Respiratory Syncytial Virus Infections in Recipients of Bone Marrow Transplants: A Systematic Review and Meta-Analysis

In the random-effect models, the risk ratio for RSV attack rate, incidence rate, and CFR by age group (adults, adults and pediatric, and pediatric) were estimated, and they interpreted that the older age group had a higher attack rate. However, since there is high heterogeneity among the studies assessed (I-squared >90%), the infection rate in the underlying population may influence the result. In addition to the subgroup analysis, the effect of age (i.e., risk/odds ratio for adults vs. child) should be analyzed for adult and pediatric group where both adult and pediatric data are available.

We totally agreed with this comment. In fact, only 3 studies provided detailed and paired data (adults vs. children). The study was amended accordingly as follows:

Results section:

As only 3 studies provided both adult and children data, OR for the occurrence of RSV in children vs. adults (i.e. individuals aged < 18 years vs. aged 18 years or older) was calculated from this smaller subset including a total of 1571 cases (62.6% aged more than 18 years) [81, 94, 101]. Overall, children were associated with an increased odds for developing RSV infection after BMT (OR 2.941, 95%CI 1.689 to 5.122). Even though the analyses were associated with seemingly reduced heterogeneity (I2 = 4.8%), corresponding 95%CI hinted to a more precautionary approach (0.0% to 90.1%).

Figure 7. Forrest plot for the Odds Ratio (OR) on the occurrence of Respiratory Syncytial Virus (RSV) infections in children (individuals aged 0 to 18 years) vs. adults (individuals older than 18 years). Eventually, a OR of 2.941 (95%CI 1.689 to 5.122) was identified, with a I2 of 4.8%, 95%CI 0.0 to 90.1) [81, 94, 101]

Sensitivity analysis:

Eventually, the removal of the study of McCarthy et al. [94] from the pooled estimate on OR led to a noticeable change in both the pooled estimates for the occurrence of RSV infections in children vs. adults (OR 2.79, 95%CI 0.92 to 8.50), and in those for I2 (34%), while the removal of the study from Small et al. [101] only affected the estimates for OR and 95%CI (OR 1.87, 95%CI 0.76 to 4.58) (Appendix A, Figure A8).

Discussion:

Where both data on adults and children were provided, the formed exhibited an increased odds for developing RSV infection compared to adults (OR 2.941, 95%CI 1.689 to 5.122).

Line 112 and 167, SARS-CoV-2: its result is not shown in the paper.

Thank you! In fact, we were unable to retrieve data on SARS-CoV-2 from included studies; therefore, the following statement was included:

Unfortunately, retrieved studies did not report data on the occurrence of SARS-CoV-2 infections in BMT cases (Row 309-310).

Line 161, 391, 638, Table 3, Table A2, (allogenic vs. heterologous), (allogenic vs. autologous) or (autologous vs. heterologous): please check if they should be consistent.

Thank you for your comment. In fact, we did fail to perceive the inconsistent reporting, and all the notations were made consistently “allogenic vs. autologous”

Line 186, attack rate: the definition of attack rate is unclear.

Table 7, Figure 4, Figure A5: Both attack rate and prevalence are used in the same context. To avoid confusion, it is better to use one term.

Thank you for your comment. In fact, not only the definition but also the use of attack rate was used inconsistently, with overlapping with the use of “prevalence rate”. We shifted the main text to “attack rate” and the entries for prevalence were removed from the text.

Moreover, we included the following definition of attack rate (row 191-193):

Attack rates were defined as the number of people who developed viral infection by the number of people at risk for the illness and reported as cases per 100 population

Lines 408-412; lines 413-415; lines 428-432: replicated from Tables 7 and 8. Please consider deleting or summarizing them.

Thank you again for your comments! Where possible, the main text was summarized and simplified, as suggested.

Line 442, p=0.178: what this p-value is for is unclear.

The main text was amended as follows:

with no substantial difference in test for subgroup difference (chi squared 3.41 , p = 0.178).

Line 464, 0% for incidence rate: is 0% for CFR (Fig A7)?

Thank you! In fact, as for the following comment, we did some mess while copying the main text from the proofs and resulting text was made unwillingly confusing. The text was amended as follows:

all analyses for attack rates and incidence rates, and around 0% for CFR (row 483)

Lines 491-500: the results of funnel plots and Egger’s test do not seem to be consistent. Was a substantial publication bias hinted for the incidence rate (not CFR)?

We totally agree: as stated for the previous point, we confused the main text while copying it from the main proofs.

Please refer to the amended text:

In our study, visual inspection of contour-enhanced funnel plots suggested that publication bias could be ascertained for incidence rates, attack rates, and CFR, as es-timates were not evenly scattered across the logit of transformed proportion.

On the contrary, point estimates were evenly scattered on both sides of regression lines in radial plots. However, Egger’s test (i.e. the linear regression analysis of the in-tervention effect estimates on their standard errors weighted by their inverse variance) hinted to substantial publication bias for both incidence rates (Figure 9b; t = 8.76, df = 28, p-value < 0.001), and CFR (Figure 9c; t = -4.78, df = 28, p-value < 0.001). On the contrary, no publication bias was reasonably associated with attack rates (Figure 9a; t = 0.65, df = 28, p-value = 0.519).

Figure 7, legend: the legend for a~f does not match for panels a~f.

Thank you again! As the main figure was made confusing by the high number of panels, we broken down it in two subfigures, 8 and 9.

§  The abstract does not explicitly mention the limitations of the study. It would be beneficial to include a brief section on the potential biases or constraints faced during the systematic review process and data extraction.

The abstract was amended by  including the following statement:

Despite the heterogeneous settings and the uneven proportion of adult and pediatric cases

§  While the abstract underscores the importance of tailored preventive strategies for recipients of bone marrow transplantation, a brief discussion on the potential implications for clinical practice and future research directions would add depth to the conclusion.

The abstract was amended by  including the following statement:

stressing the importance of specifically tailored preventive strategies and the need for effective treatment options.

§  The introduction concludes effectively by highlighting the need for an updated synthesis of the literature on RSV infections in BMT recipients. A sentence summarizing the potential impact of the study on clinical practice or policy would strengthen the conclusion.

The main text was amended as follows:

As an updated definition of the actual RSV burden of disease among BMT recipients is needed to inform health policies and both preventive and treatment guidelines, a syn-thesis of the available literature was performed to ascertain (row: 92-94)

§  The search strategy is well-documented for each database (PubMed, EMBASE, medRxiv) with clear inclusion terms. However, it would be beneficial to explicitly mention the search date range and any language restrictions applied during the search.

The main text was amended as follow:

All databases were searched from inception up to 30/10/2023, without applying any backward chronological restrictions in the following languages: English, Italian, German, French, Spanish, Portuguese. (row 132-134)

§  The items listed for data abstraction (Section 2.4) are comprehensive and cover essential aspects of the studies. However, it might be helpful to mention how missing data or ambiguous information in the retrieved studies were handled during the data abstraction process.

As stated in the previous section, “All retrieved items were independently rated by two investigators (AB, FM), and their disagreements were either resolved by consensus or through an input from the chief investigator (MR) where the preliminary consensus of investigators was not reached”. Moreover, where data were missing or ambiguous, data were reversely calculated (where possible).

The use of the OHAT ROB tool for assessing the risk of bias is appropriate, considering the focus on internal validity. Clarify whether the two independent investigators performed the risk of bias assessment independently and if there were discrepancies, how they were resolved.

The following explanatory sentence was added:

As for the screening procedures, retrieved items were preliminarily and independently rated by two investigators (AB, FM), and disagreements were either resolved by consensus or through an input from the chief investigator (MR).

The mention of the R software and specific packages used for the analysis is informative. Consider providing a brief explanation or reference for readers unfamiliar with these tools.

The following explanatory sentence was added:

The package meta provides standard methods for meta-analysis, while the package fmsb provides functions for medical statistics and the handling of demographic data

It would be beneficial to include a subsection discussing the limitations of the study. Addressing potential biases, variations in study designs, and other uncertainties will provide a balanced view of the findings.

§  The discussion on the generalizability of the study is insightful. However, it would be helpful to explicitly state the limitations related to the heterogeneity in study designs, sample sizes, and reporting strategies. Addressing the potential impact of these limitations on the robustness of the conclusions would strengthen the discussion.

Please refer to section 4.4 (4.4 Limits and implications for future studies). Following statement was also included:

For instance, demographic data were not consistently provided by all retrieved studies, impairing the accurate appraisal of individual risk factors for developing RSV and oth-er viral respiratory infections. Most notably, only 5 studies benefited from a prospective design, being therefore specifically tailored for collecting data on BMT recipients and respiratory infections [85, 89, 95, 96, 99], while only three studies provided detailed da-ta on both children and adults [81, 94, 101]. As a consequence, corresponding OR were calculated on a relatively small subset of cases, and corresponding estimates have to be only cautiously assessed.

The discussion on the higher risk of RSV infection between 2010 and 2014 compared to more recent studies is interesting. However, the reasons behind this shift are not explored. It would be beneficial to speculate on potential factors such as changes in transplant procedures, preventive measures, or RSV management strategies over time.

The following section was added:

The high risk for RSV infections between 2010 and 2014 was quite unexpected, as the global trend for RSV infections was not associated with increased occurrence of the pathogen in the general population [4,5,8,127], while the pandemic 2009/H1N1 and claims about the reduced efficacy of the 2014-2015 influenza vaccine [85,87,102,103] urged for the application of accurate preventive strategies.

The conclusion regarding the need for preventive interventions is well-founded. However, providing some specific recommendations or suggestions for future research directions could add practical value to the discussion.

Main text was amended as follows:

Therefore, a proper preventive approach to BMT cases could encompass improved testing strategies with periodic assessment of respiratory pathogens among newly admitted individuals, new and effective preventive options such as mAb, whose delivery among adults and elderly however still remains not ascertained. Moreover, because of the no-ticeable CFR, specifically designed anti-viral drugs could substantially improve the prognosis of BMT recipients affected by RSV infections.