Determinants of pneumonia among 2–59 months old children at Debre Markos referral hospital, Northwest Ethiopia: a case-control study

A Hospital based unmatched case-control study design was employed at Debre Markos Referral Hospital, Amhara National Regional State, Northwest Ethiopia. The hospital offers a full range of health care services including child care. It is expected to serve a population of 2,153,937.The hospital has 193 beds used for inpatient services [21]. The study was conducted from February 1 to March 30, 2018.

The source population was all 2–59 months old outpatient department (OPD) visitors in Debre Markos Referral Hospital. A case was defined as a child of age 2–59 months old who has attended in the hospital with a diagnosis of pneumonia within the study period. Diagnosis of pneumonia was determined by the assigned physician using the Federal Democratic Republic of Ethiopia Ministry of Health Integrated Management of Newborn and Childhood Illness (IMNCI) guideline [22] adopted from WHO. A control was defined as a child 2–59 months old who has attended in the Hospital with a diagnosis of non-pneumonia case in the study period. All 2–59 months old children who have attended the OPD of the hospital were included in the study. However, mothers or caregivers of children who were having hearing impairments or unable to talk were excluded from the study.

A sample size of 334 (167 cases and167 controls) was determined using Kelsey et al. formula [23] using the Epi-Info Version 3.5.1 software.

Data were collected using a pretested-interviewer administered structured questionnaire. The questionnaire was prepared first in English and translated in to Amharic, then back to English to check for its consistency. The Amharic version of the questionnaire was used to collect the data. Primary data were collected through face to face interview of the mother/care taker. Information regarding patient’s diagnosis based on assigned physician, patient’s weight/height, weight/age, patient’s malnutrition status (malnourished or not), HIV tested or not, and patient’s sero status (reactive or non-reactive) were collected from patient’s medical records (patient’s charts and registers). Two Nurses (first degree holder) at the hospital collected the data, and one General Practitioner supervised the whole data collection process. The data collectors and supervisor were trained for two days on the study objective and data collection process. The questionnaire was pretested on 5% of the sample size at Finoteselam Hospital, and amendments on the questionnaire were made accordingly. Intensive supervision was done by the supervisor and principal investigator throughout the data collection period.

The dependent variable in this study was Pneumonia. The independent variables were: Socio demographic factors (age of child and mother, sex of child, occupation of mother and father, educational status of mother and father, family size, marital status, age of the mother, residence and ethnicity); Home based and behavioral factors (presence of separated kitchen or not, household near to street, ventilation status of the house, number of windows and opening status, presence of cigarette smoker in the house); Nutritional factors, co- morbidities and care practice (breast feeding status of the child, vitamin A supplementation status, vaccination status of the child appropriate for age, HIV status of the child, history of diarrhea, malnutrition, URTI and co residence with URTI family member).

The data were checked for completeness and consistencies. Data were also cleaned, coded and entered into Epi-Data software version 4.2, then exported to SPSS version 25 statistical package for analysis. The crude odds ratios with 95% confidence interval were estimated in the unadjusted logistic regression analysis to assess the association between each independent variable and the outcome variable. Variables with p-value < 0.25 in the unadjusted logistic regression analysis were considered in the adjusted logistic regression analysis. The Hosmer-Lemeshow goodness-of-fit with enter procedure was used to test for model fitness. Adjusted odds ratios with 95% confidence interval were estimated to assess the strength of the association, and variables with p-value < 0.05 were considered significant factors.

Ethical clearance was obtained from Ethical Review Committee (ERC) of Debre Markos University, College of Health Sciences. Permission to conduct the study was also obtained from Debre Markos Referral Hospital administrators before data collection. Written consent was obtained from mothers after they were informed about the study objectives, expected outcomes, benefits and the risks associated with it. Confidentiality and privacy of every respondent’s information were ensured.

Out of 334 (167 cases and 167 controls) children who approached for interview, a total of 328 (164 cases and 164 controls) were included resulting in a response rate of 98.2%. The mean age of the participants was 25.5 months with a range of 2 to 59 months old. The mean age of cases and controls were 19.7 and 30.8 months respectively. More than 80% of mothers both in cases and controls were above 24 years old. About 94 (57.3%) of the cases and 109 (66.5%) of the controls were residents in the urban areas (Table 1).

The houses for most of the cases (95.7%) and the controls (97.0%) had windows. About 35 (22.3%) of cases and 141(88.7%) of controls had windows opened daily. Fifty two (42.1%) of cases and sixteen (9.8%) of controls had a kitchen attached to the living room. About 82(50.0%) of cases and 31(18.9%) of controls had only water for hand washing. Only 6(3.7%) of cases and 1(0.6%) of controls had cigarette smoking family member (Table 2).

About 152(92.7%) of cases tested for HIV and 3 (2%) of them were sero-positive while from 73(44.5%) controls tested for HIV, 2(2.7%) of them were sero-positive. Ten (6.1%) of cases and 11(6.7%) of controls had malnutrition. About 123(37.5%) of study participants, 107(65.2%) cases and 16(9.8%) controls had history of URTI for the last two weeks duration prior to the data collection period (Table 3).

As shown in Table 4, about 66(40.2%) of cases and 148(90.2%) of controls were breastfed exclusively. About 137(83.5%) of cases and 151(92.1%) of controls were vaccinated.

In the unadjusted analysis, a total of 12 variables were found to full fill the assumption of logistic regression at p-value of less than 0.25 and nominated for further analysis in the adjusted analysis. By using backward logistic regression method, 7 variables (windows daily opening status, household near to street, child care giver most of the time, hand washing material used before child feeding, birth to 6 months breast feeding practice, URTI in the last 2 weeks for the child and a child with history of co- residency with family member who had URTI) had showed significant association with pneumonia at p- value less than 0.05.

The odds of developing pneumonia among children with no windows daily opening was 6.15 times more as compared to those children with windows daily opening [AOR:6.15 (2.55, 14.83)].

The probability of having pneumonia in children with a house near to the street was 4.23 times more as compared to those children with a house not near to the street [AOR:4.23 (1.56,11.44)].

Those children who got child care mostly from the house workers and relatives were 2.97 times more likely to develop pneumonia than those children who mostly got child care from their parents [AOR:2.97 (1.11,7.93)].

The odds of developing pneumonia among children with hand washing before feeding using water only was 3.80 more as compared to those children with hand washing using both water and soap [AOR:3.81 (1.51, 9.66)].

Those children who had mixed feeding were 7.62 more likely to develop pneumonia than those children with exclusive breast feeding from birth to six months [AOR: 7.62(2.97, 19.55)].

The likelihood of developing pneumonia in children who co-reside with a person with URTI was 6.17 times more as compared to their counterparts [6.17 (2.36,16.15)].

Those children who had URTI in the last 2 weeks prior to the study period were 5.33 times more likely to have pneumonia than their counterparts [AOR:5.33 (2.16,13.19)] (Table 5).