Klebsiella pneumoniae Oropharyngeal Carriage in Rural and Urban Vietnam and the Effect of Alcohol Consumption

Introduction Community acquired K. pneumoniae pneumonia is still common in Asia and is reportedly associated with alcohol use. Oropharyngeal carriage of K. pneumoniae could potentially play a role in the pathogenesis of K. pneumoniae pneumonia. However, little is known regarding K. pneumoniae oropharyngeal carriage rates and risk factors. This population-based cross-sectional study explores the association of a variety of demographic and socioeconomic factors, as well as alcohol consumption with oropharyngeal carriage of K. pneumoniae in Vietnam. Methods and Findings 1029 subjects were selected randomly from age, sex, and urban and rural strata. An additional 613 adult men from a rural environment were recruited and analyzed separately to determine the effects of alcohol consumption. Demographic, socioeconomic, and oropharyngeal carriage data was acquired for each subject. The overall carriage rate of K. pneumoniae was 14.1% (145/1029, 95% CI 12.0%–16.2%). By stepwise logistic regression, K. pneumoniae carriage was found to be independently associated with age (OR 1.03, 95% CI 1.02–1.04), smoking (OR 1.9, 95% CI 1.3–2.9), rural living location (OR 1.6, 95% CI 1.1–2.4), and level of weekly alcohol consumption (OR 1.7, 95% CI 1.04–2.8). Conclusion Moderate to heavy weekly alcohol consumption, old age, smoking, and living in a rural location are all found to be associated with an increased risk of K. pneumoniae carriage in Vietnamese communities. Whether K. pneumoniae carriage is a risk factor for pneumonia needs to be elucidated.


Introduction
Klebsiella pneumoniae is a gram-negative pathogen that is a common cause of both community and hospital acquired pneumonia, bacteraemia, and liver abscess in Asia [1,2,3]. Historically, K. pneumoniae has been identified as a cause of severe community acquired pneumonia, known as the Friedlander syndrome, and has been found to be particularly common in alcoholics and diabetics [1,2,4,5]. Its pathogenesis is poorly understood. Alcoholics, the elderly, and diabetics can have an altered mucosal immunity, resulting in a change in the bacterial flora on their mucosal surfaces [5,6,7,8,9]. The finding that K. pneumoniae pharyngeal carriage rates are higher in elderly in China is likely an illustration of this phenomenon [10]. It is therefore possible that factors leading to altered mucosal immunity can affect K. pneumoniae colonization rates of the oropharynx. We hypothesize that oropharyngeal carriage precedes development of K. pneumoniae pneumonia, and that host factors, such as age and alcohol consumption, are significantly associated with carriage.
Although there have been previous studies on colonization of Streptococcus pneumoniae and S. aureus in relation to subsequent pulmonary infections [11,12,13], very little is known regarding oropharyngeal colonization of K. pneumoniae and other gramnegative bacilli (GNB). This study investigates pharyngeal K. pneumoniae carriage in urban versus rural Vietnamese communities and the effects of demographic and socioeconomic variables, including age, smoking habits, occupation, wealth index, history of past diseases, and alcohol consumption habits.

Study Sites
Ba Vi is a rural community near Hanoi City with a population of approximately 240,000. Longitudinal data collection has been performed at a demographic surveillance site (DSS) in Ba Vi (Field Laboratory Ba Vi -FilaBavi) since 1999 in 69 randomly selected clusters with about 11,000 households and 51,000 persons. Ba Vi has one district hospital and one health center in each of 32 communes [14].
Dong Da is an urban district in Hanoi with a population of about 352,000 and socioeconomic characteristics typical for urban Hanoi. An urban DSS was started in 2007 (DodaLab). The study population is 40,000 persons in 11,000 households in three selected communes with low, middle, and high socioeconomic levels. Dong Da has one district hospital and four commune health centers [14].

Subjects
This is a cross-sectional study, in which oropharyngeal swabs were collected for bacterial culture from quota-sampled subjects from rural (Ba Vi) and urban (Dong Da) Hanoi, located in North Viet Nam. All eligible subjects (persons enrolled in the DSS) were stratified by gender and the following age categories: 1-5 years, 6-12 years, 13-19 years, 20-29 years, 30-59 years, and . = 60 years. Subjects within each age and gender category were randomly selected. As alcohol consumption is limited in women and children, we decided to only assess the role of alcohol in a separate cohort of men (19-71 years) from the Ba Vi community. Participants completed a questionnaire to identify variables possibly associated with K. pneumoniae carriage (Questionnaire S1). Demographic data such as age, sex, ethnicity, history of past health issues, smoking habits, daily medication intake, and antibiotic intake were recorded for each patient. Socioeconomic information including occupation, level of education, household size, and socioeconomic status was also recorded (Questionnaire S1).

Sample Size
We aimed to enroll 2170 participants (female to male ratio of 1:1) from the DoDa Lab and Fila Ba Vi DSS's -the main cohort. An additional 850 males from Ba Vi were to be randomly selected from the list of DSS participants -the alcohol cohort. This sample size would allow us to estimate the prevalence rates of K. pneumoniae carriage across different age groups, sex, and socio-economic factors. To assess the impact of alcohol on carriage rates, heavy and moderate alcohol drinkers are compared to less-heavy drinkers. Previous data suggest that in rural Viet Nam, 24% of men are harmful drinkers, as determined by the AUDIT test developed by the WHO [15]. We estimated that a target population of 850 men would contain enough alcohol consumers to allow for meaningful analysis. Due to a higher than expected amount of enrolled patients electing to not participate, only 1029 participants were successfully included in the main cohort and 613 men were included in the alcohol cohort.

Microbiological Methods
Separate swabs were collected for nose and throat, using sterile commercial Dacron swabs with bacteria transport medium (Copan, Italy). Nose swabs were collected from the anterior nares. Throat swabs were collected by firmly swabbing the entire posterior pharynx and tonsillar area and leaving in place for 5-10 seconds to absorb secretions, then withdrawing the swab. All swabs were transported on the same day to the microbiology laboratory of the National Hospital of Tropical Diseases in Hanoi. Within a day of arrival, the swabs were plated on MacConkey agar (Oxoid, USA). Based on morphology and Gram staining, suspected K. pneumoniae strains and other enterobacteriaceae were identified using API20E biochemistry strips (BioMerieux, France). All morphological unique colonies were identified and tested for antibiotic susceptibility using standard microbiological techniques.
No selective media were used to enhance detection of certain resistance phenotypes.

Statistical Analysis
K. pneumoniae carriage was defined as carrying K. pneumoniae either in nose or throat. Regarding analysis of the impact of education level: Some of the subjects had been educated according to Viet Nam's old 10-year school system. In this case, their level of education was scaled to fit the current 12-year school system. Subjects with secondary vocational education were considered to have 10 years of education, subjects with college education were considered to have 14 years of education and those with university education were considered as having 16 years of education. If only literacy information was provided, illiterate subjects were considered as having one year of education, and literate subjects were considered as having two years of education. Wealth index is calculated by principle component analysis using of a set of correlated variables describing housing conditions (type of house, area, sanitation, water source, etc.) and household ownership assets (car, motobike, telephone, etc.).
Univariate analysis was done using Pearson's chi-squared test or univariate logistic regression to determine p-values. Age and level of alcohol consumption were treated as continuous variables. All other variables were treated as categorical variables. For variables with a low number of positive cases (,5), Fisher's exact test was used instead of the chi-squared test. Potential determinants of K. pneumoniae carriage were first considered in a univariate analysis. Variables reaching a univariate significance level of p,0.1 were included in multivariate analysis. Multivariate analysis was done through backward-stepwise regression using a binomial model with a logit link function. P-values ,0.05 in multivariable analysis were considered to be significant, and odds ratios and confidence intervals for those variables were calculated. The alcohol cohort (extra 613 men from Ba Vi) was analyzed separately from the main cohort. The estimated population carriage rate is calculated by direct age-adjusted standardization using the population distribution of Viet Nam in 2011 (CIA World Factbook). All analysis was done using RStudio (RStudio, Boston, Massachusetts, USA [16].

Ethics Statement
This study was approved by the ethical committees of Oxford University (Oxtrec), U.K. and Hanoi Medical University, Vietnam. Before participation, written informed consent from subjects or, in case of minors, their caregivers, was obtained on a standard study consent form. The consent process was approved by both ethical committees.

Results
The number of subjects recruited for the main cohort was 1029 (53.8% male, median age: 36 years, IQR:13-53 years), with an additional 613 men from Ba Vi (median age: 47 years, IQR: 39-52) recruited for the alcohol cohort. The subjects in the main cohort were recruited somewhat equally from the original age, sex, and site strata ( Table 1). Subjects that were selected but did not agree to participate were 52% male with a median age of 27 years. Young males tended to have lower participation rates, introducing a potential source of bias in the population.
Univariate analysis of the main cohort suggested that age (p, 0.001), years of education (p,0.05), living location (urban or rural) (p,0.05), smoking (p,0.001), sex (p,0.01), history of renal diseases (p,0.01), diabetes (p,0.05), working as a farmer (p, 0.05), working as a hired labourer (p,0.05), and wealth index (p, 0.10) were possible candidates for being independently associated with K. pneumoniae carriage ( Table 2). These variables were then included in backwards-stepwise linear regression until a minimal set of significant factors was achieved. Antibiotic use was also considered to determine its effects as a confounding variable. This was treated as a binary variable, and patients were distinguished based on whether they had reported taking antibiotics in the past four weeks. Antibiotic use was not found to be meaningful as a confounding variable. In the main cohort, 27.2% (280/1029) of subjects reported taking antibiotics in the past four weeks. Years of education was independently significant in subjects over the age of 18 (OR 0.93, 95% CI 0.87-0.99), but was found to be a confounding variable associated with age, as is explained below. The minimal set of significant factors was age (OR 1.03, 95% CI 1.02-1.04), smoking (OR 1.9, 95% CI 1.3-2.9), and living location (OR 1.6, 95% CI 1.1-2.4).
A one-year increase in age was found to be correlated with a 3% increase in odds of K. pneumoniae carriage (Figure 1). Carriage rates grew fairly steadily with age, from 1.8% carriage in subjects less than 10 years old to 32.5% carriage in subjects aged 50-54, but dropped off in subjects over age 54 (Figure 1).
In the main cohort, there were 174 smokers, 847 non-smokers, and 8 people that did not report smoking habits. Smokers were defined as subjects answering ''yes'' to the question ''Do you smoke?''. The carriage rate in smokers was 26.4% (95% CI 19.8%-33.1%), compared to just 11.6% (95% CI 9.4%-13.7%) in non-smokers (p,0.01). Multivariable analysis indicated that smoking was independently associated with a 90% increase in odds of K. pneumonia carriage. The number of years of education for each subject was only considered in subjects over the age of 18 to avoid introducing an age-dependent ceiling on education level. An increase in level of education of one year seemed to be associated with a decrease of 7% in odds of K. pneumoniae carriage. However, when examining the effect of education in age strata of subjects over 50 years of age and subjects between the age of 18 and 50, this effect was no longer present. Additionally, the AIC value for the model without level of education was close enough to the one with (367.06 vs. 366.23 for ages 18-50, 299.08 vs. 297.61 for ages .50) to give reasonable evidence for level of education as a confounding variable.
There were a number of variables that were positive with very low frequency (, = 5%), making their significance difficult to assess. These were diabetes (2.0%), history of heart disease (5.0%), malignancy (1.1%), HIV (0%), and past splenectomy (0.2%). Alcohol consumption information was only collected for subjects in the alcohol cohort, so we cannot assess the extent by which alcohol consumption may have confounded the results found in the main cohort. The cross-referenced frequencies of a number of variables are displayed in Table S1.

The Effect of Alcohol Use
The overall carriage rate of K. pneumoniae in the 613 adult men recruited to assess the effect of alcohol consumption was 28.1% (172/613, 95% CI 24.5%-31.6%). Information gathered from subjects in the alcohol cohort included type of alcohol consumed, percent alcohol content, quantity consumed per day, and days per week consumed. These categories were used to calculate the number of litres of pure alcohol consumed per week for each subject (APW), meaning that the alcohol content of the drink each subject consumed was taken into account. The median APW was 0.25 litres (range 0.0028-6.72 litres). Light drinkers were determined as subjects with APW , = 0.2, moderate drinkers had 0.2, APW , = 0.5, and heavy drinkers had APW .0.5. Using this metric, there were 235 (38.3%) light drinkers, 230 (37.5%) moderate drinkers, and 132 (21.5%) heavy drinkers. 16 (2.6%) subjects did not provide enough alcohol consumption data to calculate weekly consumption. The carriage rate was 23.0%  (Figure 2).
Univariate regression analysis suggested that age (p,0.05), alcohol consumption per week (p,0.05), smoking (p,0.10), and wealth index (p,0.10) were possible significant indicators for carriage in the alcohol cohort. Alcohol consumption per week was analyzed as a continuous variable. Multivariable analysis con- independently associated with carriage in this cohort. Consuming one more litre of alcohol per week was independently correlated with an increase of 70% in odds of K. pneumoniae carriage ( Figure 2). There were just 10 (1.6%) subjects that consumed beer on a weekly basis but not hard alcohol, so the effect of the types of drinks consumed was difficult to assess. The alcohol cohort allowed us to reject the null hypothesis that K. pneumoniae carriage rates are equal regardless of level of alcohol consumption (p, 0.05). Frequencies of K. pneumoniae carriage cross-tabulated with alcohol consumption information are reported in Table S1.

Discussion
This is one of a few studies that examines oropharyngeal carriage of K. pneumoniae in community members, particularly in lower-middle income countries and Southeast Asia. The correlation between alcohol consumption and K. pneumoniae carriage has also received little prior attention. The K. pneumoniae oropharyngeal carriage rate of 14.1% is comparable to the nasopharyngeal carriage rate of 11% determined by a recent study in Indonesia [17]. The rate of carriage of 1.6% (95% CI 0-5.1%) in children is also comparable to a study that reported a carriage rate of 1.4% amongst children aged less than 5 years attending daycare in Brazil, an upper-middle income country (The World Bank 2013) [18].
In the past, K. pneumoniae carriage has been strongly linked to hospital environment-related factors such as length of hospital stay, number of artificial interventions, and duration of interven-tions [19]. It is commonly transmitted as a hospital-acquired pathogen, often displaying high antibacterial resistance [20]. This study aimed to identify factors associated with community based nasopharyngeal carriage of K. pneumoniae pneumoniae. Antibiotic use was not related to carriage which we cannot explain. Antibiotic use in the community was high, but comparable to previous studies that assessed community antibiotic use in Vietnam [21,22,23]. Interestingly, we found two cases who carried carbapenem resistant K. pneumoniae strains, which were found to be NDM-1 positive. One case was from the rural area and the other urban. We were unable to determine the source of these resistant bacteria. The ESBL-rate appears low and maybe due to the fact no selective media were used in this study.
Multiple regression analysis indicated a predicted increase in odds ratio of 3% associated with an increase in age of one year. Decreasing mucosal immunity with aging may be one mechanism underlying the association between K. pneumoniae carriage and aging. Similarly, a study conducted in China found that gramnegative bacilli carriage rates were relatively high in the elderly [10]. Carriage rates generally peaked in subjects aged 50-54 (28.0% carriage, 95% CI 17.6%-38.4%) (Figure 1). However, carriage rates in subjects over the age of 65 are found to be 23.9% (95% CI: 15.9%-31.9%), a decline from the rate of carriage in 50-54 year old subjects. Nevertheless, this carriage rate in the elderly matches that of the oropharyngeal carriage study done in China, which found 158 strains of K. pneumoniae in 706 subjects .65 years old (22.4% carriage) [10]. The reason for the observed decrease in K. pneumoniae carriage rates in the elderly is unknown, although it is possible that a much greater proportion of elderly carriers develop pneumonia and die, leading to an apparent decrease in carriage.
The assessment of alcohol consumption in adult men indicated that weekly alcohol consumption was positively associated with K. pneumoniae carriage (OR 1.7, 95% CI 1.04-2.8) (Figure 2). A prior study has found that, in Taiwan and South Africa, communityacquired bacteremic K. pneumoniae pneumonia is significantly associated with alcoholism [2]. Although alcoholism has been correlated with the incidence of pneumonia caused by Klebsiella, prior to now, very little was known about the effects of alcohol consumption on oropharyngeal colonization by K. pneumoniae. A likely mechanism for contracting pneumonia is aspiration of K. pneumoniae from the pharyngeal mucosa into the lung. Thus, the positive relationship between carriage rates and weekly alcohol consumption suggests that a potential mechanism for the increased susceptibility of alcoholics to pneumonia is the effect that alcohol has on altering the composition of oropharyngeal mucosa. This causal analysis is beyond the scope of the present study, but is worth exploring further. A prior study by Shellito et al. has identified Interleukin-17 as an important cytokine in immune response to bacterial pneumonia. The study demonstrates that ethanol consumption in mice suppresses IL-17 release into lung tissue. It suggests that IL-17 is an important part of the immunosuppression associated with alcohol abuse [5]. Another study by Zisman et al. has identified IL-12 as another cytokine that is suppressed in the lungs of mice by alcohol consumption. They have also noticed a decreased survival rate in mice with K. pneumoniae that were given alcohol regularly [9].
The study also found that being a smoker and living in Ba Vi (rural) were associated with an increase in risk of carriage. A recent study in Indonesia identified smoke exposure from cigarettes and irritants such as mosquito coils as being significantly associated with K. pneumoniae carriage. The study also found poor food and poor water hygiene as factors significantly associated with K. pneumoniae carriage [17]. It is reasonable that the increase in carriage rate associated with living in Ba Vi is attributable to inferior sanitation conditions. The study did not gather hygiene information, although this is a possible direction for future research. These results may be somewhat confounded by alcohol consumption. However, as alcohol consumption data was not collected for subjects in the main cohort, it is unclear whether this is the case.
These findings are a first step in elucidating potential pathways leading to community acquired pneumonia with K. pneumoniae. Prospective cohort studies are needed to test the hypothesis whether K. pnaeumoniae carriage is a risk for developing pneumonia with K. pneumoniae.
The results allowed us to reject the null hypothesis that risk of Klebsiella pneumoniae oropharyngeal carriage was independent of a variety of demographic and socioeconomic factors. Age (p,0.001), living location (p,0.05), smoking (p,0.01), and weekly alcohol consumption (p,0.05) were all found to be significantly correlated with rate of K. pneumoniae carriage.

Conclusion
Oropharyngeal colonization by K. pneumoniae is common among healthy people. Moderate to heavy weekly alcohol consumption, old age, smoking, and living in a rural location are all found to be associated with an increased risk of K. pneumoniae carriage in Vietnamese communities. Whether K. pneumoniae carriage is a risk factor for pneumonia needs to be elucidated.

Supporting Information
Table S1 Cross tabulated frequencies of several variables, including K. pneumoniae carriage. K. pneumoniae carriage frequencies are displayed, categorized by age group, smoking habits, living location, level of education, and level of alcohol consumption. Demographic information including sex, smoking habits, and living location frequencies are also displayed. Subjects are from Dong Da (urban) and Ba Vi (rural), locations in Northern Viet Nam near Ha Noi. Age, smoking, living location, level of education, and level of alcohol consumption were all found to be significantly correlated with carriage. Subjects in the ''Education'' rows are all over the age of 18 to avoid introducing an age-dependent ceiling on level of education. Subjects in the ''Alcohol Consumption'' rows are from the separate cohort of 613 adult men from Ba Vi.

(TIF)
Questionnaire S1 Original questionnaire used to acquire socioeconomic and demographic information from study subjects. This questionnaire was orally given to all subjects from Ba Vi and Dong Da that were recruited for the study. A trained interviewer posed the questions and recorded the responses. The questionnaire was designed to obtain information about a wide range of socioeconomic and demographic variables, allowing for more accurate assessment of associations between variables and risk of K. pneumoniae carriage. (PDF)