published_or_final_version / Community Medicine / Master / Master of Public Health

Background There have been many multicity studies assessing health effects of the population’s exposure to PM10. They assume that there is homogeneous exposure to PM10 and the monitoring locations are representative of the population’s exposure. In a densely urban environment, like Hong Kong, street canyons are common and could alter PM10 exposure. The horizontal and vertical dispersion of PM10 in the urban environment is complex due to the interactions of street geometric, metrological and pollution source conditions. The airGIS/OSPM is a box and plume mathematical model which was designed to model this complex dispersion of PM10. Objectives This study aims to utilize the airGIS/OSPM to assess how the health impact is affected by the misclassifications of within city air pollution exposure. This helps assessing the feasibility of using the airGIS/OSPM in health studies. Methods The airGIS/OSPM was used for an urban area of Hong Kong, in the north west of Kowloon Peninsular for a period from 1998 to spring 2011. PM10 concentration’s were estimated at field measurement points, an EPD roadside monitoring site, and subjects in a subset of the elderly healthcare center cohort and for all building addresses in the modeled area. The airGIS/OSPM was validated by conducting a PM10 measurement campaign over the winter of 2010 to 2011. Also validation was conducted using the measurements of the Mong Kok EPD roadside monitoring station. A pilot study using the time stratified case-crossover analysis was conducted to explore the effects of using the airGIS/OSPM to express the subjects’ short-term outdoor residential exposure to PM10 on all cause mortality. AirGIS/OSPM PM10 estimates were compared with ambient concentrations obtained from the EPD monitoring network. The population mean exposure was calculated using the airGIS/OSPM estimated PM10 concentration at all building address points from 2007 to 2009 at the ground level and at middle building height level. The airGIS/OSPM estimate was used to identify spatial variation of PM10 within the study area. Results The airGIS/OSPM estimated well the measured PM10 concentration from the field measurement campaign and the EPD Mong Kok station. The airGIS/OSPM estimate and the ambient measure for EHC subject exposure both found an odds ratio for all cause mortality there was no difference from unity between case and control times. The airGIS/OSPM derived mean PM10 concentration at the middle height of each building was 54.8μg/m3 while that at the all EPD was 54.4μg/m3. At ground level the airGIS/OSPM PM10 estimate was 58.35μg/m3 while that at the EPD was 54.41μg/m3. Conclusion The misclassification of PM10 was negligible at the middle of buildings, but for people regularly working in ground level microenvironments are often exposed to PM10 concentrations that are higher than those measured at EPD monitors. It is feasible to use the airGIS/OSPM model to estimate PM10 exposure. The small spatial variation in exposure means the airGIS/OSPM may not be appropriate in assessing the short-term PM10 exposure, but due to the larger effect size it might be important in long-term exposure assessment. / published_or_final_version / Community Medicine / Master / Master of Philosophy

Background Night shift work is inevitable for maintaining continuous services in different sectors e.g. healthcare, financial, transport and service sectors. Night shift work increases exposure of light at night. Exposure of light at night suppresses production of a neurohormone melatonin. Melatonin has shown potential cancer protective action in animal experiments. Melatonin deficiency is suggested to be related to the development of various cancer especially breast cancer. Breast cancer incidence in Hong Kong is rising. Particular concern about association between night shift work and breast cancer is raised. Objective To collect evidence from studies of other countries with study populations of different professions and to evaluate the relationship between night shift work and breast cancer Method MEDLINE was searched to identify publications, limited to English articles, from 1973 to May 2011. Search terms include (circadian rhythm or night work or night shift or shift work or work schedule tolerance) and (cancer or neoplasm or neoplasia) and (risk or rate or incidence). No restriction was set to the publication type. Results Altogether 343 titles retrieved from MEDLINE search. Finally, 8 primary observational studies that met inclusion criteria were identified for this review. Of these, two were prospective cohort studies, one was retrospective cohort study, two were nested case-control studies and three were case-control studies. Most of the study had crude exposure assessment of night shift work, in which four studies relied on group level of exposure probability instead of individual exposure information. Six of eight studies showed positive results on the association of night shift work and breast cancer in women. Three studies found that risk of breast cancer was increased significantly for those who had engaged in night shift work in a long duration i.e. more than 20-30 years, but they were all conducted in populations of same occupational group i.e. nurse and only a moderate increase of breast cancer risk was found. The results were subject to confounding and bias. No consistent results were found for effect of shorter duration of night shift work on risk of breast cancer. Conclusion Based on the studies included, there is suggestive evidence of an association of night shift work and breast cancer. Further studies on this are needed. Involvement of population of different occupational groups, controlling confounder of hormone use and conducting exposure assessment with high reliability using individual information instead of that from group are suggested. / published_or_final_version / Community Medicine / Master / Master of Public Health

Background Vitamin D deficiency is prevalent in many places across the world. Breastfeeding has been suggested to be a significant predictor of vitamin D deficiency during infancy, which is preventable through proper supplementation. However, whether Hong Kong should adopt the international recommendation for supplementation of exclusive breastfed infants is still not yet answered. Objective: To review the available evidence regarding the association between breastfeeding and vitamin D deficiency during infancy, as well as the effectiveness of vitamin D supplementation of breastfed infants as an intervention measure. Methods A total of 5112 potentially relevant articles were searched and identified from MEDLINE (OVID, Pubmed), Science Citation Index Expanded (ISI Web of Science), Cochrane Central Register of Controlled Trials, EMBASE, and Cumulative Index to Nursing and Allied Health Literature (CINAHL) and Chinese database (CNKI) without restriction from inception to July-06-2012. 5065 articles were excluded after the initial scanning of title and abstracts. 36 were subsequently excluded due to methodological issues. A total of 11 studies were included and reviewed by two independent reviewers. Results This review pooled together a total of 1126 exclusively breast-fed infants for less than one year old from 11 studies. The pooled average prevalence of vitamin D deficiency was 54.2%. The association between breastfeeding and vitamin D deficiency during infancy has been reported consistently, and it was found to be one of the strongest predictors of vitamin D deficiency for infants less than one year old. Sunlight exposure, season, and skin pigmentation were also found to be important affecting factors. Supplementation to breastfed infants with the dosages as recommended by American Academy of Pediatrics (AAP) seem to be effective in lifting up the vitamin D levels. Conclusions The Hong Kong Government and relevant health sectors should conduct local epidemiological study to investigate the problem of concern among our breast-fed infants, and seriously consider or evaluate the AAP recommendation of supplementation. / published_or_final_version / Public Health / Master / Master of Public Health

Attention-deficit hyperactivity disorder (ADHD) is a developmental disorder of inattention, impulsivity and hyperactivity. It is the most commonly diagnosed psychosocial problems in childhood with continuing impact into adulthood. This systematic review aims to identify the role of food additives, specifically artificial food coloring, in ADHD and its symptoms. PubMed, Sciencdirect, Cochrane Library and Google Scholar databases (1991-April 2012) were searched, which was supplemented by manual searches of the reference lists. Seven randomized or quasicontrolled trials of elimination diet were summarized. Six of these studies showed artificial food coloring was positively associated with ADHD and its symptoms. This review provides some evidence that artificial food coloring plays a role in ADHD and its symptoms. However, the studies reviewed only showed behavioral improvement after eliminating certain type of artificial food coloring, which does not imply artificial food colorings cause ADHD. Additionally, small sample size, inconsistent definition of artificial food colouring, subjectively reported ADHD using various assessment tools, and possibly publication bias limit the generalizability of the findings. If causal, reducing the intake of artificial food coloring could provide an alternative treatment of ADHD in the future. Understanding the mechanism underlying the association may provide insights into the developmental origins of ADHD. / published_or_final_version / Public Health / Master / Master of Public Health

A systematic review and meta-analysis was conducted to evaluate the efficacy of different infant antiretroviral (ARV) prophylaxis regimens for prevention of mother to child transmission (MTCT) of human immunodeficiency virus (HIV) infection in breastfeeding infants who were born to HIV positive mothers but were HIV uninfected at birth. The systematic review of the literature published during January 2000 to April 2012 resulted in ten randomized and controlled clinical studies which met the study inclusion criteria. Two datasets were identified from the ten selected clinical trials. One dataset contains six studies evaluating short-course ARV prophylaxis regimens, and the second dataset contains four studies evaluating short-course versus extended ARV prophylaxis regimens. The odds ratio was used as the effect size to measure the efficacy between two comparative infant ARV prophylaxis regimens. Meta-analyses were conducted to assess the overall (pooled) treatment effect of the two comparative infant ARV prophylaxis regimens of the two datasets. The pooled ARV treatment effect was calculated as a weighted average of the effect estimated in the individual studies. If no heterogeneity was identified, a fixed-effect meta-analysis by the Mantel-Haenszel method was used. The random-effects method was used when there was heterogeneity in the meta-analysis. The inverse-variance method was used in the random-effects method of meta-analysis. Heterogeneity in the meta-analysis was accessed by the Chi-squared (χ2) test and I2 test. The combined sample size of all ten clinical trials was a total of 10,316 breastfeeding infants, and the overall postnatal HIV transmission rate regardless of ARV regimens and the timing of HIV infection status was approximately 8.7%. The overall HIV transmission rates of the short-course ARV prophylaxis regimen groups were 10.3% at 4-8 weeks and 9.0% at 6-9 months, respectively. The overall late postnatal HIV transmission rate (at 6-9 months after birth) was 5.5% in the extended ARV prophylaxis regimen group. The first dataset contains six randomized and controlled studies to evaluate the efficacy outcome (defined as the unadjusted HIV infection status at 4-8 weeks after birth) of two short-course infant ARV prophylaxis regimens, the nevirapine (NVP) regimen and the zidovudine (ZDV) with or without combination of lamivudine (3TC) or NVP regimen. Due to the existence of substantial heterogeneity, a random-effects method was used to test for the overall treatment effect. The results show that there was no significant difference between the two short-course infant ARV prophylaxis regimens (odds ratio:1.07; 95% CI: 0.69-1.66; Z=0.31, p=0.76). The results of the meta-analysis of five comparative short-course versus extended infant ARV prophylaxis regimens from four randomized and controlled clinical trials, demonstrate a favorable efficacy outcome (defined as the unadjusted HIV infection status at 6-9 months after birth), of the extended ARV regimens. There was no heterogeneity found in this dataset. There was a highly significant difference in the overall effect between the two ARV prophylaxis regimens by a fixed-effect model (odds ratio: 1.72; 95% CI:1.45-2.04; Z=0.68, p<0.00001). In summary, there was no significant difference in the overall treatment effect in reducing the early postnatal MTCT of HIV infection by infant short-course regimens of ARV prophylaxis, which include NVP, ZDV and their combination regimens. In comparison with the short-course ARV regimens, the extended ARV prophylaxis further reduced the risk of the late postnatal MTCT of HIV infection in breastfeeding infants. / published_or_final_version / Public Health / Master / Master of Public Health

Background: Pregnancy is the fertilization and development of one or more offspring. It is a period of significant importance, during the time of which, health status of the mother could have profound impact on that of the fetus. Due to the one child policy in Mainland China, ensuring the baby’s health is of the utmost priority. In the past two years the incidence rate of miscarriage has been gradually increasing. It has an impact on mothers both physically and psychologically. Formaldehyde can be released for a long time from compound and furniture products in newly decorated houses and have become a chronic source of pollution. We are interested in whether there is any association between miscarriage and formaldehyde exposure. If the association exists, the evidence could be applied to support further research to identify the sources of formaldehyde and to support formulation of environmental public health policy to reduce formaldehyde exposure. Objectives: (1) To estimate whether the serum formaldehyde concentrations were different between miscarriage women and normal pregnant women at the same stage. (2) To investigate the normal serum formaldehyde concentration in pregnant women; and (3) to identify exposure factors in miscarriage patients. Setting：Guangzhou Women and Children’s Medical Centre Method: A case-control study was carried out to estimate the association between serum formaldehyde concentrations and miscarriage. 309 pregnant women who fulfilled the inclusion criteria participated in this study, 191 normal pregnant women at term delivery as the control group and 118 women who were diagnosed with miscarriage as the case group. The main outcome was miscarriage confirmed by ultrasound and physicians. The main exposure variable was serum formaldehyde concentration. Binary logistic regression was used to estimate the crude and adjusted odds ratios for a diagnosis of miscarriage for each independent variable. The odds ratio (OR) would be used to estimate the relative risk of the serum formaldehyde concentration in miscarriage women compared with those women who were not diagnosed miscarriage. We also categorized the values of formaldehyde into quartile, and put them into logistic regression as continuous variable to test the linear dose-response relationship. Then we tested whether there would be a significant linear trend that the increasing of serum formaldehyde concentration level leads to the increased risk of miscarriage. Chi-squared tests were performed to test the association between diagnosis of miscarriage and various independent variables. And Chi-squared tests for linear trend were also used for ordinal independent to study any dose-response relationship. Results：The mean serum formaldehyde concentration was 0.0944 in the case group and 0.0239 in the control group. The difference was statistically significant. All characteristics between case group and control group were quite similar. After adjusting for age, occupation，education level, household income, home decoration situation, the following factors remained having independent impact on the miscarriage. They were formaldehyde and second-hand smoke (crude ORs=7.87 [95%CI: 4.96, 12.49]; 3.20[95%CI: 1.86, 5.52] respectively, adjusted ORs=8.06 [95%CI: 4.96, 13.09]; 3.60[95%CI: 1.58, 8.20] respectively). Exposure to formaldehyde and second-hand smoke were significantly associated with higher risk of miscarriage. The liner dose response association between formaldehyde level and miscarriage was presented with P value for trend less than 0.001. Conclusions：Our study provided some evidence of the association between the serum formaldehyde level and miscarriage. A significant linear trend was found that the increasing of serum formaldehyde concentration level lead to the increase the risk of miscarriage. And we also found that second-hand smoking was contributive to miscarriage. This association did not confound by age, occupation, education level, household income, or home decoration situation. / published_or_final_version / Public Health / Master / Master of Public Health

The Earth’s atmosphere supports microorganisms and they include potential pathogens and microbial allergens. Whilst indoor environments have been well studied, relatively little is known of bio-aerosols in outdoor locations and their potential influence on human health, particularly with regard to urban development. Hong Kong provides an ideal model system for testing hypotheses related to the impact of urbanization on bio-aerosols, with a well-defined gradient of urbanization and large population. This thesis describes work to establish the biodiversity and spatio-temporal dynamics of outdoor bio-aerosols in Hong Kong. A comprehensive study of multi-domain microbial diversity and allergen levels in urban aerosols over a contiguous annual timescale and along a gradient of urbanization was carried out. A comprehensive suite of climatic and pollutant variables were also recorded during the sampling interval. Terminal restriction fragment length polymorphism (T-RFLP) was employed to investigate variations in bacterial and eukaryal assemblages, followed by phylogenetic assessment using high-throughput sequencing. The results revealed a strong seasonality in both bacterial and eukaryal assemblages, with Archaea forming a negligible part of the urban bio-aerosols. The most abundant bacteria were proteobacteria but community shifts were seen due to increases in algae in summer, and betaproteobacteria and cyanobacteria in winter. This was most parsimoniously explained by considering the backward trajectory analysis of air mass. A greater abundance of marine-associated phylotypes such as Bacillariophyta and Chlorophyta were identified when the dominant air mass arriving in Hong Kong in the summer originated from oceanic sources. In contrast, betaproteobacteria, which indicated soil sources were prevalent when the origin of air mass was from terrestrial sources. A trend in fungal phylotypes was also apparent, with summer samples dominated by basidiomycetous Agaricales, and winter samples by the ascomycete genus Cladosporium. This was likely due to favourable climatic conditions during wetter summer months enhancing release of fungal basidiospores. A range of airborne human pathogens was also detectable at low levels including pathogenic bacteria such as Acinetobacter baumannii, Clostridium perfringens, Escherichia coli O157:H7, and Ricinus communis, and the pathogenic fungus Aspergillus terreus. Microbial allergens including bacterial endotoxins and fungal glucans were also quantified with immunological assays. These generally followed variations in biomass, and during some months were recorded at levels that may impact human health upon chronic exposure. Carbon dioxide levels were the only climatic or pollutant variable that correlated with allergen levels. Conversely changes in microbial assemblages were strongly correlated to several climatic variables including temperature, rainfall, air pressure and relative humidity, but not with the degree of urbanization or airborne pollutants. This study highlights the importance of including microbial assessments in future bio-surveillance of urban aerosols. / published_or_final_version / Biological Sciences / Doctoral / Doctor of Philosophy

Objectives : To investigate the impacts of traffic related air pollution (TRAP) on children’s respiratory health, and to investigate the policies and regulations implemented in other countries and the suggestions from the World Health Organization (WHO) to tackle this problem. Methods and Results: Using relevant keywords, 122 articles were acquired from PubMed and 15 were from MEDLINE via EBSCO host respectively. The articles were published between 1980 to 2013; all of which had examined the impacts of TRAP on children’s respiratory health. After assessing inclusion and exclusion criteria, 9 articles were finally included in this systematic review. Differences in demographic characteristics were studied and compared to determine the association between TRAP exposure and respiratory health impacts on children aged between 1 and 18. Four cross-section surveys and five cohort studies were selected in this review. The association between TRAP exposure and differences in children’s respiratory impacts were demonstrated in all studies. Multiple outcome measures were used to examine changes of children’s respiratory health such as the decline in lung function, incidences of asthma and inflammatory responses. There were 4 out of 9 articles suggesting the decline in lung function was related to TRAP exposure; 5 out of 9 articles showed the relationships between incidences of asthma and inflammatory responses and TRAP exposure and 2 articles mentioned other respiratory symptoms; notably, wheezing and sneezing, runny and stuffed nose. This review finally summarized the results of selected articles stressing that children who lived close to roadway within a 500 meter radius from home, with an early exposure to air pollution in the first year of age and a frequent exposure to heavy road traffic and outdoor air pollutants would experience greater impacts on respiratory health. Conclusion: The positive association between the impacts on children’s respiratory health and TRAP exposure has been determined through this systematic review. Stricter emission control strategies and updated guidelines are required to regulate traffic emission and improve air quality in order to reduce harm and provide a better living environment for our next generations. / published_or_final_version / Public Health / Master / Master of Public Health

Background: Childhood obesity remains a major public health concern. While preventing childhood obesity is important, the effectiveness of existing prevention strategies is indefinite. Important obesity contributors are speculated to be overlooked while the role of well-studied obesity-related factors has been questioned. Objectives: 1) Although the association between lifestyle and obesity is seemingly well-established, how an overlooked dietary factor, breakfast location, associates with obesity was investigated 2) At a time where inconsistencies in findings for the frequently studied association between TV viewing and obesity still exist, this thesis tested for the existence of the said association in the understudied Chinese population 3) The effect of mental health on childhood obesity is understudied. Using self-esteem as a mental health indicator, the effect of self-esteem on childhood obesity was investigated To understand the role of lifestyle factors in the association between self-esteem and obesity, how 4) breakfast habit and 5) TV viewing habit associates with self-esteem was explored Methods: A cohort of 83405 primary 4 (P4) Department of Health Student Health Service participants (49.7% male, mean age 9.9 years) in 1998-2000 were successfully tracked for 2 years into primary 6 (P6). A self-administered health questionnaire collected information on the breakfast and TV habits of participants alongside other lifestyle characteristics. Self-esteem was assessed using the four Culture Free Self-Esteem Inventories for Children (SEI) subscales. Body mass index (BMI) was derived using objectively measured weight and height. Weight status was classified in accordance to International Obesity Task Force standards. Logistic, multinomial logistic and linear regression were used to yield adjusted odds ratios (AOR) and β-coefficients for becoming overweight/obese in P6 (among P4 normal weight children), breakfast skipping/location, TV viewing habit and SEI subscale scores. Breakfast skipping behaviour was tested as a mediator in the association between self-esteem and obesity using standard mediation procedures. Breakfast and TV viewing habits were tested as effect modifiers in the same association by inclusion of interaction terms in models. Results: Among normal weight P4 children, those who skipped breakfast (AOR: 1.24; 95% CI: 1.02 to 1.51) or ate away from home (1.39; 1.20 to 1.60) were more likely to become overweight/obese in P6. Poor self-esteem was also associated with higher AORs for becoming overweight/obese, particularly among girls. While TV viewing in P4 did not predict overweight/obesity in P6, those who increased TV watching hours (≤2h/day to 2h/day) during the two follow-up years were 32% more likely to become overweight/obese in P6 than their counterparts who maintained watching ≤2h/day. Breakfast habit and self-esteem was bi-directionally associated. Although a curvilinear inverted J-shaped relation between P4 TV viewing and P6 self-esteem was observed, self-esteem did not predict TV viewing habit. While neither breakfast nor TV viewing habit significantly interacted with self-esteem to predict obesity, breakfast skipping was found to be a potential partial mediator in the association between self-esteem and obesity. Conclusion: Breakfast location, self-esteem and TV viewing patterns over time are important to consider in relation to childhood obesity. Additionally, the association between self-esteem and obesity could be mediated by lifestyle factors. / published_or_final_version / Community Medicine / Doctoral / Doctor of Philosophy