Objective: The specific objectives of the study are:
-
To determine the pooled prevalence of the various symptoms of long COVID from primary studies.
-
To measure summary effect sizes of the treatment approaches to long COVID from primary studies.
-
To evaluate the influence of geographic variation, race, gender and age on symptomatology, treatment approaches and the quality of life of patients with long COVID.
Review Questions:
-
What is the pooled prevalence of various symptoms of long COVID?
-
What are the various reported treatment approaches to long COVID?
-
How do factors such as geographic location, race, age, social class and gender influence symptoms of and treatments to long COVID?
Study Characteristics:
Design
This is a protocol for systematic review and meta-analysis of symptoms and treatment approaches to long COVID. It focuses on observational studies published from 2019 to present time, that are retrievable in the English Language.
Inclusion Criteria:
-
Observational studies: Cohort studies, case controls, cross-sectional studies, historic cohort studies.
-
Studies must report the primary outcome: symptoms of long COVID
-
Study must be retrievable in the English language.
Exclusion criteria:
-
Reviews, editorials, interventional studies, commentaries, methodological articles, letters to editors, case reports
-
Duplicates/ replicates of studies.
-
Studies not retrievable in the English Language.
PICOS
Populations: patients suffering long COVID.
Intervention: Various treatment approaches
Comparator: No treatment
Outcomes: The primary outcome is the proportion of patients with various symptoms of long COVID. The effect size is prevalence.
The secondary outcome is the proportion of various treatment approaches to long COVID. The effect size is prevalence.
Information sources
The search will use sensitive topic-based strategies designed for each database. The search will be carried out in the following databases: PUBMED, EMBASE, CINAHL, RESEARCHGATE, AJOL, GOOGLE SCHOLAR, WEB OF SCIENCE, SCOPUS and COCHRANE LIBRARY. Only observational studies will be included, from 2019 to present time.
Search strategy
The search strategy includes text words and entry terms. Table 1 shows the search strategy for the long COVID as used in the Pubmed. The same search strategy will be used in other databases with slight modifications.
Data Extraction and Management
Data Extraction
Data will be managed in three main softwares: DistillerSR, CMA version 3 and Microsoft Excel.
a. Screening: Identified studies will be screened independently in pairs and blindly using the DistillerSR software at 6 different levels:
i. Level 1 will involve screening of identified studies for the study design. Only observational studies would be accepted
ii. Level 2 will involve screening of identified studies in the titles and abstracts using entry terms and keywords.
iii. Level 3 will involve further screening of the contents of articles by reading the full article using the same search strategy.
iv. Level 4 will involve snowballing of literature on references from eligible studies.
v. Level 5: Studies will be screened at outcome levels to select those that reported the primary outcome with or without secondary outcomes.
vi. Level 6 will involve grey literature that report primary outcome and or secondary outcomes.
Conflicts during screening will be resolved by a third independent reviewer who serves as a tie breaker.
b. Selection Process:
Screened studies will be selected based on study charateristics: study design, inclusion/exclusion criteria and agreement between two independent and blinded reviewers. Authors of included studies with missing data will be contacted via email and telephone. After selection, studies will be deduplicated. Data items will be extracted from selected studies into predefined forms in the DistillerSR.
c. Data Collection: Data items to be extracted from selected studies include:
-
Surname of first author and year of publication
-
Symptoms of long COVID
-
Treatment approaches to long COVID
-
Socio-demographics: age, sex, race, geographic location and social class
-
Quality of life
Data items will be exported into predefined format in Microsoft Excel, to be imported into the CMA software for quantitative analysis.
Data Items/Measurable Outcomes
The key data items linked to measureable outcomes are i) various symptoms of long COVID, ii) various treatment types to long COVID, ii) socio-demographic variables, and iv) Quality of life for long COVID patients.
Risk of bias
The risk of bias (methodological quality) in the included studies will be assessed for the individual article using the National Institute of Health (NIH) Quality assessment tool for observational cohort and cross-sectional studies. The NIH Quality assessment tool has 14 questions. Studies that score 7 and above are considered good quality. This will be cross-checked with the Cochrane tool of risk of bias assessment (ROBINS-1). Publication bias in the selection of studies will be visually assessed using the funnel plot and associated variables such as trim and fill outcome, Egger’s regression intercept, Begg and Mazumdar's rank correlation and Orwin’s fail-safe N will be reported. Studies with extreme bias (NIH score less than 5) will be subjected to sensitivity testing using the include/exclude function in the CMA Software.
Assessment of Meta-bias
Meta-bias will be assessed as follows:
i) Method of reporting long COVID at outcome level. It will consider the plurality of terms.
ii) Index of reporting outcomes in studies: Studies that were reported in different indices but similar in outcome and design will be converted to the primary effect size (prevalence) based on individual case evaluation.
iii) Heterogeneity will be assessed at the study level using the Q statistics, and its p-value, I², Ʈ² (Tau squared). As a rule of thumb, I² values of less than 40% will be considered low heterogeneity while values > 40 but < 75 % will be considered moderate and values > 75% are high.
Data synthesis
Extacted data items will be used for both narrative synthesis and quantitative analysis.
The following criteria will be applied for analysis:
-
Studies that passed the methodological quality assessment using the NIH quality assessment tool will be cross-checked with the Cochrane Risk of Bias tool. The results will be presented in tabular format, indicating all the extractable data items as listed under data collection.
-
All studies with primary outcomes will be used for narrative synthesis.
-
All studies with good NIH quality scores that reported primary and or secondary outcomes will be used for quantitiative synthesis.
-
Further Analysis: Subgroup analysis will be performed using variables such as race, gender, socioenomic status, age and geographical location (country).
Meta-regression will be performed on quantitative variables such as age, proportions of treatment approaches and quality of life as explanatory variables
-
Where heterogeneity is high, sensitivity testing using include/exclude functions in the CMA software will be performed.
-
The computational model for analysis is Random effect model since the several studies across the globe will be included.
Presentation and Reporting of Results
The study selection process will be summarised in a Prisma flow chart according to the PRISMA 2015 Statement and PRISMA-P Checklist. A table of the search strategy in various databases showing text words and entry terms will be included. A list of eligible studies will be summarized in a table. Quantitative data such as prevalence of long COVID symptoms, 95 % CI, P values, and relative weights assigned to studies and heterogeneity tests will be reported in the forest plots. A table of quality scores and risk of bias of each eligible study will be included. Forest and regression plots to show sub-group analysis and meta-regression respectively will be included.