key: cord-0838104-ovd98r01 authors: Felten, Renaud; Dubois, Maxime; Ugarte-Gil, Manuel Francisco; Chaudier, Aurore; Kawka, Lou; Bergier, Hugo; Costecalde, Charlotte; Pijnenburg, Luc; Fort, Jérémy; Chatelus, Emmanuel; Sordet, Christelle; Javier, Rose-Marie; Gottenberg, Jacques-Eric; Sibilia, Jean; Fuentes-Silva, Yurilis J; Arnaud, Laurent title: Cluster analysis reveals 3 main patterns of behavior towards SARS-CoV-2 vaccination in patients with autoimmune and inflammatory diseases date: 2021-05-13 journal: Rheumatology (Oxford) DOI: 10.1093/rheumatology/keab432 sha: a00162ca213507e0d85c1496a9cb9b4e225f1059 doc_id: 838104 cord_uid: ovd98r01 INTRODUCTION: Given the COVID19 pandemic, it is crucial to understand the underlying behavioral determinants of SARS-CoV-2 vaccine hesitancy in patients with autoimmune or inflammatory rheumatic diseases AIIRD. We aimed to analyze patterns of behaviors regarding SARS-CoV-2 vaccination in AIIRD patients, as a mean to identify pragmatic actions to increase vaccine coverage in this population. METHODS: Data of 1258 AIIRD patients were analyzed using univariate and multivariate logistic regression models, to identify variables associated independently with the willingness to get vaccinated against SARS-CoV-2. Subsets of patients showing similar behaviors towards SARS-CoV-2 vaccination were characterized using cluster analysis. RESULTS: Hierarchical cluster analysis identified 3 distinct clusters of AIIRD patients. Three predominant patients’ behavior towards SARS-COV-2 vaccination: ‘voluntary’, ‘hesitant’ and ‘suspicious’ were identified. While vaccine willingness was significantly different across the 3 clusters p< 0.0001, there was no difference regarding the fear to get COVID-19 p= 0.11, the presence of co-morbidities p= 0.23, the use of glucocorticoids p= 0.21 or the immunocompromised status p= 0.63. However, patients from cluster #3 ‘suspicious’ were significantly more concerned about vaccination, the use of a new vaccine technology, the lack of hindsight regarding COVID vaccination and potential financial links with pharmaceutical companies p< 0.0001 in all than in the other 2 clusters. DISCUSSION: Importantly, the differences between patients’ behaviors are not related to the fear of getting COVID-19 or to any state of frailty, but point out to specific concerns about vaccination. This study may serve as a basis for improved communication, to increase COVID-19 vaccine coverage in AIIRD patients. The study was approved by the ethic review board of Strasbourg (#CE-2020-199) and informed consent was obtained from patients (via a dedicated question at the beginning of the online questionnaire). The data that support the findings of this study are available from the corresponding author, upon reasonable request. Given the COVID19 pandemic, it is crucial to understand the underlying behavioral determinants of SARS-CoV-2 vaccine hesitancy in patients with autoimmune or inflammatory rheumatic diseases (AIIRD). We aimed to analyze patterns of behaviors regarding SARS-CoV-2 vaccination in AIIRD patients, as a mean to identify pragmatic actions to increase vaccine coverage in this population. Data of 1258 AIIRD patients were analyzed using univariate and multivariate logistic regression models, to identify variables associated independently with the willingness to get vaccinated against SARS-CoV-2. Subsets of patients showing similar behaviors towards SARS-CoV-2 vaccination were characterized using cluster analysis. Hierarchical cluster analysis identified 3 distinct clusters of AIIRD patients. Three predominant patients' behavior towards SARS-COV-2 vaccination: 'voluntary', 'hesitant' and 'suspicious' were identified. While vaccine willingness was significantly different across the 3 clusters (p<0.0001), there was no difference regarding the fear to get COVID-19 (p=0.11), the presence of co-morbidities (p=0.23), the use of glucocorticoids (p=0.21) or the immunocompromised status (p=0.63). However, patients from cluster #3 ('suspicious') were significantly more concerned about vaccination, the use of a new vaccine technology, the lack of hindsight regarding COVID vaccination and potential financial links with pharmaceutical companies (p<0.0001 in all) than in the other 2 clusters. Importantly, the differences between patients' behaviors are not related to the fear of getting COVID-19 or to any state of frailty, but point out to specific concerns about vaccination. This study may serve as a basis for improved communication, to increase COVID-19 vaccine coverage in AIIRD patients. The SARS-CoV-2 vaccines will improve the general health situation secondary to the pandemic, by preventing future contaminations and consequently severe forms of COVID-19 [1] . The role of the rheumatologists, working together with the general practitioners as well as other specialists, is essential to increase SARS-CoV-2 vaccine coverage in patients with autoimmune or inflammatory rheumatic diseases (AIIRDs) [2, 3] which could be at increased risk of severe form of COVID-19 [4, 5] . We recently published the international VAXICOV study, which aim was to describe the expectations and potential concerns related to SARS-CoV-2 vaccination in AIIRD patients [2] . One of the main findings of VAXICOV was the limited willingness to get the SARS-CoV-2 vaccines from AIIRD patients. From a public health perspective, it is therefore crucial to understand the underlying determinants of vaccine hesitancy [6] in the context of the SARS-CoV-2 pandemic. The objective of this new study was to analyze the different types of patient behaviors regarding SARS-CoV-2 vaccination as well as their determinants, using a data-driven approach, in order to identify pragmatic means to impact positively those behaviors by the rheumatologist. The VAXICOV study consisted of an online questionnaire of 57 questions which addressed epidemiological, socio-demographic and therapeutic elements associated with expectations and potential concerns regarding SARS-CoV-2 vaccination. The online survey took place from December 12, 2020 to December 21, 2020. The questionnaire was translated from English into French and Spanish by native speakers. The detailed methods of the VAXICOV study and main findings have been already described [2] . The study included patients with a self-reported diagnosis of AIIRDs. Main study outcomes included: demographical characteristics (age, prior chronic medical conditions, marital status, children, profession, income); experiences about COVID-19; history of vaccination against seasonal influenza and pneumococcus; intention to get vaccinated if a COVID-19 vaccine were available; vaccine hesitancy; and fears and expectations about COVID-19 vaccine. Participants rated their feelings about SARS-CoV-2 vaccination using 0 to 10 scales (0: Not at all in agreement; 10: fully agree). An "immunocompromised" status was defined as participants taking at least one immunosuppressant or glucocorticoids at a dose greater than 10 mg per day of prednisone-equivalent. The study was approved by the ethic review board of Strasbourg (#CE-2020-199). Continuous data are presented as medians and their 25th-75th percentile interquartile range (IQR) and categorical data as counts and percentages. Comparisons between independent groups were made using the Mann-Whitney test for continuous outcomes and the Chi-2 test (or Fisher's exact test when appropriate) for quantitative data. Univariate and multivariate logistic regression models using a backward stepwise approach were built to identify variables associated independently with the willingness to get vaccinated against SARS-CoV-2 (selected as the dependent variable). All Likert statements with a p-value <0.10 in univariate analysis were included as independent variables in the multivariate logistic regression models. Subsets of patients showing similar specific perception about SARS-CoV-2 vaccination (based on the patient's answers to the 5 significant statements from the VAXICOV study, see results) were characterized using cluster analysis. Hierarchical cluster analysis using Ward's method was used as the principal clustering technique and the optimal number of clusters was determined using the gap/elbow method, and prespecified in a k-means cluster analysis, and the agreement between the two clustering methods was calculated using Cohen's kappa. Principal component analysis was performed using data from the statements which remained significant in the stepwise multivariable model, using a color code according to the main clustering group for each patient. All tests were bilateral and p-values <0.05 were considered statistically significant. Statistical analyses were performed with the software JMP13 (SAS institute, Cary, NC, USA). The study included 1258 patients with AIIRDs (1138 [90.5%] women and 120 [9.5%] men; median age 50 [IQR: 40-61] years), from 56 countries. The most common inflammatory or autoimmune conditions (Table 1) were Systemic Lupus Erythematosus (n=492, 39.1%), spondyloarthritis (n=174, 13.8%), rheumatoid arthritis (n=157, 12.5%) and polymyalgia rheumatic/giant cell arteritis (n=144, 11.5%). From the 14 statements assessed in VAXICOV, multivariate logistic regression models identified 5 statements independently associated with the willingness to get vaccinated against SARS-CoV-2 (#3-6-10-13-14, Table 2 ). Based on the patient's answers to these 5 significant statements, hierarchical cluster analysis identified 3 distinct clusters within the study population (labeled cluster #1, #2 and #3, 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 the 3 clusters are shown in table 3 . The concordance with k-means clustering was high (Cohen's kappa: 0.76). Vaccine willingness was significantly different across the 3 clusters ( While vaccine willingness and hesitancy were significantly different across the 3 clusters (p<0.0001, (Table 3) . Also, countries of origin (p<0.0001) and rheumatic diseases (p<0.0001) were statistically different between the 3 clusters. Importantly, there were no significant difference as to the presence of co-morbidities (p=0.23), the use of glucocorticoids (p=0.21) or the immunocompromised status (p=0.63) ( Table 3) . 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Our results based on a cluster analysis of a large international sample of patients with autoimmune and inflammatory diseases enabled the identification of 3 distinct clusters of behavior towards SARS-COV-2 vaccination. Identifying those patients' behaviors can help tailoring the approach to be adopted by healthcare professionals and governments at the individual and collective level, in order to improve SARS-CoV-2 vaccination coverage. Based on the willingness to get vaccinated and the answers to the statements presented in tables 2 and 4, cluster #1 members can be seen as voluntary, cluster #2 members as hesitant, and cluster #3 members as suspicious towards SARS-CoV-2 vaccination. Contrary to what might have been expected, willingness to get vaccinated was not statistically related to increased frailty, as evidenced by the lack of difference in terms of immunosuppression or associated comorbidities. Importantly, the fear of getting COVID19 was high among the 3 groups. While this does not mean it is an irrelevant factor for why someone would ultimately get vaccinated, the fear of getting COVID-19 or a severe form of SARS-CoV-2 infection was not statistically different across the 3 clusters therefore suggesting that there seems to be no relevance to use this element as an anchor to convince AIIRD patients to get vaccinated. This study further highlighted important differences which may exist between countries regarding vaccine willingness (Table 3 ). This had already been shown in a worldwide study showing disparities in acceptance of COVID vaccination ranging from 90% in China to less than 55% in Russia [7] . Also, another important finding was the differences in age across the three clusters (Table 3 ). In the VAXICOV study, the willingness to get vaccinated increased significantly with age [2] . Similarly, COVID-19 vaccine hesitancy and refusal were both associated significantly to age in a previous study [8] . Interestingly, we point out that this is not explained by a difference in the fear of getting COVID-19 or severe COVID- What differed the most between the 3 clusters were the concerns about vaccination in general or about COVID-19 vaccines, the lack of hindsight regarding the use of new vaccine technologies (such as RNA vaccines) and the possible financial links with the pharmaceutical companies. Patients for whom giving the medical advice to get vaccinated would be the most effective were those from cluster #2 "hesitant" and #3 "suspicious" (Table 3 ). Accordingly and based on significant concerns identified in this study, the physician's attitude towards a reluctant patient with AIIRD should be to discuss the general principles behind vaccination, including its immunological basis, the main efficacy and safety data from the large phase 3 trials [9] [10] [11] [12] and the general drug approval and licensing process in order to respond to the patients' fears on those possible points. This is an important point as it is the medical specialist who is the most trustworthy in all clusters. Another issue in 'hesitant' or 'suspicious' patients was the potential induction of a flare by the SARS-CoV-2 vaccine. This point seems worth discussing with patients, and it is important to emphasize that the benefit-risk ratio likely favors vaccination. International registries will probably provide clarification for our patients in the near future. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Table 2 . Association between the risk to decline SARS-CoV-2 vaccination and VAXICOV statements. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Severe Covid-19 Vaccination against COVID-19: Expectations and concerns of patients with autoimmune and rheumatic diseases Perspective of patients with autoimmune diseases on COVID-19 vaccination Severity of COVID-19 and survival in patients with rheumatic and inflammatory diseases: data from the French RMD COVID-19 cohort of 694 patients Factors associated with COVID-19-related death in people with rheumatic diseases: results from the COVID-19 Global Rheumatology Alliance physician-reported registry Group on Vaccine Hesitancy. Vaccine hesitancy: Definition, scope and determinants A global survey of potential acceptance of a COVID-19 vaccine COVID-19 vaccine hesitancy in a representative working-age population in France: a survey experiment based on vaccine characteristics Safety and Efficacy of the BNT162b2 mRNA Covid-19 Efficacy and Safety of the mRNA-1273 SARS-CoV-2 Safety and immunogenicity of ChAdOx1 nCoV-19 vaccine administered in a prime-boost regimen in young and old adults (COV002): a single-blind, randomised, controlled, phase 2/3 trial Safety and efficacy of an rAd26 and rAd5 vector-based heterologous prime-boost COVID-19 vaccine: an interim analysis of a randomised controlled phase 3 trial in Russia We wish to acknowledge the crucial role of the following patient associations: