This case-control study was conducted from August 2016 to October 2016 in six zonal hospitals and one tertiary hospital of Tigray region. Based on the 2007 census conducted by the Central Statistical Agency (CSA) of Ethiopia, the region has an estimated total population of 4,314,456 (2,124,853 males and 2,189,603 females), urban inhabitants account for 842,723 (19.53%) of the population. With an estimated area of 50,078.64 square kilometers, the region has an estimated density of 86.15 people per square kilometer. The region is predominantly Tegaru people with 96.55% of the population. In the region there are 712 health posts, 201 health centres and 15 hospitals (six zonal or general hospitals, one referral/tertiary and the remaining are primary hospitals). There are 3, 4, 77, 60, and 50 hospitals, health centers, medium clinics, primary clinics and specialty clinic respectively, owned by private and nongovernmental organisations [11,12].

The study design was unmatched case control. Cases included patients of CAP (having a new pulmonary infiltrate on chest radiograph plus at least one of cough, fever, hypothermia, leukocytosis, or leukopenia).

To label the cases of the study two internists in consultation with a radiologist were appointed, after history taking and diagnostic procedures like sputum, blood and X-ray examination. Controls were patients who visited the hospitals for any other reason but without CAP.

All CAP patients aged 18 years and above, in both males and females, receiving treatment during the data collection period in the government zonal hospitals and Ayder Referral Hospital were included in the study. Those patients who had history of hospital admission 14 days before the data collection period, patients who developed pneumonia 48 hours following admission, patients with Tuberculosis (TB) or previous chest X-ray which might conflict with diagnosis of CAP, chronically debilitated patients and patients with lung cancer and asthma were excluded from the study.

The sample size for the study was calculated using two proportion formula. Assuming, 26% exposure to smoking has twice risk (OR=2) for CAP, with power of 80% and significance level of 5%, a case to control ratio of 1 to 2, the required sample size was found to be 120 cases and 240 controls [13].

A semi-structured, interviewer-administered questionnaire was adapted from different literatures and organisations [14-26].

The questionnaire included structured information on the sociodemographic characteristics, life styles and habits, medical history and environmental conditions of both cases and controls. The questionnaire was prepared in English and translated into Tigrigna, and back-translated to ensure consistency in phrasing of questions. The face and content validity of the questionnaire were done by different experts in the area.

Moreover, to assure the quality of the research the adapted questionnaire was prepared using a simple and easily understandable language. Standardisation on translation was given emphasis during training. A pre-test was conducted on 10% of the study subjects to evaluate the completeness and consistency of the tools. Moreover, the stability of the measures, internal consistency of the measurement instrument, and inter-rater reliability of the instrument were assessed to maintain the higher levels of reliability of the study. Accordingly, appropriate modification and corrections were made.

Data collectors and supervisors involved were nurses and health officers. Training was given for the data collectors and supervisors for three days on the objective of the study, data collection tools and procedures to ensure consistency of interviewing and high-quality data.

At the end of every data collection day, each questionnaire was examined and necessary feedback was given to the data collectors and supervisor. Data entry was carried out using Epi info 2002 by an experienced data entry clerk with close supervision by the principal investigator. Data cleaning was done by the principal investigator and the finalised questionnaire was kept locked in a secured cabinet.

Ethical clearance was obtained from Institutional Review Board (IRB) of Addis-Ababa University, College of Health Sciences. Letter of agreement was secured from Tigray regional health bureau. All participants were informed about the purpose of the study and an individual written informed consent was solicited from the selected respondents at the time of data collection, the data collectors and the principal investigator were blinded about the cases and controls of CAP, only the physicians in charge were aware of the distribution because they were to provide all necessary management in relation to CAP. Respondents were not identified by name; they were also informed that they had the right of participation and to withdraw any time they like during the study. Furthermore, health education about prevention and treatment of CAP had been provided.

After cleaning the data, it was exported from Epi info 2002 to SPSS version 20.0 for analysis. Frequency tables, graphs and proportions were used to present the results. In addition, the associations between the exposure and outcome variables were determined using bivariable and multivariate logistic regression analysis. Odds ratios with 95% confidence intervals and p-value were calculated to measure the strength of associations.

Among the sociodemographic variables, age and education are significantly associated factors in the two sample t-test analysis and univariate logistic regression analysis: Mean difference is observed among the cases and controls based on their age, t(df), p, (t(358)= -2.2, p=0.03). Moreover, those with no formal educational status were 1.8 times more likely to develop community acquired pneumonia as compared to those who are educated, OR (95% CI), 1.8 (1.1,2.8) [Table/Fig-1].

None of the lifestyle or habit related variables were associated with CAP in the univariate logistic regression analysis [Table/Fig-2].

According to the univariate logistic regression analysis, those who had upper air way problem were two times more likely to develop CAP as compared to their counter parts, COR (95% CI), 2.1 (1.3,3.3). Likewise, the odds of developing CAP among those who had history of tonsillectomy was 3.4 as compared to those who did not have same history, COR (95% CI), 3.4 (1.1,10.5). On the other hand, those who had contact with persons who had respiratory infection were 2.6 times more likely to develop CAP as compared to their counterparts, COR (95% CI), 2.6 (1.5,4.6). Moreover, the odds of developing CAP among those who had history of confirmed pneumonia was 34 as compared to those who did not have confirmed pneumonia COR (95% CI), 34 (17.9, 64.6). Those who had history of pulmonary tuberculosis were two times more likely to develop CAP as compared with those who did not have same history, COR (95% CI), 2.2 (1.1,4.4) [Table/Fig-3].

Those who had history of contact with pets were 1.7 times more likely to develop CAP as compared to their counter parts, OR (95% CI),1.7 (1.1, 2.7). Likewise, the odds of developing CAP among those who had history of working in dusty environment was 1.8 as compared to those we didn’t have same history, COR (95% CI),1.8 (1.2,2.8) [Table/Fig-4].

In the multivariable logistic regression model, we included variables that were significantly associated with CAP at the univariate analysis, and found that only four of the variables were having statistically significant association with CAP. As shown in [Table/Fig-5], the odds of developing CAP among those working in a dusty environment was two times higher as compared to their counterparts {AOR=2.0, 95% CI 1.1-4.1)}, the odds of developing CAP among those who had history of contact with people who had respiratory infection was 2.5 times higher as compared to those who did not have history of contact {AOR=2.5, 95% CI 1. 2-5.3}, those who had history of respiratory infection were about twice more likely to develop community acquired pneumonia than those who did not have the same history {AOR=2.3, 95% CI 1.5-5.7} and the odds of developing pneumonia among those who had history of pneumonia confirmed by radiography was 39 times higher compared to those who had not {AOR=39, 95% CI 19.4-78.6} [Table/Fig-5].

The study has attempted to determine potential risk factors for CAP in the hospital setting of Tigray region in Ethiopia.

Present finding showed no significant association between smoking and CAP, while other studies conducted in different countries reported that smoking as one risk factor for the development of community acquired pneumonia [14-20]. Perhaps, only few (n=19) of present study participants reported to have been smokers which may be a small number to give a significant information.

Although, some studies have reported that alcohol consumption as being a risk factor for developing CAP, present study showed no significant association between alcohol consumption and CAP [20-22]. Similar to present study, there are findings that revealed alcohol consumption is not a risk factor for CAP [14,15,18,19]. This could be justified by difference in the number of alcohol users, type, amount and frequency of the alcohol consumption.

Having a history of respiratory infections and contact with people who had respiratory infection were found to be risk factors for the development of CAP which was in line with Almirall J et al., Almirall J et al., Teepe J et al., Antoni T et al., Rose RM et al., and Schoor M et al., [14,17,18,20,23,24]. Similar to previous findings by Almirall J et al., Almirall J et al., Shcnoor M et al., Hedlund J et al. and Hedlund J, the current study showed that patients with a history of pneumonia confirmed by radiograph had higher risk of a subsequent CAP [14,17,24-26].

In line with a study from Great Britain, working in a dusty environment was identified in present study as one of the major risk factors of community acquired pneumonia [15]. However, contrary results were also reported by others [17]. This dissimilarity could be because of different working environment, difference in exposure and the nature of the dust.

History of diabetes and heart disease were not significantly associated with CAP in this study, which was also reported by many other studies [14-16,21], however, other studies from different countries reported that heart disease and diabetes being significant risk factors for CAP [15,17,20]. These differences may be due to that many of our respondents might have not known their status or not diagnosed for diabetes and heart diseases.

Chronic bronchitis, diagnosed asthma and pulmonary TB were not associated with CAP in the current study as was also reported by others [14,16]. However, findings by Almirall J et al., Farr BM et al., Almirall J et al., and Teepe J et al., showed that chronic bronchitis, diagnosed asthma and pulmonary TB are risk factors for community acquired pneumonia among adults [14,15,17,18].

Similar to a finding by Schnoor M et al., contact with birds and pets were not significantly associated with CAP when adjusted by other variables in the present study [24], but Almirall J et al., reported that contact with birds and pets as being a risk factor for the development of CAP [17].

Since, the present study is hospital based it might not be generalised to the whole community in the region. In addition, there might be selection bias as the cases in the study had been enrolled continuously until the required sample size was met. Moreover, there might be a recall bias of the exposures in the controls as the cases might be able to recall the exposures in a better way.

In conclusion, working in a dusty environment, having a history of pneumonia, history of respiratory infection and having contact with people who had respiratory infections are confirmed as the risk factors of community acquired pneumonia in the current study. Hence, much has to be done to prevent community acquired pneumonia through health education and awareness raising interventions. On the other hand, to minimise the risk of developing community acquired pneumonia safety measures like personal protective equipments should be used when there is contact with patients having respiratory tract infections.

Whenever respiratory tract infections including pneumonia occur, proper management using standardised treatment guideline should be given so that subsequent infections or complications could be minimised. Moreover, larger studies are also needed to assess the effect of some risk factors in the general population.