Behavioural and Social Predictors of COVID-19 Vaccine Uptake among Persons with Disabilities in Kenya.

The uptake of the COVID-19 vaccine by persons with disabilities remains largely unknown in low-and middle-income countries. This evidence gap necessitates disability-focused research to inform improvements in access and inclusion in the last mile of COVID-19 vaccination programs and to support future programs for other vaccine-preventable diseases. We aimed to identify behavioural and social predictors of COVID-19 uptake among persons with disabilities in Kenya. This was a convergent parallel mixed method study that involved questionnaires (792), key informants interviews, and focus group discussions among persons with disabilities and key stakeholders (government actors and professional associations). Data were analysed using STATA statistical analysis software (version 14). Chi-square (X2) and Fishers exact tests were used to test for differences in categorical variables; multivariate regression analysis was employed to ascertain the factors that influence uptake of COVID-19 among persons with disabilities (PWDs) in Kenya. Approximately 59% of persons with disabilities reported to be fully vaccinated, with significant disparities noted among those with cognition (34.2%) and self-care (36.6%) impairments. Confidence in vaccine benefits (Adjusted odds ration [OR]; 11.3, 95% CI; 5.2-24.2), health worker recommendation (OR; 2.6, 95% CI; 1.8-3.7), employment (OR; 2.1, 95% CI; 1.4-3.1), perceived risk (OR; 2.0, 95% CI; 1.3-3.1), age and area of residence were statistically significant predictors of vaccine uptake among PWDs. The primary reasons for low uptake included perceived negative vaccine effects and lack of adequate information. No association was found between having a primary caregiver and/or assistive device, with COVID-19 vaccine uptake. Subsequent vaccination deployments should map and reach PWDs through relevant institutions of PWDs, and localized vaccination campaigns. Related communication strategies should leverage on behaviour change techniques that inspire confidence in vaccines, and on the credibility and trust in health workers to improve vaccine uptake.


Introduction
Kenya is one of the countries significantly affected by the global pandemic of coronavirus disease 2019, with reported high transmissions resulting in over 0.34 million infected cases and a death toll of over 5,668 [1].The country's COVID-19 vaccination uptake however among persons with disabilities (PWDs) in Kenya is unknown, despite the unique barriers they face [2], their special vulnerabilities including chronic conditions, and higher risk for severe outcomes.Through a national COVID-19 vaccine deployment plan, the government of Kenya prioritized vaccination as a key measure to contain COVID-19 spread that also targeted population groups including PWDs scheduled to be vaccinated in phases [3].However, disaggregated data on proportions of vaccinated PWDs has been conspicuously missing on the COVID-19 vaccination MoH update reports, making it difficult to track the progress made in reaching this cohort [1].Moreover, despite availability of several studies on drivers of vaccination among other key population groups [4], there is dismal to no evidence on the same among PWDs in low and middle income countries (LMICs).
Approximately 37% of Kenyan adults and 10% of children between ages 12 and 18 had been fully vaccinated as of December 2022 [5] .Willingness for vaccine uptake was reportedly lower among younger people compared to the older, students compared to those working, and in the Coastal and Northeastern region but higher in Nairobi and Rift Valley regions.Moreover, vaccine confidence was significantly associated with vaccine uptake with Kenyans with lower confidence in vaccine safety being more likely to refuse the vaccine [4].Such studies however miss to account for influences specific to PWDs, considering their unique vulnerabilities.
. Inadequate information on health inequalities has been identified as a distinct gap in strengthening inclusion and equity in COVID-19 responses.[6] The scanty available evidence globally shows lower COVID-19 vaccination rates compared to those without disability [7] amidst higher risk of severe illness and premature mortality from COVID-19 [8].
We sought to investigate uptake levels of COVID-19 vaccines and determine the demographic, social and behavioural predictors of COVID-19 vaccination among PWDs in Kenya.

Study Design
We conducted a convergent parallel mixed methods study, in which quantitative and qualitative elements were conducted concurrently in the same phase of the research process, the methods weighed equally, and analysed independently.We employed a cross-sectional study design where both the behaviour and social factors and uptake of COVID-19 interventions were concurrently assessed during the survey.

Study Setting
The study was conducted in four select Counties in Kenya selected on the basis of their disability prevalence, and by region to represent the main regions in the country to attain a representative rural-urban mix.The counties were namely Embu in Eastern Kenya with the highest disability prevalence of 4.4%, Siaya (4.1%) in the Westerly side, Mombasa (1.4%) in the Coastal region and Nairobi (1.1%) in the central region.

Participants
Our study participants were PWDs chosen from NCPWDs in the hearing, communication, selfcare, cognition, mobility, visual and albinism domains based on the Washington Group on Disability Statistics (WG).Albinism was included as it is classified as a disability in Kenya .because of the associated visual impairment and other vulnerabilities in society including mistreatment and exclusion [9].

Variables
The dependent variable for the study was COVID-19 vaccine uptake, whereas the independent variables included demographics, perceived risk to self, confidence in vaccine benefits, safety, and in health workers, social norms, health worker recommendation, reminder notifications, permission, knowing where to get vaccinated and ease of access.

Study process, sampling and data collection
Data was collected between 27 -31 st March 2023, about two years after rolling out of the national COVID-19 vaccination program in Kenya.
PWDs were systematically chosen using a K th of four to reach a sample size of 792 from the National Council for PWDs Registration database within Mombasa, Siaya, Embu and Nairobi counties.The respondents were drawn from men and women above 18 years of age.The respondents were contacted prior to field activity while those not found were replaced by choosing the next participant as per the skipping pattern of four.
A structured questionnaire embedded on a mobile collection tool (KOBO) was used to collect quantitative data from PWDs in the selected counties.Focus Group Discussions (FGDs) were conducted among PWDs and their care givers while Key Informant Interviews (KIIs) were done among disability service officers, social development officers, COVID-19 vaccination in-charges and NCPWDs representatives in the target counties.Behavioral and Social Drivers (BeSD) framework in Error!Reference source not found.informed the data collection tools.Research assistants were trained on data collection methods including consenting and interviewing processes, as well as on the various disability domains and related disability etiquette.Interviews were done in languages understandable to the respondents, with sign-language interpreters engaged to administer interviews to the hearing-impaired.

Data Management and Analysis
All data collected through KOBO were first transferred to the Excel for cleaning and later to STATA statistical analysis software (version 14) for analysis.
Descriptive statistics were presented in tables.Univariate analysis was done for all variables to compare outcomes of interest.Proportions were used for categorical variables and measures of central tendency and dispersion for continuous variables.Chi square and fishers test was used to determine significant differences on the key outcomes' variables.All statistically significant variables were further analyzed using logistic regression and multiple logistic regression to determine the key predictors of vaccine uptake among PWDs.
For qualitative data, all interviews were transcribed and coded for analysis using NVivo software.Data analysis was done using the Behavioral and Social Drivers (BeSD) framework.Deductive content analysis was used for this study.Though the study largely (BeSD) framework to guide the analysis, we further modified it based on findings to develop new themes not covered by the framework but mentioned in interviews.The qualitative data was initially read by (RM) who developed an initial coding framework with input from (JO).This was then shared with (RH) who read through and double coded nine transcripts, selected across the participant's category to refine the coding framework.With input from other study team members (JA, AA, LK and SK), the differences in the coding framework were then reconciled and coding was done on the themes and sub-themes identified in the final (BeSD) framework.The data was stored in a password protected shared directory on the Amref Health Africa server based on Amref Health Africa ICT data protection policy.All personal identifiers were removed from the dataset prior to archiving in the Amref Health Africa data repository.

Ethical Approval and consideration
Ethical approval was obtained from Amref Health Africa's Institutional Review Board (P1383/2023).A written Informed consent was sought from participants who went ahead and signed a copy of the consent form upon being provided with the information about the study and the potential benefits and risks of their involvement.
There were averagely 4 persons in the households visited with more than half indicating that they had a primary care giver and 50.9% used an assistive device.A majority (45.8%) had attained secondary education while 35.4% had attained primary education.Slightly more than half (52%) were unemployed with 31% engaged in self-employment activities.Moreover, over 90% were Christians with a paltry being Muslims as shown in Table 1.

Bivariate Analysis
Chi-Square test of demographic factors showed that county, residence area, age, marital status, and having certain disabilities notably cognition or selfcare disabilities were significantly associated with vaccine uptake as shown in Table 2.However, there was no significant relationship established between sex or history of COVID-19 diagnosis with vaccine uptake.

Social and Behavioural Factors Influencing Vaccine Uptake
Presence of a primary caregiver, education level, employment status, and religion were associated with vaccine uptake in the univariate analyses, however only employment was statistically significantly associated in the final model.Use of assistive devices was also not associated with vaccine uptake.Individual and societal level factors were shown to influence vaccine uptake including high risk perception, vaccine confidence, family and religious leaders' norms, recommendation by health worker, recall notification, knowing where to get vaccinated, and ease of access.Whereas community leaders' opinions not to influence vaccine uptake, qualitative findings reveal an influence of the political class on vaccine uptake.Several accounts were given of getting the vaccine upon seeing their political leaders, notably the president, getting it.No association was also found between one requiring permission to go for the vaccine and getting vaccinated as shown in Table 3. Health workers were however required to provide clear information to influence vaccine uptake, as recounted "If safety is guaranteed and the healthcare providers give clear [information] and sensitize, there will be an increase the uptake."-The Regional In-charge of the National Council for PWDs for Meru, Tharaka Nithi, Isiolo and Marsabit counties Despite high trust in health worker, the study found out that over 30% of PWDs had not been reached by health workers and the vaccine recommended to them.Four in ten PWDs reported low ease of access to vaccination services (p < 0.001) with the main reasons cited as difficulty getting to vaccination sites and long waiting times.Notably lower ease of access was reported among persons with selfcare (47.8%), cognition (48.7%) and vision (50%).A paltry (4%) were vaccinated through door-to-door, despite of home administration being the top recommended approach by PWDs on reaching them with vaccines.Other key recommendations made were working with PWD-organizations and their social networks, targeted sensitization of PWDs for informed decisions, special considerations to PWDs, equipping health workers to handle PWDs and/or making transportation arrangement for PWDs to vaccination sites.

Multivariate Analysis
Further analysis of the statistically significant factors at bivariate level showed age, county, employment status, perceived risk, confidence in vaccine benefits and health worker recommendation were statistically significant predictors of COVID-19 vaccine uptake in our final multiple logistic regression model.According to the findings, the older persons were likely to get vaccinated compared to the younger age groups.For instance, persons aged 35-44 years were two times less likely to be vaccinated compared to those aged 64 years and above.
Respondents from Embu County were less likely to be vaccinated compared to those from Siaya County.The employed according to this data were 2.63 times likely to be vaccinated compared to the unemployed.The findings also show that perceived risk, confidence in the vaccine confidence health worker recommendation are key predictors to vaccine uptake among PWDs as shown in Table 4.

Discussion
The proportion of PWDs respondents who indicated they had been fully vaccinated (59%) was higher than the general population proportion vaccinated in Kenya (36.9%) within the same period.
This points to deliberate interventions to get PWDs vaccinated, and higher acceptance levels.
However, those in the self-care and cognition domains were disproportionately vaccinated compared to other domains.This could be partly attributed to the low vaccine confidence and low ease of access among both domains, coupled with lack of information for the cognitively impaired, as established by our study.Selfcare and cognition disabilities have been linked to communication barriers, difficulty understanding information, and dependency on caregivers.A survey by PAHO found that persons with cognitive or intellectual disabilities lacked easy read, simpler text versions of public information and communication material which made decision making difficult.[12] A study among people with intellectual and developmental disabilities and their families showed that COVID-19 vaccine uptake was associated with self-reported knowledge about the vaccine and learning about the vaccine from one's doctor among other variables.[13] Accessible education and support from healthcare providers and caregivers is significant in addressing these disparities.[2] .
As in past studies [14] [15] [16],living in rural was associated with low vaccine uptake, however these finding was not statistically associated with vaccine uptake in our study's final multiple logistic regression.The study also revealed no statistically significant association between primary caregiver or usage of assistive technology by PWDs with COVID-19 vaccine uptake.The qualitative findings did however show that PWD have special reasons, such as protecting oneself against additional exposure from using supportive devices like a white cane or reading braille.
These findings imply that even if the use of assistive devices may not directly affect vaccination rates, PWDs may have particular concerns relating to their condition that affect their vaccination choices.On the low vaccine uptake among the youth, Osur et al established the main causes of vaccine hesitancy among youth in Kenya to be concerns about vaccine safety and effectiveness, with social media as the major source of information contributing to hesitancy.[17] Higher vaccine hesitancy was also reported in younger persons with intellectual and developmental disabilities in New York state individuals by [18].Some studies have also reported lower likelihood of vaccine hesitancy among the elderly [19] [20].Past related studies in Kenya have also shown low confidence in COVID-19 vaccines in the coast region (67%), Nyanza (76%) and Eastern (78%) compared to Nairobi (79%).This aligns with our findings of significantly low vaccine uptake in Mombasa in the coastal region.A study by Orangi et al on determinants of vaccine confidence in Kenya however presents contradicting results that rural counties had higher odds of reporting vaccine hesitancy (aOR=2.46;95% CI: 1.02-5.94)as compared to those in urban counties like Mombasa.[16] However, unlike many studies that show no association between employment and vaccine uptake among other population cohorts, our study's final result showed a direct significant association among PWDs.The fact that vaccination is required for access to services and for employees by the government may explain the higher vaccination rates among employed PWDs.
On the other hand, costs associated with including transport expenses would have been difficult for those without jobs.Decentralizing vaccination services to PWDs and considerations for free transportation to vaccination facilities may increase vaccine uptake rates among the unemployed PWDs.
The strong influence of high vaccine confidence and risk perception is corroborated in prior studies [21] , [22], [23], [24] with the pandemic reported to have had a positive impact on general vaccine confidence in Kenya.c Masters et al also cited low confidence in COVID-19 vaccine as the strongest correlate of not taking COVID-19 vaccines (adjusted prevalence ratio = 5.19, 95% CI = 4.93 -5.47) [25] .Orangi et al also linked risk perception to vaccine confidence with those perceiving COVID-19 as not risky having 1.8 times higher odds of being vaccine hesitant compared to those who perceived the disease as risky.[16] High risk perceptions were also attributed to vaccine acceptance (aOR = 2.76, 95% CI: 1.48-5.14)among people living with HIV [26].However, in a study by Alobaidi S, perceived severity was not significant in predicting vaccine uptake intentions.[27].Vaccination communication campaigns and advocates should inevitably consider inspiring vaccine confidence.The "Stop HPV, Stop Cervical Cancer" by the Danish Health Authority, the Danish Cancer Society and the Danish Medical Association is a classical campaign building vaccine confidence.[28] The key motivations of protecting oneself as well as significant others, as established in the study, could be capitalized on alongside the messages raising vaccine confidence.While communicating to PWDs however, the motivations should be contextualized to include routes of transmission unique to this cohort including possible infection through assistive devices and caregivers.
The major impediment to vaccine uptake was perceived vaccine effects and has also been cited in past studies [29] , [23] , [17] , [16], [27], even among health workers [30] [31] In a study among American with disabilities, vaccine effects were similarly glaring with the study highlighting higher concerns about COVID-19 vaccine safety as compared to concerns about contracting the disease.Our study revealed further that most of the purported effects were heard from external sources mainly social contacts.This points to the need to provide guidance on credible sources of information to minimize conflicting and misleading information.
Healthcare providers and family members have been similarly cited as trusted sources of information by PWDs.[18]Latkin et al also found out that close family and friends discouraging vaccination was a key predictor of low vaccine uptake (aOR = 0.26, 95% CI = 0.07-0.98).
Vaccination programs and influencers including social services officers and religious leaders should therefore not only reach PWDs, but their close social contacts too, considering the influence they hold.The high trust in health workers has been consistently attributed to [32], [33] , [34] , [35]

Generalizability
The results of our study can be extrapolated and applied to PWDs outside the sampled population to inform vaccination programs' decisions targeting the population cohort and their families.The external validity of our study is supported by the consistency of our results with those found in studies in other populations.The studies cross examined were similarly based on study designs with clearly-stated hypotheses in well-defined populations, and on a globally-applied Behavioral and Social Drivers (BeSD) of vaccination framework.

Conclusion
In conclusion, this study highlights the uneven COVID-19 vaccine uptake across disability domains, with lower rates for cognition and self-care impairments.Our study found out that confidence in vaccine benefits, being employed, perceived self-risk and health worker recommendation were associated with high vaccine uptake, whereas perceived effects were .
attributed to low uptake.The influence of health worker recommendations on vaccination choices highlights the significance of healthcare provider participation in vaccine outreach.Improving immunization rates among PWDs requires addressing issues with access, vaccination site difficulties, and long waiting times.Targeted tactics including collaborating with disability organizations, raising awareness among PWDs, and setting up transportation can be put into practice to guarantee that everyone has fair access to vaccinations.

Table 1
Socio-Demographic characteristics of PWDs participating in the survey in 4 Kenyan counties

Table 3
Social and behavioural factors influencing COVID-19 vaccination uptake among PWDs 5%) with lack of information notably higher among persons with cognition impairments.Perceived vaccine effects were also conspicuously reported among respondents in the qualitative findings, with a majority reporting to have heard of the effects from external sources, and hardly any from personal experience.Other notable reasons were distance and from the residential area to where the vaccination was taking place."-VisuallyImpaired FGD Respondent from Siakago, Embu County.Whereas PWDs believed that their close family & friends and religious leaders wanted them to get vaccinated, they trusted the healthcare workers' recommendation most (80%), and close family & friends second (39%).Other trusted social contacts were social service officers (14%), caregivers (12%) and religious leaders (8%).The qualitative findings similarly revealed that health workers were considered as credible sources of information and service provider, as recounted for instance: *The key motivations expressed by PWDs for getting vaccinated were to protect themselves (92.9%), protect family (75.5%) and gain access to spaces (30.6%).Qualitative findings further revealed unique motivations to PWDs, for instance getting vaccinated to safeguard oneself against additional exposure from using assistive devices e.g.white cane and reading braille.Low vaccine uptake was mainly attributed to perceived negative vaccine effects (35.3%) and inadequate information (20.associated cost implications.One of the respondents recounted, "The vaccination place was far from our residential area.Most of the people didn't get vaccinated because they didn't have fare to move "If I hear this information from a medical person, I would believe [it] because they have knowledge in the field.I would trust them [healthcare providers], but if I get it [information] frompeople around here I would not trust the vaccine since people have so many things going on.If it [information] comes from a religious leader, I will accept but not believe since that is not their area of specialty."-A Physically Impaired FGD Respondent from Mbeere North, Embu County.

Table 4
Logistic regression on COVID-19 vaccination uptake predictors among PWDs with the information they provide being linked to better health access[36].With vaccine The low vaccine uptake attributed to low ease of access has been shown to be bridged by localizing vaccination services which reduces the difficulty to get to vaccination centers.A study found that a community-based health effort utilizing a mobile vaccination clinic, successfully increased COVID-19 vaccine adherence among the Black population in San Bernardino County with observed lower uptake rates[41].The desire for house visits and mobile clinics to increase accessibility was overemphasized in our qualitative results too, with higher preference for doorto-door immunization expressed by PWDs.Prioritization of vulnerable populations and centralized fixed-time appointments for receiving the vaccines could potential save them from the long waiting.To improve vaccine uptake among these cohort, working with PWDs' organizations and networks, focusing awareness efforts, creating custom accommodations and specialized training for healthcare professionals have also been recommended.