Introduction: Pneumonia remains the leading infectious cause of death among children under five, and most deaths follow severe disease. Identifying modifiable determinants of severe pneumonia in local settings is essential to target prevention. Objective: To identify socio-demographic, child-related, household and maternal determinants of severe pneumonia among children aged 2–59 months. Methods: A hospital-based case-control study was conducted at a tertiary-care hospital, Maharashtra, from January 2024 to December 2025. Cases (n = 134) were children admitted with severe pneumonia as per WHO/IMNCI criteria; controls (n = 134) were age-group-matched children attending the paediatric OPD for non-respiratory illness. Data were collected using a pre-tested case record form. Crude odds ratios (OR) were estimated, and variables with p < 0.20 were entered into a multivariable logistic regression model adjusted for age and sex. Results: Mean age was 27.0 ± 17.9 months in cases and 27.2 ± 17.8 months in controls. Incomplete immunisation for age (adjusted OR [aOR] 4.36; 95% CI 2.20–8.65), history of recurrent wheeze (aOR 3.40; 1.53–7.59), previous hospitalisation (aOR 3.09; 1.50–6.35) and undernutrition (aOR 2.92; 1.62–5.29) were independent determinants of severe pneumonia. Biomass fuel use was equally common in cases and controls (44.0% each). Among cases, 77.6% had SpO₂ < 90%, 23.1% required PICU care, 6.0% needed mechanical ventilation and 2 (1.5%) died. Two-thirds of cases (66.4%) reached first medical contact more than 24 hours after symptom onset. Conclusion: Incomplete immunisation, undernutrition and prior respiratory morbidity were the strongest determinants of severe pneumonia. Strengthening full immunisation (including PCV), nutritional surveillance and follow-up of children with previous respiratory illness may reduce the burden of severe pneumonia in this region.
Pneumonia is the single largest infectious cause of death in children worldwide. In 2019 it killed 740,180 children under five, accounting for 14% of all under-five deaths and 22% of deaths among children aged 1–5 years, with the highest burden in southern Asia and sub-Saharan Africa.[1,2] India continues to contribute one of the largest absolute numbers of childhood pneumonia deaths, despite steady declines in under-five mortality.[2,3]
Most pneumonia deaths occur among children who progress to severe disease, characterised by hypoxaemia, danger signs or severe respiratory distress.[4] The risk of progression is shaped by host factors such as undernutrition, low birth weight and incomplete immunisation, and by environmental exposures including household air pollution and crowding.[4–7] A systematic review identified low birth weight, lack of exclusive breastfeeding, crowding, household air pollution, incomplete immunisation, undernutrition and HIV infection as consistent risk factors for severe acute lower respiratory infection.[4] Indian hospital-based case-control studies have reported similar associations, with notable variation in the relative importance of each factor across regions.[5,6,12]
India's Social Awareness and Action to Neutralise Pneumonia Successfully (SAANS) initiative and the inclusion of pneumococcal conjugate vaccine (PCV) in the Universal Immunisation Programme aim to accelerate reductions in pneumonia mortality.[9] Programme planning, however, depends on knowing which determinants matter most in a given population.
This study was therefore undertaken to identify the socio-demographic, child-related, household-environmental and maternal determinants of severe pneumonia among children aged 2–59 months attending a tertiary-care hospital, and to describe the clinical profile, care-seeking pattern and outcome of affected children.
A hospital-based case-control study was conducted tertiary care hospital. Data were collected from January 2024 to 2025.
Case definition: Children aged 2–59 months admitted with cough or difficult breathing plus at least one of the following, as per the revised WHO/IMNCI classification:[10] SpO₂ < 90% on room air, central cyanosis, any general danger sign (inability to drink or breastfeed, persistent vomiting, convulsions, lethargy or unconsciousness), stridor in a calm child, or severe respiratory distress (grunting, severe chest indrawing).
Control definition: Children aged 2–59 months attending the paediatric OPD during the same period for a non-respiratory complaint (acute gastroenteritis without dehydration, minor injury, skin condition or routine immunisation), with no cough or difficult breathing in the preceding 2 weeks. Controls were frequency-matched to cases by age group (2–11 and 12–59 months).
Congenital heart disease, chronic lung disease, known neuromuscular disorder, aspiration pneumonia, hospital-acquired pneumonia, hospitalisation within the preceding 14 days, and refusal of consent. Sample size was calculated for an unmatched case-control design, with biomass fuel use as the principal exposure. Assuming 30% exposure among controls, an odds ratio of 2.0, 95% confidence and 80% power with one control per case, the required sample was 141 per group (155 with 10% non-response). During the study period, 134 eligible cases and 134 controls were enrolled consecutively, which gives about 78% power for the same assumptions. After written informed consent from the parent or guardian, the primary caregiver was interviewed using a pre-tested, structured case record form. It covered socio-demographic details, birth history, feeding practices, immunisation (verified from the mother and Child Protection card or U-WIN wherever available), past illness, housing, cooking fuel and ventilation, household smoking, maternal factors and, for cases, care-seeking, clinical findings, investigations and outcome. Socio-economic status was graded using the Modified BG Prasad scale updated for the current Consumer Price Index. Weight, length or height and MUAC were measured by standard techniques. Z-scores were calculated using WHO Child Growth Standards.[11] Undernutrition was defined as any of underweight (WAZ < −2 SD), stunting (HAZ < −2 SD), wasting (WHZ < −2 SD), MAM or SAM. Incomplete immunisation was defined as being partially immunised or unimmunised for age as per the national schedule. Children whose immunisation status was based on recall only (n = 8) were excluded from analyses of this variable. Overcrowding was defined as more than 2.5 persons per living room. Biomass fuel use was defined as the use of firewood, cow-dung cakes or crop residue for cooking.
Data were entered in Microsoft Excel and analysed using Python 3 (pandas, SciPy and statsmodels). Categorical variables are presented as frequencies and percentages and compared using the chi-square or Fisher's exact test. Continuous variables are presented as mean ± SD or median (IQR) and compared using the independent t-test or Mann–Whitney U test. Crude odds ratios with 95% confidence intervals were calculated for each exposure. Variables with p < 0.20 on bivariate analysis, together with age group and sex, were entered into a multivariable logistic regression model. Where variables were closely related, one composite variable was retained; for example, "any undernutrition" was used in place of separate anthropometric indices, and low birth weight in place of prematurity. Multicollinearity was assessed using the variance inflation factor (VIF) and model fit using the Hosmer–Lemeshow test, Nagelkerke R² and the area under the ROC curve. A p-value < 0.05 was considered statistically significant. The study was approved by the Institutional Ethics Committee. Written informed consent was obtained from the parent or guardian of each child. Confidentiality was maintained using coded study IDs.
A total of 268 children were studied: 134 cases and 134 controls. Of the cases, 110 (82.1%) were admitted to the paediatric ward and 24 (17.9%) directly to the PICU. Controls attended for acute gastroenteritis without dehydration (59, 44.0%), routine immunisation (30, 22.4%), skin conditions (23, 17.2%) and minor injuries (22, 16.4%). The mother was the respondent for 232 (86.6%) children.
Baseline characteristics: Cases and controls were comparable in age, sex, residence, family type, maternal education, socio-economic class and place of delivery (Table 1). Mean age was 27.0 ± 17.9 months in cases and 27.2 ± 17.8 months in controls (p = 0.894).
Table 1. Socio-demographic characteristics of cases and controls
|
Variable |
Cases (n = 134), n (%) |
Controls (n = 134), n (%) |
p-value |
|
Age group (months) |
|
|
0.687 |
|
2–11 |
39 (29.1) |
39 (29.1) |
|
|
12–23 |
19 (14.2) |
24 (17.9) |
|
|
24–59 |
76 (56.7) |
71 (53.0) |
|
|
Sex |
|
|
0.389 |
|
Male |
71 (53.0) |
79 (59.0) |
|
|
Female |
63 (47.0) |
55 (41.0) |
|
|
Residence |
|
|
0.576 |
|
Urban |
34 (25.4) |
41 (30.6) |
|
|
Rural |
79 (59.0) |
71 (53.0) |
|
|
Tribal area |
21 (15.7) |
22 (16.4) |
|
|
Type of family |
|
|
0.117 |
|
Nuclear |
62 (46.3) |
60 (44.8) |
|
|
Joint |
56 (41.8) |
46 (34.3) |
|
|
Three-generation |
16 (11.9) |
28 (20.9) |
|
|
Mother's education |
|
|
0.734 |
|
Illiterate |
21 (15.7) |
25 (18.7) |
|
|
Primary |
25 (18.7) |
18 (13.4) |
|
|
Middle |
30 (22.4) |
25 (18.7) |
|
|
High school |
31 (23.1) |
33 (24.6) |
|
|
Higher secondary |
17 (12.7) |
19 (14.2) |
|
|
Graduate and above |
10 (7.5) |
14 (10.4) |
|
|
Socio-economic class (Modified BG Prasad) |
|
|
0.437 |
|
I |
9 (6.7) |
10 (7.5) |
|
|
II |
39 (29.1) |
44 (32.8) |
|
|
III |
43 (32.1) |
34 (25.4) |
|
|
IV |
35 (26.1) |
31 (23.1) |
|
|
V |
8 (6.0) |
15 (11.2) |
|
|
Place of delivery |
|
|
0.247 |
|
Government institution |
89 (66.4) |
77 (57.5) |
|
|
Private institution |
40 (29.9) |
53 (39.6) |
|
|
Home |
5 (3.7) |
4 (3.0) |
|
|
Mother's age (years), mean ± SD |
26.8 ± 4.4 |
26.5 ± 5.1 |
0.521 |
|
Per capita monthly income (₹), median (IQR) |
3,041 (2,254–4,477) |
3,418 (2,063–5,224) |
0.810 |
Bivariate analysis: Undernutrition was present in 55.2% of cases compared with 31.3% of controls (OR 2.70; 95% CI 1.64–4.45). Incomplete immunisation (38.0% vs 16.0%; OR 3.21), previous hospitalisation (31.3% vs 12.7%; OR 3.14) and recurrent wheeze (23.1% vs 10.4%; OR 2.58) were also significantly more common among cases. Pallor and cough or cold in a household member in the preceding 2 weeks were also associated with severe pneumonia. Biomass fuel use (44.0% in both groups), overcrowding, household smoking and maternal factors showed no significant association (Table 2). Mean WAZ, HAZ and WHZ were significantly lower in cases (−1.49, −1.20 and −1.28) than in controls (−0.69, −0.72 and −0.79; all p < 0.001). Exclusive breastfeeding was unexpectedly reported more often for cases (OR for not exclusively breastfed 0.60; 95% CI 0.37–0.99).
Table 2. Bivariate analysis of potential determinants of severe pneumonia
|
Determinant |
Cases, n (%) |
Controls, n (%) |
Crude OR (95% CI) |
p-value |
|
Child-related factors |
|
|
|
|
|
Low birth weight (< 2.5 kg) |
33 (24.6) |
22 (16.4) |
1.66 (0.91–3.04) |
0.096 |
|
Preterm birth (< 37 weeks) |
20 (14.9) |
13 (9.7) |
1.63 (0.78–3.44) |
0.193 |
|
Breastfeeding not initiated within 1 hour |
65 (48.5) |
74 (55.2) |
0.76 (0.47–1.23) |
0.271 |
|
Not exclusively breastfed for 6 months |
44 (32.8) |
60 (44.8) |
0.60 (0.37–0.99) |
0.045 |
|
Bottle feeding |
23 (17.2) |
20 (14.9) |
1.18 (0.61–2.27) |
0.618 |
|
Incomplete immunisation for age† |
49 (38.0) |
21 (16.0) |
3.21 (1.78–5.77) |
< 0.001 |
|
Vitamin A not received (last dose) |
61 (45.5) |
55 (41.0) |
1.20 (0.74–1.95) |
0.460 |
|
Previous hospitalisation |
42 (31.3) |
17 (12.7) |
3.14 (1.68–5.88) |
< 0.001 |
|
History of recurrent wheeze |
31 (23.1) |
14 (10.4) |
2.58 (1.30–5.11) |
0.006 |
|
Diarrhoea in the last 2 weeks |
20 (14.9) |
12 (9.0) |
1.78 (0.83–3.81) |
0.132 |
|
Any undernutrition |
74 (55.2) |
42 (31.3) |
2.70 (1.64–4.45) |
< 0.001 |
|
Underweight (WAZ < −2 SD) |
41 (30.6) |
18 (13.4) |
2.84 (1.53–5.27) |
0.001 |
|
Stunting (HAZ < −2 SD) |
31 (23.1) |
13 (9.7) |
2.80 (1.39–5.64) |
0.003 |
|
Wasting (WHZ < −2 SD) |
31 (23.1) |
17 (12.7) |
2.07 (1.08–3.96) |
0.026 |
|
Pallor |
43 (32.1) |
25 (18.7) |
2.06 (1.17–3.63) |
0.012 |
|
Household and environmental factors |
|
|
|
|
|
Kutcha or semi-pucca house |
86 (64.2) |
73 (54.5) |
1.50 (0.92–2.44) |
0.106 |
|
Overcrowding |
59 (44.0) |
57 (42.5) |
1.06 (0.66–1.72) |
0.805 |
|
Biomass fuel use |
59 (44.0) |
59 (44.0) |
1.00 (0.62–1.62) |
1.000 |
|
No separate kitchen |
22 (16.4) |
26 (19.4) |
0.82 (0.44–1.53) |
0.524 |
|
Child carried while cooking |
28 (20.9) |
20 (14.9) |
1.51 (0.80–2.83) |
0.203 |
|
Any smoker in household |
32 (23.9) |
36 (26.9) |
0.85 (0.49–1.48) |
0.574 |
|
Mosquito coil / incense used indoors |
54 (40.3) |
51 (38.1) |
1.10 (0.67–1.79) |
0.707 |
|
Livestock inside or attached to house |
28 (20.9) |
21 (15.7) |
1.42 (0.76–2.65) |
0.269 |
|
Household member with cough or cold in last 2 weeks |
41 (30.6) |
26 (19.4) |
1.83 (1.04–3.22) |
0.034 |
|
No own toilet |
39 (29.1) |
53 (39.6) |
0.63 (0.38–1.04) |
0.072 |
|
Maternal and socio-economic factors |
|
|
|
|
|
Mother's education primary or below |
46 (34.3) |
43 (32.1) |
1.11 (0.67–1.84) |
0.697 |
|
Socio-economic class IV–V |
43 (32.1) |
46 (34.3) |
0.90 (0.54–1.50) |
0.697 |
|
Fewer than 4 ANC visits‡ |
37 (29.4) |
35 (28.2) |
1.06 (0.61–1.83) |
0.842 |
|
Anaemia during pregnancy |
47 (35.1) |
50 (37.3) |
0.91 (0.55–1.49) |
0.703 |
|
Maternal tobacco use |
11 (8.2) |
8 (6.0) |
1.41 (0.55–3.62) |
0.475 |
|
Handwashing before feeding not always practised |
48 (35.8) |
44 (32.8) |
1.14 (0.69–1.89) |
0.607 |
† Excludes 8 children with recall-only status (cases n = 129, controls n = 131). ‡ Excludes children with missing ANC data (cases n = 126, controls n = 124).
Computed from Severe_Pneumonia_Master.xlsx · 134 cases, 134 controls, Jan 2025–Oct 2026
Multivariable analysis: After adjustment, four factors remained independent determinants of severe pneumonia (Table 3). Incomplete immunisation was the strongest (aOR 4.36; 95% CI 2.20–8.65), followed by recurrent wheeze (aOR 3.40), previous hospitalisation (aOR 3.09) and undernutrition (aOR 2.92). Pallor (aOR 1.92; p = 0.053), diarrhoea in the preceding 2 weeks (aOR 2.29; p = 0.065) and household respiratory illness (aOR 1.79; p = 0.086) showed borderline associations. The model fitted well (Hosmer–Lemeshow p = 0.874; Nagelkerke R² = 0.33). It discriminated reasonably between cases and controls (AUC 0.79), and no multicollinearity was found (all VIF < 1.1).
Table 3. Multivariable logistic regression for determinants of severe pneumonia (n = 260)
|
Determinant |
Adjusted OR |
95% CI |
p-value |
|
Incomplete immunisation for age |
4.36 |
2.20–8.65 |
< 0.001 |
|
History of recurrent wheeze |
3.40 |
1.53–7.59 |
0.003 |
|
Previous hospitalisation |
3.09 |
1.50–6.35 |
0.002 |
|
Any undernutrition |
2.92 |
1.62–5.29 |
< 0.001 |
|
Diarrhoea in the last 2 weeks |
2.29 |
0.95–5.53 |
0.065 |
|
Pallor |
1.92 |
0.99–3.71 |
0.053 |
|
Low birth weight |
1.86 |
0.89–3.90 |
0.101 |
|
Household member with cough or cold in last 2 weeks |
1.79 |
0.92–3.49 |
0.086 |
|
Kutcha or semi-pucca house |
1.25 |
0.69–2.26 |
0.453 |
|
Age 2–11 months |
1.10 |
0.59–2.05 |
0.771 |
|
Male sex |
0.72 |
0.40–1.29 |
0.274 |
|
No own toilet |
0.71 |
0.38–1.31 |
0.268 |
|
Not exclusively breastfed for 6 months |
0.67 |
0.37–1.20 |
0.176 |
Model: likelihood-ratio p < 0.001; Hosmer–Lemeshow χ² = 3.81, p = 0.874; Nagelkerke R² = 0.33; AUC = 0.79.
Multivariable logistic regression, n = 260 · Severe_Pneumonia_Master.xlsx
Care-seeking among cases: Children were ill for a median of 5 days (IQR 3–6) before admission. The first point of care was a private practitioner for 40 (29.9%), this hospital directly for 39 (29.1%), a PHC, sub-centre or ASHA for 31 (23.1%), a pharmacist for 17 (12.7%) and a traditional healer or home remedy for 7 (5.2%). The median time from symptom onset to first medical contact was 37 hours (IQR 16–51); 89 (66.4%) children reached care after more than 24 hours. Before admission, 54 (40.3%) had received antibiotics and 49 (36.6%) were referred from another facility. Among the 104 caregivers who reported a reason for delay, the commonest were failure to recognise severity (32, 30.8%) and distance or transport (29, 27.9%), followed by work or no one to accompany (16, 15.4%) and cost (15, 14.4%).
Cases only, n = 134 · Severe_Pneumonia_Master.xlsx
Clinical profile and outcome of cases: The main qualifying criterion for severe pneumonia was SpO₂ < 90% in 69 (51.5%) children, severe respiratory distress in 32 (23.9%), a general danger sign in 24 (17.9%), central cyanosis in 8 (6.0%) and stridor in 1 (0.7%). Overall, 104 (77.6%) cases had SpO₂ < 90% at admission. Two cases died (case fatality 1.5%) and 4 (3.0%) left against medical advice (Table 4).
Table 4. Clinical profile, investigations and outcome of cases (n = 134)
|
Variable |
Value |
|
Symptoms and signs, n (%) |
|
|
Fast or difficult breathing |
133 (99.3) |
|
Cough |
130 (97.0) |
|
Fever |
111 (82.8) |
|
Chest indrawing |
106 (79.1) |
|
Crepitations or bronchial breathing |
73 (54.5) |
|
Nasal flaring or grunting |
68 (50.7) |
|
Unable to drink or breastfeed |
30 (22.4) |
|
Wheeze on auscultation |
27 (20.1) |
|
Persistent vomiting |
23 (17.2) |
|
Lethargy or unconsciousness |
21 (15.7) |
|
Central cyanosis |
17 (12.7) |
|
Stridor in a calm child |
9 (6.7) |
|
Convulsions |
3 (2.2) |
|
Vital signs at admission, mean ± SD |
|
|
Respiratory rate (breaths/min) |
51.3 ± 9.5 |
|
Heart rate (beats/min) |
134.7 ± 17.7 |
|
Temperature (°F) |
100.8 ± 1.2 |
|
SpO₂ on room air (%) |
88.1 ± 3.0 |
|
SpO₂ < 90%, n (%) |
104 (77.6) |
|
Investigations |
|
|
Haemoglobin (g/dL), mean ± SD |
10.2 ± 1.6 |
|
Haemoglobin < 11 g/dL, n (%) |
88 (65.7) |
|
Total leucocyte count (/µL), median (IQR) |
14,488 (10,935–17,754) |
|
Leucocytosis > 11,000/µL, n (%) |
99 (73.9) |
|
CRP (mg/L), median (IQR) |
36.9 (9.1–58.2) |
|
CRP > 10 mg/L, n (%) |
99 (73.9) |
|
Blood culture done, n (%) |
65 (48.5) |
|
Culture positive, n (% of those done) |
17 (26.2): S. pneumoniae 6, K. pneumoniae 4, S. aureus 2, others 5 |
|
Chest X-ray done, n (%) |
106 (79.1) |
|
Patchy or bronchopneumonia, n (% of X-rays) |
45 (42.5) |
|
Lobar consolidation |
35 (33.0) |
|
Interstitial pattern |
14 (13.2) |
|
Pleural effusion or empyema |
6 (5.7) |
|
Normal |
6 (5.7) |
|
Management and outcome |
|
|
Oxygen therapy, n (%) |
103 (76.9) |
|
Duration of oxygen (days), median (IQR) |
4 (2–5) |
|
CPAP or HFNC, n (%) |
16 (11.9) |
|
Mechanical ventilation, n (%) |
8 (6.0) |
|
PICU admission, n (%) |
31 (23.1) |
|
Hospital stay (days), median (IQR) |
5.4 (3.9–7.2) |
|
Discharged |
127 (94.8) |
|
Left against medical advice |
4 (3.0) |
|
Referred |
1 (0.7) |
|
Died |
2 (1.5) |
In this hospital-based case-control study, four factors independently predicted severe pneumonia in children aged 2–59 months: incomplete immunisation for age, history of recurrent wheeze, previous hospitalisation and undernutrition. Each of these is identifiable at routine child-health contacts, so they offer practical targets for prevention in the district.
Incomplete immunisation was the strongest determinant, with more than a four-fold higher adjusted odds. This agrees with the meta-analysis by Jackson et al., which identified incomplete immunisation as a consistent risk factor for severe acute lower respiratory infection.[4] Broor et al. likewise reported inappropriate immunisation for age as a significant contributor in a Delhi case-control study,[5] and Azad et al. reported a similar association from Kolkata.[12] Measles, pertussis and Haemophilus influenzae type b vaccines and PCV directly prevent major causes of severe pneumonia. Unvaccinated or partially vaccinated children are also likely to have less contact with the health system, which further delays recognition of danger signs. S. pneumoniae was the commonest isolate among culture-positive cases in this study, which supports the importance of full PCV coverage.
Undernutrition nearly tripled the odds of severe pneumonia, and all three anthropometric indices were significantly lower among cases. Undernutrition impairs cell-mediated immunity, mucosal integrity and respiratory muscle strength. Jackson et al. reported a pooled OR of 4.47 for weight-for-age below −2 SD,[4] and Broor et al. found severe malnutrition to be an independent risk factor.[5] Because many controls attended with acute gastroenteritis, a condition that itself shares nutritional risk, the association observed here is, if anything, likely to be conservative. The high prevalence of anaemia among cases (65.7% with Hb < 11 g/dL) and the borderline association with pallor point to a broader micronutrient deficit that warrants attention.
Recurrent wheeze and previous hospitalisation reflect underlying respiratory vulnerability, from reactive airway disease, earlier infections or under-recognised chronic conditions. Mahalanabis et al. found that a history of asthma raised the odds of pneumonia more than five-fold in Kolkata and suggested it as a prognostic indicator for early action.[6] Our findings support this. A child with prior wheezing episodes or a previous admission should be flagged for closer follow-up, early review during any respiratory illness and caregiver counselling on danger signs. Cough or cold in another household member in the preceding 2 weeks was associated on bivariate analysis. Broor et al. reported the same factor as a strong risk,[5] which points to household transmission.
Contrary to our hypothesis and to most published evidence,[4,6–8] biomass fuel use was identical in cases and controls (44.0% each). Dherani et al. estimated a pooled 1.8-fold risk of pneumonia with solid-fuel exposure,[7] and a nationally representative Indian study linked solid-fuel use with both child deaths and non-fatal pneumonia.[8] Several explanations are possible. Controls were drawn from the same hospital catchment and socio-economic stratum, so they shared similar household exposures, which may have produced overmatching. Exposure was also measured crudely, by main fuel type alone. Many households now use LPG together with traditional fuels ("fuel stacking"), which a yes/no question cannot capture. Non-differential misclassification of this kind biases results toward the null. These findings should therefore not be read as evidence that household air pollution is unimportant. Future studies should measure exposure directly, using cooking duration, the child's proximity to the stove and personal monitoring.
The apparent protective association with not exclusively breastfeeding is biologically implausible. It lost significance after adjustment and most likely reflects recall bias or differential reporting by caregivers of hospitalised children. Lack of exclusive breastfeeding remains an established risk factor.[4,5]
Two-thirds of children reached their first medical contact more than 24 hours after symptoms began, and failure to recognise severity was the commonest reason for delay. Nearly one-third first consulted a private practitioner and one in eight a pharmacist, and 40% had received antibiotics before admission. These findings support the SAANS emphasis on caregiver awareness of danger signs and on early assessment and referral by front-line workers.[9] The low case fatality (1.5%) despite a high frequency of hypoxaemia reflects the availability of oxygen, CPAP and PICU care at the study centre. Similar children managed at peripheral facilities without pulse oximetry and oxygen may fare worse.
Incomplete immunisation, undernutrition, recurrent wheeze and previous hospitalisation were the independent determinants of severe pneumonia among under-five children. All of these are modifiable or identifiable early. Delayed recognition of severity by caregivers was common.