Infants Hospitalized with Lower Respiratory Tract Infections During the First Two Years of Life have Increased Risk of Pediatric Obstructive Sleep Apnea

Abstract

Rationale:

Lower respiratory tract infections (LRTI) during the first two years of life increase the risk of pediatric obstructive sleep apnea (OSA), but whether this risk varies by LRTI severity is unknown.

Methods:

We analyzed data from 2,962 children, aged 0–5 years, with early life LRTI requiring hospitalization (severe LRTI, n=235), treated as outpatients (mild LRTI, n=394) and without LRTI (reference group, n=2,333) enrolled in the Boston Birth Cohort. Kaplan-Meier survival estimates and Cox proportional hazards models adjusted by pertinent covariables were used to evaluate the risk of pediatric OSA.

Results:

Compared to children without LRTI, those with mild LRTI were at a higher risk of having OSA (HR 1.44, 95%CI: 1.01–2.05), and those with severe LRTI were at the highest risk (HR 2.06, 95CI: 1.41–3.02), independently of relevant covariables (including maternal age, race, gestational age, type of delivery). Additional risk factors linked to a higher risk of OSA included prematurity (HR 1.34, 95% CI 1.01–1.77) and maternal obesity (HR 1.82, 95% CI 1.32–2.52). The time elapsed between LRTI and OSA diagnosis was similar in mild and severe LRTI cases, with medians of 23 and 25.5 months, respectively (p=0.803).

Conclusion:

Infants with severe early life LRTI have a higher risk of developing OSA, and surveillance strategies to identify OSA need to be particularly focused on this group. OSA monitoring should continue throughout the preschool years as it may develop months or years after the initial LRTI hospitalization.

INTRODUCTION

Obstructive sleep apnea (OSA) affects 1–5% of the pediatric population^1–4. Pediatric OSA is associated with significant morbidity, especially if left untreated, including neurocognitive impairment, behavioral issues, failure to thrive, hypertension, and cardiac dysfunction, as well as systemic inflammation that can worsen other comorbidities such as obesity and asthma^3–8. There are also high healthcare-related costs for children with OSA⁹. This heightened socioeconomic burden associated with pediatric OSA can be offset by early diagnosis and treatment, which requires clinicians to be aware of which children are at the highest risk of this disease.

Previous studies have demonstrated that lower respiratory tract infections (LRTIs) in young children increase their risk of pediatric OSA^10–12. As LRTIs are the most common cause of hospitalization in young children, the development of OSA in this high-risk group could have important implications for pediatric healthcare. Snow et al. first described a significant association between RSV bronchiolitis requiring hospitalization in the first year of life and the development of OSA in 21 cases compared to 63 controls¹⁰. Similar findings have been published from a population-based study in Taiwan¹¹. More recently, in a large longitudinal birth cohort, our team identified that viral respiratory infections occurring in the first two years of life significantly increase the risk of developing OSA by age five¹². Notably, in our prior study, LRTIs occurring after the first two years did not increase OSA risk¹². Together, these results provide strong evidence that early life viral respiratory infections are an important risk factor for the development of pediatric OSA. However, there are currently no studies evaluating whether the increased risk of pediatric OSA associated with early life LRTI differs depending on disease severity, with potentially greater risk observed in severe cases requiring hospitalization.

The primary aim of this study was to test the hypothesis that experiencing early life LRTIs, specifically within the first two years of life, increases the risk of developing OSA by the age of five. Furthermore, the study aimed to investigate whether the risk of OSA is higher in children who had severe LRTIs requiring hospitalization. Our secondary aim was to define the time lapse between LRTI hospitalization and subsequent diagnosis of OSA in children. To accomplish these aims, we conducted a comprehensive analysis of clinical data obtained from a prospective birth cohort comprising 2,962 participants.

METHODS

Study population

The Boston Birth Cohort (BBC) is an ongoing prospective, longitudinal birth cohort of newborns recruited at the Boston Medical Center (BMC), encompassing a predominantly low-income, underprivileged, inner-city population in Boston, MA. A detailed description of the BBC study design and data collection has been published^12,13. In this study, we included all eligible infants with complete baseline and follow-up data available (n = 3,162). Exclusion criteria included Down Syndrome, cleft lip and/or palate, and cystic fibrosis. We also excluded infants diagnosed with OSA during the first three months of life and those who developed OSA before their first episode of LRTI (Figure 1). A total of 2,962 infants were included in the final analyses. This cohort study was approved by the institutional review boards of Johns Hopkins School of Public Health and Boston Medical Center and included written informed consent for all individuals enrolled.

Figure 1. Study Population and Outcomes.

Clinical outcomes and covariates

The main outcome of this study was a diagnosis of OSA at five years of age. Pediatric OSA was ascertained using ICD-9/ICD-10 codes extracted from the patient’s electronic medical record (EMR), as previously published by our team¹². The main predictor of this study was severe early-life LRTI, defined as the first episode of LRTI occurring in the first two years of life (0–24 months of age) and requiring hospitalization. Children with early-life LRTI not requiring hospitalization, referred to as mild LRTI, were included for comparison. Early-life LRTI information was extracted from the patient’s EMR using ICD-9/ICD-10 diagnosis codes and the healthcare setting listed for the corresponding clinical encounter (hospital admission, observation, ER visit, or ambulatory). Relevant covariables included prenatal and perinatal characteristics linked to respiratory morbidities such as maternal race, maternal level of education, pre-pregnancy maternal body mass index (BMI) category (normal, underweight, overweight, and obese), prenatal smoking (continuous smoking during pregnancy) as well as infant’s sex, gestational age, the season of birth, and breastfeeding status. We also examined other relevant maternal factors, such as age and parity. The time from early life LRTI to the diagnosis of OSA was also measured.

Statistical analysis

We first defined the baseline characteristics of our study population. Groups were compared using non-parametric Wilcoxon-Rank sum tests and single linear regression models comparison for continuous variables, and Chi-Square tests and single logistic regression for categorical data. Covariates with missing observations were imputed using the multiple imputations (mi) functions in STATA 14, and we performed sensitivity analysis comparing imputed and non-imputed data to verify that regression results were comparable. The cumulative risk of incident OSA after LRTI was estimated by Kaplan-Meier survival analysis and Cox proportional hazards models. Children entered the study at birth, and those who developed OSA left the study at the age of the first OSA diagnosis on record. Time to incident OSA was calculated as the earliest age at which a diagnosis of OSA is recorded on EMR. Children who did not develop OSA during the observation period were censored at 60 months of age or at the age of their last visit if they were lost to follow-up. Statistical analyses were conducted using the software STATA version 14 (StataCorp. Stata Statistical Software: Release 14. College Station, TX, 2015) and R studio (RStudio: Integrated Development for R. RStudio, PBC, Boston, MA, 2020)

RESULTS

Study population and baseline characteristics

The final study included 2,962 children after exclusion criteria (Figure 1). We identified that 21% of the children included (n = 629) had developed early life LRTI. Of the children who developed early-life LRTI, 37% (n = 235) had a severe LRTI, defined as requiring hospitalization. Compared to children with no early life LRTI, those who developed severe LRTI did not differ regarding maternal age, race/ethnicity, education, or smoking during pregnancy, as well as the infant’s sex (Table 1). Notably, as previously reported by our team and others, there were significant differences in the prevalence of severe prematurity (GA ≤ 32 weeks), breastfeeding status, and delivery type, as well as maternal parity, maternal pre-pregnancy obesity (BMI ≥ 30), and maternal history of asthma between children who developed severe LRTI and those that did not develop early life LRTI (Table 1).

Table 1.

Summary of maternal and infant baseline characteristics divided by study group.

Maternal baseline characteristics		No LRTI (n=2333)	Mild LRTI (not hospitalized) (n=394)	Severe LRTI (hospitalized) (n=235)
	Age in years
	Median (IQR)	28.15 (9.91)	27.82 (10.71)	28.36 (11.13)
	Race/Ethnicity (n, %)
	White	151 (6.47)	26 (6.6)	15 (6.38)
	Black	1403 (60.14)	236 (59.9)	123 (52.34)
	Hispanic	514 (22.03)	72 (18.27)	73 (31.06)
	Asian & Pacific	33(1.41)	5 (1.27)	4 (1.7)
	Other	232 (9.94)	55 (13.96)	20 (8.51)
	Education (n, %)
	No school/elementary	107 (4.59)	11 (2.79)	12 (5.11)
	Some secondary school	551 (23.62)	96 (24.37)	64 (27.23)
	High school graduate	849 (36.39)	137 (34.77)	89 (37.87)
	Some college	502 (21.52)	94 (23.86)	44 (18.72)
	College graduates & higher	304 (13.03)	55 (13.96) †	22 (9.36) †
	Unknown	20 (0.86)	1 (0.25)	4 (1.7)
	Parity (n, %)
	First child	1014 (43.46)	161 (40.86)	84 (35.74)
	> 1 child	1318 (56.49) #	233 (59.14)	151 (64.26)#
	pre-pregnancy BMI, kg/m2, median (IQR)	25.09 (7.46) #	25.98 (7.84)	26.13 (8.46)#
	BMI (n, %)
	18–25	981 (42.05)	151 (38.32)	89 (37.87)
	<18	101 (4.33)	13 (3.3)	8 (3.4)
	25–30	599 (25.68)	117 (29.7)	59 (25.11)
	>30	509 (21.82) #	97 (24.62)	68 (28.94)#
	Unknown	143 (6.13)	16 (4.06)	11 (4.68)
	Smoking during pregnancy (n, %)
	Yes	233 (9.99)	51 (12.94)	29 (12.34)
	No	2079 (89.11)	341 (86.55)	205 (87.23)
	Unknown	21 (0.9)	2 (0.51)	1 (0.43)
	Maternal history of asthma (n, %)
	No	1668 (71.50)	267 (67.77)	134 (57.02)
	Yes	254 (10.89) #	50 (12.69) †	51 (21.7)# †
	Unknown	411 (17.62)	77 (19.54)	50 (21.28)
Infant baseline characteristics		No LRTI (n=2333)	Mild LRTI (not hospitalized) (n=394)	Severe LRTI (hospitalized) (n=235)
	Sex (n, %)
	Female	1194 (51.18)	165 (41.88)	108 (45.96)
	Male	1139 (48.82) *	229 (58.12)*	127 (54.04)
	Delivery type (n, %)
	C-section	790 (33.86)	154 (39.09)	105 (44.68)
	Vaginal	1528 (65.50) #	240 (60.91)	130 (55.32) #
	Unknown	15 (0.64)	0	0
	Gestational Age
	Median (IQR)	38.85 (3.15) #	38.57 (3.14) †	37.57 (6.43)# †
	Full term	1718 (73.64)	281 (71.32)	124 (52.77)
	Mild Preterm (33–36 wks)	457 (19.59) #	88 (22.34)	57 (24.26)#
	Severe Preterm (28–32 wks)	102 (4.37) #	20 (5.08) †	24 (10.21)# †
	Extreme Preterm (<28 wks)	56 (2.4) #	5 (1.27) †	30 (12.77)# †
	Breastfed (n, %)
	No	524 (22.46)	94 (23.86)	75 (31.91)
	Yes	1546 (66.27) #	263 (66.75)	138 (58.72)#
	Unknown	263 (11.27)	37 (9.39)	22 (9.36)
	Season of birth (n, %)
	Summer	635 (27.22)	117 (29.7)	51 (21.7)
	Fall	604 (25.89)	106 (26.9)	69 (29.36)
	Winter	570 (24.43)	84 (21.32) †	60 (25.53) †
	Spring	524 (22.46)	87 (22.08)	55 (23.4)

Significant p values were expressed as follows:

^*p<0.05 No LRTI vs. Mild LRTI,

^#p<0.05 No LRTI vs. Severe LRTI, and

^†p<0.05 Mild LRTI vs. Severe LRTI

The severity of early life LRTI impacts the risk of OSA during the first five years of life

We examined the longitudinal association between the severity of early-life LRTI and the risk of developing pediatric OSA. After exclusion criteria, we identified a total of 230 children with OSA diagnosis (Figure 1). Among these children diagnosed with OSA, 31.2% (n=72) had experienced early-life LRTI, categorized as either mild (n=38) or severe (n=34) depending on whether hospitalization was required. The Log-rank test for comparison of the survival function of the 2,962 children included demonstrated a significantly increased risk of OSA in children with early-life LRTI. The highest risk for OSA was seen in children hospitalized due to severe LRTI (Figure 2, orange dashed line, p<0.001). The risk of OSA was also increased in those with mild early life LRTI relative to children without a history of LRTI (Figure 2, blue dashed line).

Figure 2. Cumulative risk of OSA during the first five years of life in children with no LRTI, mild LRTI and severe LRTI.

Kaplan-Meier analysis of the 2,962 children included demonstrates a significant difference in the incidence of OSA between children with severe early LRTI and those without LRTI during this period (p value< 0.001)

We also evaluated the longitudinal association between the severity of early-life LRTI and the risk of developing pediatric OSA after adjusting for pertinent co-variables. As shown in Table 2, we found that this association was independent of relevant covariables, including maternal race, educational level, parity, smoking during pregnancy, as well as infant sex, type of delivery, breastfeeding status, and preterm birth. The greatest risk of OSA was seen in children hospitalized with severe LRTI (adj HR 2.06, 95% CI 1.41–3.02, Table 2). The risk of OSA was also increased in those with mild early life LRTI relative to children without a history of LRTI (adj HR 1.44, 95% CI 1.01–2.05). Additional risk factors linked to the development of OSA included preterm birth (HR 1.34, 95% CI 1.01–1.77) and maternal obesity (HR 1.82, 95% CI 1.32–2.52).

Table 2.

Adjusted Cox proportional hazard model assessing the relationship between early-life LRTI and OSA during the first five years of life.

	Variables	Adjusted Hazard Ratio (95% CI)	p-value
Main Predictor	LRTI severity
	LRTI not hospitalized	1.431 (1.001 – 2.045)	0.049
	LRTI hospitalized	2.037 (1.391 – 2.982)	<0.001
Maternal	Race/Ethnicity
	White
	Black	0.93 (0.50– 1.73)	0.83
	Hispanic	1.43 (0.74 – 2.74)	0.28
	Asian & Pacific	1.58 (0.49 – 5.05)	0.43
	Other	1.26 (0.62 – 2.57)	0.51
	Education
	No school/elementary
	Some secondary school	1.11 (0.56 −2.22)	0.75
	High school graduate	1.19 (0.61 – 2.35)	0.59
	Some college	1.44 (0.72 – 2.87)	0.29
	College graduates & higher	1.05 (0.49 – 2.26)	0.889
	Multiparous	0.892 (0.68 – 1.16)	?
	BMI
	18–25
	<18	0.94 (0.433 – 2.079)	0.897
	25–30	1.15 (0.80 – 1.641)	0.441
	>30	1.82 (1.314 – 2.522)	<0.001
	Pregnancy Smoking
	Yes	0.814 (0.505 – 1.31)	0.402
Infant	Sex	1.24 (0.95 – 1.61)	0.106
	Delivery type	0.853 (0.65 – 1.11)	0.252
	Prematurity	1.339 (1.014 – 1.768)	0.039
	Breastfed	0.84 (0.62 – 1.13)	0.260

Time between LRTI and subsequent OSA diagnosis in young children

We next evaluated the period of time between LRTI and diagnosis of pediatric OSA. Overall, we found that in our study population, the average age of OSA diagnosis was 36 months, and the median time to this diagnosis after LRTI was about two years (median 23 months) The time to diagnosis of OSA did not differ by the severity of LRTI (p = 0.803, Figure 3). In those with mild LRTI, the median time to diagnosis was 23 months, compared to 25.5 months in those with severe LRTI.

Figure 3. Time between LRTI and OSA diagnosis in children with mild LRTI and severe LRTI.

The median age of OSA diagnosis was 23 months for mild LRTI and 25.5 months for severe LRTI, and no statistical difference was detected (p=0.803).

DISCUSSION

This study demonstrates that severe LRTI in the first two years of life is a critical risk factor for the development of OSA in children. In this large cohort of nearly 3,000 children, those hospitalized due to a severe LRTI had a two-fold increased risk of pediatric OSA by five years of age compared to children with no early life LRTI. Although the presence of mild LRTI was also linked to OSA risk, the children with the highest risk were those who required hospital admission. The association between LRTI and subsequent OSA development was independent of and stronger than other maternal and infant well-established risk factors for pediatric OSA, such as premature birth and maternal obesity^14–16. We also found a prolonged lapse between early-life LRTI and pediatric OSA diagnosis, which did not differ by LRTI severity. Thus, our study provides new evidence that children requiring hospitalization due to severe LRTI represent a high-risk group for OSA that may benefit from close surveillance and interventions to prevent the severe consequences of untreated OSA in children^6,17,18.

Our study expands on the existing evidence on the link between viral respiratory infections and OSA^10–12 by further stratifying children with early-life LRTI. Several mechanisms have been proposed regarding how respiratory viruses can alter both upper and lower airways^19–22. Notably, one of the main causes of OSA in children is adenotonsillar hypertrophy^2,23. Previous studies have detected viruses in children undergoing tonsillectomy or adenoidectomy, implying that persistent or recurrent viral respiratory infections may promote nasopharyngeal lymphoid proliferation^24–26. Respiratory viruses may also impact airway neuromotor control and influence neuroimmunomodulatory pathways in the upper and lower respiratory tracts^20,27–29, contributing to nasopharyngeal obstruction and airway hyperresponsiveness^10,19,29. Alternatively, infants who develop OSA after a severe LRTI might be intrinsically predisposed to both conditions^29,30. Nonetheless, the precise mechanisms underlying these observations are yet to be understood, warranting further investigation into how early life respiratory infections influence the onset of pediatric OSA.

In our study we also identified that there was a period of approximately two years between the children’ first LRTI and the diagnosis of OSA. One potential reason for this time gap is a delay in diagnosis related to current paradigms and clinical practice guidelines². While studies show that more than 10% of preschool-aged children snore regularly, OSA is much less common, estimated at 1–5%^1–3. Similarly, studies show that snoring and sleep-disordered breathing may spontaneously resolve over time or respond to courses of intranasal steroids³¹. This leads many clinicians to decide on an observation period, which can delay the diagnosis. Another potential reason for the extended interval between LRTI and OSA diagnosis may be the delayed effect of the virus on OSA pathogenesis. Nonetheless, based on our findings, we believe that clinicians should actively screen for signs and symptoms of sleep-disordered breathing in children with a history of hospitalization for LRTI in the first two years of life. This should include timely diagnostic evaluations and therapeutic interventions if the diagnosis of OSA is confirmed.

Adenotonsillectomy should be considered as the first treatment option if a child is diagnosed with OSA and has a physical exam consistent with adenotonsillar hypertrophy². Prior studies have reported that surgery is curative in 71 to 87% of pediatric OSA cases^2,32. Residual disease after adenotonsillectomy is more common among older kids (>7 yrs.), obese, or with other comorbidities such as Down syndrome or craniofacial abnormalities^33,34. Beneficial effects of early intervention include improvement of OSA symptoms, improvement of behavior, improvement of quality of life, and polysomnography (PSG) findings in school-age children³⁵. The early identification and treatment of OSA in children can also have an effect on preventing lower airway hyperresponsiveness and asthma. For instance, in a recent study³⁶, the prevalence of wheezing significantly decreased seven months after adenotonsillectomy to treat in children with OSA (47% vs. 21.6%, p < 0.001). This further highlights the potential impact of providing early intervention for OSA after severe LRTI to prevent morbidity and complications associated with this condition in children.

Our study has some limitations. First, although we included a large number of children, they are from an inner-city population, where the risk of pediatric OSA and early-life LRTI is higher than in the general population^37,38. Second, the definition of OSA was based on EMR data and was not confirmed with PSG results, nor was the severity of OSA established. However, previous studies in adult populations have shown a good correlation between EMR diagnosis and PSG confirmation³⁹. Finally, although LRTI preceded OSA diagnosis, we cannot demonstrate causality, as the precise onset of OSA symptoms could not be determined. In fact, the exact date of OSA diagnosis was unknown in our study, as is typically the case with large databases. As the EMR information was not easily accessible in all subjects (2,962 children in this study), we could not analyze additional variables that may influence LRTI and OSA risk. This included clinical features of LRTI such as the length of hospitalization, need for intensive care-level support, required respiratory treatments (e.g., antibiotics), and virus type to allow differentiation of LRTIs as bacterial or viral infections. Another limitation is that we did not have information about subsequent LRTIs or pulmonology visits. This is important because children hospitalized with LRTI might have had other comorbidities not specified in the dataset, such as history of atopy, environmental exposures, lifestyle factors, and socioeconomic information (e.g., area deprivation index). Thus, further work is needed to confirm our current findings with additional clinical variables, PSG-based diagnosis of OSA, and explore the specific viruses associated with the LRTI.

In conclusion, our results call for establishing a new paradigm in the screening and surveillance of OSA in children hospitalized for LRTI in their first two years of life. These children have a two-fold increased risk of developing pediatric OSA compared to children without early-life LRTI. Given the plethora of adverse health effects associated with OSA in children, clinicians need to be aware of its risk factors. This novel approach to the care and management of this patient population may impact the timing of pediatric OSA diagnosis and appropriate management, therefore decreasing subsequent OSA morbidity.

Abbreviations.

LRTI

Lower respiratory tract infection

LRTIs

lower respiratory tract infections

OSA

Obstructive sleep apnea

BBC

The Boston Birth Cohort

BMC

Boston Medical Center

PSG

polysomnography

Data Availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request after review and approval of the Institutional Review Board.