Beyond the funnel plot: The advantages of Doi plots and prediction intervals in meta-analyses

Shamim, Muhammad Aaqib; Dwivedi, Pradeep; Padhi, Bijaya Kumar

doi:10.1016/j.ajp.2023.103550

Cited by 9 publications

(6 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These were then synthesied using a random intercept logistic regression model [11] . To describe the variance amongst studies, the heterogeneity of the studies evaluated in the meta-analysis was assessed using the I 2 test [12] , in addition to the prediction interval [13] , tau, tausquared [14] , and Cochran's Q [15]. Heterogeneity was classified as low, moderate, and high, respectively, based on I 2 values of less than 25%, 25-50%, and more than 50%.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

WITHDRAWN: Prevalence of coronary artery disease among COVID-19 patients: a systematic review and meta-analysis

Akhtar

Padhi

Shamim

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

BackgroundThroughout the surge of the COVID-19 pandemic high rate of chronic diseases have been reported, including respiratory diseases and cardiovascular diseases. The prevalence of coronary artery disease has remained high throughout the COVID-19 pandemic, which also draws great concern towards it. This study seeks to provide a pooled estimate of the burden of coronary artery disease in COVID-19.ObjectiveTo estimate the overall prevalence of coronary artery disease among COVID-19 patientsData SourcesIn this systematic review and meta-analysis, an extensive literature search was conducted in PubMed, Scopus, Embase, EBSCO, Web of Science, Cochrane,Proquest and preprint servers (medRxiv, arXiv, bioRxiv, BioRN, ChiRxiv, ChiRN, and SSRN). References fo eligible articles, forward citation tracking, and expert opinion were used to identify other relevant articles. All published articles until 13 April 2023 were assessed as per the PROSPERO registration protocol (CRD42022367501).Study Selection, Data Extraction, and SynthesisPrimary studies that reported coronary artery disease among COVID-19 patients were included. The characteristics of the study and information on the number of cases of coronary artery disease were extracted from the included studies. Individual study estimates were pooled using the random intercept logistic regression model. The heterogeneity between the selected studies was assessed using the I2statistic, tau, tau-squared, Cochran’s Q. Prediction interval was used to identify the range into which future studies are expected to fall. Subgroup analysis based on geography (continent) was done to reduce heterogeneity. Publication bias was analyzed using doi plot and LFK index. The risk of bias in the studies was assessed as per the tools proposed by the National Institute of Health.Main outcomesThe primary outcome was the pooled prevalence of coronary artery disease among COVID-19 patients within the examined population.Results510 records were initially retrieved from electronic databases in addition to other sources like reference screening. 33 studies with 40,064 COVID-19 patients were included for quantitative synthesis. The prevalence of coronary artery disease among COVID-19 patients was 15.24% (95% CI: 11.41% - 20.06%). The prediction interval ranged from 2.49% to 55.90%. The studies were highly heterogeneous (tau-sqaured of 0.89), and subgroup analysis significantly reduced it (test of moderators: Q = 14.77, df=2, P=.002). Europe reported the highest prevalence [21.70% (14.80% - 30.65%)], and Asia has the least prevalence [10.07% (6.55% - 15.19%)]. Meta-regression for sample size was not significant (P=.11). A symmetric doi plot and an LFK index of 0.57 revealed no evidence of publication bias or small-study effects.ConclusionThe burden of coronary artery disease has been considerable, varying with geography. and further research in this area is needed. Routine cardiac screening and assessment of COVID-19 patients can help uncover undiagnosed cases, and better optimise the management of all COVID-19 patients.

show abstract

Section: Discussionmentioning

confidence: 99%

“…P<.05 has been considered statistically significant. For the assessment of publication bias and small-study effects, we used Doi plot and the accompanying LFK index since this is a meta-analysis of proportions [13]. We performed a subgroup analysis based upon and Asia has the least prevalence [10.07% (6.55% -15.19%)].…”

Section: Discussionmentioning

confidence: 99%

WITHDRAWN: Prevalence of coronary artery disease among COVID-19 patients: a systematic review and meta-analysis

Akhtar

Padhi

Shamim

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Doi plots help assess publication bias in single-group studies as they don't rely on the concept of significance. 34 They plot the individual study estimates against a folded normal quantile. Study symmetry is assessed visually, and quantified using LFK index.…”

Section: Discussionmentioning

confidence: 99%

Prevalence of hepatitis B and C infections among HIV‐positive men who have sex with men: A systematic review and meta‐analysis

Shabil,

Yadav,

Shamim

et al. 2024

Health Science Reports

View full text Add to dashboard Cite

BackgroundHuman immunodeficiency virus (HIV) infection is highly prevalent and often coexists with other infectious diseases, especially Hepatitis B virus (HBV) and Hepatitis C virus (HCV). Men who have sex with men (MSM) represent a vulnerable population in terms of HIV infection. We aimed to determine the prevalence of HCV, HBV among HIV‐infected MSM.MethodsThis systematic review and meta‐analysis searched PubMed, Cochrane, Scopus, Web of Science, and ProQuest up‐to 2023/04/22. All studies reporting the prevalence of HBV or HCV infection in MSM PLHIV were included. Meta‐analysis used random effect model for synthesis and I2 along with prediction interval for heterogeneity. Subgroup analysis based on continent and meta‐regression for study size, average age and year of publication were used to explore heterogeneity. Modified Newcastle‐Ottawa Scale was used to evaluate the quality of studies according to the protocol (PROSPERO: CRD42023428764).ResultsFifty‐six of 5948 studies are included. In 53 studies with 3,07,589 participants, a pooled prevalence of 7% (95% confidence interval [CI]: 5–10) was found for HCV among MSM PLHIV, while a 9% (95% CI: 4–18) prevalence was found for HBV infection from five studies which included 5641 MSM PLHIV. Asia reported the lowest pooled prevalence at 5.84% (95% CI: 2.98–11.13) for HCV while Europe reported the highest pooled prevalence at 7.76% (95% CI: 4.35–13.45). Baujat plot and influence diagnostic identified contributors to influence and between‐study heterogeneity. Sensitivity analyses omitting these studies result in considerably more precise estimates. Another sensitivity analysis as leave‐one‐out meta‐analysis did not change any pooled estimate significantly.ConclusionThere is a significant burden of HCV and HBV among MSM PLHIV worldwide, with varying prevalence rates. Future studies should focus on these multimorbidity clusters and investigate factors influencing disease burden, long‐term outcomes, optimal testing strategies, and tailored interventions.

show abstract

“…In sub-group analyses by DAT types, we opted not to produce summary measures due to the small number of estimates per DAT type and cut-off point. Publication bias was assessed using Doi plots in keeping with guidance for meta-analyses of proportions, using the Luis Furuya-Kanamori (LFK) index [11][12][13]. The second measure of dose-days with DAT-engagement was extracted from reports either as an average, often the arithmetic mean, of the measure across individuals or as an overall proportion (DATengagement days/all dose-days), ignoring individuals; for both measures, a standard error accounting for clustering was often not reported and could not be calculated using the available data.…”

Section: Discussionmentioning

confidence: 99%

Implementation outcomes of tuberculosis digital adherence technologies: a scoping review using the RE-AIM framework

Chilala,

Foster,

Bahukudumbi

et al. 2024

Preprint

View full text Add to dashboard Cite

IntroductionGlobally, tuberculosis (TB) remains one of the leading infectious causes of death, with 1.3 million deaths. Digital adherence technologies (DATs) have the potential to provide person-centred care and improve outcomes. Using the RE-AIM framework, we conducted a scoping review of DAT implementations for TB treatment.MethodsWe searched eight databases for papers published between January 2000 and April 2023, using keywords for ‘tuberculosis’ and ‘digital adherence technology’. Articles meeting prespecified inclusion criteria and containing data on RE-AIM domains were included (PROSPERO-CRD42022326968). We defined ‘reach’ as comprising cellphone ownership and engagement by people with TB (PWTB) with DATs, ‘adoption’ as engagement by healthcare providers with DAT programs, ‘implementation’ as the fidelity of the DAT program implemented, and ‘maintenance’ as longer-term uptake of DATs.ResultsOf 10,313 records, 105 contributed to the synthesis. DATs included SMS, phone, 99DOTS, video-supported therapy and pillboxes. For ‘reach’, across various settings, cellphone access varied from 50-100% and 2-31% of PWTB were excluded from accessing DATs due to technology challenges. 36-100% of PWTB agreed to use a DAT. The weighted mean of DAT engagement over dose-days was 81% for SMS, 85% for phone, 61% for 99DOTS, 87% for pillbox and 82% for VST. Concerning ‘implementation’, the fidelity of DAT implementations was affected by technological issues such as cellphone coverage, DAT malfunction and provider-facing issues; including failure to initiate intensified patient management following low DAT engagement. Findings related to RE-AIM dimensions of ‘adoption’ and ‘maintenance’ were limited.ConclusionOur findings suggest that the ‘reach’ of DATs may be limited by a cascade of barriers, including limitations in cellphone accessibility and suboptimal sustained DAT engagement by PWTB. Video and pillbox DATs have higher levels of engagement. Implementation challenges included technological and provider-facing issues. Improving implementation outcomes may be important for TB DATs to achieve broader public health impact.

show abstract

Beyond the funnel plot: The advantages of Doi plots and prediction intervals in meta-analyses

Cited by 9 publications

References 9 publications

WITHDRAWN: Prevalence of coronary artery disease among COVID-19 patients: a systematic review and meta-analysis

WITHDRAWN: Prevalence of coronary artery disease among COVID-19 patients: a systematic review and meta-analysis

Prevalence of hepatitis B and C infections among HIV‐positive men who have sex with men: A systematic review and meta‐analysis

Implementation outcomes of tuberculosis digital adherence technologies: a scoping review using the RE-AIM framework

Contact Info

Product

Resources

About