The QENSIU is Scotland’s sole centre for treating TSCI. Information about the Unit’s creation and funding has been previously described [10]. In 2019, the last full calendar year prior to COVID-19 reaching the UK, 41% of patients were admitted to the QENSIU within 48 h of injury, with 66% of all patients admitted within one week [11]. It is assumed all TSCIs that occur in Scotland will eventually be admitted to the QENSIU. Details of the aetiology, gender, age injury level and severity of all new admissions are entered into the QENSIU database. The database does not include information about people who die due to TSCI prior to admission.

Approval for the collection and evaluation of data within the database was granted by the Data Custodian for the Queen Elizabeth University Hospital, Glasgow, Scotland.

All patients admitted to the QENSIU during the study period (1st January 2015 to 31st August 2022) were ascertained from the clinical database. The neurological level of injury and degree of impairment after TSCI was assessed by a Consultant in Spinal Injuries on admission and defined according to the International Neurological Classification of Spinal Injury using the American Spinal Injury Association Impairment Scale (AIS) [12]. Patients who were neurologically intact, and who were recorded as an AIS E on admission, were excluded along with those under 16 years of age. TSCI aetiology was classified in accordance with the International SCI core data set as assault, fall, transport, sports and leisure (including cycling for consistency with previous work [10] and falls that occurred during sporting activities, such as rock and mountain climbing) and other traumatic (deliberate self-harm (DSH), iatrogenic [13], and industrial) [14].

All data were analysed according to the level of restrictions placed on the public at that point in time. Scotland saw four different levels of lockdown applied during the COVID-19 pandemic, with Level 1 being the least severe and Level 4 being the most severe (equivalent to a full population-level lockdown) [15]. These levels are summarised in Fig. 1. January 2015 to 31st March 2020 inclusive were denoted as the pre COVID-19 period, while September 2021 to August 2022 was classified as the post COVID-19 period. Incidence per million was calculated by comparing the incidence of TSCI in a calendar year with the corresponding years midyear population estimate [16]. As data was not yet available for 2022, the 2021 midyear population estimate was also used for 2022.

Periods of COVID-19 associated lockdowns are shown in grey dashed line. Level 0 = no lockdown measures. Level 1 = Restrictions on indoor meetings between households (maximum of 6 people from 2 households). Level 2 = As level 1, plus no indoor meeting with other households with restrictions also placed on outdoor meetings (maximum of 6 people from 2 households). Level 3 = As level 2, plus no alcohol sales indoors and outdoors with hospitality venues all to close by 6 pm. Level 4 = As level 3, plus closure of all non-essential shops, hospitality venues and gyms. The population is encouraged to only leave home for essential reasons (shopping, health care appointments etc). Equivalent to a full population-level lockdown.

Poisson regression models were used to examine the number of monthly TSCIs, adjusted for age, sex, year, season, tetraplegia, and injury completeness. The level of Covid-19 was the primary exposure variable. There is no evidence of over-dispersion. The model residual deviance/degrees of freedom was 1226.4/1462 = 0.84. This indicates mild under-inflation which could lead to slightly conservative standard error estimates. Year and month were initially modelled as penalised cubic splines in the generalised additive model framework as a preliminary analysis. The model revealed linear association for a year and non-linear association for a month (Supplementary Figure. 1). The Month variable was categorised into seasons (December-February; March-May; June-August; September-November) based on the inflection points in the spline. The exposure variable was collapsed into a binary variable: No restriction to level 2; level 3 and 4 restrictions. Additive interactions between Covid-19 restrictions and: age (<45 vs. >=45), sex (female vs. male), tetraplegia, and complete injury were examined using relative excess risk due to interaction (RERI). Under this model, we estimated the counterfactual number of SCI cases as if level 3 and 4 restrictions did not occur. These, compared with the observed SCI cases, were used to estimate the number of reduced SCI cases due to COVID-19. Incidence was also analysed with a supplementary exposure variable of COVID-19 Stringency Index [17]. This score is a measure of lockdown level summated over 9 domains - school closures; workplace closures; cancellation of public events; restrictions on public gatherings; closures of public transport; stay-at-home requirements; public information campaigns; restrictions on internal movements; and international travel controls. The index is calculated as the mean score of the nine metrics, each taking a value of between 0 and 100. A higher score indicates a stricter response (i.e. 100 = strictest response). If policies vary at the subnational level, the index is shown as the response level of the strictest sub-region. The incidence of deliberate self-harm was described in different periods since there was not sufficient power to conduct a formal analysis. R Statistical Software (version 4.2.2) was used with packages mgcv and interactionR. Descriptive data are presented as means plus standard deviations or 95% confidence intervals [14, 18]. Incidence rate ratio (IRR) and 95% CI were used to infer associations and their corresponding precision.