blood and/or cerebrospinal fluid (CSF) culture before 72 h from birth), incidence of late-onset sepsis (LOS, defined as the presence of systemic signs suggestive of infection and positive blood and/or CSF culture after 72 h from birth), incidence of hemodynamically significant patent ductus arteriosus (PDA, i.e., pharmacologically or surgically treated), and days of parenteral nutrition.

All infants were assessed at 24 ± 3 months of age by an expert pediatric neurologist using Italian-validated translation of Griffith′s Mental Developmental Scales (GMDS) (0–2 years) (16).

GMDS yields five subscales: Locomotor, Personal-Social, Hearing and Speech, Eye and Hand Coordination, and Performance. The subscales yield standardized scores for each domain and a composite developmental quotient. The cognitive developmental outcome was classified normal when Griffiths’ developmental quotient (GDQ) was > 85; borderline, when GDQ was from 70 to 85; and delayed when GDQ was less than 70 (17).

Conversely, the motor outcome was classified as normal development, “minor neurological dysfunction” (MND, according to Touwen) (18), and cerebral palsy (CP, according to Bax) (19).

The onset of epilepsy was also recorded.

The severity of Retinopathy of prematurity (ROP) was defined as stage of ≥ 3, according to ICROP criteria (20), whereas all infants underwent an Auditory Brain Response test at 3 months CA to identify hearing impairment. Mild hearing loss was defined as the auditory threshold of 15–40 dB, moderate hearing loss if 40–70 dB, severe if 70–90 dB, and deafness if > 90 dB (21).

Disability was classified according to a previous scheme utilized in the EPICure studies (22), classifying outcomes as severe, moderate, and mild or no impairment using defined categories in motor, developmental, sensory, and communication domains.

Data are presented as numbers and percentages for categorical variables. Continuous variables are expressed as mean ± standard deviation (SD) if they were normally distributed or as median and interquartile range if normality could not be accepted. Categorical variables were compared using the Fisher’s exact test or Chi-square with Yates correction. T-test was used to compare cases and controls in terms of Griffiths’ development quotient (GDQ) and neurodevelopment impairment (NDI). A p-value < 0.05 was considered significant.

Multivariate analysis (by means of logistic regression in case of binary outcome or linear regression in case of continuous variable outcome) was considered when appropriate, after the correction by GA and the presence of major brain lesions (IVH ≥ III grade or cPVL).

Statistical analysis was performed using software programs Microsoft Excel (2016 for Windows) and STATA/IC (version 15.1 for Windows).

The study was carried out in compliance with the Declaration of Helsinki and its later amendments, approved by the Ethics Committee of Fondazione Policlinico Universitario “Agostino Gemelli,” Rome, Italy (in the context of the protocol number 0036181/20 - ID 3244), and written informed consent was obtained from parents for any clinical research purpose about clinical data.

Maternal characteristics and neonatal characteristics of the study patients are reported in Table 1. Among 254 preterm infants, only thirty-two infants (12.6%) were reported to be exposed to U./M. colonization during pregnancy: among these, 26 were exposed to Ureaplasma (10 U. urealyticum, 4 U. parvum, 12 U. spp. Not specified) and 6 were exposed to Mycoplasma spp. Mycoplasma cases were not typed.

However, only 91 mothers were screened for U./M. colonization. If we only consider them, the colonization rate increases up to 32/91 cases (64.8%).

Mean GA at the time of U./M. detection was 19.3 ± 7.3 weeks GA (range: 7–29). The two groups of exposed and unexposed infants were similar in all maternal characteristics, except for the administration of antenatal corticosteroids (significantly higher in the exposed group) and twin pregnancies (higher in the unexposed group).

None of the preterm infants had a respiratory culture obtained after birth that resulted positive for U./M. spp. Exposed infants had a higher incidence of EOS, but not statistically significant. Furthermore, we identified no cases of early-onset and late-onset sepsis due to U./M. spp.

In Table 2, we show how neurodevelopmental outcomes at 24 months CA were similar in exposed and unexposed infants to U./M. colonization during pregnancy.

Indeed, exposed infants and unexposed ones had a similar GDQ (106 ± 27.2 vs. 108.9 ± 19.5, respectively), without significant differences (p = 0.46). Analyzing Griffith′s subscales, exposed infants had a significantly poorer outcome than their unexposed peers in terms of locomotor abilities (subscale A, 100.7 ± 28.3 exposed vs. 111.5 ± 26.1 unexposed, p = 0.03). Conversely, unexposed infants achieved similar results in other subscales: Personal-Social interactions (subscale B), Hearing and Language assessment (subscale C), Eye and Hand Coordination (subscale D), and Performance scale (subscale E) (Table 3). We tested the result of locomotor abilities with linear regression to verify if the effect of maternal colonization was independent of confounding factors. Maternal colonization was still significantly associated with the locomotor score after correcting with GA, IVH ≥ III grade, and cPLV (OR –10.3; CI –20.5/–0.04; p = 0.04).