Background

The development of highly efficacious vaccines against the SARS-CoV-2 virus in 2020 was heralded as a potential end to the pandemic that has caused more than 6.39 million deaths globally as of August 2022 [1]. However even where vaccines were made widely available, roll-out progressed slowly or stagnated, in part due to persistent vaccine hesitancy [2, 3]. Vaccine hesitancy is defined by the World Health Organization (WHO) as “delay in acceptance or refusal of vaccination despite availability of vaccination services” [4]. Varying levels of vaccine hesitancy are reported globally; a review from February 2021 reported vaccine hesitancy across more than 30 assessed countries ranging from 10.0 to 57.8% [5, 6]. Among countries in the Middle East, data on vaccine hesitancy is sparse, or relies on data from web-based surveys that may exclude certain population groups, potentially resulting in biased estimates. A non-representative, web-based survey in Kuwait found vaccine hesitancy at 26% [7], while another online survey from December 2020 reported much higher levels of vaccine hesitancy—68.2% in Saudi Arabia, 71.6% in Jordan, and 76.4% in Kuwait [8]. Investigating and addressing the drivers of hesitancy may support efforts to increase uptake of the vaccine among populations who remain skeptical, misinformed, or fearful of vaccine risks. Surveys conducted to date have identified factors such as trust in government, sex, education level, risk perception, and use of social media for COVID-19 information, among others, to be associated with vaccine hesitancy [6, 8, 9].

Gaza Strip, the 362 sq km Palestinian territory bordering Israel and Egypt, faces immense challenges with vaccine roll-out given its fractured and poorly resourced health system and political obstacles to the importation of medical supplies [10, 11]. The first vaccine shipments arrived in Gaza in February 2021, however quantities were limited and directed to priority groups at risk of severe COVID-19 disease (older adults, people with cancer, and people with kidney disease) and healthcare workers [12]. In May 2021, a blockade of the territory during a two-week period of renewed conflict between the Israeli military and Hamas led to a temporary cessation of vaccine imports that further reduced availability [12]. By February 2022, the WHO estimated nearly 2 million people across the occupied Palestinian territory (oPt) had received at least one dose of the vaccine. The data also highlighted potential disparities in access to and uptake of the vaccine. While 44.82% of the Gazan population (aged 12 years and older) had been vaccinated and 32.51% were fully vaccinated, coverage was substantially higher in the West Bank, at 65.93% and 61.82% respectively [13].

Notably, vaccination levels have stagnated across the oPt in 2022 despite continued donations ensuring the availability of supplies [13]. Fewer than 100,000 people were newly vaccinated across the oPt between December 2021 and July 2022, compared to 1.91 million vaccinated in the previous 10 months [14]. The plateau suggests initial demand for vaccines was already saturated in 2021. A recent publication highlighted rumors, misinformation, and conspiracy theories as the main reasons for Palestinians’ “reluctance or unwillingness” to take the COVID-19 vaccine [12]. Limited data exist quantifying levels of vaccine hesitancy and its risk factors in Gaza; a web-based survey conducted in October 2020 found 63% of respondents would accept the vaccine (37% expressed hesitancy), with women and young adults less likely to be vaccine hesitant [11]. However to our knowledge no recent data was collected since the vaccination campaign began, and no population-representative study had been conducted.

The international non-governmental organization International Medical Corps (IMC) commissioned an external consultant epidemiologist to conduct a mixed methods vaccine hesitancy study in the Gaza Strip. The study was carried out between October 1 and 30, 2021 across all governorates in Gaza. The study aimed to investigate community members’ and healthcare workers’ perceptions of the COVID-19 vaccine; understand barriers and enablers associated with vaccine acceptance; and obtain updated estimates of COVID-19 vaccination coverage along with predictors of uptake. The results of the study provide new evidence on HCW and population perceptions that can inform COVID-19 service delivery and information campaigns in Gaza.

Methods

The vaccine survey targeted two study populations—the general population of Gaza (community component) and a sub-population of HCWs (HCW component). For the community component of the survey, we carried out a population-based, cross-sectional survey of households across all five governorates. For the HCW component, a non-random survey design was used with a combination of purposive and convenience sampling.

Survey instrument

A closed-ended questionnaire (see Additional file 1) was developed by an IMC expert working group in April 2021 based in part on the WHO recommended vaccine hesitancy assessment methodology [15]. Survey elements included current vaccination status, demographic characteristics, information sources, and perceived risks related to COVID-19. The questionnaire was translated into different languages by native speakers and standardized for organizational use, as the intention was to conduct vaccine hesitancy studies in different humanitarian contexts to draw cross-country comparisons. The Arabic version of the questionnaire was successfully field-tested in comparable settings, including Lebanon and Iraq, prior to Gaza.

Sample size

Based on Gaza’s population demographics (1.99 million residents in 2019), we estimated the study population size to be 1,034,800 adults 18 years and older [16]. The sample size was calculated using EpiInfo (CDC) (v7.2.4.0) for a population survey with parameters set at 95% level of confidence, 50% estimated proportion, 5% margin of error, and a design effect of 2. Given lack of data from previous population surveys in Gaza, a conservative non-response rate of 15% was added. The resulting sample size was 884 households, while 906 were actually surveyed. For the HCW component, we targeted a sample size of between 30 and 40 HCWs from each facility based on the number of eligible staff, study timeline, and anticipated homogeneity of responses. 169 HCWs were ultimately surveyed.

Sampling for population survey

Multi-stage stratified sampling was used to select households for the community component of the study (see Additional file 2). First, we compiled a list of geo-locationsFootnote 1 and the type of location (urban, rural, and refugee camp) for all five Gaza Strip Governorates (North, Gaza, Middle, Khan Yunis, and Rafah) using administrative data. From this list, we used stratified random sampling to select two geo-localities from each of the three strata (urban, refugee camps, and rural). We selected two geo-localities from each stratum largely due to insecurity and access constraints in Gaza; this approach reduced the number of areas data collection teams visited while maximizing the representativeness of the geographic types included in the study. In the second stage of sampling, we used simple random sampling to select four clusters (administrative units comprised of blocks in the camp settings and neighborhoods in the urban and rural areas) from each geo-location for a total of 24 clusters. Population data was not available at the cluster level in Gaza, so we calculated the number of households to be sampled in each cluster using probability proportional to size of governorates from which the clusters were selected [16] (see Additional file 3). Sampling weights are used in analysis to partially correct for population differences at the level of the geo-locality.

At the third stage, systematic random sampling was used to select households in each identified cluster. We used population data on households from the Palestinian Central Bureau of Statistics (PCBS) to estimate the total number of sampling units in the selected clusters and to calculate the sampling interval. Sampled households were replaced if the house was unoccupied. Enumerators approached the household selected for the survey and generally interviewed the first point of contact who was available (at home), willing (consented), and eligible (adult 18 years or older) to participate. In households where multiple individuals met these criteria and someone other than the first point of contact expressed a preference to participate, that person was interviewed instead. The targeting of adults through convenience sampling within households was largely based on time and resource constraints. Given a fixed number of days for data collection, enumerators could not consider household members who were not present.

Sampling for HCW survey

For the HCW component of the study, five health care facilities were purposively selected as entry points: Caritas Gaza Health Center, Ahli Arab Hospital, Saint John’s Eye Hospital, Karama Hospital, and European Gaza Hospital. Four of the facilities had existing agreements with IMC to provide different levels of healthcare. The only facility not supported by IMC, European Gaza Hospital, is a public hospital mandated to receive COVID-19 patients experiencing moderate to severe symptoms that require in-patient care. Selection of these health facilities was primarily due to operational constraints in Gaza. Bi-lateral institutions that fund humanitarian aid in Gaza generally prohibit international non-governmental organizations from engaging with public sector service providers under the de-facto administration, as well as with non-vetted non-profit and private sector service providers. As a result, accessing health facilities outside of existing partnership networks poses certain risks, so we prioritized facilities already receiving technical and financial support from IMC.

Respondents were recruited to participate in the survey through a combination of purposive and convenience sampling. The study aimed to include a broad cadre of HCWs holding technical or clinical positions and commonly interfacing with patients. First, IMC enumerators consulted administrators at the five selected facilities to identity eligible staff (purposive component). Respondents were considered eligible HCWs if they held a position as either a medical doctor, nurse, dentist, midwife, pharmacist, lab technician, radiologist, or physiotherapist. A total of 655 staff were eligible. Second, facility administrators contacted eligible HCWs to describe the survey and request voluntary participation. HCWs who were available and willing to be surveyed were contacted by IMC enumerators to schedule interviews (convenience component).

Ultimately one-quarter (25.8%) of eligible staff were reached in the survey. The final sample consisted of 169 HCWs—17 HCWs from Caritas Gaza Health Center, 26 from Ahli Arab Hospital, 20 from Saint John’s Eye Hospital, 20 from Karama Hospital, and 86 from European Gaza Hospital.

Data collection

Ten survey enumerators (5 men, 5 women) were recruited to conduct the survey. They were separated into five teams, with each team comprised of one female and one male enumerator to ensure culturally appropriate interviews could be conducted with women and men in the communities. The recruited enumerators were all young adults holding at least a bachelor’s degree and who had prior experience with data collection in Gaza. The research consultant provided a one-day training to enumerators on the data collection tool, interview techniques, sampling protocols, and COVID-19 prevention measures. Enumerators were also trained to phrase the survey question on vaccine hesitancy in a uniform way in Arabic based on local availability of the vaccine.

The survey was deployed using the electronic data collection application KoBoCollect (UNOCHA). Enumerators used tablets to collect and submit data on a daily basis. The teams were required to adhere to COVID-19 preventive measures recommended by the WHO and IMC (physical distancing, mask-wearing, and hand hygiene) during data collection. Informed consent was sought from all survey participants using standard formats. As data were recorded electronically and to avoid separate paper records, enumerators obtained consent orally rather than in writing.

Two staff members from IMC’s monitoring, evaluation, accountability and learning (MEAL) department supervised all data collection in the field, providing daily hands-on support and data quality assurance. The survey consultant also followed up regularly during data collection to ensure quality and integrity of the data.

Data analysis

Table 1 shows the variables considered in the analysis. The primary outcomes used in the analysis were “COVID-19 vaccination status” and “COVID-19 vaccine hesitancy”. Individuals were classified as vaccinated if they responded “yes” to the question, “Have you already received COVID-19 vaccine?”, with at least one dose of any vaccine considered vaccinated.

Table 1 Thematic areas investigated in analysis
Full size table

Vaccine hesitancy was classified in two ways for primary analysis and sub-analysis. For the primary analysis, vaccinated and unvaccinated respondents were pooled to increase statistical power, and vaccinated individuals were classified as non-hesitant. Individuals who were not vaccinated were classified as hesitant based on their response to the question “If the vaccine was available, would you get it?” according to the SAGE Working Group definition of vaccine hesitancy [4]. Unvaccinated individuals who responded “no” or “don’t know” to the question were classified as hesitant. In sub-analysis, data were restricted to the unvaccinated population, and risk factors for vaccine hesitancy were analysed within this sub-group. Hesitancy among unvaccinated individuals was classified using the same criteria as in the primary analysis.

Analyses were conducted using R version 4.0.5 [17] and RStudio version 1.4.1106 [18] with the packages survey [19] for all statistics and gtsummary [20] for the tables. All statistics described in the Results section represent population-weighted estimates. Data were weighted to the population using 2017 administrative data provided to IMC by the Palestinian Central Bureau of Statistics, using cluster sampling weights and household weights, and to account for stratification by type of geo-location (rural, urban, and refugee camp). As population in each cluster was not available, cluster weights within each selected geo-location were calculated using average cluster population in the selected geo-location. See Additional file 3 for the population of geo-localities. See Additional file 4 for details on finite population correction and sampling weights.

Demographic characteristics of the survey respondents are presented in Table 2. Demographic characteristics are presented as both unweighted and weighted univariable statistics, stratified by vaccination status (Table 3) and vaccine hesitancy status (Table 4). Univariable descriptive statistics with both unweighted and weighted results are presented for all other variables in Tables 5 (vaccination status) and 6 (vaccine hesitancy). Distributions of demographic characteristics and risk factor variables were compared using chi-square test with Rao & Scott's second-order correction.

Table 2 Demographic characteristics of survey respondents
Full size table
Table 3 Demographic characteristics by vaccination status, community members only
Full size table
Table 4 Demographic characteristics by vaccine hesitancy status, community members only
Full size table
Table 5 Knowledge, beliefs, and other factors associated with vaccination status, community members only
Full size table
Table 6 Knowledge, beliefs and other factors associated with vaccine hesitancy, community members only
Full size table

Unadjusted odds ratios (ORs) were estimated using logistic regression (Tables 7, 8). Adjusted odds ratios (aORs) (Tables 9, 10) were estimated using multivariable logistic regression models, adjusted for sex, age, highest education received, and hypothesized risk factors based on literature, which  included having enough information on the vaccine, receiving information from healthcare workers, receiving information from family, receiving information from social media, trusting information from healthcare workers, perception of vaccine safety, and believing in personal risk of serious illness or hospitalization from COVID-19. For each, estimates of unadjusted and adjusted ORs, p values and 95% CI’s are presented; a p value of < 0.05 is considered significant.

Table 7 Unadjusted odds ratios of vaccine receipt, community members only 
Full size table
Table 8 Unadjusted odds ratios of vaccine hesitancy, community members only
Full size table
Table 9 Adjusted odds ratios of vaccine receipt, community members only
Full size table
Table 10 Adjusted odds ratios of vaccine hesitancy, community members only
Full size table

Results

Respondent characteristics

1075 people (906 community members and 169 health professionals) participated in the survey (Table 2). The demographic and geographic distribution of community members respondents generally mirrors the population of Gaza. Females represented 45.03% of respondents, compared to 49.3% of the global population of Palestinians estimated to be women and girls [21]. The largest proportion of respondents were from Gaza City (34.55%) and the lowest from Rafah (10.71%), which is similar to the PCBS population projections for Gaza Strip where it is estimated 34% of the population reside in Gaza City and 12% in Rafah [22]. Younger people may be over-represented in our sample. 58.17% of respondents were between 18 and 39 years-old, while the PCBS estimated the 18–29 age group comprised just 21.8% of the population [23]. Although not a random sample, the HCWs surveyed resemble the demographics of the Palestinian health workforce. In our study 78.11% of HCW respondents were 18–39 years-old and 42.60% were female. A recent Palestinian healthcare labor mapping survey (2021) found 74% of HCWs were under 45 years-old, while the proportion of female healthworkers varied substantially according to the professional category but was generally balanced [24].

Vaccination status and characteristics

Table 3 presents the unweighted and weighted distribution of the characteristics of the population by vaccination status. Table 5 presents the unweighted and weighted distribution of risk factors by vaccination status and Table 7 presents the unadjusted odds ratios. Population-weighted vaccination coverage was 49.08% (95% CI 43.10–55.08). Males were more likely to be vaccinated than females (54.87% vs. 41.45%, p = 0.018). Older adults (40 years and older) were more likely to be vaccinated than younger adults (52.96% vs. 46.17%, p = 0.032). There was an increasing trend in vaccination with higher levels of education; respondents with university degrees were twice as likely to be vaccinated compared to those with primary or no education (OR 2.12, 95% CI 1.85–2.42), and those with secondary or college/vocational education were 27% more likely to be vaccinated than those with primary or no formal education (OR 1.27, 95% CI 1.03–1.57).

Certain self-reported perceptions of the risks and trade-offs of COVID-19 vaccination were associated with vaccination status. The vaccinated had 15 times higher odds of considering the vaccines safe (OR 15.4, 95% CI 12.2–19.60). They were also less likely to express concerns about side effects and less likely to prefer natural immunity to the vaccine. The unvaccinated group reported concerns specifically about physical disability (27.80%) and death (59.20%) at higher levels than the unvaccinated. Concerns about fever and body aches were also common and reported by more than 50% of both the vaccinated and unvaccinated population.

More than 80% of both groups believed they were at risk of COVID-19 infection, while less than one-third believed they were at risk of severe illness and hospitalization (Table 5). This suggests perceptions of risk of infection are universally high; however, people do not largely believe they are at risk of developing serious disease, and these views do not appear to vary significantly among vaccinated and unvaccinated individuals (p > 0.05).

Perceptions of adequate vaccine information and the channels used to access information differed among the vaccinated and unvaccinated population. 54.62% of vaccinated individuals reported having enough information about the vaccine, while 31.30% of the unvaccinated believed they received sufficient information. Those who reported having inadequate information had around 60% lower odds of being vaccinated (OR 0.38, p < 0.001, 95% CI 0.28–0.52). Among both vaccinated and unvaccinated individuals, the most commonly requested information was on side effects and effectiveness of the vaccines, as reported by  more than 80% of both groups. Notably, slightly more than 11% of the unvaccinated expressed interest in information on how to register for the vaccine (Table 11).

Table 11 Factors associated with vaccine hesitancy, non-vaccinated community members only (n = 449)
Full size table

Sources of information on COVID-19 vaccines varied among vaccinated and unvaccinated respondents. Social media was the most common source of information overall (61.64% among the vaccinated and 61.33% among the unvaccinated). Among the respondents receiving information from social media, nearly all used Facebook and around half referenced Instagram. However when asked about information sources on the vaccine, media types including social media, television, radio and newspapers were not associated with vaccination status in the weighted analysis (p value > 0.05). There was no difference in social media use among vaccinated and unvaccinated individuals or in the use of specific platforms.

Other common sources of vaccine information were healthcare workers (65.31% of the vaccinated vs. 55.83% of the unvaccinated, p = 0.21); neighbors, friends, and colleagues (35.40% of the vaccinated vs. 49.73% of the unvaccinated, p < 0.01); and family (30.26% of the vaccinated vs. 37.18% of the unvaccinated, p = 0.12). Those who were vaccinated were more likely to report commonly receiving information on the vaccine from newspapers, religious leaders, civil society organizations, and to trust the information received from healthcare providers (OR p values < 0.05). The unvaccinated were more likely to receive information from mass events, family, and neighbors, friends, and colleagues (OR p values < 0.05).

Multivariable regression provided further evidence for demographic characteristics as independent determinants of vaccination status (Table 9). In the adjusted model, vaccination was around twice as likely among males (aOR 1.88, p = 0.006, 95% CI 1.20–2.95), older compared to younger adults (aOR 1.92, p < 0.001, 95% CI 1.73–2.13), and the university educated compared to those with primary or no education (aOR 2.19, 95% CI 1.51–3.17). The vaccinated were less likely to report having insufficient information on the vaccine (aOR 0.62, p < 0.001, 95% CI 0.48–0.80) as well as more likely to consider the vaccines safe (aOR 13.8, 95% CI 10.1–18.8) or somewhat safe (aOR 5.41, 95% CI 3.69–7.94). However perception of personal risk of developing severe symptoms of COVID-19 was not independently significantly associated with vaccination status (aOR 0.86, 95% CI 0.59–1.24), nor were sources of information, after controlling for demographic characteristics and other risk factors.

Vaccine hesitancy and risk factors

Vaccine hesitancy was 34.08% (95% CI 28.14–40.56) in the pooled population of vaccinated and unvaccinated individuals,Footnote 2 and 67.24% (95% CI 49.04–81.41) among the unvaccinated sub-group (n = 449).Footnote 3 Table 4 presents the unweighted and weighted distribution of the characteristics of the population by vaccine hesitancy status. Among the overall population, males were less likely to report hesitancy (p = 0.006). Other demographic characteristics were not significantly associated with vaccine hesitancy (p > 0.05).

Table 6 presents the unweighted and weighted distribution of the risk factors for vaccine hesitancy and Table 8 provides the unadjusted Odds Ratios. The main factors the vaccine hesitant reported as barriers to getting the vaccine were stress associated with the vaccine (51.03%), lack of social acceptance (31.03%), and lack of information on the vaccines (20.45%). However, these barriers were not specifically associated with vaccine hesitancy in unadjusted analysis (p > 0.05 for ORs comparing hesitant and non-hesitant individuals), suggesting they were non-specific constraints to vaccine uptake although we were underpowered for this analysis.

Common sources of vaccine information among the hesitant population were social media (58.18%); HCWs (53.29%); friends, neighbors and colleagues (52.52%); and family (41.55%). Notably, those classified as hesitant were more likely to report family (OR 1.68, p < 0.001, 95% CI 1.29–2.17); friends, neighbors and colleagues (OR 1.83, p = 0.006, 95% CI 1.19–2.82); radio (OR 1.48, p = 0.017, 95% CI 1.07–2.03); and mass events (OR 2.28, p = 0.023, 95% CI 1.12–4.62) as important sources of information, and less likely to mention HCWs (OR 0.64, p = 0.004, 95% CI 0.47–0.86); religious leaders (OR 0.51, p = 0.025, 95% CI 0.28–0.92); and civil society organization (OR 0.41, p < 0.001, 95% CI 0.24–0.69). They were also more likely to trust family (OR 1.75, p < 0.001, 95% CI 1.28–2.40); neighbors, friends and colleagues (OR 3.05, p = 0.006, 95% CI 1.37–6.77); and radio (OR 2.87, p < 0.001, 95% CI 1.99–4.14) for information on COVID-19 vaccines. They were less likely to trust HCWs (OR 0.54, p < 0.001, 95% CI 0.46–0.65); local leaders (OR 0.02, p < 0.001, 95% CI 0.00–0.18) and civil society organizations (OR 0.43, p < 0.001, 95% CI 0.34–0.56) for accurate vaccine information.

Perceptions of vaccine safety and effectiveness also appear to impact vaccine hesitancy. Those classified as hesitant were far less likely to consider the vaccines safe (2.33% vs. 30.34%, OR 0.03, 95% CI 0.01–0.16). Around 80% of both hesitant and non-hesitant groups believed they were at risk of getting COVID-19; however the hesitant were less likely to consider themselves at risk of severe disease or hospitalization if they contract COVID-19 (OR 0.69, 95% CI 0.50–0.97).

In multivariable regression, the adjusted models showed demographic factors as well as certain information sources and risk perceptions were independently associated with vaccine hesitancy (Table 10). The vaccine hesitant were less likely to be male (aOR 0.58, p = 0.004, 95% CI 0.40–0.84), older compared to younger adults (aOR 0.47, p = 0.005, 95% CI 0.28–0.80), and university educated compared to primary or no education (aOR 0.32, 95% CI 0.25–0.42). The vaccine hesitant were more likely to mention family as a common source of information on the vaccines (aOR 1.29, p = 0.051, 95% CI 1.00–1.67), and less likely to trust HCWs for vaccine information (aOR 0.58, p < 0.001, 95% CI 0.49–0.68). Perception of vaccine safety and serious personal risk from COVID-19 are also independently related to vaccine hesitancy. Those who consider the vaccine safe had 97% lower odds of being hesitant (aOR 0.03, 95% CI 0.01–0.15), while those who believed in the possibility of serious illness from COVID-19 had 38% lower odds of being hesitant (aOR 0.62, 95% CI 0.37–1.04).

Vaccine hesitancy among the non-vaccinated

In sub-analysis, risk factors for vaccine hesitancy were considered for the population of unvaccinated individuals (n = 449) (Table 11). Vaccination outreach efforts will need to specifically target the currently unvaccinated population who are hesitant (~ 67% of this sub-group). In this population, there was not enough evidence that sex, age, or educational background were associated with hesitancy (p value > 0.05). There was little variation in the information requested and information sources accessed among hesitant and non-hesitant individuals. The most common sources of information among those classified as hesitant were social media (58.32%), HCWs (53.21%), neighbors, friends and colleagues (52.29%), and family (43.28%).

Perception of vaccine safety and concern about risks of side effects were marginally not significant in the weighted analysis. However the study was not powered to investigate risk factors for vaccine hesitancy among the unvaccinated sub-group. These are likely to be true risk factors despite being underpowered, given the effect size of the estimate and significance in the primary analysis that used the pooled population of vaccinated and unvaccinated individuals. Perception that the virus posed severe individual health risks did vary with hesitancy status. 23.20% of hesitant individuals versus 41.90% of non-hesitant individuals believed they were at risk of serious disease or hospitalization if they contracted COVID-19 (p = 0.010).

Vaccination and vaccine hesitancy among healthcare workers

Table 12 presents the descriptive results for vaccine uptake and risk factors for hesitancy among HCWs. 89.35% (n = 151) of the HCWs surveyed were vaccinated and 10.65% (n = 18) were unvaccinated. Among the non-vaccinated HCWs, 50% (n = 9) were classified as hesitant. Given the small sample of non-vaccinated HCWs, few inferences can be made. However the explanations for hesitancy appear to mirror that of the population. HCWs who were hesitant were less likely to consider the vaccines safe (89% did not consider them safe compared to 11% of non-hesitant HCWs) and were more likely to express concerns about side effects (89% compared to 26% of non-hesitant HCWs).

Table 12 Factors associated with vaccine hesitancy, HCWs only
Full size table

Discussion

The beginning of the global roll-out of COVID-19 vaccines in 2021 was met by calls to ensure equitable access for less developed countries, refugees, and internally displaced populations [25]. Disparities nonetheless became apparent within the first year [26]. According to the United Nations, as of November 2021 only 4% of COVID-19 vaccine doses had been administered in the 30 countries prioritized for multilateral humanitarian assistance [27]. The evidence base on COVID-19 vaccines from these types of settings also remains limited. Most reports publicly available consist of basic vaccine administration metrics, or otherwise provide descriptive or semi-qualitative results such as a recent study in Yemen [28].

This is one of the first analytical studies of COVID-19 vaccine coverage and hesitancy from a humanitarian setting. We estimated vaccine coverage in Gaza to be 49.09% (95% CI 43.10–55.08) of the adult population. This estimate is comparable, although slightly higher, than the 44% figure reported by the WHO in its February 2022 situation report [13]. The difference could be largely attributed to the methods, whereby the WHO estimates the percentage of the population vaccinated by comparing routine vaccine administration data to population statistics. Our survey was also restricted to adults 18 years and older, while the WHO data encompasses adolescents 12 years and older who were not eligible in the early phases of vaccine roll-out. Although there is no data on parental hesitancy in Gaza, adolescents may also be less likely to receive the COVID-19 vaccine compared to adults due to parental reluctance that has been observed globally and varies across contexts [29, 30].

The study’s large sample size also enabled reasonable sub-group estimates of vaccination coverage. This is useful because routine data on COVID-19 vaccine distribution, especially from lower income countries, has generally not been disaggregated, which prevents analysis of within-country inequities [31]. In Gaza, we found moderate to large differences in vaccine coverage for each demographic characteristic assessed. Women were less likely to be vaccinated than men, with 41.45% of females vaccinated compared to 54.87% of males. Routine data from vaccine administration in the oPt corroborates this trend, as females comprised just 44% of people who received a vaccine dose as of February 2022 [13]. People with lower levels of education were also less likely to be vaccinated—37.30% of those with primary or no education compared to 43.12% of those with secondary or college education and 55.75% of those with university education. These findings suggest certain systemic inequities observed in global vaccination programs may be occurring with COVID-19 vaccination, at least in Gaza, and warrants attention in other settings. A recent meta-analysis on vaccine equity in low and middle-income countries found lower likelihood of vaccination among females, households with lower education, and poorer households [32].

In terms of hesitancy, we estimated the vaccine hesitant consist of 34.08% (95% CI 28.14–40.56) of the adult population of Gaza, which is similar to the estimate from the web-based survey (37%) conducted in Gaza in October 2020 [11]. Temporal data from vaccine hesitancy studies have generally shown levels of hesitancy reported within countries to decrease over time as vaccines are rolled out and the populations gain first-hand experience [5]. While limited comparison can be drawn to the web survey, our similar figures suggest COVID-19 vaccine confidence in Gaza has been persistently low following the initial vaccine roll-out. Follow-up surveys are warranted to monitor the coverage and effectiveness of vaccine promotion efforts.

Among the non-vaccinated, 67.24% were classified vaccine hesitant, suggesting they will be difficult to reach through improved vaccine supply alone. Characteristics of the vaccine hesitant in Gaza mirror global findings, with less education as well as lower perception of serious risk from COVID-19 linked to vaccine hesitancy [33, 34]. Among the non-vaccinated group, those classified as hesitant were less likely to consider the vaccine safe and less likely to believe they were at risk of severe disease—suggesting their perceived risk of vaccination outweighed perceived risk of infection. They are also more likely to rely on family and friends for vaccine information—a finding that mirrors the global scoping review conducted by Biwas et al. [5]. Conversely, perception of personal benefit and effectiveness of the vaccines, as well as lower concern about side effects, appears to positively influence vaccination.

Globally, there remains uncertainty about the associations between demographic characteristics and vaccine hesitancy. Research is conflicting on relative levels of vaccine hesitancy among different demographic groups and appears to be context specific. Our study was not powered to specifically investigate risk factors among non-vaccinated and vaccine hesitant sub-groups. Recent studies from the Middle East have reported females more likely to be vaccine hesitant, which our study similarly found in the primary analysis [7, 8, 34]. However, vaccination rates were also notably lower among women compared to men, and sex was not associated with vaccine hesitancy among the non-vaccinated sub-group alone. The difference in the sub-analysis suggests unvaccinated women may encounter unique barriers to accessing vaccines compared to women who are vaccinated. Additional research is warranted to assess potential explanatory factors such as economic empowerment, social freedoms, and access to information. Age was also not associated with vaccine hesitancy among the non-vaccinated in Gaza, whereas other studies have found younger groups to be more vaccine hesitant or less motivated to pursue vaccination [33]. As our study was not powered for sub-group analyses, though, it is possible that larger surveys would identify further variations.

We found high COVID-19 vaccination coverage (89%) among HCWs, similar to global reports [35, 36]. Despite the smaller number of HCWs interviewed for our study and use of purposive sampling, the high level of uptake is likely a valid representation given the vaccine mandates for this group. From June 2021, the de facto government in Gaza Strip required staff of the Ministry of Health (MoH) and other agencies to be vaccinated against COVID-19. Therefore among this group, vaccination is not necessarily a valid measure of vaccine hesitancy; a non-negligible proportion of HCWs expressed doubts about certain aspects of the vaccine. More than 15% of HCWs stated they do not consider the vaccine safe or do not know, while one-quarter of HCW respondents reported concern about side effects. Among non-vaccinated HCWs, perception of insufficient vaccine safety and efficacy appear to be major determinants of hesitancy. Recent articles by El Kibbi et al. and Heyerdahl et al. raised concerns on “unspoken vaccine hesitancy” among HCWs in the Middle East, which our own study’s findings modestly support as a subject worth monitoring [37, 38].

A major strength of our study is its broad scope which includes vaccine coverage estimates, vaccine hesitancy, comparisons among sub-groups of the populations, and risk factors associated with both vaccination and vaccine hesitancy. The findings can be used to develop more persuasive and effective information campaigns, and to proactively address the concerns of vaccine hesitant sub-groups through the information channels they most commonly use and trust. Subsequent research in Gaza could examine changes in vaccine hesitancy over time and evaluate the effectiveness of outreach campaigns and social messaging based on the emerging evidence of risk factors for vaccine hesitancy.

Limitations

The main limitation of the study is the non-standard approach to cluster sampling, whereby the clusters were selected with probability proportional to size of governorates rather than size of clusters, and the number of geo-localities selected per strata was limited to two in each stratum. We partly account for sampling bias in analysis through the sampling weights. The reliance on convenience sampling of primary survey respondents within households rather than purely random sampling of adult household members has potential for selection bias. Characteristics of people who were available and willing to participate in the survey may have differed from a random sample of eligible household members and the general population. In areas where socio-cultural norms reflect greater male dominance, men may have been more likely to answer the door or respond to the survey. The survey also excluded people who were working or moving outside the house at the time of the survey. While the direction of potential bias related to the sampling design is unclear, prevalence values and model results should be interpreted with caution.

Similarly, we collected hesitancy information from healthcare workers to investigate coverage and hesitancy among this at-risk group of frontline responders. Operational constraints in Gaza necessitated a non-random selection of health facilities that were accessible to the researchers. Due to the sampling design, the findings of the HCW survey should not be considered representative of the health workforce of Gaza. While the facilities included in our study reasonably reflect the different types of primary and secondary health services available in the territory, they are likely to differ from a random sample in terms of their financial means, workforce training, and connectedness to humanitarian aid systems. The respondents within each facility were also a self-selected sample of eligible HCWs, who may differ systematically from staff who chose not to take the survey. For example, stigma regarding vaccine hesitancy among HCWs may have driven a lower survey response rate among staff who were hesitant.

Another limitation was the use of a standard global tool for assessment of vaccine hesitancy and confidence. As such, not all reasons for hesitancy were factored into the survey and questions were not customized to the context of the Gaza Strip. We did not inquire about previous SARS-CoV-2 infection and history of COVID-19 disease, comorbidities, movement restrictions in Gaza, the recent conflict, and other individual and local factors that could influence vaccine hesitancy. The standard vaccine hesitancy tool was also not customized for the interviews with healthcare workers, so certain factors more relevant to this population—such as clinical and epidemiological knowledge, patient care experience, and employer mandates—that correlate with vaccine acceptance were not assessed here.

Conclusion

The recurrent and rapid emergence of COVID-19 variants reinforces the urgency to accelerate vaccination efforts in under-resourced and conflict-prone settings like the Gaza Strip. Scientists have linked recent variants such as Delta and Omicron to low and inequitable vaccination coverage globally, indicating the need for both supply- and demand-oriented strategies to combat the pandemic [39, 40]. Our study complements previous research on the latter through an investigation of vaccination coverage and vaccine hesitancy in the difficult public health setting of the Gaza Strip. It is one of the first population-representative studies of these topics in the Middle East and is also the first carried out in Gaza since the vaccine roll-out.

Just half of adults in Gaza had received at least one dose of a vaccine by October 2021. Disparities in vaccination coverage were apparent across the territory’s social and demographic groups; vaccination campaigns should subsequently target those who are less likely to be vaccinated, including women and those with less education. HCWs are a widely trusted information channel on COVID-19 vaccines among the vaccinated and unvaccinated, although less trusted among those who are hesitant. The latter reported that they accessed information from family, friends, neighbors, and colleagues at higher levels than the vaccinated and non-hesitant. This suggests the importance of disseminating accurate messages through community-based channels to permeate social networks of the unvaccinated and to address misinformation from these sources. Health promotion strategies to encourage vaccination should emphasize vaccine effectiveness and minimal risk of serious side effects, both of which were major concerns voiced by the unvaccinated and vaccine hesitant. The optimal approach is likely to be a combination—mobilizing people who are already well-trusted by vaccine hesitant individuals and training these potential influencers on specific topics linked to vaccine hesitancy.