INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The association between pneumonia in early life and the subsequent development of chronic respiratory symptoms and decreased levels of lung function has been a matter of considerable attention (1). Results of several retrospective studies suggest that both infants (2, 3, 5) and young children (1, 4) who are diagnosed as having pneumonia subsequently show lower levels of several measures of maximal expiratory flow. A possible explanation for these findings is that severe respiratory infections that occur at the time of rapid lung development may impair lung growth (1). This could then lead to increased risk for the development of chronic obstructive lung disease later in life (6).

Explanations other than injury caused by respiratory infection are possible for the association between lower respiratory tract illness (LRI) in early life and subsequently diminished lung function. We have reported that children with LRIs causing wheezing during the first year of life had significantly lower levels of lung function before any LRI occurred (9). Others have reported similar results (10). No attempt has been made to determine whether episodes of illness associated with the diagnosis of pneumonia were also associated with lower premorbid lower levels of lung function.

Most epidemiologic studies of the association of pneumonia in early life with subsequent respiratory function and morbidity have been based either on parental questionnaires administered years after the episode of pneumonia occurred (1) or on reports by home health visitors who obtained the relevant data from the parents at different times during the subjects' early lives (2). In none of these studies was the diagnosis of pneumonia based on radiologic evidence, despite the frequent difficulty in distinguishing pneumonia clinically from other LRIs occurring early in life (11). In these circumstances, a diagnosis of pneumonia that is not based on radiologic ascertainment may not be reliable, because signs and symptoms usually associated with pneumonia may overlap those present in subjects who have other forms of LRI in this age group. Only a few hospital-based follow-up studies, involving small numbers of children with LRI, are available, but in none of these was pneumonia studied separately from other LRIs (14).

In the present study we assessed respiratory symptoms, asthma diagnosis, and lung function at ages 6 and 11 yr in children with and without a radiologic diagnosis of pneumonia as ascertained directly by a physician during the first 3 yr of life. Infant pulmonary function tests were also performed before the development of any LRI in a subgroup of these children.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The children included in the study were part of a birth cohort of 1,246 children enrolled from 1980 to 1984 in the Tucson Children's Respiratory Study, a large longitudinal study of respiratory illnesses in children (18). Detailed information about the enrollment process and study design have been published elsewhere (18, 19). Birth weight, level of maternal education, maternal smoking, and data on family history of asthma and allergies were obtained from questionnaires administered to parents shortly after their child's birth.

The study was approved by the Human subjects Committee of The University of Arizona, and informed consent was obtained from the subjects' parents.

LRIs were diagnosed by the subjects' pediatricians at the time of the subjects' acute illness during the first 3 yr of life (19). Parents were requested to take their children to their pediatricians whenever the children had any of the following symptoms: deep or "wet" chest cough, wheezing, hoarseness, stridor, barking cough, or shortness of breath. The pediatricians obtained a detailed history at the time of such illnesses, and recorded all relevant signs and symptoms (including wheezing on auscultation) on a study form.

Of the original 1,246 children enrolled in the study, 888 were followed for LRIs by the original study pediatricians for the entire first 3 yr of life. A comparison between children followed for LRIs and those who dropped out of follow-up has been reported elsewhere (20). Briefly, children not followed were of lower socioeconomic status and had mothers with a higher prevalence of maternal smoking (20).

Children were classified as having had pneumonia if the pediatrician specified this diagnosis on the study form and if there was radiographic evidence compatible with this diagnosis. Radiographic criteria considered compatible with the diagnosis of pneumonia were the presence of infiltrates and/or a radiologist's diagnosis of bronchopneumonia or pneumonitis. Chest radiograms were not required in the study, and were ordered for clinical reasons only. Children who had more than one LRI during the first 3 yr of life and had a diagnosis of pneumonia and radiologic evidence of pneumonia for one or more of these LRIs were included in the pneumonia group. Children with a diagnosis of pneumonia but without radiologic evidence compatible with this diagnosis were included in a separate group (LRI/no pneumonia). Children who had other LRIs but did not have either a diagnosis of pneumonia or radiologic evidence of pneumonia were also included in the LRI/no pneumonia group. Only the first episode of pneumonia or the first LRI (in the case of the LRI/no pneumonia group) was used to describe LRI characteristics in this study.

At the time of a subject's LRI, the study nurses who worked in the pediatrician's office obtained additional historical information and nasopharyngeal swabs for isolation of pathogenic agents, including respiratory viruses (respiratory syncytial virus [RSV], parainfluenza virus, adenovirus, influenza virus, and others), mycoplasma, and chlamydia. Culture and immunofluorescence techniques used to identify pathogenic agents have been described in detail elsewhere (19). Not all children who developed LRIs had microbiologic tests, owing to lack of nurse or laboratory availability at the time of the LRI.

Asthma and Wheezing Data

Questionnaires were completed when the subjects were aged 6 and 11 yr (mean ± SD: 6.3 ± 0.9 yr and 10.9 ± 0.6 yr, respectively). Parents were asked whether they had ever been told by a doctor that their child had asthma ("MD asthma") and how many asthma attacks or episodes of asthma the child had had during the previous year. A child was considered to have "current MD asthma" if he/she had a diagnosis of asthma and one or more asthma episodes in the previous year. Parents were also asked whether the child's chest had sounded wheezy or whistling during the past year, and how often this had occurred. Children were considered to have "current infrequent wheezing" if they had had from one to three episodes of wheezing during the previous year. Children were considered to have "current frequent wheezing" if they had had more than three episodes of wheezing during the previous year.

Markers of Atopy

Skin prick tests were performed with extracts of allergens common in the Tucson area. At age 6 yr, subjects were tested with house dust, Bermuda grass, olive, careless weed, Alternaria alternata, mesquite, and mulberry, and at age 11 yr Dermatophagoides farinae and cat dander were added (allergens provided by Hollister-Stier Laboratories, Everett, WA). Two perpendicular diameters of all wheals were measured at 20 min. Skin test-positive subjects were defined as those who had at least one positive reaction (wheal size measuring 3 mm or more after substraction of the control value).

Blood for serum IgE analysis was obtained from subjects at a median age of 9.3 mo (the "9 mo" sample), and again at ages 6 and 11 yr. Total serum IgE levels were measured with a paper radioimmunosorbent test (Pharmacia, Diagnostics, Piscataway, NJ). All samples were assayed in duplicate, and results were expressed as international units per milliliter.

Pulmonary Function Tests

Pulmonary function testing was done on 176 children during the first year of life. A detailed description of the selection criteria and the medical and demographic characteristics of these infants, as compared with those who were not tested, has been reported (9); prevalence of a family history of asthma or allergies did not differ significantly for the infants who underwent pulmonary function testing and those who were not tested. Partial expiratory flow-volume curves were obtained by the chest compression technique (20, 21). Of the 176 children initially tested, 133 were tested before any LRI occurred and were included in the study. The subjects' age (mean ± SD) at the time of testing was 2.4 ± 2.0 mo.

Partial expiratory flow-volume curves were also obtained at the time of the 6-yr survey (age: 6.4 yr ± 0.5 yr [mean ± SD]). Voluntary maneuvers were used as described originally by Taussig (22). Maximal expiratory flow at FRC (maxFRC) was calculated as described previously (20) both for tests performed during the first year of life and for those obtained at age 6 yr.

At the time of the 11-yr survey (age 10.8 yr ± 0.5 yr), FVC, FEV₁, and FEF_25-75 were measured with standard spirometry. Baseline values were first obtained, and the best curve was chosen according to predefined criteria (23, 24). Subsequently, two puffs of albuterol (180 µg) were administered with a metered dose inhaler having an Aerochamber spacing device (Monaghan Medical, Plattsburgh, NY). Postbronchodilator values were obtained 15 min after the albuterol inhalation.

Statistical Analysis

Total serum IgE values were log-normally distributed, and results were expressed as geometric mean and 95% confidence intervals (CIs).

Values for maxFRC were logarithmically transformed and adjusted for length or height; results were standardized to the children's average length (57.4 cm) before the age of 1 yr or to their average height (110.3 cm) at the age of 6 yr. Values for FVC, FEV₁, and FEF_25-75 obtained at age 11 yr were also logarithmically transformed and adjusted for height. Results were standardized to the children's average height (143.2 cm for boys and 144.9 cm for girls) at the age of 11 yr. Results of transformed and adjusted maxFRC, FVC, FEV₁, and FEF_25-75 were expressed as geometric means and 95% CIs.

Analysis of variance (ANOVA) and Duncan's multiple-comparison test were used to compare means, and the chi-square test was used to compare proportions (25). The 95% CIs for odds ratios (ORs) were calculated with standard algorithms (26). Multiple regression models were used to determine whether pneumonia and LRI/no pneumonia remained associated with the dependent variables (pulmonary function test [PFT] parameters at ages 6 and 11 yr) after controlling for gender, current wheezing at age 6 yr, maternal prenatal smoking, maternal education, and parental ethnicity; for lung function at age 11 yr, current wheezing at age 11 yr was also considered. Results of the multiple regression analysis were expressed as differences between the no LRI group (assigned a value of zero) and the two LRI groups. Statistical significance was defined by a two-sided  level of 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Of the 888 eligible children who were followed for LRIs during the first 3 yr of life, 613 (69.0%) had at least one PFT measurement. There were no significant differences between these 613 children with PFTs and 275 children without PFTs with respect to prevalence of asthma or wheezing at age 6 or 11 yr (Table 1). However, the children with PFTs belonged to families of a significantly higher socioeconomic status and with a lower prevalence of maternal smoking (Table 1). Of the 888 eligible children, 66 (7.4%) had pneumonia, 397 (44.7%) had LRI/no pneumonia, and 425 (47.9%) had no LRI. As with the prevalence of asthma or wheezing, the cumulative incidence of pneumonia and of LRI/no pneumonia was also similar in children with and without PFT measurements (Table 1). Among children with LRI/no pneumonia, a chest radiogram that showed no evidence of pneumonia (as described in METHODS) was obtained from 37 children. We found no differences in any of the parameters studied between this group and children classified as having LRI/no pneumonia but for whom no chest film was requested by the pediatrician. For this reason, these two groups of subjects were combined in the LRI/no pneumonia category.

Similar proportions of male and female children were found among subjects with pneumonia (54.5% males) and with LRI/no pneumonia (51.1% males). The subjects' ages (mean ± SD) at the first episode of pneumonia (pneumonia group) and at the first LRI (LRI/no pneumonia group) were 1.00 ± 0.73 yr and 1.00 ± 0.74 yr, respectively. Reports of wheezing (by history or physical examination) during the first episode of pneumonia or during the first LRI were very similar in the two groups (65.4% versus 69.8%, respectively). A report of tachypnea during the first LRI was significantly more prevalent in the pneumonia group than in the LRI/no pneumonia group (62.0% versus 34.9%, respectively; OR: 3.0; 95% CI: 1.7 to 5.6; p < 0.001). Among 66 children with pneumonia, 42 had radiologic evidence of infiltrates, 22 had bronchopneumonia, and two had pneumonitis. The proportion of children from whom RSV, parainfluenza, or other agents were isolated for the first episode of pneumonia (pneumonia group) or for the first LRI (LRI/no pneumonia group) was not significantly different for the two groups (Table 2).

Among 66 children with pneumonia, 22 had only one episode of LRI, 43 had two episodes (one of which was diagnosed as pneumonia and the other as either pneumonia [n = 1] or other LRI/no pneumonia [n = 42]). One subject had three episodes of LRI (two diagnosed as pneumonia and the other as LRI/no pneumonia). Only two children with pneumonia required hospitalization.

Current MD Asthma and Current Wheezing at Ages 6 and 11 yr

Current MD asthma was significantly more likely at ages 6 and 11 yr among children who had pneumonia or LRI/no pneumonia in early life than in children with no LRI (Table 3). Current infrequent wheezing (from one to three episodes during the previous year) and current frequent wheezing (more than three episodes during the previous year) at age 6 yr were also more likely in the two LRI groups than in the reference group (Table 4). Conversely, children with either pneumonia or LRI/no pneumonia in early life were not at increased risk for having infrequent wheezing at age 11 yr, although those with pneumonia were significantly more likely to have frequent wheezing at age 11 yr than was the reference group (Table 4).

Markers of Atopy

There was no significant difference in prevalence of at least one positive skin test at age 6 yr among the study groups (34.7%, 40.1%, and 41.7% for the pneumonia, LRI/no pneumonia, and reference groups, respectively; p = 0.65). Similarly, prevalence of at least one positive skin test at age 11 yr did not differ among the groups (53.3%, 60.8%, and 63.7%, respectively; p = 0.38). Also, no significant association was found between the different LRI groups and any individual allergen assessed at age 6 and 11 yr (data not shown). The geometric mean (95%CI) for total serum IgE levels at 9 mo was 4.2 IU/ml (3.1 to 5.7) for the pneumonia, 3.8 IU/ml (3.3 to 4.4) for the LRI/ no pneumonia, and 4.1 IU/ml (3.5 to 4.7) for the reference group (p = 0.74). Likewise, there was no significant difference in total serum IgE levels among groups either at age 6 yr (p = 0.57) or at age 11 yr (p = 0.55).

PFTs

Length-adjusted mean maxFRC values before any LRI (at age < 1 yr) were lower in children with pneumonia and with LRI/no pneumonia than in children with no LRI, although these results only reached statistical significance for the LRI/ no pneumonia group (Table 5). Significantly lower mean height- adjusted maxFRC values at age 6 yr were observed both in children in the pneumonia group and in those in the LRI/no pneumonia group than in children in the no LRI group (Table 5). At age 11 yr, mean unadjusted values for FEV₁ and FEF_25-75 before albuterol were significantly lower in children who had pneumonia and in children who had LRI/no pneumonia than in children who had no LRI. Children in the pneumonia group also had a significantly lower mean FEF_25-75 value before albuterol than children in the LRI/no pneumonia group. Unadjusted mean values for FVC before albuterol were not significantly different among groups. After use of albuterol, mean unadjusted values for FVC, FEV₁, and FEF_25-75 were not significantly different among groups (Table 5).

Differences in lung function between groups at age 6 and at age 11 yr remained substantially unchanged after adjusting for possible confounders (Table 6). Although both children with a history of pneumonia and those with a history of LRI/no pneumonia had significant deficits in baseline maxFRC at age 6 yr and in baseline FEV₁ and FEV_25-75 at age 11 yr, these deficits were considerably larger for subjects with a history of pneumonia. However, after administration of albuterol, the deficits observed for both groups at age 11 yr were markedly reduced, and this was especially true for the group with a history of pneumonia. As a consequence, no significant deficits in postbronchodilator lung function were observed for either children with a history of pneumonia or for those with a history of LRI/no pneumonia.

None of the associations between type of LRI and the different outcomes described here showed significant heterogeneity after stratifying by the etiology of the LRI (data not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

To our knowledge, this is the first study in which sequelae in later childhood of radiologically confirmed pneumonia in the first 3 yr of life were assessed in a large population sample of children enrolled at birth. The great majority of these children were seen by their pediatricians as outpatients, and did not require hospitalization.

The cumulative incidence of pneumonia during the first 3 yr of life in this study (7.4%) was similar to that described by Murphy and associates (27) for a general pediatric practice in North Carolina. The etiologic agents for episodes of pneumonia were very similar to those of other LRIs without a diagnosis of pneumonia. Almost 60% of all such episodes were linked to viral isolates, and the main agent detected was RSV, which accounted for almost one-half of all isolated agents. This finding is in agreement with several other epidemiologic studies in which the etiologic agent associated with pneumonia was assessed during the first 3 yr of life (28, 29).

The association between pneumonia in early life and the subsequent development of asthma has not been extensively explored. The main findings of this study indicated that children with a diagnosis of pneumonia during the first 3 yr of life were three times more likely to have current physician-diagnosed asthma at ages 6 and 11 yr than were children who had no LRIs during the same period. Similarly, children with a diagnosis of pneumonia were almost four times as likely to have frequent wheezing at age 6 yr and twice as likely to have frequent wheezing at age 11 yr than were children who had no LRIs during the first 3 yr of life. Conversely, children with a history of LRI/no pneumonia during the first 3 yr of life were twice as likely to have current asthma and frequent wheezing at age 6 yr but were only marginally more likely to have current asthma (OR = 1.6) and frequent wheezing (OR = 1.4) at age 11 yr than were children who had no LRIs during the first 3 yr of life. More than 65% of all children who had pneumonia were also found to have wheezing at the time of their first episode of pneumonia, and the great majority of these episodes were due to RSV. These results suggest that the factors that determine the association between pneumonia early in life and asthma later in life are similar to those that determine the association between bronchiolitis early in life and asthma later in life. It is thus plausible to speculate that the same disease mechanisms that determine LRIs in children with pneumonia are also present in those who develop LRI/no pneumonia. As in other studies (30), the diagnosis of pneumonia in this cohort was related to the degree of tachypnea, considered an index of severity of LRI. Children who develop roentgenographic evidence of pneumonia during LRIs early in life may be at the extreme of severity within the same basic mechanism of disease that causes most LRIs in this age group.

We have previously shown (9) that a major determinant of LRIs in early life is the level of lung function prior to the occurrence of LRI. Children with a history of pneumonia might then be expected to have had lower levels of lung function before any pneumonia occurred, if the hypothesis is correct that pneumonia early in life represents a more severe form of LRI. Unfortunately, only eight such children had their lung function measured during the first months of life in our study. These children tended to have lower mean levels of maxFRC, an index derived from partial flow-volume curves obtained through the chest rapid compression technique (21). The levels of lung function in these eight children were similar to those observed in children who subsequently developed LRI/no pneumonia, but difference did not reach statistical significance when compared with lung function in the no LRI group, probably because of the small number of subjects involved. When PFTs were done later in childhood in the same group of children with a history of pneumonia during the first 3 yr of life, significant reductions were observed for maxFRC at age 6 yr and for both FEV₁ and FEF_25-75 at age 11 yr. These reductions were particularly noticeable for FEF_25-75 at age 11 yr, with children who had a history of pneumonia having significantly lower levels for this index than those observed for children with a history of LRI/no pneumonia during the first 3 yr of life (Table 5). Interestingly, no significant differences between groups in either FEV₁, or FEF_25-75 were observed after administration of a bronchodilator (albuterol). These results were independent of known confounders including wheezing at both ages 6 and 11 yr.

There are two possible explanations for these findings. It is plausible that the lower levels of lung function observed both in children with pneumonia and those with LRI/no pneumonia at ages 6 and 11 yr were the consequence of premorbid alterations in airway structure or in airway smooth-muscle tone. This explanation is supported by the evidence of diminished lung function present before any LRI occurred both in children with a subsequent history of pneumonia and in those with a history of LRI/no pneumonia. That children with a history of pneumonia had larger deficits in maxFRC and in FEF_25-75 at ages 6 and 11 yr, respectively, than those observed among children with a history of LRI/no pneumonia may offer clues about why the diagnosis of pneumonia was applied to one group and not to the other. Since both maxFRC and FEF_25-75 are believed to be indices of small-airway function (34, 35), it is possible that preexisting lung function is even more diminished in children who will subsequently develop pneumonia than in those who will develop LRI/no pneumonia. As a consequence, subjects with pneumonia could more readily develop airway occlusion and foci of microatelectasis or localized pneumonitis. In addition, the greater degree of airway obstruction in these children would be associated with more severe symptoms, and these two factors would suggest the diagnosis of pneumonia to the pediatrician.

A second explanation for the previously described findings relating to lung function is also possible and is not incompatible with the possibilities previously described. LRIs associated with viral infections may produce alterations in lung structure and in lung function, and these alterations may in part be responsible for the lower levels of lung function observed subsequently in these children. More severe infections (associated with the diagnosis of pneumonia) would result in more severe deficits in lung function. Animal models suggest that viral infections in immature animals may be associated with significant and persistent alterations in nonadrenergic, noncholinergic regulation of airway tone (36). Whether these alterations also occur in humans with viral LRIs early in life remains to be determined.

Results of spirometric tests performed at age 11 yr in our study subjects after the administration of a bronchodilator (albuterol) provide important new information about the nature of the putative alteration in airway function present in children with a history of LRI. Both subjects with a history of pneumonia and those with a history of LRI/no pneumonia showed significant positive responses to bronchodilator challenge. As a result, postbronchodilator levels of lung function were slightly but not significantly lower in children with a history of pneumonia or with a history of LRI/no pneumonia than those observed in children with no LRI during the first 3 yr of life. These results strongly suggest that the lower levels of lung function present during the school years in subjects with a history of pneumonia or of LRI/no pneumonia are, to a significant extent, due to alterations in the regulation of airway tone. The observed deficits (particularly for FEF_25-75), were, however, not completely reversed by bronchodilator administration (see Tables 5 and 6). In addition, we have no information about reversibility of lung function deficits observed shortly after birth. Therefore, we cannot exclude the possibility that both anomalies in the regulation of airway tone and structural alterations of the lungs and airways may be present in subjects who will subsequently develop more severe respiratory tract illnesses.

In summary, our results suggest that pneumonia during the first 3 yr of life is part of a continuum of severity of viral LRIs in this age group. These pneumonias are associated with a significant risk of subsequent persistent wheezing and with significant deficits in lung function. These deficits can already be detected during the first months of life, before the development of any LRI. Elucidating the factors that determine diminished lung function early in life may be crucial to any strategy for preventing pneumonia and other LRIs in infancy and early childhood.