Residential crowding and severe respiratory syncytial virus disease among infants and young children: A systematic literature review

A systematic literature search was conducted for studies assessing the association of residential crowding with severe RSV disease in children younger than 5 years, including populations considered at high risk of developing severe RSV disease.

High-risk populations were defined as premature infants, children with chronic lung disease (CLD) of prematurity or hemodynamically significant congenital heart disease, or members of an ethnic group believed to be at increased risk of respiratory disease (i.e. Alaska native, Inuit). Other study populations were defined as mixed-risk and consisted predominantly of children from the general population not meeting the high-risk definition. Children in the mixed-risk population may have had varying levels of risk because of other factors that increase the risk of severe RSV disease, such as young age (≤1 year) [2]. Studies limited to populations with severe health conditions, such as cancer or organ transplantation were excluded.

The literature search included studies published and indexed in the PubMed (including MEDLINE) and EMBASE databases between January 1, 1985, and October 8, 2009 (see Additional file 1 for the search strategies used for each database). The PubMed search used the National Library of Medicine medical subject heading (MeSH) terms “respiratory syncytial virus infections” or “bronchiolitis, viral” or “respiratory tract infections” combined with the words “syncytial” or “RSV” or “acute” plus “lower” anywhere in the database record. To focus the search on studies evaluating risk factors, MeSH terms were limited to the subheadings “epidemiology,” “etiology,” and “complications.” To identify studies of residential crowding as a risk factor for severe RSV disease, the search used the MeSH term “crowding” or any of the following words describing crowding: “resident,” “residents,” “residential,” “bedroom,” “bedrooms,” “household,” “households,” “number of children,” “number of people,” “per room,” “> and people,” “sibling,” “siblings,” “sharing,” “share,” or “overcrowding.”

Studies that examined multiple risk factors for severe RSV disease were also included in the full-text review, whether or not residential crowding variables were mentioned in the abstract. This approach ensured that the present study avoided the bias caused by overlooking nonsignificant findings related to crowding (which may not be highlighted in a study’s abstract when a set of risk factors are examined). To identify these studies, titles or abstracts were searched for a respiratory disease–related term, as described above, and any of the text words describing analysis of multiple risk factors: “multivariate,” “multifactorial,” or “regression.” This broader search was limited to the population of interest by also requiring use of any of the words “child,” “children,” “infant,” or “infants.”

The titles and abstracts of articles published in English and identified in the database searches were reviewed for potential relevance (Figure 1). For relevant sources, the full articles were obtained and reviewed to determine whether residential crowding was assessed. Only studies in which the children had lower respiratory tract illness due to laboratory-confirmed RSV and were younger than 5 years were included. Any treatment setting for RSV was accepted; however, the majority of the studies were on populations hospitalized with an RSV infection. Few studies examined patients with RSV lower respiratory tract infection in an outpatient setting. In contrast to RSV hospitalization, outpatient RSV illness reflects mild disease. Each relevant article was read by several authors. Study details were extracted into tables by one author, and the content of the tables was verified by a second author not involved in the data extraction. The authors discussed each article to reach consensus regarding the study details. For each study, the following data were extracted: reference, publication year, country origin, study design, study population size and description, residential crowding definition and categories, disease outcome evaluated, results and significance of unadjusted analyses, adjusted analyses, and other statistical analyses. The principal summary measures were adjusted odds ratio (aOR) or adjusted relative risk (aRR) for multivariate analyses and odds ratio (OR) for bivariate analyses.

The risk of bias was assessed using the Cochrane risk of bias tool [13]. The assessment focused on likely sources of bias in observational studies. The reviewed studies were examined for risk of selective reporting bias and confounding bias that pertained to the association between residential crowding and RSV disease. Selective reporting bias can occur if outcomes for all studied risk factors are not reported. Not reporting outcomes of all variables could bias the reader’s interpretation of the outcomes. For example, reporting only outcomes for a measure of crowding that was significant but not reporting the negative outcomes of alternative definitions of crowding could overstate the importance of residential crowding. Alternatively, reporting only negative outcomes of a single crowding variable in multivariate analysis, but not the positive associations of numerous measures of crowding in unadjusted analysis, may obscure potential explanations for the negative finding. Confounding bias may have been introduced if a study did not control for the effects of other variables in multivariate analysis and may lead to either over- or underestimation of the impact of residential crowding on risk of severe RSV disease, depending on how correction for other variables affects the outcome with the crowding variable. A more detailed explanation regarding bias assessment is presented in Additional file 2.

Table 1 provides an overview of these 20 studies, including whether they reported a significant association between RSV disease and a residential crowding variable in the adjusted or unadjusted analysis. The studies were grouped by the risk status of the study populations: 9 studies were conducted in high–risk children, and 11 studies were conducted in populations with mixed risk. Most of the studies examined laboratory-confirmed RSV hospitalization; a few studies based their analyses on laboratory-confirmed RSV infection diagnosed or treated in the emergency department [14], outpatient clinic [9], combination of hospital and outpatient settings [15], or pediatrician’s office [16, 17]. All of the studies were observational. Most of the studies were large, having hundreds or in some cases thousands [18‐20] of subjects, although a couple of studies were small [21, 22]. The studied populations were geographically diverse.

Bias assessments for each study are presented in Table 1. Overall, there was a low risk of selective reporting or confounding across the majority of the 20 studies. Most of the studies carefully detailed the list of variables evaluated. Most studies also had a low risk of confounding bias because most reported an adjusted or multivariate analysis.

Among all the studies, crowding was measured in numerous ways. Some crowding variables assessed the number of residents per household, room, or square meter. Others measured the number or age of children or siblings in the home or how many persons shared the case child’s bed.

Most of the studies in high-risk populations found that residential crowding significantly increased the risk of severe RSV disease as measured by at least one crowding variable (Table 2). Adjusted outcomes for crowding variables ranged from aOR 1.45 (P = 0.035) [15] to aOR 2.85 (P < 0.000001) [26].

Several large studies analyzed the risk factors for severe RSV disease in high-risk populations. A case–control study in Alaska natives of children <3 years of age examined numerous and varied crowding-related factors among 204 cases and 338 controls [23]. Most of the crowding variables were significantly associated with increased risk of RSV hospitalization. Most of the residential crowding variables were significant for the infants younger than 6 months compared with infants 6 months or older [23].

Among the many large epidemiological studies of RSV hospitalization conducted in Spain were populations of premature infants born at gestational ages of ≤32 weeks [24, 25], 32 to 35 weeks [18], or 33 to 35 weeks [26]. In all of these studies, having one or more school-aged siblings was associated with a risk of RSV hospitalization in unadjusted [18] or adjusted [24‐26] analysis. The smaller case–control study by Figueras-Aloy et al. found an association when the crowding variable was having four or more household inhabitants, not including school-aged siblings or the study subject [26]. However, this variable was not significant in the larger 2-cohort study [18].

The Pediatric Investigators Collaborative Network on Infections in Canada (PICNIC) study, which also examined children born at 32–35 weeks gestation, found that when changes in risk status following discharge from the hospital were considered, infants in homes with five or more residents were significantly at risk of hospitalization due to RSV [19]. The presence of preschool-aged siblings also was significantly associated with RSV hospitalization in both unadjusted and adjusted analyses [19].

A study in the United Kingdom examined lower respiratory tract illness due to RSV that was either diagnosed in a general practitioner’s office or resulted in hospitalization [15]. This study found an association between the number of school-aged siblings and lower respiratory tract illness due to RSV in preterm infants born at gestational ages younger than 32 weeks.

All of the studies in mixed-risk populations found that at least one residential crowding variable was associated with severe RSV disease in multivariate analyses (Table 3) [20, 29‐34]. In these studies, adjusted outcomes for crowding variables ranged from aOR 1.23 (95% confidence interval 1.01–1.56) [20] to aOR 9.1 (P < 0.001) [32].

Two studies conducted in Africa examined multiple crowding-related risk factors for severe RSV disease [9, 32]. In a Kenyan study, the crowding variables that were associated with RSV-related pneumonia after adjusting for other variables were the number of children in the family and the number of siblings younger than 6 years [9]. The Gambian study found that the number of people living in the house and the number of children aged 3 to 5 years significantly increased the risk of RSV hospitalization after adjusting for other factors [32].

Older siblings and the related variable birth order were associated with RSV hospitalization in studies in Australia [29], Denmark [31], and Italy [30]. A second study in Denmark of children younger than 2 years found that the presence of older or younger siblings alone was not a risk factor for RSV hospitalization, but the presence of siblings with an age difference of less than 5 years from the case child was a risk factor [20].

Of the studies in mixed-risk populations that did not report adjusted outcomes, two examined risks in hospitalized children—one in American Samoa [21] and one in Italy [28]—and two evaluated RSV disease diagnosed in settings not limited to the hospital [14, 16]. In the studies of hospitalization, the Italian study found a significant risk with crowding, whereas the association was less clear in the study in American Samoa.

No associations of crowding variables were demonstrated in the two studies reporting only unadjusted outcomes for RSV disease diagnosed in the emergency department or hospital in a study in Portugal [14] or in pediatric outpatient clinics in Iraq [16]. However, a large US cohort study of healthy infants followed from birth to 12 months of age found that sharing a room with two or more persons increased the risk of RSV disease diagnosed in pediatricians’ offices even after adjusting for other factors [17].