Investigating the Role of Vitamin D in the Morbidity of COVID-19 Patients

STUDY DESIGN Arm 1: Hospital in-patients with a clinical diagnosis of COVID-19 This will be an observational cohort study of patients hospitalised at Tameside General Hospital with a clinical diagnosis of COVID-19, as defined by the clinical coding department. Patient identifiers will be extracted by the coding department and held on a secure server at Tameside Hospital. Identifiers will be used to ensure serum samples already taken as part of routine clinical management of these patients can be processed for a vitamin D level. No additional blood samples will need to be taken.

Patient demographic details will be extracted from their clinical notes and electronic patient records, then placed on an anonymised database for analysis. Data will be analysed to determine the following:

1. Are vitamin D levels associated with outcomes/prognosis in patients hospitalised with COVID-19?
2. Are vitamin D levels associated with ethnicity in patients hospitalised with COVID-19?
3. Are vitamin D levels associated with comorbidity in patients hospitalised with COVID-19?

Projected recruitment is 200 patients to be opportunistically recruited who are undergoing usual clinical care for COVID-19.

Arm 2: Primary care patients prospectively studied with reference to baseline vitamin D and risk of COVID-19 Patients will be prospectively recruited from local GP surgeries within the Tameside and Glossop Clinical Commissioning Group. Patients will be consented by Good Clinical Practice (GCP)-certified practitioners. A serum sample will be collected on recruitment for a baseline vitamin D level to be obtained. Patient contact details will be collected, as well as baseline health status and comorbidities, then stored on an encrypted database on a secure server at Tameside Hospital. Vitamin D levels will be communicated to GP surgeries and GPs can decide whether to treat patients according to local treatment guidelines for suboptimal vitamin D. After 6 months, patients will be followed-up via telephone interview, in order to ascertain whether they have developed COVID-19, any treatment outcomes for this, and whether they have been treated with vitamin D in the interim in primary/secondary/tertiary care.

Data will be analysed once all recruited patients have been followed-up after 6 months. At follow-up, patients who have developed COVID-19 will be deemed to be cases, and those who have not will be deemed to be controls, for a case-control analysis using vitamin D levels as an exploratory covariate. It is anticipated that there will be some patient drop-out, but compliance with follow-up is likely to be high, due to the use of a telephone interview. Analysis will aim to determine whether vitamin D levels are associated with the risk of developing COVID-19, with reference to additional covariates, such as ethnicity and comorbidity.

Projected recruitment is 800 patients to be opportunistically recruited who are attending their GP surgery for routine care, not necessarily related to COVID-19.

INCLUSION CRITERIA Arm 1

Inclusion criteria:

• Aged >=18 years.
• Inpatient admitted to Tameside General Hospital.
• Clinical diagnosis of COVID-19 - not necessary for SARS-CoV-2 swab to be positive.
• Patients on vitamin D treatment to be included, but this will be adjusted for in analysis.

Exclusion criteria:

• Aged <18 years.
• Final clinical diagnosis NOT COVID-19.

Arm 2

Inclusion criteria:

• Aged >=18 years.
• Patients on vitamin D treatment to be included, but this will be adjusted for in analysis.

Exclusion criteria:

• Aged <18 years.
• On active treatment for malignancy/diagnosis of cancer within 6 months prior to enrolment.
• Patients on immunosuppression.

ANALYSIS Arm 1

Mean values of age and gender, as well as proportions of different ethnic groups, will be used to assess group comparability. Primary analysis will involve linear regression of vitamin D levels (as a continuous variable, with adjustment for age and gender) against variables representing COVID-19 treatment outcomes:

• Mode of oxygen therapy (low-flow <10L/min, high-flow>=10L/min, nasal high-flow, CPAP, endotracheal intubation).
• Treatment history (excluding oxygen therapy).
• Discharge.
• Death.

Primary analysis will also include logistic regression of ethnic groups against the above treatment outcome variables, with adjustment for age, gender and vitamin D status.

The following variables will be included as additional predictors of severity:

• Admission SpO2.
• Chest x-ray findings on admission.
• CRP on admission.
• D-dimer on admission.
• Creatinine on admission.
• Adjusted calcium on admission.
• Treatment with vitamin D.
• Ethnicity.

The following variables will be included as covariates:

• Age.
• Gender.
• Comorbidities.
• BMI.
• Presence of diabetes.
• SARS-CoV-2 swab result.

Linear regression will be reported as a coefficient with a 95% confidence interval. Logistic regression will be reported as an adjusted odds ratio with a 95% confidence interval. Missing data will be imputed. Analysis will be carried out once all data has been collected for n=200 patients and serum vitamin D levels are available.

Arm 2

Mean values of age and gender, as well as proportions of different ethnic groups, will be used to assess group comparability. Once all data has been collected, patients who developed COVID-19 will be defined as "cases" and those who did not will be defined as "controls." A case-control study including vitamin D as a predictor variable will then be carried out, with statistics reported as an adjusted odds ratio with a 95% confidence interval. Further subanalysis will be carried out on the following outcome measures (again, being defined as a "case"):

• Hospitalisation with COVID-19.
• Discharge from hospital following treatment for COVID-19.
• Death from COVID-19.

The following variables will be included as additional predictors of severity:

• Treatment with vitamin D.
• Ethnicity.

The following variables will be included as covariates:

• Age.
• Gender.
• Comorbidities.
• BMI.
• Presence of diabetes.

Missing data at enrolment will be imputed. Patients lost to follow-up will have baseline data included and they will be treated as a "control." Analysis will be carried out once all follow-up data has been collected for n=800 patients recruited.