Role of Primary Health Care in child hospitalization due to pneumonia: a case-control study 1

Objective: to evaluate the association of primary health care and other potential factors in relation to hospitalization due to pneumonia, among children aged under five years. Method: epidemiological study with a case-control, hospital-based design, which included 345 cases and 345 controls, matched according to gender, age and hospital. Data were collected using a pre-coded questionnaire and the Primary Care Assessment Tool, analyzed by means of multivariate logistic regression, following the assumptions of a hierarchical approach. Results: the protective factors were: family income >US$216.12 (OR=0.68), weight gain during pregnancy ≥10 kg (OR=0.68), quality of Primary Health Care (OR for scores >3.41=0.57; OR for scores >3.17 and ≤3.41=0.50), gastro-esophageal reflux (OR=0.55), overweight (OR=0.37) and birth interval ≥48 months (OR=0.28). The risk factors included: parity (2 childbirths: OR=4.60; ≥3 childbirths: OR=3.25), out-of-date vaccination (OR=2.81), undernutrition (OR=2.53), history of wheezing (≥3 episodes OR=2.37; 1 episode: OR=2.13), attendance at daycare center (OR=1.67), and use of medicines over the past month (OR=1.67). Conclusion: primary health care and its child health care practices, such as nutritional monitoring, immunization, care to prevalent illnesses, prenatal care and family planning need to be prioritized to avoid child hospitalization due to pneumonia.


Introduction
The importance of pneumonia to child morbidity and mortality has been widely addressed in the literature (1)(2) . The most recent world estimate for children aged under five years is 120 million new episodes per year, along with severe cases and high rates of hospitalizations (2) . In Brazil, 1.5 million new cases of community-acquired pneumonia (CAP) occur in this age group every year, and the disease ranks first on the list of hospitalization causes of these children in all regions of the country (3) .
Knowledge of the risk factors for hospitalization due to pneumonia makes it possible to identify priorities for preventing and managing this disease.
These factors are: gender, age, type of birth, birth weight, gestational age, birth order, breastfeeding, malnutrition, previous and current morbidities, immunization, use of medications, parent and socioeconomic characteristics, basic sanitation, indoor and outdoor air pollution, child care outside the home and access to health care services (1) .
Primary Health Care (PHC) can be considered as a factor that influences hospitalization due to pneumonia. The role of PHC has been investigated in relation to controlling pneumonia in regard to having access to health care services and clinical management of the disease (4) . PHC is the level of health care that is focused on early diagnosis, treatment and interventions, the purpose being to decrease exposure to risk factors and/or increase exposure to protective factors associated with pneumonia (3) . The inclusion of PHC as an explanatory factor of this disease should take all of the characteristics of its health actions into consideration, guided by the following attributes: first contact access, longitudinality, comprehensiveness, health care coordination, family orientation and community orientation (5) . It is important to take not only the care given during the acute episodes into account, because factors related to health care and health communication provided routinely for child, mother and family can be reflected in child health (6) .

Factors involved in child hospitalization due
to pneumonia can represent interventions by professionals working in PHC (such as immunization, nutritional counseling, preventive care at home); in addition, aspects related to family income and parents' education can influence access to PHC services which, in turn, can influence the risk of hospitalization due to the disease (1) . This understanding supports the construction of a hierarchical conceptual framework for child hospitalization due to pneumonia, which is based on methods with a hierarchical approach to evaluate the relationships between exposure and outcome, considering that the risk factors can be influenced by those located in preceding levels, either directly or indirectly (7) . From a hierarchical point of view, however, we did not find any studies that included PHC as an explanatory variable for hospitalization due to pneumonia.
Thus, this study aimed to evaluate the association of PHC and other potential factors in relation to hospitalization due to pneumonia, among children aged under five years .

Method
This is an analytical epidemiological study with a hospital-based case-control design. The study was developed at three public hospitals in the city of Ribeirão Preto, Brazil.
Brazil is a South American developing country with a unified free-of-charge public health care system.
The Southeast is home to 43% of Brazil's population and accounts for 56% of the Gross Domestic Product (GDP). Ribeirão Preto has the tenth largest GDP in the state of São Paulo. The weather is characterized by one mild, cool and dry and another hot and wet season, with moderately high temperatures (8) .
The study was developed at three hospitals within the framework of the Brazilian Unified Health System -SUS (Hospital A -state institution, national reference for clinical specialties, pediatric clinic with 58 beds; Hospital B -municipal institution, 18 pediatric beds; Hospital C -municipal institution, 20 pediatric beds).
The study was conducted only with children living in Ribeirão Preto, users of the primary health care network through the SUS.

Selection of Cases and Controls
With the aim of detecting OR ≥2.0 (two-tailed tests), with a statistical power of 80% and α=0.05, and considering 10% as the lowest probability of exposure among the controls, a total of 345 cases and 345 controls were selected. Cases were children aged under five years who had been hospitalized due to CAP, as confirmed through radiological examination, at the participating hospitals. Each case was arranged Pina JC, Moraes AS, Freitas ICM, Mello DF.

Minimizing Berkson's bias involved including incident
cases and selecting hospital controls with a variety of admission diagnoses, including those with diseases of the upper respiratory tract.
Living in the city for less than six months, having a recent history of liquid or foreign body aspiration (due to the diagnostic hypothesis of aspiration pneumonia), and being aged under two months were used as  (9) . The essential PHC score (mean of the essential attribute items) and the overall PHC score (mean of the essential and derived attribute items) were obtained (9) . The parents or legal guardians of the child answered all questions from the PCA Tool and those included in the precoded questionnaire. Therefore, all the explanatory variables of the study were collected during the interviews, except the anthropometric measurements, which were collected as described below.
For children aged under two years, weight was measured using an infant digital scale (precise to 5 grams) and height was measured using a horizontal stadiometer. For children aged older than two years, a portable digital scale (100-gram precision) and a vertical stadiometer were used. The techniques for anthropometric measurement and the cut-offs of the anthropometric indexes followed World Health

Organization (WHO) recommendations.
A conceptual hierarchical model ( Figure 1) was composed, based on the literature (7) , including the outcome and the explanatory variables, which guided the statistical techniques for data analysis.
Some study variables require further explanation regarding their definition and classification, such as:

Statistical Analysis
Cases and controls were characterized according to sociodemographic variables. The groups were compared using the chi-square likelihood ratio. During the analytical phase, crude and adjusted Odds Ratios (OR), with their respective 95% confidence intervals, were estimated using non-conditional multivariate logistic regression (10) , following the assumptions of hierarchical models.
Through univariate models, the inclusion criterion for being adjusted for the variables of the preceding levels (10) . The use of unconditional analysis was based on two assumptions: 1) the results were very close to those obtained in the conditional analysis; and 2) matching according to gender and age group does not characterize close matching (10) . The goodness of fit of the final model was evaluated based on the Hosmer-Lemeshow chisquare test (10) , while the sensitivity (se) and specificity

Results
The sociodemographic characteristics of the study population are presented in Table 1.
Pina JC, Moraes AS, Freitas ICM, Mello DF. There was a greater proportion of controls with a higher family income and parent education level, although these differences were not statistically significant. There were no significant differences either between the two groups regarding maternal age, marital status or whether the mother worked away from home.
There was a greater proportion of homes with six or more people living in the same home. Regarding the number of rooms in the home, however, there was no significant difference between the groups. In addition, factors related to family income, maternal obstetric history, child and mother nutritional status, morbidity and child health care were associated with hospitalization due to pneumonia in the population studied.

Discussion
This study serves to fill a gap in the literature, demonstrating the contribution of PHC to prevent child hospitalization due to pneumonia. Therefore, the study goes beyond the access received during the disease, an aspect other authors have already pointed out (4) . The association identified in relation to the essential PHC score signals the importance of PHC organization as a regular source of health care (5,9) , and the link over time and families (11) . In childcare, the systematic follow-up of vaccination, child nutrition, childhood illnesses and habits, such as the use of medication is carried out (1,11) , which contributes to improving integral child care.
One of the most important strategies for controlling pneumonia within PHC is vaccination (1,12) . In this study, situations in which out-of-date vaccination was used were seen more frequently among the cases, and these lead to a higher risk of hospitalization due to pneumonia.
Vaccines against the influenza virus (Haemophilus influenza type b) and against Streptococcus pneumoniae grant specific protection (1) . Also, adequate immunization is considered as a proxy for childcare. Strategies need to be implemented to vaccinate all children who visit health services for other interventions (13) .
In relation to nutritional variables, there was a higher risk of hospitalization due to pneumonia among children who were undernourished and a lower risk among overweight children. The effect of undernourishment on the occurrence and severity of pneumonia has already been reported in the literature (14) . Pneumonia, undernutrition and micronutrient deficiencies have common risk factors, such as not exclusively breastfeeding infants (< 6 months), zinc deficiency, and measles infection (1)(2) .
A protective effect of being overweight against pneumonia among adults has previously been described (15) . Although there is evidence that obese individuals are more susceptible to inflammatory and viral pulmonary diseases, the impact of obesity on bacterial pneumonia is still uncertain (16) . These findings suggest that it might be a mechanism to explain the protective effect of obesity against those pneumonias.
In this study, despite the aforementioned evidence from the literature, it is important to highlight that the abrupt acute weight loss among the cases, caused by lack of appetite and dehydration secondary to the respiratory condition, may have contributed towards the greater proportion of controls with overweight individuals and a consequent inverse association between obesity and child hospitalization due to pneumonia.
Regarding prevalent childhood illnesses, in this study, a history of wheezing was associated with the outcome studied, thus demonstrating the importance of identifying and following up individuals with recurrent wheezing, who might be undiagnosed as asthmatic individuals. Asthma leads to greater vulnerability to the pathogens that cause pneumonia, which in turn exacerbates asthmatic symptoms and contributes towards a worsening of the clinical conditions and the need for hospitalization (17) .
Also in relation to prevalent illnesses, there was a negative association observed between gastroesophageal reflux and child hospitalization due to pneumonia. There is evidence that chronic aspiration or microaspiration of gastric content may cause pulmonary harm, thus contributing to the genesis of aspiration pneumonia (18) .
Reflux is not necessarily associated with aspiration or complications though, due to the protective mechanism of the laryngeal chemoreflex, which is triggered by the presence of acidic solutions (such as gastric juice) in the larynx (19) . Since the acidity of the gastric juice forms part of the individuals' innate immunity, small amounts of the liquid in the pharynx, without any associated aspiration, could neutralize some pathogens that cause CAP, thus contributing to the protective effect found. In this study, Another factor relating to childcare that increased the risk of hospitalization due to pneumonia was the use of medicines during the 30 days prior to hospitalization.
PHC professionals should analyze any abusive use of medicines in pediatrics. Non-recommended antibiotics predispose to bacterial resistance, while antipyretic drugs, acid-suppressants and glucocorticoids may interfere with the immune system, thereby increasing susceptibility to pneumonia (20) . However, one limitation Rev. Latino-Am. Enfermagem 2017;25:e2892.
in this study was that the medicines used were not specified. Thus, new studies designed to provide this information would be extremely useful to verify this discovered association.
Child nutritional status is influenced by current and previous nutrient intake, which includes the prenatal period. Multiparity and short intervals between pregnancies cause damage to the maternal body and have an influence on weight gain during pregnancy, which may result in low weight at birth and during childhood (21) . In this study, the variables parity, birth interval, weight gain during pregnancy and birth weight were associated with the outcome in the univariate analysis. In the multivariate model, however, low birth weight lost its statistical significance. Thus, the results suggest that the maternal reproductive factors acts, at least in one way, independently from the perinatal factors (such as the birth weight), possibly because the ability for maternal self-care would positively influence childcare (21) .
During the PHC activities, in addition to those in childcare, prenatal follow-up permits adequate weight gain during pregnancy, and care for women of reproductive age promotes family planning actions (11) , thereby providing guidance regarding parity and the birth interval, which are variables associated with the outcome studied.
The results of this study indicate attendance at a daycare center as a risk factor for the outcome, and this finding was consistent with the literature (22) . The concentration of children within these institutions may facilitate disease transmission and result in permanent exposure to pathogens that may cause repetitive episodes, thereby contributing towards worsening the medical condition and exacerbating the need for hospitalization (22) . Attendance at a daycare center is not directly the object of attention for the PHC professionals, but it is extremely important to articulate these two scenarios (health and childhood education) for child care, health promotion and disease prevention.
Another variable that composes the final model of this study, and which tends to go beyond the scope of the health sector, is family income. The effect of family income on the outcome studied was independent of marital status and parents' education, which is consonant with another study (23) . It should be pointed out that the family needs and income deserve to be seen in an intersectoral way. Intersectoral collaboration and resource mobilization are among the priority actions needed to achieve the goal of ending preventable deaths from pneumonia (24) .
Social inequalities lead to differences in exposure to risk factors and in access to effective interventions in pneumonia (2,12) . These differences may exist within the same community, which seems to be the case in our study population. Although the study region is among the most prosperous in the country (8)

Conclusion
PHC and its child health care practices, such as nutritional monitoring, immunization, attention to prevalent diseases, prenatal care and family planning, constitute actions to be prioritized so that child hospitalization due to pneumonia can be avoided.
The hierarchical approach permitted understanding the phenomenon of child hospitalization due to pneumonia in the Brazilian population studied. It depicted the influence of the characteristics related to child, mother and family, along with the care practices and qualified PHC. These findings should direct rational planning of actions to avoid child hospitalization, especially in the context of PHC.