This study comprised a prospective cross-sectional sample survey of individuals living with SCI in Bangladesh over a 3-month period, who were in active in-patient rehabilitation.

The study took place in two tertiary rehabilitation centres in Bangladesh, namely the Spinal Cord Injury Rehabilitation Unit of the Centre for the Rehabilitation of the Paralysed (CRP) and the National Institute of Traumatology and Orthopaedic Rehabilitation (NITOR). In Bangladesh, only these two rehabilitation centres were provided rehabilitation treatment for the people living with SCI.

Inclusion criteria for the study were identified as a confirmed medical diagnosis of SCI and required admission to these Rehabilitation centres for active rehabilitation intervention From July to September 2020, all patients admitted for rehabilitation in CRP and NITOR were invited to take part in this study.

This enabled the inclusion of a wide age range of participants. Individuals who agreed and consented to participate in this study were admitted during the said time period to take part in the study. Exclusion criteria of this study were those who were not admitted in the mentioned time period and who did not give their consent, aged <15 years, and second-time rehabilitation or follow-up patients. According to inclusion and exclusion criteria, there were a total number of 207 participants being recruited.

The study used a hospital-based random sampling procedure where participants were selected based on the inclusion and exclusion criteria. From July 2020 to September 2020, a total of 171 patients were admitted at CPR and 117 in NITOR. A total of 153 patients from CRP and 104 patients from NITOR matched the inclusion criteria. However, consent for data collection was given by 149 and 73 patients from CRP and NITOR before starting the data collection, prior permission from CRP and NITOR was taken.

A clear flow diagram of the study procedure has been given in Figure 1 to meet the quality by Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) requirements (13). From July to September 2020, a group of trained data collectors collected the data by face-to-face interview and documented the answers given by participants.

The questionnaire was a mixed type (semi-structured and structured), validated, translated, and customised to fulfil the objectives of the study. American Spinal Injury Association (ASIA) impairment scales A, B, C, D, and E had been used to categorise the patients with SCI (14) which were collected from the record file of hospital patients. The knowledge, attitude, and practise (KAP) questionnaires were 16-score pretested questionnaires taken from a Chinese study (15), customised, translated, and used in a population-based study in Bangladesh (11) where the knowledge section had 12 score questions, attitude, and practise both had 2 scored questions each. There were categorical answers to attitudes and beliefs towards the control of the pandemic, and practises of wearing masks and avoiding public gatherings. The health status and co-morbidities questionnaire (16) (heart disease, blood pressure, respiratory disease, diabetes, ulcer, kidney disease, liver disease, anaemia pressure sore, depression, urinary tract infection, postural hypotension, bowel bladder dysfunction, oedema, muscle spasm, and autonomic dysreflexia) were modified for SCI respondents and taken from a population-based study (17). The depression, anxiety, and stress have been assessed by Depression, Anxiety, Stress Scale (DASS-21), which is validated (18) to use in SCI subjects and had an overall Cronbach alpha score of α0.743 and depression, anxiety, and stress each subscale had α score 0.98, 0.964, and 0.919, respectively, in our study which were indicated satisfactory level of internal consistency with these specific samples. DASS-21 has 7 items for each depression, anxiety, and stress subscales. The sum score of DASS-21 each subscale was divided into cut-off points for the severity rating scale which was also used in a previous study (19). For language validation, Bengali and the linguistic validation process was followed as per WHO guidance (20), first, the questionnaire was translated in forwarding and backward translation to English and Bengali by a group led by a bilingual British researcher. Later it was evaluated and commented on by a bilingual epidemiologist, a psychiatrist, and a clinician experienced in working in SCI-related research.

The knowledge score was calculated out of 12 scores as per the correct answers provided and considered as continuous data. The attitude and practise answers were categorical data and DASS-21 was calculated separately in depression, anxiety, and stress out of 21 scores; those were continuous data. Descriptive statistics employed in socio-demographic data, to measure relationships among continuous and categorical data-independent sample t-test to compare the means for two groups, one-way ANOVA test for categorical data and continuous data. Further Binary logistic regression has been employed using dichotomous variables as the dependent variable and KAP and DASS 21 as covariates. The Pearson test has been used to measure the relationship between two continuous variables. Data analysis was done using the IBM-SPSS version 20. The alpha level of significance was set at P < 0.05.

The socio-demographic data of the SCI respondents is shown in Table 1. Out of a total of 207 respondents, 180 were men and 27 were women. The mean age of participants was 34.18 ± 12.9 years, with the majority of participants (70.5%) aged between 19 and 40 years and a smaller group (21.25%) were aged between 41 and 60 years. The majority of participants had completed secondary education (27%) and 20% of participants were identified as illiterate. Furthermore, 70% of participants were receiving rehabilitation treatment at CRP and 67% resided in a rural area. Participants resided within and across all the administrative divisions of Bangladesh. The majority of the participants (63%) were diagnosed with an ASIA level of SCI and 137 (66%) individuals were paraplegic, and 70 (34%) participants were tetraplegic.

In the International Standard for Neurological Classification of Spinal Cord Injury (ISNCSCI) the mean motor score was 45.38 ± 19.5 out of 100 and the sensory score mean was 97.2 ± 52 out of 253. Oxygen saturation (SpO₂) mean was 95.07 ± 3.3 and the maximum rate of oxygen consumption (Vo₂max) found was 35.7 ± 3.7 ml/kg/min. From multiple response analysis, respondents suffering from comorbidities or SCI related complications had bowel bladder complication 95.65%, existing pressure score 93.24%, muscle spasm 70.05%, depressive symptoms 60.87%, postural hypotension 35.75%, respiratory disease 25.24%, high blood pressure 25.12%, and urinary tract infection 22.22% (Details in Figure 2).

The overall mean knowledge score was 8.59 ± 2.3 out of 12. Women had a greater mean (12.15 ± 9.4) than men (11.04 ± 7.8). There was a higher knowledge score in the age group 19–40 years (9.06 ± 2) but all age groups had a mean of more than 7. The graduates have more knowledge (10.11 ± 1.9) but the illiterate people also have sufficient knowledge of COVID-19 (8.21 ± 2.0). CRP (8.53 ± 2) and NITOR (8.7 2 ± 2.3) have similar knowledge scores, also the knowledge found satisfactory in rural and urban dwellers. Participants from the Chittagong administrative division had the highest knowledge score of 9.67 ± 0.99, then participants from Rajshahi division, Sylhet division, Khulna division, and Dhaka division had knowledge scores of 9.14 ± 4.3, 9.11 ± 1.2, 8.59 ± 2.1, and 8.27 ± 2.1, respectively. On the other hand, the Rangpur division has the lowest knowledge score 2.5 ± 2.3. Participants with the different ASIA diagnostic criteria had nearly similar knowledge scores. Statistical associations were observed with gender and knowledge (t = 2.06, p < 0.05), age with knowledge (F = 7.905, p < 0.05), education and knowledge score (F = 5.87, p < 0.01) and administrative divisions with knowledge score (F = 2.68, p < 0.01). Details are supplied in Table 1.

In Table 2, binary logistic regression found positive relationships between knowledge scores and gender, wherein men had 6.6 times more knowledge score than women (β = 1.89, odds ratio [OR] = 6.6, p < 0.01). Knowledge scores and age group of 19 to 40 years were negatively associated than others, wherein this age group has 0.41 times less score than other age groups (β = −0.879, OR = 0.41, p < 0.01). Literate participants and knowledge scores were positively associated wherein literate respondents had 3.8 times higher knowledge scores than illiterate respondents (β = 1.33, OR = 3.8, p < 0.01). Respondents from NITOR had negatively associated with knowledge score, wherein knowledge of respondents from NITOR had 0.41 times less knowledge score compared to respondents from CRP (β = −0.873, OR = 0.418, p < 0.05), Rural people knowledge scores were negatively associated than urban residency, wherein they were 0.48 times less compared to urban (β = −0.715, OR = 0.48, p < 0.01). Tetraplegic respondents' knowledge score was negatively associated, wherein their knowledge score was 0.511 less then paraplegic respondents (β = −0.671, OR = 0.511, p < 0.01). Table 3 shows that no statistically significant relationship was found between knowledge scores and family income (r = 0.107, P = 0.126), depression (r =−0.06, P = 0.39), anxiety (r = 0.04, P = 0.49), and stress (r = 0.05, P = 0.42).

The majority of the participants regardless of age, gender and location of geographical residency had agreed on positive attitude (belief in control and world can win) towards COVID-19. Table 4 shows the statistical relationship between education and belief in control (X² = 14, p < 0.05); residential living area with belief in control (X² = 9.4, p < 0.05) and world can win (X² = 8.84, p < 0.05); and geographical area of residence with world can win (X² = 4.19, p < 0.05).

The majority of participants (66%) did not go into crowds and wore a face mask during every exposure (80%) (Table 4). 32% of the participants who were men went into a crowd at least once. Furthermore, 18% of participants who were men as compared with 17% of women did not wear a face mask during exposure to outside people. There was a statistically significant relationship found between age and the practise of wearing a face mask (X² = 17.6, p < 0.05), geographical location of residence, and practise of solation (X² = 6.4, p < 0.05). In addition, there was a statistically significant relationship found between participants diagnosed with complete SCI and the behavioural practise of wearing a face mask (X² = 17.9. p < 0.05) (Details are in Table 4). Participants also reported that they and their caregivers followed health advice by consulting health professionals (65.7%), implementing isolation (63.8%), taking droplet precaution care (87.4%), and hygiene care (90.3%).

Among the participants, 52.7% had mild to extremely severe levels of depression, whereas 6.8% and 5.8% had severe and extreme depression, respectively (Figure 3). The overall mean depression score was 11.18 ± 8 out of 21. Women had a higher depression score (12.15 ± 9.4) than men (11.04 ± 7.8). Elderly people, more than 60 years old had a higher depression score (13.8 ± 7.8) than other age groups. Urban people had a higher depression score (12.76 ± 10) than rural or semi-urban dwellers. There was a relationship between depression with living area (F = 3.55, p < 0.05) (Table 1). Table 2 shows that men had a higher depression score compared to women where men respondents depression score was 6.6 times higher than women respondents (β = 1.89, OR = 6.6, p < 0.01), and the younger aged people between 19 and 40 years were negatively associated with depression score where 19 to 40 years old participants had 0.41 less depression score than the other age ranges (β = −0.873, OR = 0.418, p < 0.01). Literate participants had 3.8 times higher depression scores when compared to illiterate participants (β = 1.33, OR = 3.8, p < 0.05) and were positively associated with depression score; Depression score from respondents of CRP was negatively associated, wherein respondents from CRP had 0.41 times less depression compared to the respondents from NITOR (β = −873, OR = 0.418, p < 0.05); Participant's residents in rural areas showed negative association with depression scores where rural respondents had 0.48 times less depression compared those residents in urban areas (β = −0.715, OR = 0.48, p < 0.05). Tetraplegic participants showed a negative association with depression scores where tetraplegic respondents had 0.51 times less depression than paraplegic respondents (β = −0.671, OR = 0.511, p < 0.01). Table 3 represents statistical relationships among depression and anxiety (r = 0.45, p < 0.01), and depression with stress (r = 0.58, p < 0.01).

Among 207 respondents, 48.8% had a different level of anxiety, of which 7.2 and 3.4% reported that they had severe and extreme severe anxiety, respectively (Figure 3). The overall mean anxiety score was 7.72 ± 5.1 out of 21. Table 1 represents women (8.81 ± 6.6) who had higher anxiety scores than men (7.56 ± 4.9). The associated characteristics of participants who had more anxiety were those aged 41 to 60 years, those who had completed higher secondary education (8.46 ± 5.4), those who were resident in urban areas (9 ± 7.8), those who were diagnosed ASIA-B (8.4 ± 6.7), and those with having two comorbidities (8.33 ± 5.3). Binary logistic regression found a negative association with anxiety score and tetraplegic respondents where tetraplegic respondents had 0.51 times less anxiety than paraplegic respondents (β = −0.671, OR = 0.511, p < 0.01). Pearson correlation found highly statistical significance (p < 0.01) of the relationship between anxiety with depression (r = 0.45) and anxiety with stress (r = 0.47) (Table 3).

Among the participants, 16.9% reported that they had anxiety where 3.4% had a moderate level of stress, 1% had a severe level of stress, and 1.4% had an extremely severe level of stress (Figure 3). The overall mean stress score was 9.32 ± 6.7 out of 21. Women had higher stress (10.67 ± 11.2) than men (9.12 ± 5.8) (Table 1). A statistical association was found between stress with living area (F = 7.07, p < 0.01) where urban people had more stress scores than others (13.12 ± 11.1). In Binary logistic regression results, men respondents showed a positive association with stress scores whereas men respondents had a 6.6 times higher stress score compared to women respondents (β = 1.89, OR = 6.6, p < 0.01) and tetraplegic respondents had shown a negative association with stress score where tetraplegic respondents had.51 times less stress score than paraplegic respondents (β = −0.671, OR = 0.511, p < 0.01) (Table 2). In Table 3, a relationship was found between stress and depression (r = 0.58, p < 0.01) and anxiety (r = 0.476, p < 0.01).

In the perspective of Bangladesh, there is no other study on the knowledge, attitudes, behavioural practises, and psychological impact relating to COVID-19 for vulnerable people with SCI during the COVID pandemic. This study also offers endorsement to the premise that rehabilitation institutions can play an important and effective role in the protection, health promotion, and rehabilitation of vulnerable people, such as those with diagnosed SCI during the COVID-19 pandemic. However, the study population was limited to those who were receiving active in-patient rehabilitation in CRP and NITOR. As a result, the data do not reflect those people with SCI who stayed at home during this time. Because the data was only collected once, there was no way to compare or measure the changes in knowledge, attitude, and practise before and after the rehabilitation phase.