Theses and Dissertations

Viral Suppression among Men Who Have Sex with Men Living with HIV Who Were on Risk Reduction Interventions and Control Group in Mvita Sub-County, Mombasa County, Kenya

Thu, 11 Jun 2026 00:00:00 GMT

Viral Suppression among Men Who Have Sex with Men Living with HIV Who Were on Risk Reduction Interventions and Control Group in Mvita Sub-County, Mombasa County, Kenya Karanja, Jane Wangui Globally, men who have sex with men (MSM) are more vulnerable to HIV as compared to men in the general population. However, if commenced promptly on ART and adherence to treatment is observed their viral load can be suppressed hence unable to transmit the virus to their partner. The aim of this study was to determine the effectiveness of the risk reduction interventions on viral load suppression levels among men who have sex with men (MSM) living with HIV versus those in the control group in Mvita sub-County, Mombasa County, Kenya between December 2020 and June 2021. A quasi-experimental study design using quantitative methods was adopted and the study population was the MSM residing in Mvita sub-County in Mombasa County. Inclusion criteria were, being biological male-sex assigned at birth, 18 years and above, had anal sex with at least one male partner in the past 3 months, a resident of Mvita sub-County, and willingness to give informed consent in writing. Exclusion criteria were those who declined to give consent or were intoxicated with alcohol or drugs. The respondent-driven sampling (RDS), a semi- probabilistic method, in the absence of a sampling frame was used to obtain the sample of respondents. A total of 114 HIV positive MSM participated in the study. Half (57) of MSM living with HIV were actively followed and risk reduction interventions (adherence to ARVs, general counseling and nutritional care), and HIV prevention measures (prompt treatment of STI/OI and condom use) were offered after every 2 months for six months. The control group (57) received no risk reduction interventions. Semi-structured questionnaire and laboratory investigations were employed to collect data. Log-binomial univariate and the multivariate regression analysis model were used to identify the variables which were associated with undetectable viral load. Undetectable viral load was defined as having an HIV viral load of less than 50 copies/ml. Data collected was cleaned, coded and analysed using STATA software, Version 17. Level of significance was fixed at 5%. Ethical approval was received from the relevant bodies. Majority of MSM were between ages 19 to 29 (54%), were single (77%), were Christian (61%), never married to a female (39%) and had an income of <5,000 ksh (65%). MSM in the control group, who reported ever use of PEP/PrEP (44%), were likely to have detectable viral load while MSM who received interventions, who reported condom break more than once during anal sex (61%), who had more than one regular anal sex partners (61% both groups), and who drunk more than 2 bottles of beer (33% both groups) were likely to have undetectable viral load. MSM in the control group who reported being always high on alcohol during anal sex (19%), were likely to have detectable viral load. MSM who received interventions and reported feeling uneasy while seeking health services (75%) had detectable viral load while those in the control group who attended private clinics (42%), had undetectable viral load. Overall, after six months, the proportion of MSM achieving viral load suppression was significantly higher in the intervention group as compared to control group by 60% (95% CI= 49‒70, p < 0.001). In conclusion, MSM in the control group had low knowledge of HIV prevention hence detectable viral load. However, high risk behaviours such as condom break were associated with undetectable viral load in the group that received risk reduction interventions. MSM in the control group who were attended in private clinics received better HIV care, thus had undetectable viral load. MSM who received risk reduction intervention had better treatment outcome with many MSM xix achieving undetectable viral load. From the study it is recommended that peer-led HIV services to be integrated in all HIV programmes targeting MSM, effective substance use intervention program be re-designed to rehabilitate the MSM on alcohol and drugs, improvement on HIV care in public health facilities and risk reduction interventions to be offered to all MSM. PhD Public Health

Effect of Retail Investor Profile on Individual Portfolio Choice at the Nairobi Securities Exchange, Kenya

Thu, 11 Jun 2026 00:00:00 GMT

Effect of Retail Investor Profile on Individual Portfolio Choice at the Nairobi Securities Exchange, Kenya Maritim, Francis Kipsanai The retail investor profile plays a major role in determining the choice of an investment portfolio composed of one or more assets that an investor receives over a certain period of time. Although there are numerous controversial arguments about the factors that determine the level of individual portfolio choice, the multidimensional perspectives on investors' characteristics have received little attention. The general objective of this study was to establish the effect of retail investor profile on individual portfolio choice at the Nairobi Securities Exchange (NSE). The specific objectives to guide this study was to determine the effect of investor’s attitude towards risk on individual portfolio choice at the NSE; to establish the effect of lifestyle characteristics on individual portfolio choice at the NSE; to examine the effect of investor’s specific needs on individual portfolio choice at the NSE; to assess the effect on investor’s investment avenues on individual portfolio choice at the NSE and to investigate the moderating effect of investor’s age on the correlation between retail investor profile and individual portfolio choice at the NSE. The study is anchored on four theories, namely, modern portfolio theory, prospect theory, expected utility theory and risk aversion theory. The population consisted of individual investors estimated at 2.4 million as of 30 th September 2025, based on Central Depository and Settlement Corporation Limited (CDSC). A target population of 873,980 active retail investors who also form the accessible population at the NSE was used to draw a sample size of 385 active individual retail investors, out of which 320 participated. Both stratified and convenience sampling were used to select the required number of respondents. A structured questionnaire was used to collect the data, whereby a drop-and-pick approach was used by the researcher and research assistants. Pilot testing of the instruments was performed to assess their reliability. The descriptive research design was used in descriptive analysis for the variables under study up to a period of five years from January, 2021 to December, 2025, using frequency, percentages, means and standard deviation Tables. Further, correlational and multiple regression analyses were used to analyze the data obtained, which was presented using Pearson correlation, model summary, ANOVA, moderated regression and hypothesis analysis tables. The study findings revealed that attitude towards risk and investors’ specific needs are positively correlated and significant on individual portfolio choice, while lifestyle characteristics and investment avenues were negatively correlated. As such, a unit increase in a predictor variable lead to an increase in investment in stocks, bonds and treasury bills. Further, the results of the study indicated that investors’ ages are positively correlated but do not moderate the relationship between retail investor profile and individual portfolio choice. When the independent variables interacted with the moderator, the results indicated that some were positively correlated, others were negatively correlated, some were significant, yet some were not significant. The study would benefit investors to better understand their investment goals, the regulator the Capital Markets Authority to improve regulatory policies and make the market more attractive to both the existing and potential traders, and Nairobi Securities Exchange to assist individual investors in their portfolio choices and also to the academicians to advance the conceptual arguments of the lack of moderating effect of age on the relationship between retail investor profile and individual portfolio choice. PhD Business Administration

Vaccination Coverage and its Associated Factors among Children Aged 2 - 5 Years in Eldas Sub County, Wajir County

Thu, 11 Jun 2026 00:00:00 GMT

Vaccination Coverage and its Associated Factors among Children Aged 2 - 5 Years in Eldas Sub County, Wajir County Abdi, Cathra Childhood vaccination is crucial intervention to reduce the morbidity and mortality of vaccine-preventable diseases. It has been estimated that vaccination campaigns prevent almost 3 million deaths globally each year. However, in 2018, 19.4 million children worldwide missed all necessary vaccinations. In Africa, routine vaccination coverage remains particularly low. In Kenya, the North Eastern region has the lowest vaccination coverage; (41%). The study aimed to establish vaccination coverage and its associated factors among children aged between 2 and 5 years in Eldas Sub- County, Wajir County. The specific objectives entailed establishing individual level factors associated with complete vaccination coverage, assessing status of routine vaccination and to determine health system level factors associated with complete vaccination coverage among children aged 2 to 5 years in Eldas sub county, Wajir County. A cross-sectional design was employed on sample size of 367 caregivers. Respondents were chosen using probability proportion to size and systematic random sampling strategies. A structured questionnaire and key informants’ guide were deployed for data collection. Data analysis was performed using Statistical Package for the Social Sciences (SPSS) version 25.0 with descriptive and inferential statistics generated. Chi-square test examined the relationship between categorical variables at confidence interval of 95%. Demographic findings showed that households were majorly male led with majority practicing Islami religion (98.1%) and 8.37% never attended school. A considerate percentage (40.1%) was in monogamous marriage arrangement, with only 23.2% in polygamous marriage. Sixty seven percent (67.9%) were unemployed while 23.2% earned between Ksh 6,001 and 12,000 a month. Radio ownership among caregivers was quite low. More than 21% (78 caregivers) attended to non-biological children. Caregivers with formal education were 3.47 times more likely to comply and have their children complete vaccination schedule (95% CI = 1.18 – 9.57, P = 0.032). Children of those earning more than Kshs 24000 a month were 3.18 times more likely to complete vaccination (95% CI = 1.27 – 11.67, P < 0.001) while those earning g between Kshs. 12000 - 24000 per month were 2.96 times more likely to complete vaccination (95% CI = 1.15 - 10.39, P = 0.002). Children with caregivers aware of vaccination's purpose were 4.51 and 2.57 times more likely, respectively, for complete vaccination (95% CI = 1.36 – 12.75, P < 0.001; 95% CI = 1.18 – 8.62, P = 0.031). Children born with skilled birth attendants were 5.36 times more likely for complete vaccination (95% CI = 2.15 – 13.21, P < 0.001). Children from places 1 – 5 km or 5 – 10 km from the hospital were 4.28 and 2.57 times more likely, respectively, for complete vaccination (95% CI = 1.28 – 14.87, P = 0.009; 95% CI = 1.54 – 6.88, P = 0.024). Content analysis was adopted in summarizing qualitative findings. More than ninety six percent (96.2%) of the children had been vaccinated atleast once. Distant location and lack of knowledge on immunization time were cited as major barriers to complete vaccination. Around 78.5% of the studied children reported experiencing a medical issue, 42.9% opined that atleast they were late on immunization schedule. Absence of personal issues among caregivers was associated with 2.29 increased likelihood of completing vaccination (95% CI = 1.41 - 3.75, P < 0.001). Knowledge on immunization further improved completion of immunization schedule by 5.53%. Qualitative findings also revealed a considerably high coverage attributed to strong community engagement and healthcare team dedication. The high coverage implied xv increasing level of access, awareness and system strength in delivering immunization solutions to the children. The notable health system factors included lifestyle, information, access and logistics, income level, education and awareness. However, misinformation, societal beliefs, low economic capacity, language barrier and nomadic lifestyle contributed to vaccine hesitancy among caregivers. The study concluded that individual level factors associated with complete vaccination status were education, occupation, income level, and ownership of electronic devices significantly influenced complete vaccination status among children aged 2 to 5 years in Eldas sub-county, Wajir County. However, vaccination completeness performance was below the recommended set target by WHO and the Ministry of Health. Health system related predictors that explained complete vaccination status included availability of skilled healthcare professional, presence of well-equipped public healthcare facilities and caregivers’ proximity to those facilities. It was suggested that the County Health Department should implement education outreach programs, support caregivers, enhance vaccination accessibility for lower-income families, use media for awareness campaigns, and address clinic shortages to improve vaccination coverage. It also recommends implementing subsidies or incentives to bridge economic gaps, leveraging electronic devices for vaccination knowledge, and enhancing healthcare infrastructure. MSc in Public Health

A Framework for Distributed Denial-of-Service Attack Detection in Internet of Things Environments

Thu, 11 Jun 2026 00:00:00 GMT

A Framework for Distributed Denial-of-Service Attack Detection in Internet of Things Environments Wawire, Silas Amisi Internet of Things (IoT) networks are ubiquitous across industries, homes, and critical infrastructure due to their automation capabilities. However, IoT devices remain highly vulnerable to Distributed Denial-of-Service (DDoS) attacks owing to their limited computational power and inherent heterogeneity. Also, IoT systems often favor usability over security. While deep learning has shown promise for detecting such attacks, existing approaches have largely been validated on conventional network datasets using centralized architectures that create single points of failure. Furthermore, the centralized approaches raise privacy concerns by requiring raw data transmission to the cloud and lack preprocessing tailored to IoT-specific protocols such as MQTT. This study aimed to develop a distributed deep learning framework for DDoS detection in IoT environments. The specific objectives were to design and implement a distributed detection framework based on deep learning, and to evaluate and compare its performance against a centralized paradigm using accuracy, precision, recall, and F1-score. A quantitative experimental research design using the DoS/DDoS-MQTT-IoT dataset was employed. The proposed framework integrated three components: a CNN-BiLSTM ensemble model, a three-tier edge-fog-cloud architecture incorporating federated learning where edge devices performed preprocessing and fog nodes conducted local training while raw data remained on premises, and preprocessing adapted to MQTT's publish-subscribe semantics. Model hyperparameters were held constant across both experimental conditions, and ten-fold cross-validation was applied. Statistical significance was assessed using a paired t-test with an alpha level of 0.05. The distributed detection framework achieved 99.68% accuracy, 99.02% precision, 99.28% recall, and 99.10% F1-score. The distributed framework demonstrated a 64% lower error rate than the centralized approach (1.35% versus 3.76%). This improvement was statistically significant as confirmed by a paired t test over ten-fold cross-validation (p < 0.001). The distributed framework also outperformed traditional machine learning methods and standalone deep learning models including CNN alone, BiLSTM alone, and RNN alone. Overall, the distributed approach exhibits superior accuracy and adaptability compared to centralized detection. It also provides privacy benefits through federated learning. Future work should validate the framework on additional IoT protocols and datasets and explore model compression techniques to further reduce computational requirements on resource-constrained edge devices. MSc in Computer Systems