Study selection (PRISMA-style flow)

The PRISMA 2020 flow diagram below traces the review from database searching through to the studies included in the meta-analysis. The literature search identified 2,367 records — 492 from PubMed, 971 from Embase/OVID, and 904 from Scopus — leaving 1,560 records after de-duplication (807 duplicates removed). Screening was managed in Covidence: 172 records were screened at title/abstract and 52 taken forward to full-text review, of which 48 were assessed and 21 studies (22 reports) met inclusion criteria — 6 studies (6 reports) enter the quantitative (pairwise) meta-analysis of echinocandin + TMP-SMX vs TMP-SMX monotherapy. A network meta-analysis was pre-specified “if data permit”, but the current evidence is a single connected contrast (k = 6) and does not support one, so the synthesis is pairwise. The diagram is generated with the PRISMA2020 package from a filled copy of its CSV template (analysis/prisma2020_data.csv).

Show code

library(PRISMA2020)
d <- read.csv("analysis/prisma2020_data.csv", stringsAsFactors = FALSE)
PRISMA_flowdiagram(PRISMA_data(d), previous = FALSE, other = FALSE,
                   meta_analysis = TRUE, detail_databases = TRUE, fontsize = 10)

Note

Upstream counts come from the database searches (PubMed, Embase/OVID, Scopus) and the Covidence export. Of the 1,560 de-duplicated records, 172 were carried into Covidence for title/abstract screening; the intervening 1,388 records are shown as removed before screening. That step — along with the 52 → 48 step (reports not retrieved, n = 4) — is provisional pending final confirmation of the exact Covidence import path.

Analytic selection flow (echinocandin contrast)

The review-level diagram above covers all 21 included studies. The narrower flow below shows the analytic selection for the echinocandin-versus-monotherapy mortality comparison specifically — the 6 studies (3 primary + 3 sensitivity) used in this site’s pairwise synthesis.

flowchart TD
  A["Full-text studies in extraction corpus<br/>(n = 48)"] --> B["Comparative antifungal-regimen<br/>studies screened for the echinocandin contrast"]
  A --> A1["Excluded: background / context-only,<br/>single-arm, or no eligible contrast (n = 26)"]
  B --> C["Echinocandin + TMP-SMX vs TMP-SMX<br/>monotherapy, mortality reported"]
  C --> E1["Excluded — HIV-positive cohort:<br/>Tian2020 (n = 1)"]
  C --> E2["Excluded — duplicate cohort:<br/>Yanmeng2026 = Qi2025 (n = 1)"]
  C --> E3["Excluded — no monotherapy comparator:<br/>Wang2019 pilot, Zhang2018, Wang-ZG-2019 (single-arm)"]
  C --> E4["Excluded from primary — back-calculated events /<br/>administrative dataset: Taniguchi"]
  C --> D["Included in meta-analysis<br/>(n = 6)"]
  D --> D1["Primary — in-hospital / ~30-day:<br/>Qi2025, Xu2025, Lu2017"]
  D --> D2["Sensitivity — longer timepoints:<br/>Jin2019, Qi2023, Li2024a"]

Six studies met criteria for the echinocandin-versus-monotherapy mortality comparison. The primary synthesis uses the three studies reporting in-hospital or ~30-day mortality; the pre-planned sensitivity synthesis adds three studies reporting longer-timepoint mortality (one of which, Li2024a, is confounded and flagged).