Part I Meta-analysis
Literature search in Pubmed with terms “meta-analysis” and “ COVID-19 mortality”, I found a good article to review and summarize data entitled, “ Is diabetes mellitus associated with mortality and severity of COVID-19 ? A meta-analysis”.1
Definitions, terms, description and purpose within meta-analyses:-
Forest plot is a visual tool for a quick assessment of study heterogeneity and a pooled result from combining individual studies. It contains the study or subgroup (first author and the year of publication), the effect estimates information (log odds ratio, or risk ratio and confidence interval), overall statistics (heterogeneity and overall effect) and a graphical representation of odd ratio or risk ratio with a vertical line with number 1 in x axis (a line of no change in the effect measure), fixed or random model, 95% confidence interval (horizontal line) and a dark diamond shape (overall all 95% CI and overall effect measure (OR or RR). 2
Funnel plot is a visual tool for investigating publication and other bias in meta-analyses the ideal funnel plot is the one that has the scattered plot on either side of the overall effect on the log of odd ration on x axis and standard of error on the y axis. This is called symmetry in funnel plot. The potential sources of asymmetry in funnel plots include selection biases, publication biases, language bias, citation bias, multiple publication bias, true heterogeneity, data irregularities, artifact due to poor choice of effect measure and chance.3
Random effects are that the true effect size varies from one study to the next and that the studies representing a random sample of the effect sizes that is the estimate of the mean of the summary effects.4 The random effects model allow the true effect sizes to differ from study to study.5 The use of regression test in the funnel plot showed that it is symmetric for random effects model while for mixed effects models; they are found to be asymmetric.6
Fixed effect is that the true effect size is the same in all studies, and the summary effect is the estimate of this common effect size. There is one true effect size of all studies ad the all differences in observed effects are due to sampling error.4,5,6
Heterogeneity is the measure of difference in summary effects among the chosen studies in meta-analysis showing the variance among the collected studies. This variation can be assessed by using I2 statistics denoting as low variation (25%), moderate variation (50%) and high variation (75%) in conjunction with Cochrane’s Q statistic (significance level < 0.05)7
Effect size is based on the sample size of the studies especially if sample sizes are small making the effect size estimates vary from one study to the other study. 8 The effect estimates of meta-analyses based exclusively on the published literature might be an overestimation of the true effect size (publication bias).9 Effect size describes the magnitude of the quantitative relationship between a treatment group and another specific outcome.12
The evaluation of “Is diabetes mellitus associated with mortality and severity of Covid-19? A meta-
1. Did the research questions and inclusion criteria for the review include the components of PICO?
Yes, these criteria were fully met.
Population: One hundred diabetic adult patients with Covid-19 were categorized into two or group depending on the severity, clinical course, or mortality of Covid-19.
Intervention: No intervention- literature review for meta-analysis.
Comparator group: comparing survival and mortality and severity of Covid-19 in diabetic patients.
Outcome: Survival, mortality and severity of Covid-19 patients with diabetes.
2. Did the report of the review contain an explicit statement that the review methods were established prior to the conduct of the review and did the report justify any significant deviations from the protocol?
Yes, the criteria were fully met. There were no deviations from the protocol. The PRISMA and MOOSE guidelines were used during the stages of design, analysis, and reporting of meta-analysis. The protocol was registered in PROSPERO and posted on NIHR website.
3. Did the review authors explain their selection of the study designs for inclusion in the review? Yes, this criterion was met. Authors only searched study articles published from January 1, 2010 to April 22, 2020 in PUBMED database only. They also wanted the publications to be in English language only.
4. Did the review authors use a comprehensive literature search strategy?
No. The authors only chose PUBMED database: the only source of literature search. The language of choice for literature search also was limited to English language only. The authors justified the use of English language only and the literature search from PUBMED database to avoid the duplication of publications that many authors may recycle the materials to different medical or research journals.
5. Did the review authors perform study selection in duplicate?
No. Three authors were independently searching, screening and selecting the studies according to the search, inclusion and exclusion criteria. From reading the author contributions, it seems that one author not participating in this meta-analysis fully. He was simply a supervisor to this study. There seems to be doing this study independently with no group meeting to build consensus on which literature search to exclude or include in the study.
Rating overall confidence in the results of the review of this article:
From reading this article, I would rate the overall confidence in the results of the review as low rating. This is evidenced by the search strategy done on publication from January 1, 2020 to the last searched performed on April 22,2020. Pubmed database search of 5834 articles were done and left with 33 articles to do the meta-analyses. From article information, the article was received on April 24, 2020, revised on April 27, 2020 and accepted for publication on April 28, 2020. This was really a quick turnaround time for up-to-date information. I am just curious how three authors could screen and selected 5834 articles (identification and screening process), ran through eligibility process and down with 33 articles that included the established criteria to do the meta-analysis ; the last search performed on April 22,2020. The publisher received on April 24, 2020. It took only 2 days to process 5834 articles to get 33 articles and doing statistical analysis in a rapid sequence.
It seems that these authors had the platform to build the case format to place in content, doing the appropriate statistics and submitting for publication in a speedy way. I detected the variation in summary effect of heterogeneity in Fig 3 Forest plot with I2 =63% meaning of moderate variations in the study of pooled odds ratio of diabetes mellitus associated with severe clinical course including mortality.
The critical flows in the review include the limitation of literature search only one source (PUBMED), lack of double check each other work and a speedy submission of this art
1. Kumar, A., Arora, A., Sharma, P. et al. “Is diabetes mellitus associated with mortality and severity of Covid-19 ? A meta-analysis” Diabetes and metabolic syndrome: clinical research and reviews, 14(2020), 535-545. https://doi.org/10.1016/j.dsx.2020.04.044
2. Stephenson, J. “ Explaining the forest plot in meta-analyses” Journal of wound care, (2017), 26, No 11, 611-612.
3. Sterne, J. and Harbord, R., “Funnel plots in meta-analysis” , The Stata Journal (2004), 4, No 2, 127-141.
4. Borenstein, M, Hedges, L., Higgins, J., and Rothstein, H. Introduction to meta-analysis. 2009. John Wiley & Sons, Ltd.
5. Borenstein, M., Hedges, L., Higgins, J., and Rothstein, H. “ A basic introduction to fixed-effect and random-effects models for meta-analysis”. Research Synthesis Methods, 2010,1, 97-111. (wileyonlinelibrary.com) DOI: 10.1002/jrsm.12
6. Jain, S., Sharma, S., and Jain, K. “Meta-analysis of fixed, random and mixed effects models”, International Journal of Mathematical, Engineering, and Management Sciences,(2019) 4, No1, 199-218.
7. Ssentongo, P., Ssentongo, A., Hellbrunn, E., Ba, D., and Chinchilli, V. “Association of cardiovascular disease and 10 other pre-existing comorbidities with Covid-19 mortality: A systematic review and meta-analysis.” Plos One 15(8):e0238215, August 26, 2020. https://doi.org/10.1371/journal.pone.0238215
8. Lakens, D., Hilgard, J., and Staaks, J. “On the reproducibility of meta-analyses: six practical recommendations.” BMC Psychology (2016) 4:24 DOI 10.1186/s40359-016-0126-3
9. Mueller, K., Meerpohl, J., and Briel, M. et al “Detecting, quantifying and adjusting for publication bias in meta-analyses: protocol of a systematic review on methods” Systematic Reviews, 2013, 2:60 https://www.systematicreviewsjournal.com/content/2/1/60
10. Shea, B., Reeves, B. and Wells, G. et al “Amstar 2: a critical appraisal tool for systematic reviews that include randomized or non-randomized studies of healthcare intentions, or both. BMJ 2017,358,j4008 doi:10.1136/bmj.j4008
11. Zheng, X., Song, Y., and Jiang, T. et al “Interventions to reduce burnout of physicians and nurses”. Medicine 2020,99,26(e20992) https://dxdoi.org/10.1097/MD.0000000000020992
12. Schober, P., Bossers, S., and Schwarte, L. “Statistical significance versus clinical importance of observed effect sizes: what do p values and confidence intervals really represent?” Anesthesia &Analgesia March 2018, 126, No3, 1068-1072
Part 2 Screening for disease.
Sensitivity is defined as the proportion of infected people who are correctly identified as “positive” by the test.1
Specificity is defined as the proportion of noninfected people who are correctly identified as “negative” by the test.1
Positive predictive value is the proportion of patients who test positive actually have the disease in question.1
Negative predictive value is the proportion of patients who test negative actually not having the disease.1
False positive is defined as a group of people who test positive for the particular disease but not having that disease.1
False negative is when a person has the disease but informed in error that the test result is negative.1
There are five test principle to detect Covid-19 nucleic acid detection namely RT-PCR (reverse transcriptase-polymerase chain reaction), real-time RT-PCR, CRISPR( Clustered regularly interspaced short palindromic release), isothermal nucleic acid amplification technology and SHERLOCK (specific high sensitivity enzymatic reporter unlocking).2,3,4
RT-PCR has been used to detect Covid-19 nucleic acids in nasopharyngeal swabs, lower respiratory tract secretions, sputum, blood, feces and other specimens. It can detect Covid-19 within less than 45 minutes. Samples are collected by nasopharyngeal swab or nasal wash and prepared in less than a minute. This is developed by Fred Hutchinson Cancer Research Center, Washington; commercially available as Xpert® Xpress by Cepheid Roche Molecular System. The RT-PCR based rapid test device has the sensitivity for Covid-19 of 84.6%2
Isothermal nucleic acid amplification technology has more advanced nucleic acid detection techniques for Covid-19 testing. It was found suitable to detect Covid-19 in low quantity of viral RNA and can be used as the alternative to RT-PCR based rapid test device. ID Now® is manufactured by Abbott that was studied by Wuhan Institute of Virology, China. It claimed the sensitivity of test about 95%2
In my opinion, I would choose ID Now® as a Covid-19 test of choice due to its superior sensitivity and can detect the Covid-19 virus in a small quantity. One can identify true positive patients easily with this test. There will be more Covid-19 test that offered convenient to clients in terms of privacy and ease of doing self-testing. University of California at San Diego has implemented the vending machines for Covid-19 testing on campus for students to do self-testing in their convenience. The results will be also tracked and reported to their on-campus computer dashboard for Covid-19 similar to the one that Johns Hopkins University has offered.
A health clinic pharmacist has screened 10,000 people for Covid-19. Assume that the true prevalence of Covid-19 at the time of screening is 4.5% Assume that one of the Covid-19 tests has the sensitivity of 90% and the specificity of 90%.
Total people with disease = prevalence X screen population =0.045 x 10,000 = 450
True positive = sensitivity x disease population =0.9 x 450 = 405
Total people without disease = 10,000 – 450 = 9550
True negative = specificity x disease-free population = 0.9 x 9550 = 8595
False negative = total people with disease – true positive = 450 -405 = 45
False positive = total people disease-free – true negative = 9550 – 8595 = 955
Test positive = true positive + false positive = 405 + 955 = 1360
Test negative = true negative + false negative = 8595 +45 = 8640
COVID TEST DISEASE NON-DISEASE TOTAL
POSITIVE 405 TRUE POSITIVE 955 FALSE POSITIVE 1360 TEST POSITIVE
NEGATIVE 45 FALSE NEGATIVE 8595 TRUE NEGATIVE 8640 TEST NEGATIVE
450 TOTAL DISEASE 9550 TOTAL NON-DISEASE 10,000 TOTAL SCREENS
1. Gordis, L. Epidemiology, Third Edition. 2004 Elsevier Saunders
2. Jamshaid, H., Zahid, F., and Din, I. “Diagnostic and treatment strategies for Covid-19.” AAPS PharmSciTech July 13,2020. Doi:10.1208/s12249-020-01756-3
3. Joung, J, Ladha, A. and Saito, M. et al “Detection of SARS-CoV2- with SHERLOCK one-pot testing”. The New England Journal of Medicine September 16,2020 DOI:10.1056/NEJMc2026172
4. Rhoads, D., Cherian, S. and Roman, K. “Comparison of Abbott ID Now, Diasorin Simplexa, and CDC FDA emergency use authorization methods for the detection for SAR-Cov-2 from nasopharyngeal and nasal swabs from individuals diagnosed with Covid-19.
5. Kumleben, N. Bhopal, R. and Czypionka, T. “Test, test, test for Covid-19 antibodies: the importance of sensitivity, specificity and predictive powers,” The Royal Society for Public Health. June 11, 2020. Elsevier Ltd. https://doi.org/10.1016/j.puhe.2020.06.006