Risk Factors and Predictors of Pneumonia Among Under-Five Children in Sana’a City, Yemen: A Multicenter Case-Control Study Authors: Mohammed Al-Shehab, Rashad Abdul-Ghani, Muneera Muthana Zaid, Ahmed A Azazy, Sayaf A Alsoofi, Reham S Al-Areqi, Rawan A Al-Akwaa, Mohammed H Al-Arashani, Kafa Alwan Ali Alkhder, Sahar A Thabet, Ahlam Abdulmalik Esmail, Ahmed S Hudna Published: 2026 ABSTRACT BackgroundPneumonia is a major cause of preventable morbidity and mortality among under-five children, particularly in low-resource and conflict-affected countries. However, evidence from Yemen remains scarce. This study examined child-related, maternal, and environmental factors possibly associated with pneumonia among under-five children in Sana’a City, Yemen.MethodsAn unmatched multicenter case-control study was conducted among 302 children aged 2–59 months at five tertiary hospitals from August to December 2023. The study included 151 children admitted to pediatric wards with pneumonia and 151 controls attending outpatient clinics for non-respiratory conditions and without clinical evidence of pneumonia. Data on potential factors associated with pneumonia were collected using a structured interviewer-administered questionnaire. Univariable binary logistic regression was used to assess associations, and variables with p < 0.20 were entered into a multivariable binary logistic regression model to identify independent predictors.ResultsIn univariable analysis, low birthweight, prematurity, acute malnutrition, recent upper respiratory tract infection (URTI), having a household contact with respiratory infection, and poor ventilation in the cooking area were significantly associated with pneumonia. In multivariable analysis, acute malnutrition (adjusted odds ratio [AOR] = 1.9, 95% CI: 1.12–3.36; p = 0.020), recent URTI within the preceding week (AOR = 4.4, 95% CI: 2.49–7.80; p < 0.001), and having a household contact with respiratory infection (AOR = 1.9, 95% CI: 1.08–3.34; p = 0.027) remained independent predictors.ConclusionAcute malnutrition, recent URTI, and household contact with respiratory infection show independent associations with pneumonia among under-five children attending tertiary hospitals in Sana’a. These findings highlight child nutrition, early recognition and management of URTI, and household infection-prevention practices as priorities in this resource-limited, conflict-affected setting. However, these hospital-based case-control findings require confirmation in longitudinal, population-based studies before they can inform context-specific prevention strategies. Introduction Pneumonia is a leading preventable cause of morbidity and mortality among under-five children worldwide, particularly in low- and middle-income countries (LMICs).1,2 The World Health Organization (WHO) estimated that pneumonia accounted for 14% of deaths in under-five children in 2019, causing more than 740,000 deaths worldwide.3 Determinants of childhood pneumonia are multifactorial and reflect interactions of child-related, maternal, and environmental factors.4 These determinants influence susceptibility to respiratory infections by affecting immune function, nutritional status, pathogen exposure, and living conditions. Child-related factors associated with increased pneumonia risk include lack of exclusive breastfeeding during the first 6 months of life, low birthweight (LBW), prematurity, malnutrition or undernutrition, and incomplete or lack of immunization.4–8 Key maternal factors associated with childhood pneumonia include low maternal education, maternal smoking, and extremes of maternal age.7–10 Environmental exposures associated with childhood pneumonia include indoor air pollution, tobacco-smoke exposure, household overcrowding, inadequate household ventilation, and the use of unclean cooking fuel.5,8,11–14 Identifying modifiable risk factors for childhood pneumonia is essential for developing targeted interventions to reduce pneumonia burden among under-five children.4 For example, exclusive breastfeeding during the first 6 months of life, immunization, and reduced exposure to household air pollutants can decrease pneumonia incidence and mortality in children.15 Between 2000 and 2023, 170 million children died before reaching five years of age, with conflict-affected and fragile settings experiencing disproportionately high child mortality.16 Furthermore, infectious diseases remain a major contributor to mortality among under-five children, with approximately 20% of deaths attributed to lower respiratory tract infections (LRTIs), including pneumonia.16 Yemen has experienced prolonged conflict and humanitarian crises that have significantly disrupted healthcare infrastructure and access to essential child health services. Consequently, preventable infectious diseases, including pneumonia, continue to contribute substantially to under-five morbidity and mortality in the country. Pneumonia is a major public health problem among under-five children in Yemen, accounting for an estimated 17% of under-five deaths in 2021.15 In 2023, Yemen had an estimated under-five mortality rate of 39 deaths per 1,000 live births.16 Nevertheless, published studies on childhood pneumonia in the country remain scarce and primarily predate the escalation of conflict in 2015. Two large cohorts of children hospitalized with severe pneumonia—one including 2,554 children in 1991–1995 and another including 529 children in 1997—documented case fatality rates of 8.7% and 9.8%, respectively, with more than half of deaths occurring in infants younger than 6 months of age.17,18 Although these studies identified associations of LBW, rickets, anemia, cyanosis, and heart failure with mortality from severe pneumonia,17,18 they did not assess risk factors for pneumonia. A 10-year analysis (2000–2010) confirmed very severe pneumonia as the leading cause of severe morbidity and death in children aged 2–60 months, with only modest reductions in hospitalization and death after the introduction of Haemophilus influenzae type b (Hib) vaccine in 2005.19 Despite the recognized burden of pneumonia in Yemen, evidence on the risk factors and predictors of childhood pneumonia in the local context is scarce. For instance, younger age, incomplete immunization, the presence of household smokers, and use of non-gas cooking fuels have been associated with severe acute respiratory viral infections among Yemeni children.20 Identifying determinants of childhood pneumonia is essential for developing effective prevention and control strategies. Addressing modifiable determinants could contribute to reducing preventable childhood mortality and support progress toward Sustainable Development Goal 3 (SDG 3): “to ensure healthy lives and promote well-being for all at all ages.”21 Therefore, this study aimed to assess child-related, maternal, and environmental factors associated with pneumonia and identify its predictors among under-five children in Sana’a City, the capital of Yemen. Methods Study Design, Population and Setting An unmatched multicenter case-control study was conducted among under-five children at five tertiary hospitals in Sana’a City from August to December 2023. These hospitals serve as major referral centers for pediatric patients in the city and were purposively selected based on administrative permission from hospitals, the availability of functional pediatric wards, and high pediatric admission volumes documented in hospital records. Three public (Al-Sabeen Maternity and Childhood Hospital, Al-Thawra Modern General Hospital, and Al-Jomhouri University Teaching Hospital) and two private (University of Science and Technology Hospital and Modern European Hospital) hospitals were included in the study. Children of either sex and aged 2–59 months who met the eligibility criteria were enrolled at the selected hospitals during the study period. Children were excluded if they had chronic medical conditions or if their mothers or legal guardians declined to provide informed consent for participation. Cases were children aged 2–59 months admitted with pneumonia to the pediatric wards of the participating hospitals. Pneumonia was diagnosed according to WHO Integrated Management of Childhood Illness (IMCI) criteria as cough and/or difficulty breathing accompanied by tachypnea (≥50 breaths/min in children aged 2 to <12 months and ≥40 breaths/min in those aged 12 to 59 months) and/or chest indrawing.22 Children with severe pneumonia or very severe disease, defined by stridor while calm or any WHO IMCI general danger sign, including inability to drink or breastfeed, vomiting everything, convulsions, lethargy, or unconsciousness,22 were excluded. Controls were children aged 2–59 months who attended the outpatient pediatric clinics of the same hospitals for non-respiratory conditions during the study period and had no clinical evidence of pneumonia. Study Sample and Sampling Method The sample size was calculated using the open-source OpenEpi software (Version 3.01; www.openepi.com), assuming a 95% confidence level, 80% statistical power, a 1:1 case-to-control ratio, an odds ratio of 2.0, and an exposure prevalence of 50% among controls because no prior local estimate was available. The minimum required sample size was 296 participants. However, the final sample included 302 children (151 cases and 151 controls). Cases and controls were recruited consecutively from eligible children at the study hospitals until the target sample size was reached. Data Collection A structured interviewer-administered questionnaire was used to collect data from the mothers or legal guardians of the children. The questionnaire was developed by the research team after reviewing relevant literature and was assessed by experts for content validity. It included child characteristics (sex, age, weight, height, prematurity, breastfeeding, immunization status, complementary feeding, and malnutrition), maternal characteristics (age, literacy status, and employment status), and environmental factors (household size, number of persons sharing the child’s bedroom, other under-five children in the household, household smokers, presence of a household contact with respiratory infection, use of solid fuels for cooking or heating, household ventilation, exposure to volatile chemicals, number of windows per habitable room, and the presence of windows and a chimney in the cooking area). The household-contact item recorded any reported respiratory infection and did not distinguish the clinical syndrome or etiologic agent. Definitions of Study Variables A premature baby was defined as one delivered before 37 completed weeks of gestation, and LBW as birthweight <2500 g.23 Exclusive breastfeeding was defined as providing breast milk only, with no additional food or fluids.24 Complementary feeding was defined as giving infants aged 6–23 months other foods and liquids in addition to breast milk.25 It was therefore assessed only among children aged ≥6 months. Nutritional status was assessed using weight-for-height/length Z-scores (WHZ/WLZ) according to WHO 2006 child growth standards.26 Children with Z-scores below −2 standard deviations (SDs) from the WHO reference median were classified as having acute malnutrition, whereas those with Z-scores ≥−2 SD were classified as having normal nutritional status. Data Analysis Data were analyzed using IBM SPSS Statistics, version 23 (IBM Corp., Armonk, NY, USA), with statistical significance set at p < 0.05. Missing values were not imputed because missingness was limited for most variables. Complementary feeding was analyzed only among children aged ≥6 months, and observations from younger infants were classified as not applicable rather than missing. Therefore, its analysis was restricted to this subgroup. The number of missing observations for each variable was reported in the corresponding table footnotes. Univariable binary logistic regression was used to examine associations between potential risk factors and pneumonia, with crude odds ratios (CORs) and 95% confidence intervals (CIs) estimated. For each univariable analysis, missing data for the factor being examined were excluded only from that analysis. The exclusive breastfeeding variable, which had the highest number of missing observations, was included only in the univariable analysis but was eligible for inclusion in the multivariable model. Variables with p < 0.20 were entered into a multivariable binary logistic regression model to identify predictors of pneumonia, and adjusted odds ratios (AORs) and 95% CIs were estimated. The seven variables entered were maternal age ≥ 30 years, prematurity, LBW, acute malnutrition, URTI during the week preceding presentation, household contact with respiratory infection, and poor ventilation in the cooking area. To account for missing values, only 300 children with complete data for all seven variables (151 cases and 149 controls) were included in the multivariable analysis. Before fitting the multivariable model, multicollinearity was assessed using variance inflation factors (VIFs) and tolerance values. The multicollinearity diagnostics showed acceptable tolerance values > 0.6 and VIF values < 2, indicating no problematic multicollinearity among the included variables. Furthermore, the Hosmer-Lemeshow goodness-of-fit test did not indicate lack of model fit (χ2 = 9.707; p = 0.206). Ethical Considerations This study adhered to the ethical standards of the Declaration of Helsinki for research involving human participants. In addition, it was ethically approved by the Research Ethics Committee (REC) of the University of Science and Technology, Sana’a, Yemen (Ethical clearance No.: 1447/00101/UREC/UST). Informed consent was obtained from the mothers or legal guardians of all enrolled children before participation. When written consent could not be obtained, verbal informed consent was secured with prior approval from the REC. Mothers or guardians were informed about the study objectives and their right to refuse participation of their children or withdraw at any stage without giving a reason. Measures were taken to ensure participant confidentiality and protect the privacy of collected data. Results Demographic Characteristics of Participating Children Of the 302 children included in the study, the majority were male (62.3%), representing 63.6% of cases and 60.9% of controls. Most children were aged ≤ 2 years, accounting for 81.5% of cases and 76.8% of controls (Table 1).Table 1Demographic characteristics of participating childrenCharacteristicsCases (N = 151)Controls (N = 151)Total (N = 302)n (%)n (%)n (%)Sex Male96 (63.6)92 (60.9)188 (62.3) Female55 (36.4)59 (39.1)114 (37.7)Age (years) ≤ 2123 (81.5)116 (76.8)239 (79.1) > 228 (18.5)35 (23.2)63 (20.9) Table 1Demographic characteristics of participating childrenCharacteristicsCases (N = 151)Controls (N = 151)Total (N = 302)n (%)n (%)n (%)Sex Male96 (63.6)92 (60.9)188 (62.3) Female55 (36.4)59 (39.1)114 (37.7)Age (years) ≤ 2123 (81.5)116 (76.8)239 (79.1) > 228 (18.5)35 (23.2)63 (20.9) Demographic characteristics of participating children Child-Related and Maternal Risk Factors for Pneumonia Among Under-Five Children Univariable analysis showed significant associations between pneumonia and prematurity (COR = 2.9, 95% CI: 1.41–5.86; p = 0.003), LBW (COR = 2.1, 95% CI: 1.12–4.01; p = 0.019), acute malnutrition (COR = 1.9, 95% CI: 1.16–2.98; p = 0.009), and URTI during the week preceding presentation (COR = 4.8, 95% CI: 2.93–7.74; p < 0.001). Conversely, no significant associations were observed between pneumonia and child’s male sex or age of ≤ 2 years, maternal age ≥ 30 years, maternal illiteracy, maternal employment, incomplete immunization, lack of exclusive breastfeeding during the first 6 months, or lack of complementary feeding (Table 2).Table 2Child-related and maternal risk factors for pneumonia among under-five children attending hospitals in Sana’a City, Yemen, 2023VariablesCases(N = 151)Controls(N = 151)COR (95% CI)p-valuen (%)n (%)Male sex96 (63.6)92 (60.9)1.1 (0.70–1.78)0.635Age ≤ 2 years123 (81.5)116 (76.8)1.3 (0.76–2.32)0.322Maternal age ≥ 30 yearsa141(93.4)132 (88.0)1.9 (0.86–4.32)0.108Maternal illiteracy31 (20.5)34 (22.5)0.9 (0.51–1.54)0.674Maternal employment5 (3.3)5 (3.3)1.0 (0.28–3.53)1.000Incomplete immunizationa80 (53.0)69 (46.0)1.3 (0.84–2.01)0.226Prematurity30 (19.9)12 (7.9)2.9 (1.41–5.86)0.003LBW32 (21.2)17 (11.3)2.1 (1.12–4.01)0.019Lack of exclusive breastfeeding during the first 6 monthsb81 (63.8)85 (57.8)1.3 (0.79–2.09)0.314Lack of complementary feedingc5 (4.8)6 (4.8)1.0 (0.29–3.35)0.991Acute malnutrition (WHZ/WLZ < −2 SD)69 (45.7)47 (31.1)1.9 (1.16–2.98)0.009URTI during the week preceding presentation106 (70.2)50 (33.1)4.8 (2.93–7.74)< 0.001Notes: aMissing data for one control; bmissing data for 24 cases and four controls; ccalculated for children aged ≥ 6 months (104 cases and 124 controls).Abbreviations: COR, crude odds ratio; CI, confidence interval; LBW, low birthweight; WHZ, weight-for-height Z-score; WLZ, weight-for-length Z-score; SD, standard deviation; URTI, upper respiratory tract infection. Table 2Child-related and maternal risk factors for pneumonia among under-five children attending hospitals in Sana’a City, Yemen, 2023VariablesCases(N = 151)Controls(N = 151)COR (95% CI)p-valuen (%)n (%)Male sex96 (63.6)92 (60.9)1.1 (0.70–1.78)0.635Age ≤ 2 years123 (81.5)116 (76.8)1.3 (0.76–2.32)0.322Maternal age ≥ 30 yearsa141(93.4)132 (88.0)1.9 (0.86–4.32)0.108Maternal illiteracy31 (20.5)34 (22.5)0.9 (0.51–1.54)0.674Maternal employment5 (3.3)5 (3.3)1.0 (0.28–3.53)1.000Incomplete immunizationa80 (53.0)69 (46.0)1.3 (0.84–2.01)0.226Prematurity30 (19.9)12 (7.9)2.9 (1.41–5.86)0.003LBW32 (21.2)17 (11.3)2.1 (1.12–4.01)0.019Lack of exclusive breastfeeding during the first 6 monthsb81 (63.8)85 (57.8)1.3 (0.79–2.09)0.314Lack of complementary feedingc5 (4.8)6 (4.8)1.0 (0.29–3.35)0.991Acute malnutrition (WHZ/WLZ < −2 SD)69 (45.7)47 (31.1)1.9 (1.16–2.98)0.009URTI during the week preceding presentation106 (70.2)50 (33.1)4.8 (2.93–7.74)< 0.001Notes: aMissing data for one control; bmissing data for 24 cases and four controls; ccalculated for children aged ≥ 6 months (104 cases and 124 controls).Abbreviations: COR, crude odds ratio; CI, confidence interval; LBW, low birthweight; WHZ, weight-for-height Z-score; WLZ, weight-for-length Z-score; SD, standard deviation; URTI, upper respiratory tract infection. Child-related and maternal risk factors for pneumonia among under-five children attending hospitals in Sana’a City, Yemen, 2023 Notes: aMissing data for one control; bmissing data for 24 cases and four controls; ccalculated for children aged ≥ 6 months (104 cases and 124 controls). Abbreviations: COR, crude odds ratio; CI, confidence interval; LBW, low birthweight; WHZ, weight-for-height Z-score; WLZ, weight-for-length Z-score; SD, standard deviation; URTI, upper respiratory tract infection. Environmental Risk Factors for Pneumonia Among Under-Five Children Univariable analysis showed statistically significant associations between pneumonia and having a household contact with respiratory infection (COR = 2.7, 95% CI: 1.71–4.33; p < 0.001) and poor ventilation in the cooking area (COR = 2.4, 95% CI: 1.12–5.04; p = 0.022). In contrast, no significant associations were found between pneumonia and household size ≥ 5 members, presence of other under-five children in the household, presence of indoor smoker(s), exposure to solid fuels, exposure to volatile substances, poor household ventilation, or sharing a child’s bedroom with more than two persons (Table 3).Table 3Environmental risk factors for pneumonia among under-five children attending hospitals in Sana’a City, Yemen, 2023VariablesCases(N = 151)Controls(N = 151)COR (95% CI)p-valuen (%)n (%)Household size ≥5 members118 (78.1)115 (76.2)1.1 (0.65–1.92)0.681Other under-five children in the household84 (55.6)82 (54.3)1.1 (0.67–1.66)0.817Having a household contact with respiratory infection93 (61.6)56 (37.1)2.7 (1.71–4.33)< 0.001Presence of indoor smoker(s)a79 (52.3)81 (54.4)0.9 (0.59–1.45)0.723Exposure to solid fuelsb60 (39.7)59 (39.3)1.0 (0.64–1.61)0.943Exposure to volatile substances (paints, petrol, kerosene, air fresheners, etc.)c89 (60.5)97 (64.2)0.9 (0.53–1.37)0.510Poor household ventilation (≤1 window per habitable room)d9 (6.0)6 (4.2)1.4 (0.50–4.14)0.501Poor ventilation in the cooking area (≤1 window and/or absence of chimney)a24 (15.9)11 (7.4)2.4 (1.12–5.04)0.022Sharing a child’s bedroom with more than two persons82 (54.3)76 (50.3)1.2 (0.75–1.84)0.489Notes: amissing data for two controls; bmissing for one control; cmissing data for four cases; dmissing data for nine controls.Abbreviations: COR, crude odds ratio; CI, confidence interval. Table 3Environmental risk factors for pneumonia among under-five children attending hospitals in Sana’a City, Yemen, 2023VariablesCases(N = 151)Controls(N = 151)COR (95% CI)p-valuen (%)n (%)Household size ≥5 members118 (78.1)115 (76.2)1.1 (0.65–1.92)0.681Other under-five children in the household84 (55.6)82 (54.3)1.1 (0.67–1.66)0.817Having a household contact with respiratory infection93 (61.6)56 (37.1)2.7 (1.71–4.33)< 0.001Presence of indoor smoker(s)a79 (52.3)81 (54.4)0.9 (0.59–1.45)0.723Exposure to solid fuelsb60 (39.7)59 (39.3)1.0 (0.64–1.61)0.943Exposure to volatile substances (paints, petrol, kerosene, air fresheners, etc.)c89 (60.5)97 (64.2)0.9 (0.53–1.37)0.510Poor household ventilation (≤1 window per habitable room)d9 (6.0)6 (4.2)1.4 (0.50–4.14)0.501Poor ventilation in the cooking area (≤1 window and/or absence of chimney)a24 (15.9)11 (7.4)2.4 (1.12–5.04)0.022Sharing a child’s bedroom with more than two persons82 (54.3)76 (50.3)1.2 (0.75–1.84)0.489Notes: amissing data for two controls; bmissing for one control; cmissing data for four cases; dmissing data for nine controls.Abbreviations: COR, crude odds ratio; CI, confidence interval. Environmental risk factors for pneumonia among under-five children attending hospitals in Sana’a City, Yemen, 2023 Notes: amissing data for two controls; bmissing for one control; cmissing data for four cases; dmissing data for nine controls. Abbreviations: COR, crude odds ratio; CI, confidence interval. Independent Predictors of Pneumonia Among Under-Five Children In the multivariable binary logistic regression model, acute malnutrition (AOR = 1.9, 95% CI: 1.12–3.36; p = 0.020), URTI during the week preceding presentation (AOR = 4.4, 95% CI: 2.49–7.80; p < 0.001), and having a household contact with respiratory infection (AOR = 1.9, 95% CI: 1.08–3.34; p = 0.027) were independent predictors of pneumonia among under-five children (Table 4).Table 4Multivariable analysis of factors associated with pneumonia among under-five children attending hospitals in Sana’a City, Yemen, 2023aFactors Included in the ModelAOR (95% CI)p-valueMaternal age ≥ 30 years2.4 (0.96–6.16)0.060Prematurity2.0 (0.92–4.43)0.079LBW2.0 (0.97–4.23)0.059Acute malnutrition1.9 (1.12–3.36)0.020URTI during the week preceding presentation4.4 (2.49–7.80)< 0.001Having a household contact with respiratory infection1.9 (1.08–3.34)0.027Poor ventilation in the cooking area2.3 (0.99–5.29)0.054Notes: aThe analysis included all 151 cases and 149 controls. Two controls were excluded: one had missing data for both maternal age and poor ventilation in the cooking area, and the other had missing data only for poor ventilation in the cooking area.Abbreviations: AOR, adjusted odds ratio; CI, confidence interval; LBW, low birthweight; URTI, upper respiratory tract infection. Table 4Multivariable analysis of factors associated with pneumonia among under-five children attending hospitals in Sana’a City, Yemen, 2023aFactors Included in the ModelAOR (95% CI)p-valueMaternal age ≥ 30 years2.4 (0.96–6.16)0.060Prematurity2.0 (0.92–4.43)0.079LBW2.0 (0.97–4.23)0.059Acute malnutrition1.9 (1.12–3.36)0.020URTI during the week preceding presentation4.4 (2.49–7.80)< 0.001Having a household contact with respiratory infection1.9 (1.08–3.34)0.027Poor ventilation in the cooking area2.3 (0.99–5.29)0.054Notes: aThe analysis included all 151 cases and 149 controls. Two controls were excluded: one had missing data for both maternal age and poor ventilation in the cooking area, and the other had missing data only for poor ventilation in the cooking area.Abbreviations: AOR, adjusted odds ratio; CI, confidence interval; LBW, low birthweight; URTI, upper respiratory tract infection. Multivariable analysis of factors associated with pneumonia among under-five children attending hospitals in Sana’a City, Yemen, 2023a Notes: aThe analysis included all 151 cases and 149 controls. Two controls were excluded: one had missing data for both maternal age and poor ventilation in the cooking area, and the other had missing data only for poor ventilation in the cooking area. Abbreviations: AOR, adjusted odds ratio; CI, confidence interval; LBW, low birthweight; URTI, upper respiratory tract infection. Discussion To our knowledge, this study is among the first to examine the risk factors and predictors of pneumonia among under-five children in Yemen, where epidemiological evidence remains limited. The absence of significant associations with child sex or age in this study is consistent with findings from Ethiopia and Indonesia.7,27 Conversely, findings from Egypt showed that male sex and younger age were independent predictors of pneumonia among under-five children.28 These differences may reflect variations in population characteristics, environmental exposures, healthcare access, and study design across settings. Child sex and age are non-modifiable risk factors that may influence susceptibility to respiratory infections through physiological and immunological mechanisms but cannot be directly altered through public health interventions. Prematurity and LBW were significantly associated with pneumonia in this study, but they did not remain independent predictors after adjustment for confounders. This pattern is consistent with findings from Egypt, Indonesia, and Pakistan.7,28–31 By contrast, both lower birthweight and small-for-gestational-age status were independent predictors of neonatal pneumonia in Shijiazhuang, China.32 Infants born prematurely or with LBW often have immature immune systems, reduced pulmonary defense mechanisms, and underdeveloped lung structure and function, which increase susceptibility to respiratory pathogens during infancy and early childhood.33–35 The non-significant association between pneumonia and lack of exclusive breastfeeding during the first 6 months of life in the present study contrasts with well-established evidence that exclusive breastfeeding protects against respiratory infections, including pneumonia. This protection is mediated through passive immunity, improved nutrition, and bioactive immune components of breast milk that reduce the risk of LRTIs.36–39 Similar non-significant findings have been reported in studies from Brazil, Nepal, and India.6,40,41 Non-exclusive breastfeeding was significantly associated with pneumonia among Egyptian under-five children, although the association disappeared after adjustment for confounders.28 In contrast, several case-control studies from developing countries have identified the lack of exclusive breastfeeding as an independent predictor of pneumonia.7,42–44 The non-significant association in the present study may be partly attributed to recall bias in maternal reporting of breastfeeding practices or the influence of other interacting determinants, such as nutritional status, environmental exposures, and household transmission of respiratory infections, which may have weakened the observed effect of breastfeeding. Nevertheless, this finding does not diminish the well-established importance of exclusive breastfeeding for optimal child health and protection against infections. Malnourished children in the present study had significantly higher odds of pneumonia than those with normal nutritional status, and acute malnutrition remained an independent predictor after adjustment. This finding is consistent with case-control studies conducted in several LMICs.6,7,29,42,45,46 Malnutrition compromises innate and adaptive immune responses,47 increasing the incidence, severity, and mortality associated with pneumonia.48–50 These findings reinforce the importance of nutritional interventions in reducing the burden of childhood pneumonia. The non-significant association between incomplete immunization and pneumonia among children in this study agrees with reports from Nepal and Ethiopia.40,45 In contrast, case-control studies have reported significant associations in other countries, including Egypt,28 Indonesia,7,46 Pakistan,31 India,41 Tanzania,42 and Ethiopia.43 In some of these studies, however, the association lost significance after adjustment for confounders. A systematic review and meta-analysis also found incomplete immunization to be significantly associated with acute LRTIs, including pneumonia.5 Differences may reflect vaccination coverage, healthcare access, study design, sample size, and the influence of other interacting determinants. Maintaining high vaccination coverage remains essential for preventing vaccine-preventable respiratory infections and reducing the burden of childhood pneumonia.51 A recent history of URTI was an independent predictor of pneumonia in this study, with approximately fourfold higher adjusted odds among children with URTI during the week preceding presentation. This finding is consistent with reports from other settings.27,43,45 URTIs may progress to LRTIs and predispose children to pneumonia by disrupting mucosal defenses, impairing mucociliary clearance, and facilitating secondary bacterial infections of the lower respiratory tract.52–54 The presence of a household contact with respiratory infection was also an independent predictor of pneumonia in this study, with nearly twofold higher adjusted odds among children exposed to family members with a reported respiratory infection. Close living arrangements and frequent interpersonal contact may facilitate respiratory pathogen transmission. However, respiratory infection was recorded as a broad respondent-reported category, precluding identification of specific infections or pathogens and estimation of their relative contributions. This finding highlights the role of household transmission of respiratory pathogens, particularly in settings where close contact among family members facilitates the spread of infectious agents.55 Similar findings have been reported from other resource-limited settings.56,57 Respiratory infections can easily spread within households through droplets, close physical contact, and shared living spaces, thereby increasing children’s susceptibility to more severe LRTIs, including pneumonia. Poor ventilation in the cooking area was significantly associated with pneumonia in the univariable analysis. However, this association did not remain independent after adjustment, suggesting that other determinants may have confounded the observed association. In comparison, poor home aeration was identified as an independent predictor of pneumonia among under-five children in Egypt.28 Other environmental exposures examined in the present study, including indoor smoking, solid-fuel use, and household crowding, were not significantly associated with pneumonia. These null findings may reflect limited variability in exposure or misclassification arising from self-reported measures. In contrast, case-control studies from other countries have reported significant associations between pneumonia among under-five children and household exposure to air pollution, with some of these factors identified as independent predictors.7,41,44 These exposures include solid-fuel use, indoor cooking smoke, poor ventilation, lack of a chimney, and keeping the child near the cooking area.7,41,44 Overall, these findings underscore the importance of improving household air quality and reducing indoor pollution as part of strategies to prevent pneumonia among under-five children. Although the present study focused on factors associated with pneumonia occurrence, clinical risk stratification complements the identification of associated factors by helping clinicians determine which children require imaging, hospital admission, oxygen therapy, or closer monitoring. A validation study in a pediatric emergency department showed that some models could distinguish children with radiographic pneumonia reasonably well, although their performance varied and validation in other pediatric populations remains necessary.58 In hospitalized under-five children from developing countries, hypoxemia and markedly elevated procalcitonin were associated with mortality.59 These findings support standardized clinical assessment and routine pulse oximetry, particularly in resource-limited settings where laboratory biomarkers may not be readily available. Because children with severe pneumonia or very severe disease were excluded and severity outcomes were not analyzed, future research should validate simple pediatric risk-stratification tools appropriate for Yemeni healthcare settings. This multicenter study provides important local evidence on risk factors and predictors of pneumonia among under-five children in a resource-limited, conflict-affected setting where epidemiological data are scarce. Among its strengths is its multicenter design, with children enrolled from three public and two private tertiary hospitals, reducing dependence on the practices and patient characteristics of a single facility or health sector and capturing variation across healthcare settings in the city. Including both public and private health sectors also broadened referral pathways, service-delivery contexts, and patient populations represented. The use of standardized WHO IMCI diagnostic criteria promoted consistent case identification across participating hospitals and facilitated comparison with studies using similar definitions. Moreover, assessing child-related, maternal, and household factors allowed associations across multiple domains to be examined within a single analytic framework and relevant variables to be considered simultaneously in the adjusted analysis. The identification of acute malnutrition, recent URTI, and household contact with respiratory infection as factors independently associated with pneumonia highlights clinically recognizable and potentially actionable areas. Collectively, these findings provide locally relevant evidence for pediatricians, general practitioners, and program planners working in a resource-limited and conflict-affected setting. Several limitations should be disclosed for transparent interpretation of findings. First, the case-control design does not allow the establishment of causal relationships between the identified predictors and pneumonia. Second, hospitalized pneumonia cases were compared with outpatient controls because recruiting community controls was logistically impractical, potentially introducing selection bias because these controls may not adequately represent the underlying source population from which cases arose. Although controls with respiratory conditions were excluded to minimize bias, differences between the two groups in healthcare-seeking behavior, socioeconomic status, and underlying health status may have contributed to residual selection bias. The study sample may also overrepresent hospital-attending cases and limit generalizability to children in the community or those treated at primary healthcare facilities. Furthermore, the exclusion of children with severe pneumonia or very severe disease limits generalizability across the full clinical spectrum of childhood pneumonia. Third, because missing values were not imputed, the univariable analyses were based on available observations. If missingness was not random for the exclusive breastfeeding variable, missing values may have reduced statistical precision and introduced selection bias. Therefore, the observed non-significant association with pneumonia should be interpreted cautiously. Complementary feeding was assessed only among children aged ≥6 months, resulting in a smaller eligible subgroup of 104 cases and 124 controls. This restriction may also have reduced statistical precision, and the corresponding estimate applies only to children aged ≥ 6 months. Neither exclusive breastfeeding nor complementary feeding met the prespecified criterion of p < 0.20 in the univariable analysis and, therefore, neither variable was included in the multivariable model. By contrast, the multivariable analysis used complete-case data, excluding only two controls with missing observations for one or more of the seven included variables. Accordingly, any reduction in precision or potential selection bias in the multivariable analysis was likely minimal. Fourth, several exposures were based on maternal self-report and proxy environmental measures, introducing the possibility of recall, reporting, and exposure misclassification biases. Finally, although multivariable analysis was used to control for confounding, residual confounding by unmeasured factors, such as detailed socioeconomic indicators, hygiene practices, or pneumonia etiology, cannot be ruled out. Conclusion Acute malnutrition, recent URTI, and household contact with respiratory infection show independent associations with pneumonia among under-five children attending hospitals in Sana’a City. Strengthening interventions that promote optimal child nutrition, early diagnosis and management of respiratory infections, and household infection-prevention practices may help reduce the burden of childhood pneumonia in this low-resource and conflict-affected setting. However, the hospital-based case-control design precludes causal and population-level inferences. Population-based longitudinal studies are therefore warranted to confirm these associations and inform context-specific prevention strategies.