1. In Table 2: The US shows 1,166 citations in the VOSviewer panel (46 documents) but 12,686 citations in the Scopus panel (12 documents). China similarly jumps from 8,805 to 37,628 citations. This ~10-fold difference suggests a serious data extraction or reporting error. How do you explain this impossibility?
2. Figure 4: You report a “gradient change of 9.826 (SE=0.725)” from pre- to post-COVID. But with only 3-4 post-COVID data points (2021-2025), your standard error is likely severely underestimated. Did you test for autocorrelation or use Newey-West standard errors? Without this, your significant p-value is meaningless.
3. Your PRISMA flow diagram (Fig. 3) shows 204 documents retained, but your country analysis in Table 2 (first panel) only accounts for 46+37+25+9+12+… summing to far fewer than 204 documents. Where did the remaining ~100 countries’ publications go?
4. The “55 most-cited articles” selection introduces fatal bias. You claim to reduce bias against recent publications, but a 2024 paper cannot mathematically compete with a 2015 paper in raw citations. Why not use citation velocity (citations/year) or percentile rank to avoid systematically excluding methodologically superior recent research?
5. You report two researchers independently coded themes, but provide no inter-rater reliability coefficient (e.g., Cohen’s kappa). With subjective constructs like “psychological dimensions of caregiving,” what was your agreement rate?
As a reviewer, I have additional critical concerns:
6. Eq. (1) is mathematically invalid. y=1.3983x−2805.8y=1.3983x−2805.8 gives y≈−7.8y≈−7.8 for 2001 and y≈25.8y≈25.8 for 2025, which directly contradicts Fig. 4. Please provide the correct regression equation and clarify how xx was coded.
7. Fig. 5 is not a citation-based ranking. The text says “top ten journals based on citations counts” with a 100-citation threshold, but the y-axis is “Number of documents.” PLOS ONE cannot be called the “most influential” source without citation counts.
8. Table 2 first panel lists only 9 countries, yet you state 10 countries met the threshold. Which country is missing, and why?
9. Table 3 cluster items sum to 75 (30+17+13+10+5), but you state 77 keywords met the threshold. Where are the missing 2 keywords?
10. Fig. 9 contains arithmetic errors. Research-method counts sum to 53/54, not 55; thematic percentages sum to 101%; thematic study counts sum to 56, not 55. These need correction.
11. Table 4 rank 55 has only 8 citations, while ranks 38–48 all have 33 citations. How can an 8-citation 2025 paper be among the “55 most-cited” of 204 papers? What tie-breaking rule was used?
12. Direct contradiction: Abstract and limitations state developing societies/LMICs remain underrepresented, but the comparison table claims the review “Covers global contexts, including both high-income and LMIC regions.” Reconcile.
13. Theme-category mismatch: Fig. 9 includes “Emerging Technologies and Health Tech” and “Risk Management and Safety,” but the theme-definition table omits or fails to name these categories. Clarify.
14. Unsupported theory claim: You state most existing research uses UTAUT/UTAUT2/TAM, but Table 4 repeatedly lists “Not Mentioned.” Provide counts to support this assertion.
15. Why is “adolescent” a keyword in Cluster 3 of a review on older adults, and how does it inform RQ1?
16. Fig. 7 contradicts itself: banking/mobile banking are called niche themes, then said to have evolved to basic themes. Clarify.
17. Sensitivity analysis says “two variations,” but you report three keyword thresholds (3, 5, 7) and three citation thresholds (80, 100, 120). Clarify.