Incidence of respiratory virus illness and hospitalizations in a Panama and El Salvador birth cohort, 2014-2018.

Background: Respiratory viruses remain a key cause of early childhood illness, hospitalization, and death globally.The recent pandemic has rekindled interest in the control of respiratory viruses among paediatric populations. We estimate the burden of such viruses among children <2 years.
Methods: Enrolled neonates were followed until two years of age. Weekly active symptom monitoring for the development of acute respiratory illnesses (ARI) defined as cough, rhinorrhoea, difficulty breathing, asthenia, anorexia, irritability, or vomiting was conducted. When the child had ARI and fever, nasopharyngeal swabbing was performed, and samples were tested through singleplex RT-PCR. Incidence of respiratory viruses was calculated by dividing the number of laboratory-confirmed detections by the person-time accrued during weeks when that virus was detectable through national surveillance then corrected for under-ascertainment among untested children. Findings: During December 2014-November 2017, 1567 enrolled neonates contributed 2,186.9 person-years (py). Six in ten (64·4%) children developed ARI (total 2493 episodes). Among children <2 years, incidence of respiratory syncytial virus (RSV)-associated ARI episodes (21·0, 95%CI 19·3-22·8, per 100py) and rhinovirus-associated (20·5, 95%CI 20·4-20·7) were similar and higher than parainfluenza 1-3-associated (14·2, 95%CI 12·2-16·1), human metapneumovirus-associated (9·2, 95%CI 7·7-10·8), influenza-associated (5·9, 95%CI 4·4-7·5), and adenovirus-associated ARI episodes (5·1, 95%CI 5·0-5·2). Children aged <3 months had the highest rates of RSV ARI (49·1, 95%CI 44·0-54·1 per 100py) followed by children aged 3-5 (25·1, 95%CI 20·1-30·0), 6-11 (17·6, 95%CI 13·2-21·9), and 12-23 months (11·9, 95%CI 10·8-12·9). One in ten children with RSV was referred to the hospital (2·5, 95%CI 2·1-2·8, per 100py). Interpretation: Children frequently developed viral ARI and a substantive proportion required hospital care. Such findings suggest the importance of exploring the value of new interventions and increasing uptake of existing prevention measures to mitigate burden of epidemic-prone respiratory viruses. Funding: The study was supported by the Centers for Disease Control and Prevention.

Evidence before this study

Respiratory viruses are commonly identified in young children globally. The burden of such viruses has been well described in high-income countries, where the burden of respiratory syncytial virus (RSV) was shown to be high among young children and a major cause of hospital admissions. There is comparatively little data on respiratory virus burden from low- and middle-income tropical countries (LMICs).

Added value of this study

In two LMICs, Panama and El Salvador, rhinovirus was the most common (35.7%) respiratory virus detected among children aged <2 years with acute respiratory illness (ARI), followed closely by RSV (25.8%). RSV infection was highest among the youngest children (aged <3 months) and caused the greatest number of ARI episodes and hospital referrals.

Implications of all the available evidence

Our study shows the role of RSV in causing ARI morbidity among very young children (<3 months) in two LMICs and should inform investment in the acceleration of nonpharmaceutical interventions and pharmaceutical therapies.

Alt-text: Unlabelled box

Introduction

COVID-19 has provided public health officials with first-hand experience of nonpharmaceutical interventions, accelerated the production of vaccines, monoclonal antibodies, and therapeutics, and escalated interest in the control of respiratory viruses. To better understand the value proposition of nonpharmaceutical interventions and pharmaceutical therapies (i.e., vaccines, monoclonal antibodies, and therapeutics), data on incidence of respiratory viruses among young children are needed. While there is ample evidence of such burden from high income temperate countries, there is comparatively little available from low- and middle-income tropical countries (LMICs). Furthermore, available data are typically about respiratory syncytial viruses (RSV) and influenza, but seldom about other viruses which frequently sicken and hospitalize very young children.

Before the COVID-19 pandemic, focusing on RSV and influenza was sensible because these were the only two respiratory viruses with available pharmaceuticals for prevention. Premature and other high risk infants can be provided prophylaxis against RSV with monoclonal antibodies. Such antibodies can help mitigate the large disease burden of RSV infections among infants. RSV-associated hospitalizations and deaths affect an estimated 55,000–125,000 and 200–500 children each year, respectively. Caregivers of children aged <6 months can also be vaccinated against influenza to cocoon infants before their immune systems are ready to benefit from vaccination. Children aged ≥6 months can be vaccinated and those ≥2 weeks treated for influenza with antivirals. In addition, companies that manufacturer mRNA and other newly implemented vaccines are racing to test products, some of which elicit substantial immunogenicity against RSV in phase 1 clinical trials. While the public health community awaits these products, it is opportune to better describe the burden associated with respiratory viruses.

In this manuscript, we leverage findings from a multi-year birth cohort in two LMICs to quantify the burden of selected respiratory viruses among young children. Starting in 2014, the Centers for Disease Control and Prevention (CDC) established a cooperative agreement with Gorgas Memorial Institute to develop the mother/baby cohort in Panama and El Salvador known as INFLUMIKA for its Spanish acronym () which documented the negative association between influenza and infant cognitive development. In the current analysis, we estimate the incidence rates of RSV, influenza viruses, rhinoviruses, adenoviruses, parainfluenza 1–3, human metapneumoviruses (HMPV), bocaviruses, and seasonal coronaviruses among children aged <2 years participating in INFLUMIKA. As a secondary objective, we explore risk factors for bronchopneumonia, bronchiolitis, or hospitalization. Study findings are intended to inform value proposition analyses for existing (RSV monoclonal antibodies, influenza vaccines, influenza antivirals), and new interventions (e.g., mRNA vaccines against RSV and coronaviruses) especially where their opportunity cost might seem high a priori to local officials (e.g., in LMICs).

Methods

Enrolment & follow-up

Mothers

Women in their first or second trimester seeking prenatal care at government-subsidized health clinics that manage prenatal care and provide influenza vaccination free-of-charge in Panama (i.e., Tocumén and 24 de Diciembre Health Centres) and El Salvador (Tomás Pineda and San Rafael Health Centres) were eligible for enrolment into INFLUMIKA as described previously. Briefly, the study teams approached eligible women and among those who signed informed consent, data were collected about their demographics, prenatal history, ultrasound-estimated gestational age, and influenza vaccination status. Within two weeks of childbirth, mothers were asked to consent to the enrolment of their neonates in the birth cohort.

Children

At birth, children were weighed, and their length and head circumference measured using calibrated infantometers. Neonates with TORCH (Toxoplasmosis, Other Agents, Rubella, Cytomegalovirus, and Herpes Simplex) syndrome, chromosomal abnormalities, or other congenital illnesses were excluded. Enrolled neonates were followed once per week by phone until their second birthday to identify acute respiratory illnesses (ARI) or other acute illnesses that may have developed in the preceding week.

ARI were defined as cough, rhinorrhoea, difficulty breathing, asthenia, anorexia, irritability, or vomiting; non-specific symptoms associated with ARI were included to increase the sensitivity of our case definition. Staff asked parents to bring children to study clinics for nasopharyngeal swabbing when they had ARI and feverishness (i.e., subjective or measured fever ≥38° C). Due to resource constraints, only children with both ARI and feverishness, rather than all children with ARI, were swabbed. Evidence indicates that the probability of a positive test for most respiratory viruses increases during the presence of fever and adjustments for under-ascertainment were made as described below.

Swabs were inserted into viral transport media and kept refrigerated or in cool boxes at 4–8 °C until transport within 24–48 h to each country's national reference laboratory (i.e., Gorgas Memorial Institute for Health Studies in Panama and the Laboratorio Nacional de Salud Pública in El Salvador). Samples were then tested through singleplex RT-PCR to detect influenza virus, RSV, parainfluenza 1–3, adenovirus, HMPV, and rhinovirus RNA using CDC testing protocols. Study staff also systematically recorded laboratory results, imaging studies, and vaccinations obtained through children's routine preventive clinical care. Unscheduled visits to clinics or hospitals within the study catchment were also recorded whenever possible.

Statistical analysis

Incidence of each respiratory virus was calculated by dividing the number of ARI with laboratory-confirmed findings by the person-time each child accrued during weeks when that virus was detectable by each country's national surveillance. We corrected for under-ascertainment among afebrile children who were not swabbed using previously described methods,, and report these corrected estimates as virus-associated below. In brief, we assumed untested children had a similar probability of testing positive for each virus as afebrile children of the same age-group and country, who were tested the same epidemic week through national surveillance, after correcting for differences in the probability of testing positive for each virus by age-group and by afebrile vs. febrile status. Confidence intervals for these attributions incorporated the variance in the proportions of 1. persons of any age positive for each virus among all persons tested that epidemic week, 2. children of a given age-group positive for a virus among persons of any age positive for that virus, and 3. afebrile vs. febrile children positive for influenza and RSV.

Cox proportional hazards regression was used to evaluate associations between incident observed influenza ARI, RSV ARI, or adverse outcomes (bronchopneumonia, bronchiolitis, and hospitalization) and potential risk factors adjusting for sex of the child. Child's age (i.e., <3, 3–5, 6–11, and 12–23 months) was examined as an effect modifier. Time-varying covariates were assessed using a Kolmogorov-type Supremum test for proportional hazards assumption and for collinearity. When variables were collinear (r ≥ 0.70), the variable with the strongest univariate association was included in the model.

This study was reviewed and approved by the Panama and El Salvador IRB; CDC relied on Panama and El Salvador ethical review. Eligible mothers signed informed consent for their neonates’ participation.

Role of the funding source

The study was funded through CDC cooperative agreement 5U01IP000791-05; its staff provided technical assistance but did not otherwise have contact or access to identifying information from participants.

Results

Enrolment, demographics, and follow-up

From December 2014 until November 2017, we identified 1584 neonates and enrolled 1567 which became our analytic sample. Approximately half (651, 46·1%) were female (Table 1). Children in El Salvador were born to self-described mestizo (692, 99·7%), indigenous (1, 0·1%) or black (1, 0·1%) mothers and in Panama to mestizo (444, 51·3%), indigenous (225, 26·0%), or black (52, 6·0%) mothers. One-quarter of children (337, 22·3%) were born to mothers with less than a middle-school education. Each child contributed an average of 1·4 (SD 0·6) person-years (py) for a cumulative 2186·9py among all children. Rhinovirus was detectable through the national surveillance system during the entire 2186·9py risk period, adenovirus during 91%, parainfluenza 1–3 during 75%, RSV during 74%, HMPV during 67%, and influenza during 65% of the risk period. Six hundred and thirty-six children completed the study; 430 (27·4%) were lost to follow-up during their first year, and 495 (31·6%) were lost to follow-up during their second year typically because they missed more than two control follow-up appointments or moved out of the study area.

Table 1

Mother and child demographic characteristics by country.

	Panama	El Salvador	P-value	Overall
Maternal characteristics
Number	865	694	-	1559
Age at enrollment (years) [median (IQR)]	23 (20–28)	22 (19–27)	0·02	23 (19–27)
Primigravid	254 (29·4%)	323 (46·7%)	<0·001	577 (37·1%)
Race/ethnicity			<0·001
Indigenous	225 (26·0%)	1 (0·1%)		226 (14·5%)
Mestizo	444 (51·3%)	692 (99·7%)		1136 (72·9%)
White	141 (16·3%)	0 (0%)		141 (9·0%)
Black	52 (6·0%)	1 (0·1%)		53 (3·4%)
Other	3 (0·4%)	0 (0%)		3 (0·2%)
Married	64 (7·4%)	126 (18·2%)	<0·001	190 (12·2%)
Primary school or less	148 (17·6%)	189 (28·2%)	<0·001	337 (22·3%)
Pre-existing condition	106 (12·5%)	25 (3·6%)	<0·001	131 (8·5%)
Income < $400 USD	233 (38·1%)	533 (90·2%)	<0·001	766 (63·7%)
Consumes alcohol	25 (2·9%)	2 (0·3%)	<0·001	27 (1·7%)
Smoker	6 (0·7%)	3 (0·4%)	0·74	9 (0·6%)
Influenza vaccinationa	595 (68·8%)	292 (42·1%)	<0·001	887 (56·9%)
Child characteristics
Number	871	696		1567
Female	321 (44·4%)	330 (47·8%)	0·20	651 (46·1%)
Twins	12 (1·4%)	4 (0·6%)	0·08	16 (1·0%)
Gestational age (weeks) [median (IQR)]	39 (38–40)	39 (37–40)	0·19	39 (38–40)
Preterm	78 (9·2%)	104 (15·1%)	0·002	182 (11·8%)
Weight Z<-2b	15 (3·3%)	23 (4·3%)	0·45	38 (3·9%)
Head circumference Z<-2b	21 (4·9%)	26 (4·9%)	0·98	47 (4·9%)
Length Z<-2b	8 (1·9%)	19 (3·6%)	0·11	27 (2·8%)
Person-years contributed (total)	1119·6	1067·3	0·001	2186·9
Completed study	326 (37·6%)	310 (45·5%)	0·002	636 (41·1%)

Vaccination with inactivated trivalent vaccine ≤1 year of child's 6-month anniversary.

Adjusted for gestational age in week.

Maternal and child characteristics were compared between countries using the Wilcoxon Rank-Sum test for nonparametric continuous variables or chi-square for categorical data.

ARI clinical presentation and swabbing

Six in ten (1015, 64·8%) children developed ARI (with or without fever) at least once (total 2493 episodes); 387 (24·7%) had ARI once, 237 (15·2%) twice, 152 (9·7%) three, 103 (6·6%) four, 76 (4·8%) five, 33 (2·1%) six, 10 (0·6%) seven, 12 (0·8%) eight, and 1 (0·1%) ten times during their first two years. Children with ARI frequently had cough (884, 87·6%), rhinorrhoea (854, 84·6%), fever (750, 74·3%) of a median of 3 day duration, hyporexia (216, 21·4%), and lassitude (76, 7·5%) (Table 2). Nearly one in ten (118, 7·5%) of the 1567 children were diagnosed with bronchopneumonia or bronchiolitis and 125 (8·0%) were referred to the hospital. Staff swabbed 589 children during 1009 (40·5%) of 2493 episodes of ARI, corresponding to all episodes of ARI with fever.

Table 2

Acute respiratory illnesses (ARI) presentation, and severe outcomes associated with ARI in the INFLUMIKA cohort, overall and by country.

	Panama (n = 871)	El Salvador (n = 696)	P-value	Overall (n = 1567)
ARI episodes
Any ARI (unique persons)	587 (67·4%)	428 (61·5%)	0·02	1015 (64·8%)
Any ARI during first year of life	530 (90·3%)	376 (87·9%)	0·22	906 (89·3%)
Total ARI episodes	1520	973	–	2493
Total ARI episodes tested	522 (34·3%)	487 (50·1%)	–	1009 (40·5%)
Symptoms
Cough	501 (85·3%)	383 (89·5%)	0·06	884 (87·1%)
Fever or feverishnessa	413 (70·4%)	337 (78·7%)	0·003	750 (73·9%)
Median days of fever (IQR)	3 (2–4)	3 (2-5)	0·14	3 (2–5)
Rhinorrhea	501 (85·3%)	353 (82·5%)	0·22	854 (84·1%)
Hyporexia	160 (27·3%)	56 (13·1%)	<0·001	216 (21·3%)
Fatigue	49 (8·3%)	27 (6·3%)	0·11	76 (7·5%)
Chest pain	6 (1·0%)	20 (4·7%)	0·0003	26 (2·6%)
Headache	1 (0·2%)	3 (0·7%)	0·32	4 (0·4%)
Myalgias	1 (0·2%)	1 (0·2%)	0·99	2 (0·2%)
Prostration	7 (1·2%)	0 (0%)	0·02	7 (0·7%)
Irritability	201 (34·2%)	159 (37·2%)	0·34	360 (35·5%)

Subjective or measured fever ≥38 °C.

Laboratory results and estimated cases by aetiology

Of these 589 children swabbed (with ARI and fever), 210 (35·6%) tested positive for rhinoviruses, 152 (25·8%) for RSV, 99 (16·8%) for parainfluenzas 1–3, 72 for influenza (12·2%), 69 (11·7%) for HMPV, and 40 (6·8%) for adenoviruses at least once (Table 3). We estimated an additional 234·8 (95%CI 231·9–237·6) rhinovirus, 183·2 RSV (95%CI 155·0–211·5), 130·4 parainfluenza 1–3 (95%CI 99·3–161·6), 65·5 HMPV (95%CI 43·1–87·9), 60·9 adenovirus (95%CI 59·0–62·8), and 12·0 influenza (95%CI 0–33·4) episodes would have been detected through RT-PCR during the follow up period if all ARI had been tested. Using data from the national reference laboratory (NRL), we estimated 0·6 (95%CI 0·2–1·0) seasonal coronavirus and 0·6 (95%CI 0·2–1·0) bocavirus ARIs during 2018; the year the NRL started routinely testing for these viruses.

Table 3

Number of laboratory-confirmed acute respiratory illnesses, risk periods, and incidence rates by viral etiology among children enrolled in the INFLUMIKA cohort.

	Panama (n = 871)	El Salvador (n = 696)	Overall (n = 1567)
Laboratory-confirmed ARI
Total children swabbed	316	273	589
Influenza	42 (13·3%)	30 (11·0%)	72 (12·2%)
Respiratory syncytial virus	83 (26·3%)	69 (25·3%)	152 (25·8%)
Rhinovirus	112 (35·4%)	98 (35·9%)	210 (35·7%)
Human metapneumovirus	31 (9·8%)	38 (13·9%)	69 (11·7%)
Parainfluenza 1–3	52 (16·5%)	47 (17·2%)	99 (16·8%)
Adenoviruses	24 (7·6%)	16 (5·9%)	40 (6·8%)
Estimated virala ARI in untested children
Influenza	10·0 (0–28·6)	2·0 (0–12·6)	12·0 (0–33·4)
Respiratory syncytial virus	131·7 (106·6–156·7)	51·6 (38·5–64·6)	183·2 (155·0–211·5)
Rhinovirus	204·6 (201·9–207·3)	30·2 (29·2–31·1)	234·8 (231·9–237·6)
Human metapneumovirus	63·4 (41·5–85·4)	2·1 (0–6·5)	65·5 (43·1–87·9)
Parainfluenza 1–3	108·4 (80·5–136·3)	22·1 (8·3–35·8)	130·4 (99·3–161·6)
Adenoviruses	52·1 (50·4–53·8)	8·8 (8·0–9·5)	60·9 (59·0–62·8)
Total risk accruedb (years)
Influenza	39732 (69·2%)	33745 (61·1%)	73480 (65·2%)
Respiratory syncytial virus	47341 (82·4%)	35531 (64·4%)	82875 (73·6%)
Rhinovirus	57428 (100%)	55196 (100%)	112627 (100%)
Human metapneumovirus	38689 (67·4%)	37185 (67·4%)	75874 (67·4%)
Parainfluenza 1-3	53733 (93·6%)	30571 (55·4%)	84304 (74·9%)
Adenoviruses	52234 (91·0%)	50203 (91·0%)	102437 (91·0%)
Adjustedc incidence rates (per 100py)
Influenza	6·8 (4·4–9·2)	4·9 (3·3–6·6)	5·9 (4·4–7·5)
Respiratory syncytial virus	23·6 (20·8–26·3)	17·6 (15·7–19·6)	21·0 (19·3–22·8)
Rhinovirus	28·7 (28·4–28·9)	12·1 (12·0–12·2)	20·5 (20·4–20·7)
Human metapneumovirus	12·7 (9·7–15·6)	5·6 (5·0–6·2)	9·2 (7·7–10·8)
Parainfluenza 1–3	15·5 (12·8–18·2)	11·8 (9·4–14·1)	14·2 (12·2–16·1)
Adenoviruses	7·6 (7·4–7·7)	2·6 (2·5–2·6)	5·1 (5·0–5·2)

Seasonal coronaviruses and bocaviruses were not measured because the study was initiated before such primers and probes were commonly used at the National Reference Laboratories.

Person-time each child accrued during weeks when that virus was detectable by each country's national surveillance.

Adjusted for under-ascertainment among afebrile children who were not swabbed. In brief, we assumed untested children had a similar probability of testing positive for each virus as afebrile children of the same age-group and country, who were tested the same epidemic week through national surveillance, after correcting for differences in the probability of testing positive for each virus by age-group and by afebrile vs. febrile status. Confidence intervals for these attributions incorporated the variance in the proportions of 1. persons of any age positive for each virus among all persons tested that epidemic week, 2. children of a given age-group positive for a virus among persons of any age positive for that virus, and 3. afebrile vs. febrile children positive for influenza and RSV.

Corrected incidence rate of ARI by viral aetiology

The incidence of ARI episodes among children aged <2 years was 114·6 episodes per 100py. The incidence of RSV-associated ARI (21·0, 95%CI 19·3–22·8, per 100py) and rhinoviruses-associated (20·5, 95%CI 20·4–20·7 per 100 py) among children aged <2 years were similar and higher than the incidence of parainfluenza 1–3-associated (14·2, 95%CI 12·2–16·1), HMPV-associated (9·2, 95%CI 7·7–10·8), influenza-associated (5·9, 95%CI 4·4–7·5), adenovirus-associated (5·1, 95%CI 5·0–5·2), bocavirus-associated (1·8, 95%CI 0·7–3·0), and seasonal coronavirus-associated (1·6, 95%CI 0·6–2·7) (Figure 1). Children aged <3 months had the highest rates of RSV-associated ARI (49·1, 95%CI 44·0–54·1 per 100py) followed by children aged 3–5 (25·1, 95%CI 20·1–30·0), 6–11 (17·6, 95%CI 13·2–21·9), and 12–23 months (11·9, 95%CI 10·8–12·9) (Figure 2a, Table 4). The corrected incidence of other viruses peaked between 3 and 11 months of age (e.g., influenza-associated Figure 2b, Table 4).

Figure 1

Acute respiratory illness incidence rate attributable to different viral etiologies, expressed per 100 person-years.

Adjusted for under-ascertainment among afebrile children who were not swabbed. In brief, we assumed untested children had a similar probability of testing positive for each virus as afebrile children of the same age-group and country, who were tested the same epidemic week through national surveillance, after correcting for differences in the probability of testing positive for each virus by age-group and by afebrile vs. febrile status. Confidence intervals for these attributions incorporated the variance in the proportions of 1. persons of any age positive for each virus among all persons tested that epidemic week, 2. children of a given age-group positive for a virus among persons of any age positive for that virus, and 3. afebrile vs. febrile children positive for influenza and RSV.

Figure 2

(a) Acute respiratory illness incidence rate attributable to respiratory syncytial viruses by age group, expressed per 100 person-years (b) Acute respiratory illness incidence rate attributable to influenza viruses by age group, expressed per 100 person-years.

Adjusted for under-ascertainment among afebrile children who were not swabbed. In brief, we assumed untested children had a similar probability of testing positive for each virus as afebrile children of the same age-group and country, who were tested the same epidemic week through national surveillance, after correcting for differences in the probability of testing positive for each virus by age-group and by afebrile vs. febrile status. Confidence intervals for these attributions incorporated the variance in the proportions of 1. persons of any age positive for each virus among all persons tested that epidemic week, 2. children of a given age-group positive for a virus among persons of any age positive for that virus, and 3. afebrile vs. febrile children positive for influenza and RSV.

Table 4

Incidence of laboratory-confirmed acute respiratory illnesses by viral etiology, stratified by age group and country, among children enrolled in the INFLUMIKA cohort.

	Panama (n = 871)	El Salvador (n = 696)	Overall (n = 1567)
RSV-associated ARI
<3 months	47·0 45·6–48·4)	53·0 (49·4–56·6)	49·1 (44·0–54·1)
3–5 months	24·5 (23·5–25·4)	26·2 (24·6–27·7)	25·1 (20·1–30·0)
6–11 months	24·3 (23·5–25·1)	8·4 (7·9–8·8)	17·6 (13·2–21·9)
12–23 months	11·3 (10·9–11·7)	12·5 (12·0–12·9)	11·9 (10·8–12·9)
Rhinovirus-associated ARI
<3 months	31·4 (30·7–32·0)	8·6 (8·4–8·7)	20·9 (14·4–15·1)
3–5 months	28·6 (28·0–29·3)	13·6 (13·4–13·7)	21·4 (21·1–21·8)
6–11 months	30·9 (30·4–31·5)	14·7 (14·5–15·0)	23·0 (22·7–23·2)
12–23 months	25·6 (25·2–26·0)	10·7 (10·5–10·8)	18·0 (17·8–18·2)
Parainfluenza 1–3-associated ARI
<3 months	17·0 (12·9–21·2)	8·4 (7·9–8·9)	14·0 (10·3–17·7)
3–5 months	17·1 (16·5–17·7)	15·6 (14·9–16·2)	16·6 (11·4–21·7)
6–11 months	17·5 (17·0–17·9)	19·1 (17·0–21·2)	18·0 (13·1–22·9)
12–23 months	12·7 (12·3–13·0)	6·8 (5·8–7·8)	10·4 (8·4–12·4)
HMPV-associated ARI
<3 months	11·6 (11·2–12·1)	7·6 (7·3–7·8)	9·8 (6·4–13·1)
3–5 months	18·2 (17·4–18·9)	8·9 (8·6–9·1)	13·7 (9·8–17·6)
6–11 months	14·1 (13·4–14·8)	8·0 (7·9–8·2)	11·1 (7·1–15·2)
12–23 months	9·7 (9·4–10·0)	2·0 (1·9–2·0)	5·8 (4·1–7·4)
Influenza-associated ARI
<3 months	3·5 (3·3–3·7)	2·2 (2·0–2·3)	2·9 (1·6–4·2)
3–5 months	6·6 (6·1–7·1)	6·5 (6·3–6·8)	6·6 (4·7–8·4)
6–11 months	9·8 (8·4–11·1)	3·8 (3·6–3·9)	7·0 (4·8–9·3)
12–23 months	6·3 (6·1–6·4)	6·3 (6·2–6·3)	6·3 (2·5–10·0)
Adenoviruses-associated ARI
<3 months	3·7 (3·5–4·0)	0·3 (0·3–0·4)	2·2 (2·0–2·3)
3–5 months	8·8 (8·4–9·2)	2·5 (2·4–2·6)	5·8 (5·6–6·0)
6–11 months	8·4 (8·0–8·7)	2·8 (2·6–2·9)	5·6 (5·4–5·8)
12–23 months	8·0 (7·7–8·3)	3·2 (3·1–3·3)	5·6 (5·4–5·7)

Adjusted for under-ascertainment among afebrile children who were not swabbed. In brief, we assumed untested children had a similar probability of testing positive for each virus as afebrile children of the same age-group and country, who were tested the same epidemic week through national surveillance, after correcting for differences in the probability of testing positive for each virus by age-group and by afebrile vs. febrile status. Confidence intervals for these attributions incorporated the variance in the proportions of 1. persons of any age positive for each virus among all persons tested that epidemic week, 2. children of a given age-group positive for a virus among persons of any age positive for that virus, and 3. afebrile vs. febrile children positive for influenza and RSV.

Corrected incidence rate of ARI hospitalization by viral aetiology

One in ten children aged <2 years with RSV-associated ARI were referred to the hospital (2·5 referrals per 100py, 95%CI 2·1–2·8 per 100py). The incidence of hospital admission attributed to an RSV-associated ARI was highest for children aged <3 months (10·1, 95%CI 8·7–11·5 per 100py), followed by children aged 3–5 months (2·6, 95%CI 1·5–3·8 per 100py), 6–11 months (1·7, 95%CI 1·3–2·0 per 100py), and 12–23 months (0·3, 95%CI 0–0·7 per 100py) (Figure 3). The incidence of hospital referrals among children aged <2 years was lower than for RSV-associated ARI but remained notable for rhinovirus-associated (0·56, 95%CI 0·54–0·58 per 100py), parainfluenza 1–3-associated (0·8, 95%CI 0·4–1·2 per 100py), HMPV-associated (0·4, 95%CI 0·1–0·7 per 100py), adenovirus-associated (0·15, 95%CI 0·14–0·16 per 100py), influenza-associated (0·1, 95%CI 0–0·7 per 100py), bocavirus-associated (0·1, 95%CI 0–0·4 per 100py), and seasonal coronavirus-associated (0·1, 95%CI 0–0·4 per 100py).

Figure 3

Hospitalization admission incidence rate attributable to RSV-associated ARI per 100 person-years by age group.

Adjusted for under-ascertainment among afebrile children who were not swabbed. In brief, we assumed untested children had a similar probability of testing positive for each virus as afebrile children of the same age-group and country, who were tested the same epidemic week through national surveillance, after correcting for differences in the probability of testing positive for each virus by age-group and by afebrile vs. febrile status. Confidence intervals for these attributions incorporated the variance in the proportions of 1. persons of any age positive for each virus among all persons tested that epidemic week, 2. children of a given age-group positive for a virus among persons of any age positive for that virus, and 3. afebrile vs. febrile children positive for influenza and RSV.

Risk of ARI by demographic characteristic

Children aged <3 months born to indigenous mothers were more likely to be diagnosed with RSV ARI (HR 9·4, 95%CI 1·3–68·8, p = 0·03) compared with those born to non-indigenous mothers, adjusted for country of residence, maternal age at enrolment, education, and the child's sex, gestational age, and breastfeeding exclusivity. Children aged <3 months living in households with four or more family members were 5-times more likely to develop RSV ARI (HR 5·1, 95%CI 1·6–16·3, p = 0·01) compared with those living in households with three or fewer members, adjusting for the aforementioned covariates and maternal race/ethnicity. The risk of influenza ARI was higher among children aged <3 months born to mothers with pre-existing health conditions compared with those with none (HR 7·6, 95%CI 1·2–48·2, p = 0·03), independent of country of residence, maternal age at enrolment, race/ethnicity, education, receipt of influenza vaccine ≤1 year of child's 6-month anniversary or prior natural infection, and the child's sex, gestational age, and breastfeeding exclusivity. Children aged 3–6 months who were born premature (gestational age <37 weeks) were more likely to be diagnosed with influenza ARI (HR 6·4, 95%CI 1·1–36·3, p = 0·04), adjusting for country, maternal age at enrolment, race/ethnicity, education, receipt of influenza vaccine or natural infection, and the child's sex and breastfeeding exclusivity.

ARI complications by demographic characteristic

The incidence of bronchopneumonia or bronchiolitis was 5·7 episodes per 100py and of referral to hospital care for an ARI was 6·1 referrals per 100py. Approximately two in three (64·3%) bronchopneumonia and bronchiolitis diagnoses or hospital referrals occurred on the first day of symptom onset. Although boys and girls had a similar risk of ARI, girls aged <3 months were 70% less likely to be diagnosed with bronchopneumonia or bronchiolitis or to be referred to the hospital because of their ARI compared with boys (hazard ratio [HR] 0·3, 95%CI 0·1–0·7, p = 0·003), adjusting for country of residence, maternal age at enrolment, race/ethnicity, education, and the child's sex, gestational age, and breastfeeding exclusivity. There was no difference in the risk of these severe ARI outcomes by sex among older children aged 3–23 months. Although not statistically significant, children breastfed or whose mothers planned to breast feed to at least 6 months of age had a 55% lower risk of severe outcomes (HR 0·45, 95%CI 0·16–1·2, p = 0·12) compared with those not breastfed to 6 months, independent of country, maternal age at enrolment, race/ethnicity, education, and the child's sex and gestational age.

Influenza vaccination

Overall, few children (217, 13·8%) were fully vaccinated against influenza during the two-year follow-up (191 [21·9%] in Panama, and 26 [3·7%] in El Salvador) and among vaccine-eligible children, 100 (46·1%) of influenza vaccinations occurred after the peak of influenza activity each year. Most mothers (56·9%; 68·8% in Panama and 42·1% in El Salvador), however, had received influenza vaccines within ≤1 year of child's 6-month anniversary. Children with any influenza vaccination between the ages of 6 to 24 months had a lower incidence of influenza than children who were never vaccinated against influenza during the follow-up period (58·9 episodes per 1000py vs. 74·2 episodes per 1000py, respectively). Of the 217 fully vaccinated children, 61 (28·1%) were born to indigenous mothers compared to 155 (71·4%) born to mestizo, black, or white mothers.

Discussion

Respiratory virus infections were common among children in Panama and El Salvador who often had multiple acute respiratory illnesses during their first two years of life. While rhinoviruses were the most frequently detected respiratory viruses, RSV infection caused the greatest number of ARI episodes and hospitalization referrals. RSV incidence was greatest among the youngest children, when ARI are most likely to progress to lower respiratory tract illnesses and merit supportive hospital care. These findings are consistent with results from a prospective birth cohort in Nicaragua, where children <3 months had the highest rate of severe RSV-associated ALRI. There are currently no licensed therapeutics or vaccines to protect against RSV, but there are several in development including an mRNA vaccine, similar to that used for SARS-CoV-2, with demonstrated immunogenicity in phase 1 clinical trials. Panama and El Salvador have access to RSV monoclonal antibodies, which can be used as prophylaxis for infants born premature and others at high risk of RSV complications. Current monoclonal antibodies, however, are costly and infrequently used in LMICs. The development of new vaccines, monoclonal antibodies, and therapeutics for RSV might fundamentally change the value proposition for the mitigation of RSV and respiratory virus illnesses among infants worldwide.

Unlike RSV, influenza and other respiratory viruses had higher incidence of ARI at older age groups, when we observed the need for hospital care is also lower. The incidence of laboratory-confirmed influenza ARI, for example, increased from birth to 5 months of age, peaked at 6–11 months, and declined steadily thereafter. It is possible that the rise in influenza incidence at approximately 6 months of age coincided with a nadir in passively conferred maternal antibodies against influenza, and when children were too young to complete the two dose influenza vaccination series. It is also not surprising that cohort children aged ≥6 months who were immunized against influenza had a lower risk of influenza illness thereafter when compared to unimmunized children. Our findings underscore the value of annual influenza vaccination among caregivers of infants aged <6 months and of children aged ≥6 months of age.

Despite the well documented effectiveness of standard dose influenza vaccines to protect young children against influenza illnesses, few cohort children were fully vaccinated against influenza (<15%). As recommended by PAHO, the governments of Panama and El Salvador provide influenza vaccines free-of-charge to children aged 6 months to 5 years cared for in public sector clinics to mitigate their influenza attributed morbidity and mortality. Despite this forward-thinking policy, fewer cohort children were immunized against influenza than their pregnant mothers. Subregional data suggest low vaccine coverage might be associated with low healthcare provider recommendations for influenza vaccines because of uncertainty about their benefit to individuals and inattention to their cumulative benefit to the public. Furthermore, among the low proportion of cohort children who were immunized, half were vaccinated after the start of their seasonal influenza epidemics, when they would have already been at risk of developing illnesses. To optimize the value of these investments in illness prevention, Panama and El Salvador authorities could deploy World Health Organization Influenza Vaccine Post-Introduction Evaluations to pinpoint processes that could be improved to strengthen vaccination coverage.

Risk of ARI and its complications occurred disproportionately throughout the cohort. Infants of indigenous mothers, for example, were more likely to have laboratory-confirmed RSV than other children. Although there have been attempts to relate human leukocyte antigen haplotype with pathogen diversity among indigenous communities,, other factors may be responsible for the increased hazard of severe respiratory illnesses including relative scarcity of running water and sanitation facilities in affected households, exposure to smoke from open flames, and greater distance from providers., Indeed, our finding that infants living in crowded households (4+ family members) were more likely to be diagnosed with an RSV ARI than infants in less crowded households, suggests that social determinants of disease may contribute to observed disparities.

One exception to the disproportionate risk may be the finding that infant girls aged <3 months were as likely as infant males to have ARI, but 70% less likely to be diagnosed with bronchopneumonia, bronchiolitis or be referred for hospital care. These findings did not persist into older age groups (i.e., 3–23 months). In our previous cohorts, we worried that similar findings might have been an artifact of subtle parental or provider preference towards observant vs. active care depending on the child's sex. Nevertheless, additional findings generated during the COVID-19 pandemic substantiate the hypothesis that oestradiol may be an important modulator of cytokine dysregulation during viral illness. It is therefore possible that feminizing hormones responsible for sex differentiation during foetal development and early childhood may also modulate respiratory virus illness cytokine dysregulation during the very first months of life among girls more than among boys.

In addition to RSV and influenza, birth cohort children also had substantial rates of rhinovirus, parainfluenza 1–3, HMPV, adenovirus, bocaviruses, and seasonal coronaviruses. While there are currently no vaccines or therapeutics for these seasonal viruses, several are in development. The diversity of respiratory viruses that place young children at risk merits investment in sustainable pan-respiratory surveillance to alert health authorities when respiratory virus epidemics occur and to trigger nonpharmaceutical intervention campaigns such as those used during the COVID-19 pandemic. While not all COVID-19 nonpharmaceutical interventions can be readily adopted outside the context of public health emergencies, others such as risk communication campaigns to promote frequent handwashing, avoidance of close contact with ill persons, self-isolation and quarantine, and cloth face covers while indoors may be cost-effective ways to empower communities to adopt such behaviours during respiratory virus epidemics.

Strengths and limitations

Our cohort findings are derived from active weekly surveillance during multiple years in two countries. We used nasopharyngeal swabs and sensitive molecular diagnostics to confirm viral aetiology using Collaborating Centre testing guidelines and previously used methods to correct for under-ascertainment. Furthermore, we sought to capture prenatal exposure by incorporating similarly collected data from the maternal cohort and present all findings using STROBE guidelines. Nevertheless, our cohort had important limitations. The number of children lost to follow-up may have resulted in attrition bias. For example, children born to indigenous mothers were less likely to remain in the study (Supplemental Table), potentially biasing our results away from these populations and limiting generalizability. We were unable to afford the testing of all ARI. Consequently, children with ARI, but without fever, were not swabbed for respiratory virus testing. Adjustments for under-ascertainment were made, though these estimates were not corrected for potential misclassification introduced by imperfect diagnostic test performance (e.g., sensitivity = 0.98, 95% CI: 0.91–1.00). We did not systematically differentiate referrals to the hospital for evaluation or observation versus admissions from the outpatient clinic notes. Nevertheless, hospitalization rates were similar to those described by cohorts able to do so. We did not, however, test cohort children for seasonal coronaviruses and bocaviruses because the study was initiated before such primers and probes were commonly used at the NRLs.

Conclusions

Respiratory virus illnesses were common among children in Panama and El Salvador. One in four children aged <2 years developed RSV-associated ARI, of those one in ten were referred to the hospital. RSV-associated ARI and hospitalization incidence was high, similar to estimates observed globally. Future evaluations would help to understand which children in Panama and El Salvador, who were at high risk of complications from RSV, received monoclonal antibodies and the value proposition of novel prophylactics and vaccines to protect children against RSV in LMICs. Children were also frequently ill with influenza. While fully vaccinated children aged ≥6 months were at lower risk of influenza than unvaccinated children, relatively few mothers and infants were fully vaccinated against influenza. Future program evaluations might explore opportunities to optimize influenza vaccination coverage among providers, pregnant women, and very young children to protect them from influenza illness complications. Lastly, children were frequently ill with other respiratory virus coronaviruses at an incidence similar to that estimated for SARS-CoV-2 during the COVID-19 pandemic. Whether or not SARS-CoV-2 becomes a seasonal coronavirus, it will be important to determine the value of non-pharmaceutical interventions, immunizations, and novel therapeutics to decrease this additional burden of acute viral respiratory illnesses and their potential sequela.,

Contributors

All authors contributed to and approved the final manuscript. LMD and EAB (the lead authors) affirm that the manuscript is in an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as originally planned have been explained. EAB, RG, AC, SCKD, VV, RD, RR, JA, DF, JMP conceptualized and designed the study, RG, AC, RD, AV, RR, AMR, JA, DF, JMP were involved in data collection, EAB, LMD, NO, VV conducted data analysis, and all authors were involved in data interpretation as well as review and editing of the manuscript. The contents are those of the authors and do not necessarily represent the official views of, nor an endorsement, by the Centres for Disease Control and Prevention/Health and Human Services, or the U.S. Government.

Data sharing statement

Individual participant data will not be made available. Relevant aggregate data are provided in Table 3. De-identified data will be made available upon request.

Declaration of interests

NO received travel support from QLife (Ecole Normale Superieure) to attend the Quantitative Viral Dynamics Across Scales Winter School Workshop. All other authors declare that they have no potential conflicts of interest to disclose.