Study design and setting

STRONGER is an international, investigator-initiated and conducted, multicentre, prospective, randomised, open label, blinded endpoint study, with a fixed time point for outcome assessments. Open label indicates that participants and researchers are aware of the treatment allocation. Participants are centrally randomised (1:1) to either the intervention group to receive 40 mg atorvastatin for 12 months on top of usual care or to the control group to receive usual care. Recruitment began on 5 May 2022 and ended in July 2024. The follow-up of participants is until July 2025, and the results are planned to be presented in October 2025.

The study was approved for conduct at three medical clinic sites: the Brain and Mind Centre of the University of Sydney, Monash University in Australia (central ethics approval from Sydney Local Health District Royal Prince Alfred Hospital Ethics, numbers X21-0113 and 2021/ETH00777 10, date of first approval 19 May 2021) and at the Clínica Alemana Universidad del Desarrollo, Santiago, Chile (Comité ético científico: Facultad de Medicina, Clínica Alemana de Santiago, Universidad del Desarrollo, number 2021–75, date of first approval 20 August 2021; government approval: number EC1707587, date: 17 November 2021) as outlined in online supplemental file 1. All subjects provided written informed consent according to the forms shown in online supplemental file 2.

Eligibility of study participants

Eligible participants are aged ≥18 years, have a history of COVID-19 that is confirmed by a positive PCR test, a rapid antigen test or as per local guidelines for COVID-19 diagnosis at the time of screening and have ongoing neurological symptoms as a result of COVID-19. These symptoms included problems with memory, concentration, sleep disturbance and fatigue and were systematically identified through administration of the Somatic and Psychological Health Report-34 item questionnaire34 and of any reported loss of smell (anosmia). They must be able to participate in all procedures and provide written informed consent.

Participants were excluded if they had any of the following: evidence of dementia and/or significant cognitive impairment on screening; a severe comorbid medical (eg, renal failure) or psychiatric condition (ie, drug or alcohol dependence and schizophrenia) that prevented participation; history of traumatic brain injury with loss of consciousness (>30 min) within the last 2 years; ongoing long-term use or clear indication or contraindication for statin use; evidence of severe or significant liver disease; creatine kinase levels more than twice the upper limit of normal; being female of childbearing potential and unable or unwilling to use a reliable method of contraception, currently breastfeeding or planned pregnancy. For the subgroup of participants who agree to undergo MRI, they must have had no contraindication due to metallic body parts or claustrophobia. Finally, participants were excluded if their medical history might, in the opinion of their treating physician, put them at significant risk if they were to participate in the trial.

Intervention

Participants who meet the eligibility criteria were randomised to receive standard care or atorvastatin 40 mg on top of standard care for a period of 12 months (figure 1). The randomisation was stratified by country, time (<6 vs ≥6 months since acute COVID-19 illness), age (<60 vs ≥60 years), current anosmia (yes vs no) that has occurred with temporal relation to COVID-19 and participation in the MRI/biomarker substudy. The randomisation allocation was blinded to researchers conducting the cognitive assessments and endpoint adjudication, and participants, physicians and other study team members were aware of the treatment allocation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Flow chart of study design.

Visit summary

The study incorporates a blend of inperson visits and telephone interviews (figure 1). Preliminary screening was via telephone or teleconference to obtain informed consent and for review of the inclusion and exclusion criteria. Inperson visits encompass baseline assessments, a clinical evaluation at 6 months and an end-of-study assessment at 12 months. During the baseline visit, various evaluations were performed, including fasting blood tests, clinical assessment, cognitive tests, health questionnaires and physical activity assessments (table 1). Randomisation was undertaken after these assessments were completed, with the participants only being informed of their allocation when all the baseline evaluations were finished, to mitigate the risk of differential withdrawal between the study arms. Furthermore, participants who were enrolled in the substudy are required to undergo MRI scans and blood biomarker analyses at baseline and the end of follow-up. The 6-month visit focused on clinical evaluation and adverse event assessment within the treatment group. On the 12-month follow-up, the baseline assessments are repeated.

Telephone interviews complement the inperson visits, particularly at the 6-week mark, to evaluate adverse events, medication adherence and changes in concomitant medication. Every effort is made to conduct assessments remotely for participants unable to attend inperson visits.

Outcomes

Primary outcome is processing speed, assessed by the SDMT,35 36 which evaluates frontal lobe executive processing. Participants match numbers to symbols based on a cue of 9 abstract symbols paired with numbers. Performance is measured by the number of correct symbols within 90 s.

Secondary outcomes include a comprehensive battery of assessments covering executive functions, memory, processing speed and other domains, alongside evaluations of health status, MRI and blood biomarkers. A cost-effectiveness analysis, relative to standard care, will be executed. The table provides a detailed overview of all evaluations.

Cognitive assessments are administered by trained research psychologists specialised in cognitive measures. Importantly, these assessments can be conducted either in person or via videoconference.

Safety outcomes

This study documents expected adverse reactions to the study medication of special interest (AESI), including myalgia, nausea, elevated blood glucose, elevated creatine kinase, abdominal pain, new-onset diabetes mellitus and rhabdomyolysis. These AESIs are closely monitored at specific visits to assess participants’ tolerability to study drugs, noting whether they are new or continuing from baseline, regardless of severity. Additionally, all serious adverse events (SAEs)37 and adverse events (AEs) are reported. If treatment is discontinued as a result of any AE, serious or not, the site study team is to provide details of all events that led to the discontinuation of treatment.

Sample size

A sample of 410 patients was estimated to provide 80% power (α=0.05) to detect at least a 0.3 SD effect size difference between groups, assuming equal group participation. These calculations assume a modest 5% non-compliance and 5% dropout over 12 months of follow-up. The SDMT age-adjusted mean score is estimated at 60 (SD 13) at baseline (based on healthy control data).35 The effect is based on trials of the use of statins for the prevention of dementia and trials of treatments for multiple sclerosis, where achieved effect sizes of 0.3–0.4 are clinically meaningful and likely to confer public health benefits.38 For the substudy of participants undergoing MRI, a sample size of 220 (110 per group) is estimated to provide 80% power (α=0.05) to detect an effect size of relative difference of 5.0–6.5 (variance between groups/variance within groups) for the imaging endpoints, assuming a 20% drop out. The study statistician (XW) used the software package PASS V.15 (NCSS, Kaysville, Utah, USA; http://www.ncss.com/software/pass) for these power calculations.

Data collection methods

All data entry is completed via a secure web-based data management system (IBM Clinical Development) at The George Institute for Global Health. Data entry is performed at the participating sites by authorised site staff who have completed training and been given appropriate role-based access to the system. Data logic and consistency checks are programmed into the data entry forms so that data entry errors are captured in real-time and queries are auto-generated. Authorised electronic signatures are used to lock completed data entry forms once all data queries have been resolved within the system. Data entry and all subsequent changes or deletions are captured in an accessible audit trail. Coding of outcomes is centrally performed either automatically via the IBM coding module or manually by the Central Coordinating Centre (CCC). All coding is reviewed and verified by a medical monitor. Reporting within the system is used for regular data reviews and overall trial monitoring. Data are stored and backed up on the IBM cloud servers in the USA.

Statistical analysis plan

The study will follow the intention-to-treat principle for analysis. Baseline characteristics will be summarised by treatment group. Continuous baseline characteristics will be described by means (SD), if approximately normally distributed, or by medians (IQR), if normality cannot be assumed. Categorical baseline characteristics will be presented by frequency per category. Statistical comparisons of baseline characteristics between treatment and control groups are not planned.

The primary endpoint, SDMT score, will be summarised by means (SD) if normally distributed, or by medians (IQR) if normality cannot be assumed. The primary analysis will be conducted using linear regression, with the dependent variable being SDMT and the independent variable being group allocation (treatment vs control) with adjustment for the covariates of age, sex and site. A sensitivity analysis will be undertaken with adjustment for any meaningful baseline imbalances and clinically confounders. Another sensitivity analysis will be conducted by dichotomising SDMT according to population norms (average scores of 54 in Australia and 53.2 in Chile for people aged 44 years) and use of logistic regression models with adjustment for age, sex and study site.

Continuous secondary endpoints will be analysed using linear regression, adjusting for age, sex, study site, unbalanced baseline characteristics and clinically meaningful confounders. Binary secondary endpoints will be analysed using binary logistic regression, adjusting for age, sex, study site, unbalanced baseline characteristics and clinically meaningful confounders. Ordinal secondary endpoints with more than two categories will be analysed using a cumulative logit model, adjusting for age, sex, study site, unbalanced baseline characteristics and clinically meaningful confounders and with testing of the proportional odds assumption.

Descriptive statistics will be provided for safety data, where SAEs and treatment discontinuation will be tabulated using standard terminology. Heterogeneity of the treatment on the primary endpoint will be assessed in the predefined subgroups of age (<60 vs ≥60 years), time since COVID-19 diagnosis (<6 vs ≥6 months), baseline C-reactive protein levels (0–9 vs ≥10 mg/L), ethnicity (white Caucasian vs other) and prior cardiovascular risk (no vs yes for any of a history of hypertension, hyperlipidaemia, current smoker and body mass index (height (cm)/weight (kg) ≥10).2 The hypothesis is that there will be a larger relative treatment effect in younger people, in people with higher levels of inflammation by virtue of an earlier time from the acute COVID-19 infection, in people with raised inflammatory markers, in non-white individuals and those with elevated cardiovascular risk. A detailed outline of the statistical analyses will be specified a priori in a full statistical analysis plan prior to unblinding of the data. A modelled cost-utility analysis using trial data (health-related quality of life, captured by EuroQoL Group 5-Dimension 5-level self-report questionnaire; drug costs; health service utilisation costs, including AEs) will be conducted by comparing use of atorvastatin with standard of care. A 5-year time horizon will be undertaken, with analyses conducted in line with accepted Australian standards for use of economic evaluation in decision-making.

Data monitoring

Trial data are monitored, using central risk-based monitoring principles, to detect any unusual patterns of data that would require further investigation. During the study, representatives of the CCC monitor site performance and quality via remote methods, including via videoconference, to ensure that the study is conducted in accordance with the protocol, International Conference on Harmonisation guidelines for Good Clinical Practice (ICH-GCP) guidelines and relevant ethical and regulatory requirements.

Areas of potential bias

As this is a clinical trial, a major source of bias is selection, so we plan to compare the characteristics of participants with those in published cohorts and between those who were screened. Being an open trial, another main source of bias relates to differential drop-out, for example, participants in the control ‘no treatment’ group failing to return to follow-up assessments or withdrawing from the study. We will try to reduce this by ensuring that all participants fully understand the importance of participation and follow-up, irrespective of ceasing the study medication, and in maintaining good communication and relationships with participants. A final source of bias is behavioural, where participants in either group may change their lifestyle and attitude simply due to participation or in response to factors outside of the trial. We will monitor this by asking questions about use of concomitant medications, diet, smoking and exercise and through use of actigraphy as a proxy measure of sleep and physical activity. There is a low likelihood that the control ‘no treatment’ group will receive atorvastatin (ie, treatment crossover) as this can only be prescribed for people who are at high risk of cardiovascular disease.

Patient and public involvement

The STRONGER study protocol was presented to the Brain Health Consumer Panel of The George Institute for Global Health at meetings in March and November of 2021. This panel comprises a dozen people with lived experience of brain health conditions or their caregivers. Feedback was sought regarding the protocol and patient-facing documents. Those with lived experience provided comments on the feasibility of the study, mode of recruitment, activities and information dissemination. This feedback was incorporated into the study where appropriate, and there was agreement to ensure that the findings of the study, along with a plain language summary, will be disseminated to participants of the trial after the study is published.