Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1) The main question addressed by the research is related to the effects of photobiomodulation therapy on pain and healing of episiotomies and grade 2 and 3 perineal lacerations after vaginal delivery. 2) The topic is marginally original addressing a specific gap in the field because it examines a relatively new method for pain and healing of episiotomies and grade 2 and 3 perineal lacerations after vaginal delivery. 3) Compared with other published material, it adds to the field that, among other “classic” methods for pain and healing of episiotomies and grade 2 and 3 perineal lacerations after vaginal delivery, photobiomodulation can have similarly effective results. 4) The specific improvements the authors should consider regarding the methodology are related a) to the terminology: “low-level laser therapy” is a term broadly used, and b) some more comments and/comparisons with other “classic” methods. 5) The conclusions are consistent with the evidence and arguments presented and they address the main question posed because the conclusions are strictly related to the effectiveness of the method on pain and healing of episiotomies and grade 2 and 3 perineal lacerations after vaginal delivery. 6) The references are appropriate. 

Author Response

Response Letter,

Point-by-point response to the reviewer’s comments

 

Dear Editor and Reviewers

 

Thank you very much for taking the time to review this manuscript. Please find the detailed responses in this letter and the corresponding revisions/corrections highlighted/in track changes in the re-submitted files

 

Reviewer 1:

 

1) The main question addressed by the research is related to the effects of photobiomodulation therapy on pain and healing of episiotomies and grade 2 and 3 perineal lacerations after vaginal delivery.  The topic is marginally original addressing a specific gap in the field because it examines a relatively new method for pain and healing of episiotomies and grade 2 and 3 perineal lacerations after vaginal delivery.  Compared with other published material, it adds to the field that, among other “classic” methods for pain and healing of episiotomies and grade 2 and 3 perineal lacerations after vaginal delivery, photobiomodulation can have similarly effective results.  The specific improvements the authors should consider regarding the methodology are related a) to the terminology: “low-level laser therapy” is a term broadly used,

 

Dear reviewer, thank you for carefully reading our manuscript and for your suggestions.

 

In fact, the term low-level laser therapy is historic and widely used to designate the use of low-intensity therapeutic light. However, we chose to use and standardize the term photobiomodulation (PBM) throughout the manuscript because in 2014, a nomenclature consensus meeting was organized under the auspices of the joint conference of the North American Association for Light Therapy and the World Association for Laser Therapy in September, 2014. It was attended by 15 international participants to discuss the best nomenclature for these therapies. Photobiomodulation was considered by many participants to be the term of choice to describe it’s use and therfore the term photobiomodulation therapy was added to the MeSH database for its 2016 version as an entry term to the existing record of laser therapy, low-level. Since then, researchers from all over the world have made efforts to maintain the use of this term recommended by these important institutions, giving reliability to the studies. And currently the term PBM is the most widely used. The use of this term is a key point, as it distinguishes photobiomodulation therapy, which is nonthermal, from the popular use of light-based devices for simple heating of tissues as can be accomplished using colored lamps, or other applications of light energy that rely on thermal effects for all or part of their mechanism of action.

 

However, we understand your concern and therefore included the mention of the old term low-level laser therapy in the introduction (line 64) and also the references about this term (Anders et al 2015 and Hamblin 2016).

 

 

and b) some more comments and/comparisons with other “classic” methods 

 

Dear reviewer, thank you for carefully reading our manuscript and for your suggestions. 

 

We agree with your observation and included this aspect in the discussion (Lines 367-397):

 

The management of perineal pain and the promotion of wound healing following episiotomies and lacerations are critical components of postpartum care, with evidence supporting a combination of pharmacological and non-pharmacological strategies. Non-pharmacological approaches such as cryotherapy and transcutaneous electrical nerve stimulation (TENS) have been identified as promising conservative therapies for mana-ging early postpartum pain, offering alternatives with limited adverse effects [11]. Emerging evidence also supports the use of specific technologies. Ultimately, optimal outcomes depend on accurate diagnosis, appropriate surgical repair techniques, and the integration of these evidence-based interventions into a comprehensive, individualized care plan. A recent meta-analysis by Kurnaz et al. [18] highlights that interventions performed within the first 24 h after episiotomy did not reduce pain. However, the effects of the interventions were observed on the second day, with cold application identified as the most effective method. Additionally, interventions did not affect healing during the first three days, but a more pronounced improvement was noted in the intervention group by the fifth day. Healing began around the 7th-10th days even without inter-vention. The REEDA score (redness, edema, ecchymosis, discharge, and approximation) decreased most significantly in the patients that received perineal education (diet, Kegel exercise, infection symptoms, and perineal hygiene).

Infrared therapy has been shown in phase II studies to significantly improve epi-siotomy wound healing and reduce pain compared to standard care alone. Constant et al. [19] compared photobiomodulation and cryotherapy in the immediate postpartum period among women with grade I and II lacerations and/or episiotomy, observing superiority of PBM in pain reduction and improved healing after 24 hours. These results suggest that laser therapy can have similar or even superior results to other non-pharmacological therapies, and that dosimetry and timing of application are key determining factors in obtaining the best results. Nonetheless, data in obstetrics remain limited. The hetero-geneity of protocols (including dose parameters, timing, and treatment frequency) and the lack of controlled studies comparing laser therapy with other conventional techniques have been identified as possible explanations for these inconclusive results, highlighting the need for standardization in future studies to generate higher-quality evidence and support the development of informed clinical guidelines.

 

The conclusions are consistent with the evidence and arguments presented and they address the main question posed because the conclusions are strictly related to the effectiveness of the method on pain and healing of episiotomies and grade 2 and 3 perineal lacerations after vaginal delivery.

The references are appropriate.

 

Dear reviewer, thank you for carefully reading our manuscript and for your valuable suggestions.

 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Based on the article you uploaded, here is a structured analysis of what is original/relevant and what specific gap in the field the paper addresses. Based on a detailed review of the manuscript, the study addresses an important clinical question. However, several methodological improvements and additional controls would substantially strengthen internal validity, reduce bias, and improve interpretability.

1. Clearly define and justify all covariates included in the PSM model and also provide a balance table.
2. Model expected natural pain decline using the no-laser group as trajectory reference.
3. Address Baseline Imbalances More Thoroughly, including these as covariates in all outcome models. Also, adjust for baseline laceration extent and Apgar differences.
4. You could control for analgesic timing and labor variables.

In summary, while the study is methodologically solid with its prospective, controlled, and randomized design, integrating these improvements would significantly enhance the internal and external validity, reduce bias, and provide stronger evidence for clinical recommendations.

Author Response

Response Letter,

Point-by-point response to the reviewer’s comments

 

Dear Editor and Reviewers

 

Thank you very much for taking the time to review this manuscript. Please find the detailed responses in this letter and the corresponding revisions/corrections highlighted/in track changes in the re-submitted files

 

 

Reviewer 2

 

Based on the article you uploaded, here is a structured analysis of what is original/relevant and what specific gap in the field the paper addresses. Based on a detailed review of the manuscript, the study addresses an important clinical question. However, several methodological improvements and additional controls would substantially strengthen internal validity, reduce bias, and improve interpretability.

 

 

1. Clearly define and justify all covariates included in the PSM model and also provide a balance table

 

We thank the reviewer for the opportunity to discuss this important topic. After your review we also provided these aspects described below and presented on the Supplementary Material File 1.

 

 

In accordance with the prespecified Statistical Analysis Plan, the outcomes included pain (Numeric Pain Scale – NPS) and wound healing (REEDA scale), both assessed over the first three days of hospitalization. The primary predictor (treatment exposure) was defined as receipt of photobiomodulation therapy during hospitalization and categorized into three groups: (A) two laser sessions (Days 1 and 2), (B) one laser session (Day 1 or Day 2), and (C) no laser sessions.

Covariates included in the propensity score models were selected a priori based on their theoretical and empirical relevance to both treatment assignment and outcome prediction. To ensure adequate statistical power (considering loss to follow-up) and clinical interpretability, the following comparisons were performed: A (Day 2) vs. B, A (Day 2) vs. C, and A (Day 2) vs. A (Day 3).

Propensity scores were estimated using multiple model specifications. Covariate balance was assessed using standardized mean differences (SMD), empirical cumulative distribution function (eCDF) statistics, and standardized pairwise distances. Models demonstrating improved balance across these metrics were selected for matching. Matching was subsequently performed using nearest-neighbor or full matching approaches, depending on the comparison, and post-matching balance was evaluated through both quantitative diagnostics and graphical methods (e.g., density plots, eCDF plots, and Love plots). Regions of common support were also examined.

Following matching, a reduced analytic dataset including only matched observations was constructed. The balance tables for each comparison, along with graphical diagnostics, have been provided in the Supplementary Material (Supplementary File 1).

 

Comparison: A d2 vs B

 

The table below presents the diagnostics for eight propensity score models, evaluated using standardized mean differences (SMD), empirical CDF statistics, and standardized paired distances. Models with lower values in these metrics indicate better covariate balance.

Model	Method	Matching	Mean SMD After	Max SMD After	eCDF Mean	eCDF Max	Std. Pair Dist	Discarded?	Notes
1	GAMlogit	Nearest Neighbor	1.78	1.73	High	High	High	No	Poor balance, high SMDs
2	GAMlogit	Optimal	0.0048	0.1344	0.00	0.01	0.12	No	Best model — excellent balance
3	Non-linear Logit	Optimal	0.07	0.28	0.02	0.09	0.28	No	Very good balance
4	Non-linear Logit	Nearest Neighbor	0.39	0.38	Moderate	Moderate	Moderate	Yes	Worse than Model 3
5	Non-parametric Logit (no END/REEDA)	Optimal	0.06	0.17	0.01	0.06	0.24	No	Good model, but less info
6	Non-parametric Logit (no END/REEDA)	Nearest Neighbor	0.28	0.32	0.04	0.14	0.54	No	Worse than Model 5
7	Random Forest	Optimal	0.06	0.28	0.01	0.07	0.32	No	Competitive model
8	CART	Nearest Neighbor	0.14	0.33	0.03	0.14	0.44	No	Acceptable, but not best

 

Selected model: GAMlogit (optimal matching)

Model specification:

m.out_gam <- matchit(

  treatment ~ cov_END_pre + cov_REEDA_pre, data = data_compare_a_b,

  method = "nearest",  distance = "GAMlogit")

 

Comparison:  A d2 vs C

Summary Table (Post-Matching SMDs for Key Covariates) (A vs C)

Model	Matching	Distance	Max SMD (Post)	eCDF Max	Balance
m.out_1	Nearest	Logit	2.38	0.83	Poor
m.out_gam_1	Nearest	GAMlogit	2.38	0.83	Poor
m.out_gam_optimal_1	Optimal	GAMlogit	0.18	0.29	Excellent
m.out_non_linear_optimal_1	Optimal	Logit+Poly	0.21	0.20	Strong
m.out_non_linear_1	Nearest	Logit+Poly	2.38	>0.8	Poor
m.out_non_linear_gam_1	Nearest	GAMlogit	2.38	>0.8	Poor
m.out_non_linear_gam_opt_1	Optimal	GAMlogit	0.18	0.17	Excellent
m.out_non_linear_gam_full_1	Optimal	GAMlogit+	<0.15	<0.20	Best
m.out_cart_1	Optimal	Tree	~0.18	~0.17	Good

 

Selected model: Nonlinear GAMlogit with full matching

Model specification:

m.out_non_linear_gam_full_1 <- matchit(

  treatment ~ cov_END_pre_d1 + cov_REEDA_pre_d1 + I(cov_END_pre_d1^2) +

    I(cov_REEDA_pre_d1^2) + cov_END_pre_d1:cov_REEDA_pre_d1, data = data_compare_a_c,   method = "full", distance = "GAMlogit")

 

Comparasion A d2 vs A d3

Summary Table (Post-Matching SMDs for Key Covariates)

Model	Distance	Matching	Mean SMD After	Max SMD After	eCDF Mean	eCDF Max	Std. Pair Dist	Discarded?	Notes
1	Logit	Nearest	0.1327	0.2631	0.1513	0.3875	1.342	No	Poor balance
2	GAMlogit	Nearest	0.1056	0.2388	0.1056	0.369	1.211	Yes	Slightly better, but still weak balance
3	GAMlogit	Optimal	0.0643	0.1516	0.0918	0.3203	1.109	Yes	Improved, moderate balance
4	Logit+Poly	Optimal	0.0686	0.1752	0.1009	0.3562	1.151	No	Moderate balance
5	Logit+Poly	Nearest	0.122	0.2599	0.1249	0.3562	1.253	Yes	Poorer than model 3, not ideal
6	GAMlogit	Nearest	0.0406	0.1047	0.0479	0.1813	0.673	No	Excellent balance, best model
7	GAMlogit	Optimal	0.0415	0.105	0.048	0.1822	0.694	No	Excellent, very close to model 6
8	GAMlogit+	Optimal	0.0785	0.153	0.0612	0.1618	1.099	No	Moderate balance, not ideal
9	Tree	Optimal	0.0426	0.1139	0.0515	0.1988	0.71	No	Excellent, second best

 

Selected model: GAMlogit (nearest-neighbor matching)

Model specification:

m.out_non_linear_1 <- matchit(

  treatment ~ cov_END_pre_d1 + cov_REEDA_pre_d1 + I(cov_END_pre_d1^2) +

    I(cov_REEDA_pre_d1^2) + cov_END_pre_d1:cov_REEDA_pre_d1, data = data_compare_ad3_ad2,

  method = "nearest",  distance = "logit").

 

 

1. Model expected natural pain decline using the no-laser group as trajectory reference

 

We thank the reviewer for the opportunity to  discuss this theme.

 

To account for the expected natural decline in pain over time following delivery, we incorporated the no-photobiomodulation group (Group C) as a reference trajectory for spontaneous recovery in the longitudinal analysis.

Specifically, pain scores (NPS) measured over the first three days of hospitalization were modeled using Group C (no laser exposure) to represent the natural healing trajectory in the absence of intervention. This approach allowed us to distinguish treatment-related effects from the expected physiological reduction in pain over time.

Following propensity score matching, longitudinal comparisons were conducted between treatment groups and the matched no-laser group to estimate deviations from this natural pain trajectory. In this framework, the observed temporal evolution of pain in Group C served as the counterfactual reference against which the pain trajectories of Groups A and B were evaluated. Thus, any statistically significant differences in pain reduction between the treated groups and the matched no-laser group may be interpreted as the effect of photobiomodulation therapy beyond the expected spontaneous recovery process.

This modeling strategy was implemented to reduce bias associated with time-dependent improvements in pain and to enhance causal interpretability of treatment effects.

 

The corresponding analyses and trajectory comparisons have been incorporated into the revised Supplementary material File 1

 

 

1. Address Baseline Imbalances More Thoroughly, including these as covariates in all outcome models. Also, adjust for baseline laceration extent and Apgar differences.

 

We thank the reviewer for the thoughtful suggestion to include additional covariates in our analysis. We completely agree that exploring a wider range of potential predictors can provide valuable insights. Several baseline clinical and obstetric variables potentially associated with the outcomes were initially presented in Table 2, including mode of delivery, extent of laceration, and neonatal indices such as the Apgar score. A statistically significant association was observed between assisted vaginal delivery and greater postpartum pain intensity, as well as between lower Apgar scores and higher reported pain during the postpartum period.

 

However, after careful methodological consideration, we believe that incorporating numerous additional covariates into the current PSM model presents certain statistical challenges that could compromise the validity of our findings. We chose not to include all these adjustments in the models presented in this manuscript in order to parsimoniously preserve caution in the presentation of results. First, increasing the number of covariates in the statistical model elevates the risk of Type I errors (false positives) due to multiple testing, particularly when these analyses were not pre-specified. Second, given the observational nature of our study and the fact that our sample size was not predetermined but rather reflected the real-world characteristics of the service and patient flow during the study period, we must exercise caution when expanding the number of covariates in our models. With a relatively small sample size, the inclusion of numerous covariates would lead to model overfitting, resulting in unstable estimates and inflated standard errors. This could produce spurious associations that are not reproducible. In observational studies with limited samples, each additional covariate reduces the degrees of freedom and increases the risk of Type I errors, particularly when these analyses were not pre-specified in a protocol. Our sample size was determined by the natural patient flow (convenience sample), meaning it provides adequate power only for the primary analyses and the limited set of covariates originally planned. Expanding the model post-hoc to include multiple new covariates would violate key statistical assumptions and could generate misleading conclusions.

 

 Therefore, to maintain the statistical rigor and interpretability of our study, we have opted to retain the focus on the covariates pre-defined in our original protocol. We have, however, acknowledged the potential influence of other unmeasured covariates as a limitation in the discussion section (Line 488-510) and suggested this as a direction for future research (Lines 601-606).

 

1. You could control for analgesic timing and labor variables.

 

We thank the reviewer for the thoughtful suggestion to include additional covariates in our analysis. We completely agree that exploring a wider range of potential predictors can provide valuable insights. Information related to the timing of analgesic administration and labor-related variables, recognized as potential factors influencing the evaluated outcomes, was collected and presented.

 

However, after careful methodological consideration, we believe that incorporating numerous additional covariates into the current model presents certain statistical challenges that could compromise the validity of our findings. We chose not to include all these adjustments in the models presented in this manuscript in order to parsimoniously preserve caution in the presentation of results. First, increasing the number of covariates in the statistical model elevates the risk of Type I errors (false positives) due to multiple testing, particularly when these analyses were not pre-specified. Second, given the observational nature of our study and the fact that our sample size was not predetermined but rather reflected the real-world characteristics of the service and patient flow during the study period, we must exercise caution when expanding the number of covariates in our models. With a relatively small sample size, the inclusion of numerous covariates would lead to model overfitting, resulting in unstable estimates and inflated standard errors. This could produce spurious associations that are not reproducible. In observational studies with limited samples, each additional covariate reduces the degrees of freedom and increases the risk of Type I errors, particularly when these analyses were not pre-specified in a protocol. Our sample size was determined by the natural patient flow (convenience sample), meaning it provides adequate power only for the primary analyses and the limited set of covariates originally planned. Expanding the model post-hoc to include multiple new covariates would violate key statistical assumptions and could generate misleading conclusions.

 

 Therefore, to maintain the statistical rigor and interpretability of our study, we have opted to retain the focus on the covariates pre-defined in our original protocol. We have, however, acknowledged the potential influence of other unmeasured covariates as a limitation in the discussion section (Line 488-510) and suggested this as a direction for future research (Lines 601-606).

 

In summary, while the study is methodologically solid with its prospective, controlled, and randomized design, integrating these improvements would significantly enhance the internal and external validity, reduce bias, and provide stronger evidence for clinical recommendations.

 

Dear reviewer, thank you for carefully reading our manuscript and for your suggestions. We agree with your observation. Therefore, to maintain the statistical rigor and interpretability of our study, we have opted to retain the focus on the covariates pre-defined in our original protocol. We have, however, acknowledged the potential influence of other unmeasured covariates as a limitation in the discussion section (Line 488-510) and suggested this as a direction for future research (Lines 601-606).

 

 

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Comments to Authors

1. Explain how does the observational cohort design with propensity score matching is responsible for influencing the strength of the causal conclusions about PBM effectiveness when compared with some randomized controlled trial?
2. How do various biological mechanisms are responsible for the 808 nm photobiomodulation contribution to both analgesia and the enhanced tissue regeneration in perineal wounds scenarios?
3. Explain how the Numerical Pain Scale (NPS) and REEDA scale acts as an appropriate tool for evaluating various postoperative recovery processes, and what are the limitations of the performed scales?
4. What is the main objective of combining the propensity score matching (PSM) technique with the two-way ANOVA application, and how does this combination strengthen the comparison between the PBM and non-PBM groups for various applications mentioned in the article?
5. In Fig 2 what does the y label stand please mention in the plot.
6. In Table 2 you are presenting results of Mean, Standard deviation etc., so can you please mention the expression of these terms how you are calculating these values.

Author Response

Response Letter,

Point-by-point response to the reviewer’s comments

 

Dear Editor and Reviewers

 

Thank you very much for taking the time to review this manuscript. Please find the detailed responses in this letter and the corresponding revisions/corrections highlighted/in track changes in the re-submitted files

Reviewer 3

1. Explain how does the observational cohort design with propensity score matching is responsible for influencing the strength of the causal conclusions about PBM effectiveness when compared with some randomized controlled trial?

 

Dear reviewer, thank you for carefully reading our manuscript and for your valuable comment.

While randomized controlled trials (RCTs) operating under strictly controlled ideal conditions remain the gold standard for efficacy analysis, this 'real-world' observational design offers a good external validity by demonstrating the effectiveness and feasibility of PBM within a routine hospital workflow. This approach respects patient autonomy and pragmatic clinical practice, thereby aligning the results with clinical effectiveness. Although RCTs excel at isolating the specific effects of a therapy, our observational study contributes insights often overlooked by controlled trials, as it evaluates the intervention precisely as it occurs in a university hospital setting, facilitating a direct and rapid translation of findings into clinical practice. By allowing patients to transition between groups (migration between laser and conventional treatment), the study captures the acceptability and perceived need for the therapy among women—nuances that the forced randomization of an RCT might mask. Furthermore, whereas RCTs often employ highly restrictive inclusion criteria, this study demonstrates that PBM is effective in a diverse population with varying degrees of laceration extent and baseline pain levels. It also confirms that PBM implementation caused no adverse events and was successfully integrated with conventional therapies (analgesics and anti-inflammatories) without requiring supplemental analgesia. Moreover, to reinforce the robustness of these conclusions within a real-world setting, Propensity Score Matching (PSM) was utilized to balance the main baseline covariates between groups, thereby minimizing the selection bias inherent in observational designs. By demonstrating that pain reduction was statistically significant even after propensity score matching, our results suggest a robust association that approaches the internal validity of an RCT while maintaining the external validity of an already implemented clinical protocol.

 

Following your suggestion, we highlighted this explanation in the manuscript (lines 566-587).

 

1. How do various biological mechanisms are responsible for the 808 nm photobiomodulation contribution to both analgesia and the enhanced tissue regeneration in perineal wounds scenarios?

 

Dear reviewer, thank you for carefully reading our manuscript and for your valuable suggestions. Following your suggestion, we included in the discussion some aspects about biological mechanisms responsibles for the 808 nm photobiomodulation contribution to both analgesia and the enhanced tissue regeneration in perineal wounds scenarios (Lines 431-444).

 

The literature has well-documented the effects of PBM in various clinical contexts. The therapeutic effects of 808 nm PBM on pain and tissue repair are mediated by complex biomodulatory pathways. At this infra-red wavelength, photons penetrate deeply into the tissue and are primarily absorbed by cytochrome c oxidase in the mitochondrial respiratory chain. This interaction enhances ATP synthesis and leads to the photodissociation of nitric oxide (NO), which promotes local vasodilation and improves oxygen delivery to the damaged tissue. For analgesia, 808 nm PBM modulates the inflammatory response by downregulating pro-inflammatory cytokines (e.g., TNF-α and IL-6) and increasing anti-inflammatory mediators like IL-10, effectively reducing perineal edema and pressure-induced nociception. Regarding tissue regeneration, the therapy stimulates fibroblast proliferation and activates the TGF-β signaling pathway, which is essential for collagen deposition and extracellular matrix remodeling. These combined mechanisms explain the significant reduction in NPS scores and the healing rates observed, particularly in patients receiving multiple sessions.

 

1. Explain how the Numerical Pain Scale (NPS) and REEDA scale acts as an appropriate tool for evaluating various postoperative recovery processes, and what are the limitations of the performed scales?

 

We would like to thank the reviewer for the opportunity to discuss these topics and we have included these aspects in the Methods and in the study limitations

 

We agree that Numerical Pain Scale (NPS) and REEDA scale instruments have limitations. The NPS is a subjective measure dependent on individual perception and may be influenced by emotional, cultural, and clinical context factors, in addition to not capturing qualitative aspects of pain such as functional or emotional impact. The REEDA scale also has a subjective component, as the assessment of items such as redness and edema may vary between examiners, potentially leading to interobserver variability. Furthermore, REEDA primarily evaluates superficial aspects of healing and may not fully reflect deeper tissue changes.  Despite these limitations, when used in a standardized and combined manner, both scales constitute complementary and useful instruments for a comprehensive evaluation of postoperative recovery (lines 588-598).

 

The Numerical Pain Scale (NPS) is a commonly used instrument for pain measurement. It is quick and easy to administer and easily understood by patients, making it an appropriate method for estimating pain intensity (Martines, Jr 2011). The NPS is printed on the patient assessment form, and participants are instructed to mark a point on a 10-cm line indicating their pain intensity before and after the procedure. It consists of a straight line with the marking “no pain” at one end and “worst possible pain” at the other (Bijur et al., 2001). The REEDA scale is an instrument for assessing perineal healing developed by Davidson (Davidson 1974) and later revised by Carey (Carey ILP 1971). It includes five items related to the healing process: redness, edema, ecchymosis, discharge, and approximation of the wound edges. This scale can be used to evaluate all types of postpartum perineal trauma. The REEDA score ranges from 0 (indicating complete healing and closure of the lesion) to a maximum score of 15, which corresponds to the worst perineal healing with failure of approximation of the skin, subcutaneous tissue, and musculature (Lines145-151; 155-161)

 

The Numerical Pain Scale (NPS) and the REEDA scale are appropriate tools for evaluating different dimensions of postoperative recovery because they allow simple and standardized measurement of key aspects of recovery, such as pain and tissue healing. The NPS acts as a sensitive instrument for assessing the intensity of pain perceived by the patient, enabling monitoring of clinical evolution over time and evaluation of the response to the instituted treatment, as well as facilitating comparisons across different time points and groups. The REEDA scale complements this assessment by providing a structured clinical evaluation of perineal healing, including inflammatory signs and wound integrity, allowing early identification of changes in the tissue repair process and systematic monitoring of healing progression (Lines 199-208).

 

Naturally, the follow-up period limited to hospitalization, up to 72 hours, does not allow evaluation of tissue remodeling or complete healing of second- and third-degree perineal lacerations, which in fact occur over several weeks. This assessment was primarily to analyze the early aspects of the healing process, such as early inflammatory signs, edema, ecchymosis, hyperemia, wound edge approximation, and pain evolution, which are clinically relevant parameters in the immediate postpartum period and can be captured by instruments such as the REEDA scale and the NPS. In addition, the early treatment assessment was precisely to demonstrate that there was no worsening of inflammatory signs in patients treated with PBM, as these signs may worsen in the first days in the absence of specific wound care.

We acknowledge that this short-term assessment of tissue healing should be interpreted only as an initial signal of the inflammatory phase and the early repair process rather than as an endpoint of definitive healing. A clinically meaningful time horizon for evaluating perineal healing would be a long-term follow-up until complete wound healing, since the first 24–72 hours, to approximately 6 weeks postpartum, aligning with routine postpartum review and the expected timeline of tissue repair. Clinically relevant complications also include wound infection, suture dehiscence, persistent pain with functional impact, dyspareunia, urinary symptoms, and anorectal symptoms. Therefore, in addition to clinical instruments such as the REEDA scale, future studies should include patient-reported outcomes, quality-of-life and treatment satisfaction questionnaires, and evaluation of return to sexual activity and urinary/anorectal symptoms.

Despite these limitations, these preliminary results show a trend toward lower REEDA scores (indicating better outcomes) in the laser group, as suggested by the negative coefficient—an effect that could be confirmed in larger studies (Lines 517-540).

 

 

 

1. What is the main objective of combining the propensity score matching (PSM) technique with the two-way ANOVA application, and how does this combination strengthen the comparison between the PBM and non-PBM groups for various applications mentioned in the article?

 

 

We would like to thank the reviewer for the opportunity to highlight this aspect. We additionally performed the PSM after ANOVA to reassess the effect of photobiomodulation on outcomes. The PSM is a very robust, rigorous, and more reliable test in situations where the sample size is small or the groups being compared are unbalanced. PSM offers advantages over traditional ANOVA, particularly in small samples where ANOVA assumptions are rarely met, including the flexibility to use different probability distributions and covariates analysis. We have highlighted this aspects in the Methods item (Lines 225-229; 232-234)

 

 

1. In Fig 2 what does the y label stand please mention in the plot

 

Dear reviewer, thank you for pointing out this flaw. We corrected the figure 2 inserted in the manuscript as well as, its caption, making it clearer

 

1. In Table 2 you are presenting results of Mean, Standard deviation etc., so can you please mention the expression of these terms how you are calculating these values

 

Dear reviewer, thank you for pointing out this flaw. We corrected the figure inserted in the manuscript as well as, its caption, making it clearer. We also insert the description of calculations in the methods statistical analysis (Lines 213-216). Continuous variables are expressed as mean ± standard deviation (SD). The mean was calculated as the arithmetic average of the observations (Σx / n), and the SD was calculated as the square root of the sample variance [√Σ(xᵢ − x̄)² / (n − 1)]. Categorical variables are presented as absolute frequencies (n) and proportions when appropriate.

 

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

This prospective observational cohort study evaluates whether routine postpartum photobiomodulation (PBM) is associated with reduced perineal pain and improved healing after episiotomy or second-/third-degree tears during a short inpatient follow-up. The authors report statistically significant pain reduction (especially with two PBM sessions) and a non-definitive trend towards better healing.

The Introduction explains PBM mechanisms and cites general postoperative/dentistry evidence, but needs a clearer obstetric-specific gap: what is already known for perineal trauma postpartum, what is uncertain (dose/parameters, timing, outcomes, comparative effectiveness), and why a real-world cohort design is appropriate for this setting.

The Methods state women could move between PBM/no-PBM based on pain and discharge timing, meaning exposure is time-varying and confounded by indication. Please pre-specify how participants were classified for each analysis (e.g., intention-to-treat by day-1 choice vs per-day exposure vs cumulative sessions) and justify why those approaches answer the stated objective. You mention PSM, Welch tests, OLS regression, and Monte Carlo validation, but you should explicitly list: covariates included in the propensity model (and why), matching method (ratio, caliper, replacement), balance assessment approach (standardised mean differences/diagnostics), and how repeated measurements across days were treated (independence assumptions vs clustered/longitudinal structure).

In results, early discharge and switching create varying denominators and potential informative missingness. Please report, in text, how many observations contributed to each key comparison (by day and subgroup), and whether any participants had incomplete NPS/REEDA measurements and how this was handled (complete-case vs imputation vs model-based). Furthermore, baseline differences (e.g., pain severity driving PBM uptake; other differences noted between subgroups) mean raw comparisons can be misleading. Present results in a way that clearly distinguishes: (a) within-person pre/post PBM change, (b) between-group differences after adjustment, and (c) which is primary. Also report whether the observed NPS differences meet a clinically meaningful threshold, not only statistical significance.

The Discussion sometimes reads causally (“PBM was effective…”) despite the observational design, self-selection, and migration based on pain. Reframe to “associated with” and expand on confounding by indication, regression to the mean, natural recovery, placebo/context effects, and analgesic/anaesthetic carryover as alternative explanations.

Furthermore, the REEDA signal is described as a trend with p-values not meeting the conventional threshold; the very short inpatient follow-up also limits inference about clinically important healing outcomes. Discuss what outcome horizon would be clinically meaningful (e.g., 1–6 weeks), what complications matter (infection, dehiscence, dyspareunia, continence symptoms), and how this hospital’s routine PBM workflow affects generalisability to other maternity settings.

Author Response

Response Letter,

Point-by-point response to the reviewer’s comments

 

Dear Editor and Reviewers

 

Thank you very much for taking the time to review this manuscript. Please find the detailed responses in this letter and the corresponding revisions/corrections highlighted/in track changes in the re-submitted files

Reviewer 4

This prospective observational cohort study evaluates whether routine postpartum photobiomodulation (PBM) is associated with reduced perineal pain and improved healing after episiotomy or second-/third-degree tears during a short inpatient follow-up. The authors report statistically significant pain reduction (especially with two PBM sessions) and a non-definitive trend towards better healing.

 

1. The Introduction explains PBM mechanisms and cites general postoperative/dentistry evidence, but needs a clearer obstetric-specific gap: what is already known for perineal trauma postpartum, what is uncertain (dose/parameters, timing, outcomes, comparative effectiveness), and why a real-world cohort design is appropriate for this setting

 

Dear reviewer, thank you for carefully reading our manuscript and for your valuable suggestions. Following your suggestion, we highlighted these aspects in the manuscript (Lines 71-85).

 

 

Despite growing interest in PBM for postpartum perineal care, the current evidence remains limited and heterogeneous. Previous studies evaluating PBM after episiotomy or perineal lacerations have generally involved small samples, short follow-up periods, and variability in treatment parameters such as wavelength, energy dose, timing, and number of sessions, making it difficult to establish standardized clinical recommendations. Although still underexplored in the obstetric field, PBM has shown potential in relieving pain and promoting healing after perineal trauma, incisions, and wounds [11]. Furthermore, there is limited evidence on PBM effectiveness in real-world obstetric settings, where variations in patient characteristics, clinical practices, and healing processes may influence outcomes. Important uncertainties therefore remain regarding optimal treatment protocols and the magnitude of PBM benefits on pain reduction and tissue repair in routine care. In this context, a real-world cohort design allows the evaluation of PBM under everyday clinical conditions, providing pragmatic evidence on its effectiveness and safety and supporting its potential incorporation into obstetric practice.

 

 

2. The Methods state women could move between PBM/no-PBM based on pain and discharge timing, meaning exposure is time-varying and confounded by indication. Please pre-specify how participants were classified for each analysis (e.g., intention-to-treat by day-1 choice vs per-day exposure vs cumulative sessions) and justify why those approaches answer the stated objective

 

We would like to thank the reviewer for the opportunity to discuss thia aspect. We have highlight this explanation on manuscript.

 

In our hospital, PBM is routinely offered to all patients with episiotomies and perineal lacerations after vaginal delivery. The decision to accept the treatment is made by the patient (Lines 164-167). The primary objective of the study was to evaluate PBM effectiveness under real-world conditions rather than efficacy under controlled exposure; therefore, classification based on the initial treatment decision was considered the most appropriate approach to reflect pragmatic clinical use (Lines 218-219).

 

For all analyses regarding outcomes, comparisons were made between subgroups of participants who chose to accept adjuvant laser associated with conventional treatment and the subgroup of participants who chose only conventional treatment. Since participants have the right to refuse or accept the laser according to their own wishes, they voluntarily transitioned between subgroups during hospitalization primarily based on their pain levels and perceived need for additional analgesic treatment. In this way, the cohort composition changed daily due to participant migration between laser option. This situation posed a methodological challenge for analysis due to changes in treatment groups, resulting in multiple possible analytical perspectives for the subgroups formed. To address this issue, participants were pre-specified and classified according to their treatment choice on the first evaluation day (day 1), which reflects the initial clinical decision context and minimizes bias related to changes occurring after treatment initiation. This approach was chosen to approximate an intention-to-treat framework, preserving baseline comparability and allowing evaluation of PBM effectiveness as implemented in routine clinical practice. In addition, secondary exploratory analyses considered the cumulative number of PBM sessions received to assess potential dose–response effects, acknowledging variability in exposure due to hospital discharge timing and clinical evolution. The primary objective of the study was to evaluate PBM effectiveness under real-world conditions rather than efficacy under controlled exposure; therefore, classification based on the initial treatment decision was considered the most appropriate approach to reflect pragmatic clinical use. The outcomes were analyzed by comparing pre-treatment NPS and REEDA scores between the first, second and third  day and also between subgroups (laser or no laser) by analyzing daily measurements described. A comparison between before and after PBM moments for NPS was also done within the subgroup of participants who accepted PBM. For comparisons about measurements taken across the three days of hospitalization, we also chose to compare REEDA and NPS scores among three other subgroups (that emerged from participants’ spontaneous choice): participants who chose not to undergo the laser on any day; participants who underwent only one laser session throughout their three days of hospitalization; and women who underwent two laser sessions throughout three days of hospitalization (Lines 167-198).

 

This situation posed a methodological challenge for analysis due to changes in treatment groups, resulting in multiple possible analytical perspectives for the subgroups formed. To address this issue, participants were pre-specified and classified according to their treatment choice on the first evaluation day (day 1), which reflects the initial clinical decision context and minimizes bias related to changes occurring after treatment initiation. This approach was chosen to approximate an intention-to-treat framework, preserving baseline comparability and allowing evaluation of PBM effectiveness as implemented in routine clinical practice. In addition, secondary exploratory analyses considered the cumulative number of PBM sessions received to assess potential dose–response effects, acknowledging variability in exposure due to hospital discharge timing and clinical evolution (Lines 173-188).

 

 

 

3. You mention PSM, Welch tests, OLS regression, and Monte Carlo validation, but you should explicitly list: covariates included in the propensity model (and why), matching method (ratio, caliper, replacement), balance assessment approach (standardised mean differences/diagnostics), and how repeated measurements across days were treated (independence assumptions vs clustered/longitudinal structure).

 

We thank the reviewer for the opportunity to discuss this important topic. After your review we also provided these aspects discussed below on the Supplementary Material File 1.

 

 

In accordance with the prespecified Statistical Analysis Plan, the outcomes included pain (Numeric Pain Scale – NPS) and wound healing (REEDA scale), both assessed over the first three days of hospitalization. The primary predictor (treatment exposure) was defined as receipt of photobiomodulation therapy during hospitalization and categorized into three groups: (A) two laser sessions (Days 1 and 2), (B) one laser session (Day 1 or Day 2), and (C) no laser sessions. Covariates included in the propensity score models were selected a priori based on their theoretical and empirical relevance to both treatment assignment and outcome prediction. To ensure adequate statistical power (considering loss to follow-up) and clinical interpretability, the following comparisons were performed: A (Day 2) vs. B, A (Day 2) vs. C, and A (Day 2) vs. A (Day 3).

 

Propensity scores were estimated using multiple model specifications. Covariate balance was assessed using standardized mean differences (SMD), empirical cumulative distribution function (eCDF) statistics, and standardized pairwise distances. Models demonstrating improved balance across these metrics were selected for matching. Matching was subsequently performed using nearest-neighbor or full matching approaches, depending on the comparison, and post-matching balance was evaluated through both quantitative diagnostics and graphical methods (e.g., density plots, eCDF plots, and Love plots). Regions of common support were also examined.

Following matching, a reduced analytic dataset including only matched observations was constructed. The balance tables for each comparison, along with graphical diagnostics, have been provided in the Supplementary Material (Supplementary File 1).

 

Comparison: A d2 vs B

The table below presents the diagnostics for eight propensity score models, evaluated using standardized mean differences (SMD), empirical CDF statistics, and standardized paired distances. Models with lower values in these metrics indicate better covariate balance.

Model	Method	Matching	Mean SMD After	Max SMD After	eCDF Mean	eCDF Max	Std. Pair Dist	Discarded?	Notes
1	GAMlogit	Nearest Neighbor	1.78	1.73	High	High	High	No	Poor balance, high SMDs
2	GAMlogit	Optimal	0.0048	0.1344	0.00	0.01	0.12	No	Best model — excellent balance
3	Non-linear Logit	Optimal	0.07	0.28	0.02	0.09	0.28	No	Very good balance
4	Non-linear Logit	Nearest Neighbor	0.39	0.38	Moderate	Moderate	Moderate	Yes	Worse than Model 3
5	Non-parametric Logit (no END/REEDA)	Optimal	0.06	0.17	0.01	0.06	0.24	No	Good model, but less info
6	Non-parametric Logit (no END/REEDA)	Nearest Neighbor	0.28	0.32	0.04	0.14	0.54	No	Worse than Model 5
7	Random Forest	Optimal	0.06	0.28	0.01	0.07	0.32	No	Competitive model
8	CART	Nearest Neighbor	0.14	0.33	0.03	0.14	0.44	No	Acceptable, but not best

 

Selected model: GAMlogit (optimal matching)

Model specification:

m.out_gam <- matchit(

  treatment ~ cov_END_pre + cov_REEDA_pre, data = data_compare_a_b,

  method = "nearest",  distance = "GAMlogit")

 

Comparison:  A d2 vs C

Summary Table (Post-Matching SMDs for Key Covariates) (A vs C)

Model	Matching	Distance	Max SMD (Post)	eCDF Max	Balance
m.out_1	Nearest	Logit	2.38	0.83	Poor
m.out_gam_1	Nearest	GAMlogit	2.38	0.83	Poor
m.out_gam_optimal_1	Optimal	GAMlogit	0.18	0.29	Excellent
m.out_non_linear_optimal_1	Optimal	Logit+Poly	0.21	0.20	Strong
m.out_non_linear_1	Nearest	Logit+Poly	2.38	>0.8	Poor
m.out_non_linear_gam_1	Nearest	GAMlogit	2.38	>0.8	Poor
m.out_non_linear_gam_opt_1	Optimal	GAMlogit	0.18	0.17	Excellent
m.out_non_linear_gam_full_1	Optimal	GAMlogit+	<0.15	<0.20	Best
m.out_cart_1	Optimal	Tree	~0.18	~0.17	Good

 

Selected model: Nonlinear GAMlogit with full matching

Model specification:

m.out_non_linear_gam_full_1 <- matchit(

  treatment ~ cov_END_pre_d1 + cov_REEDA_pre_d1 + I(cov_END_pre_d1^2) +

    I(cov_REEDA_pre_d1^2) + cov_END_pre_d1:cov_REEDA_pre_d1, data = data_compare_a_c,   method = "full", distance = "GAMlogit")

 

Comparasion A d2 vs A d3

Summary Table (Post-Matching SMDs for Key Covariates)

Model	Distance	Matching	Mean SMD After	Max SMD After	eCDF Mean	eCDF Max	Std. Pair Dist	Discarded?	Notes
1	Logit	Nearest	0.1327	0.2631	0.1513	0.3875	1.342	No	Poor balance
2	GAMlogit	Nearest	0.1056	0.2388	0.1056	0.369	1.211	Yes	Slightly better, but still weak balance
3	GAMlogit	Optimal	0.0643	0.1516	0.0918	0.3203	1.109	Yes	Improved, moderate balance
4	Logit+Poly	Optimal	0.0686	0.1752	0.1009	0.3562	1.151	No	Moderate balance
5	Logit+Poly	Nearest	0.122	0.2599	0.1249	0.3562	1.253	Yes	Poorer than model 3, not ideal
6	GAMlogit	Nearest	0.0406	0.1047	0.0479	0.1813	0.673	No	Excellent balance, best model
7	GAMlogit	Optimal	0.0415	0.105	0.048	0.1822	0.694	No	Excellent, very close to model 6
8	GAMlogit+	Optimal	0.0785	0.153	0.0612	0.1618	1.099	No	Moderate balance, not ideal
9	Tree	Optimal	0.0426	0.1139	0.0515	0.1988	0.71	No	Excellent, second best

 

Selected model: GAMlogit (nearest-neighbor matching)

Model specification:

m.out_non_linear_1 <- matchit(

  treatment ~ cov_END_pre_d1 + cov_REEDA_pre_d1 + I(cov_END_pre_d1^2) +

    I(cov_REEDA_pre_d1^2) + cov_END_pre_d1:cov_REEDA_pre_d1, data = data_compare_ad3_ad2,

  method = "nearest",  distance = "logit").

 

 

 

4. In results, early discharge and switching create varying denominators and potential informative missingness. Please report, in text, how many observations contributed to each key comparison (by day and subgroup), and whether any participants had incomplete NPS/REEDA measurements and how this was handled (complete-case vs imputation vs model-based)

 

We thank the reviewer for highlighting this important point. Due to early discharge and treatment switching during hospitalization, the number of observations contributing to each comparison varied across groups and time points. The sample sizes for each matched comparison are reported below and Supplementary File 1.

 

For the comparison between Group A (Day 2) and Group B, 124 participants were included in Group A (Day 2) and 35 in Group B. After propensity score matching, all 35 participants in Group B were matched to 35 participants from Group A (Day 2), resulting in 35 matched pairs (n = 70). The remaining 89 participants in Group A (Day 2) were unmatched and therefore excluded from this specific comparison.

 

For the comparison between Group A (Day 2) and Group C (no laser), 124 participants were included in Group A (Day 2) and 24 in Group C. Following matching, all participants from both groups were retained in the matched analytic sample.

 

For the comparison between Group A (Day 2) and Group A (Day 3), 124 participants were included in Group A (Day 2) and 58 in Group A (Day 3). After matching, 58 participants from Group A (Day 2) were matched to all 58 participants in Group A (Day 3), while the remaining 66 participants from Group A (Day 2) were excluded from this analysis. Incomplete NPS or REEDA measurements were identified prior to analysis. Observations with missing outcome data were excluded using a complete-case approach. No data imputation techniques were applied. Consequently, all analyses were conducted using matched samples with complete outcome information.

 

Comparison: A d2 vs B

 Sample Sizes:

                              Control Treated

All                               35     124

Matched                    35      35

Unmatched                0       89

Discarded                   0        0

 

Comparison:  A d2 vs C

Sample Sizes:

                                          Control Treated

All                                         24       124

Matched                             24       124

Unmatched                          0         0

Discarded                             0         0

 

Comparasion A d2 vs A d3

 

Sample Sizes:

                                   Control Treated

All                                     124     58

Matched                          58      58

Unmatched                     66       0

Discarded                         0        0

 

 

5. Furthermore, baseline differences (e.g., pain severity driving PBM uptake; other differences noted between subgroups) mean raw comparisons can be misleading.

 

We thank the reviewer for the thoughtful suggestion to include additional covariates in our analysis. We completely agree that exploring a wider range of potential predictors can provide valuable insights. Information related to differences noted between subgroups, recognized as potential factors influencing the evaluated outcomes, was collected and presented.

 

However, after careful methodological consideration, we believe that incorporating numerous additional covariates into the current model presents certain statistical challenges that could compromise the validity of our findings. We chose not to include all these adjustments in the models presented in this manuscript in order to parsimoniously preserve caution in the presentation of results. First, increasing the number of covariates in the statistical model elevates the risk of Type I errors (false positives) due to multiple testing, particularly when these analyses were not pre-specified. Second, given the observational nature of our study and the fact that our sample size was not predetermined but rather reflected the real-world characteristics of the service and patient flow during the study period, we must exercise caution when expanding the number of covariates in our models. With a relatively small sample size, the inclusion of numerous covariates would lead to model overfitting, resulting in unstable estimates and inflated standard errors. This could produce spurious associations that are not reproducible. In observational studies with limited samples, each additional covariate reduces the degrees of freedom and increases the risk of Type I errors, particularly when these analyses were not pre-specified in a protocol. Our sample size was determined by the natural patient flow (convenience sample), meaning it provides adequate power only for the primary analyses and the limited set of covariates originally planned. Expanding the model post-hoc to include multiple new covariates would violate key statistical assumptions and could generate misleading conclusions.

 

Therefore, to maintain the statistical rigor and interpretability of our study, we have opted to retain the focus on the covariates pre-defined in our original protocol. We have, however, acknowledged the potential influence of other unmeasured covariates as a limitation in the discussion section (Line 488-510) and suggested this as a direction for future research (Lines 601-606).

 

6. Present results in a way that clearly distinguishes: (a) within-person pre/post PBM change, (b) between-group differences after adjustment, and (c) which is primary.

 

Dear reviewer, thank you for this valuable comment. Following your suggestion, we included these denominations throughout the text in the entire presentation of the results and statistical analysis to clearly distinguishes: (a) within-person pre/post PBM change, (b) between-group differences after adjustment (Lines 286, 289, 299, 305, 311, 321, 331, 339, 343, 350, 353). The primary comparision is the between-group diferences (clearly statemented in Lines 218-219).

 

7. Also report whether the observed NPS differences meet a clinically meaningful threshold, not only statistical significance.

 

Dear reviewer, thank you for this valuable comment. Following your suggestion, we included these aspects on manuscript (Lines 465-482)  

 

The within-group differences in the mean NPS pain scores among patients who underwent laser therapy were both statistically significant and clinically relevant. For the subgroup of patients who received a single laser session, the difference in NPS pain scores over the three-day assessment period clearly exceeded the minimal clinically important difference (MCID ~2 points), although this finding was obtained from a small sample. The subgroup that received two laser sessions also demonstrated a clinically significant, albeit borderline, improvement (~1.35 points). Furthermore, at the end of follow-up, patients who received two sessions presented better outcomes than those who received no laser sessions. The main reported effect was a mean difference of 1.6761 points in NPS scores (two sessions vs. zero sessions), with a p-value of 0.000191. In terms of clinical relevance, this mean effect of 1.67 points is likely clinically meaningful and closely approaches the more conservative Minimal Clinically Important Difference (MCID) threshold (~2 points). Moreover, since the 95% confidence interval includes values above 2, it is plausible that the true (population) effect may be clinically strong in certain scenarios. In contrast, for the subgroup that received no laser sessions, the 0.75-point difference over the three days was clearly not clinically significant. These data reinforce and corroborate the statistical difference found suggesting that the laser, in this population, is related to a greater improvement in pain than conventional treatment.

 

 

8. The Discussion sometimes reads causally (“PBM was effective…”) despite the observational design, self-selection, and migration based on pain. Reframe to “associated with” and expand on confounding by indication, regression to the mean, natural recovery, placebo/context effects, and analgesic/anaesthetic carryover as alternative explanations.

 

 

Dear reviewer, thank you for this valuable comment. Following your suggestion, we included these denominations throughout the text in the entire discussion,  conclusion  and abstract (Lines 43-46, 407, 450, 559, 609). Furthermore, we expand our discussion and limitations in this regard (Lines 457-464; 488-510; 599-607).

 

9. Furthermore, the REEDA signal is described as a trend with p-values not meeting the conventional threshold; the very short inpatient follow-up also limits inference about clinically important healing outcomes. Discuss what outcome horizon would be clinically meaningful (e.g., 1–6 weeks), what complications matter (infection, dehiscence, dyspareunia, continence symptoms), and how this hospital’s routine PBM workflow affects generalisability to other maternity settings

 

Dear reviewer, thank you for this valuable comment.

 

Naturally we acknowledge that p-values above the conventional threshold do not constitute conclusive evidence and should be interpreted with caution as suggestive findings, possibly limited by statistical power, heterogeneity of exposure, and the short observation period and sample. Moreover, the follow-up period limited to hospitalization, up to 72 hours, does not allow evaluation of tissue remodeling or complete healing of second- and third-degree perineal lacerations, which in fact occur over several weeks. This assessment was primarily to analyze the early aspects of the healing process, such as early inflammatory signs, edema, ecchymosis, hyperemia, wound edge approximation, and pain evolution, which are clinically relevant parameters in the immediate postpartum period and can be captured by instruments such as the REEDA scale and the NPS. In addition, the early treatment assessment was precisely to demonstrate that there was no worsening of inflammatory signs in patients treated with PBM, as these signs may worsen in the first days in the absence of specific wound care.

We acknowledge that this short-term assessment of tissue healing should be interpreted only as an initial signal of the inflammatory phase and the early repair process rather than as an endpoint of definitive healing. A clinically meaningful time horizon for evaluating perineal healing would be a long-term follow-up until complete wound healing, since the first 24–72 hours, to approximately 6 weeks postpartum, aligning with routine postpartum review and the expected timeline of tissue repair. Clinically relevant complications also include wound infection, suture dehiscence, persistent pain with functional impact, dyspareunia, urinary symptoms, and anorectal symptoms. Therefore, in addition to clinical instruments such as the REEDA scale, future studies should include patient-reported outcomes, quality-of-life and treatment satisfaction questionnaires, and evaluation of return to sexual activity and urinary/anorectal symptoms.

In spite of these limitations, these initial early results could suggest that PBM may be effective for perineal healing, an effect that could be confirmed in larger studies.

 

Following your suggestion, we included these aspects on discussion  (Lines 514-540; 591-598)

 

 

 

Author Response File: Author Response.pdf

Reviewer 5 Report

Comments and Suggestions for Authors

The study addresses a clinically relevant issue (perineal trauma management) and employs a sound methodological approach, using Propensity Score Matching (PSM) to mitigate the limitations of its observational design. The Photobiomodulation (PBM) parameters are well-defined, ensuring high reproducibility. However, to meet publication standards, several critical points regarding the study design, statistical interpretation, and conclusions must be addressed.

Please provide detailed responses and revise the manuscript according to the following points:

1. Justification of cohort design: Why was an observational study chosen instead of a Randomized Controlled Trial (RCT)? Given that patients with higher initial pain levels (NPS 4.07 vs 1.71) were the ones who opted for laser therapy, please discuss how this self-selection bias might have influenced the perceived effectiveness, even after PSM.

2. Short-term follow-up: The study is limited to the hospital stay (max 3 days). How do the authors justify the assessment of "tissue healing" in such a short window, considering that remodelling and complete healing of Grade 2 and 3 lacerations take significantly longer?

3. Clarification of significance: The results for the REEDA scale showed a p-value of 0.076. By standard scientific convention (p < 0.05), this is not statistically significant. How do you justify the statement in the Abstract and Conclusion that PBM was effective in improving healing? Please revise the text to describe this as a trend toward improvement or a non-significant clinical observation.

4. Session variability: There was a range of 1 to 3 sessions due to early hospital discharges. Did the authors perform a sub-analysis to see if there was a dose-response relationship (e.g., did 3 sessions show significantly better results than 1)?

5. Analgesic interference: Since both groups received standard pharmacological care, was a sensitivity analysis conducted to ensure that the use of conventional analgesics did not mask or confound the isolated effect of the PBM?

6. Internal consistency: There is a discrepancy between the Abstract (which claims effectiveness) and the Results section (which shows non-significant p-values for healing). Please align the entire manuscript to reflect the statistical reality of the data.

7. Originality vs clinical context: While PBM is documented for pain, its application in Grade 3 lacerations in a real-world setting is valuable. Please expand the Discussion to compare your findings with existing literature, specifically for high-grade tears.

Author Response

Response Letter,

Point-by-point response to the reviewer’s comments

Dear Editor and Reviewers

Thank you very much for taking the time to review this manuscript. Please find the detailed responses in this letter and the corresponding revisions/corrections highlighted/in track changes in the re-submitted files

Reviewer 5

The study addresses a clinically relevant issue (perineal trauma management) and employs a sound methodological approach, using Propensity Score Matching (PSM) to mitigate the limitations of its observational design. The Photobiomodulation (PBM) parameters are well-defined, ensuring high reproducibility. However, to meet publication standards, several critical points regarding the study design, statistical interpretation, and conclusions must be addressed.

 

 

Please provide detailed responses and revise the manuscript according to the following points:

 

1. Justification of cohort design: Why was an observational study chosen instead of a Randomized Controlled Trial (RCT)? Given that patients with higher initial pain levels (NPS 4.07 vs 1.71) were the ones who opted for laser therapy, please discuss how this self-selection bias might have influenced the perceived effectiveness, even after PSM.

 

 

Dear reviewer, we agree with your observation that the gold standard for evaluating efficacy is the randomized controlled clinical study. We also agree that patients chose the use of laser according to their baseline pain intensity.

 

In our hospital, PBM is routinely offered to all patients with episiotomies and perineal lacerations after vaginal delivery. The decision to accept the treatment is made by the patient (Lines 165-167).

For all analyses regarding outcomes, comparisons were made between subgroups of participants who chose to accept adjuvant laser associated with conventional treatment and the subgroup of participants who chose only conventional treatment. Since participants have the right to refuse or accept the laser according to their own wishes, they voluntarily transitioned between subgroups during hospitalization primarily based on their pain levels and perceived need for additional analgesic treatment. In this way, the cohort composition changed daily due to participant migration between laser option. This situation posed a methodological challenge for analysis due to changes in treatment groups, resulting in multiple possible analytical perspectives for the subgroups formed. To address this issue, participants were pre-specified and classified according to their treatment choice on the first evaluation day (day 1), which reflects the initial clinical decision context and minimizes bias related to changes occurring after treatment initiation. This approach was chosen to approximate an intention-to-treat framework, preserving baseline comparability and allowing evaluation of PBM effectiveness as implemented in routine clinical practice (Lines 167-181).

We also did not analyze the influence of other covariates—such as the type and number of previous labours, differences noted between subgroups, analgesic medications used, baseline laceration extent or Apgar differences, nor the timing of laser use (i.e., whether women chose treatment on the first or second day), natural recovery, placebo/context effects, and analgesic/anaesthetic carryover—on outcomes. Future exploratory post-hoc analyses of these factors may clarify their effects on treatment response and provide new insights for further research (Lines 601-607).

In addition, secondary exploratory analyses considered the cumulative number of PBM sessions received to assess potential dose–response effects, acknowledging variability in exposure due to hospital discharge timing and clinical evolution. The primary objective of the study was to evaluate PBM effectiveness under real-world conditions rather than efficacy under controlled exposure; therefore, classification based on the initial treatment decision was considered the most appropriate approach to reflect pragmatic clinical use.The outcomes were analyzed by comparing pre-treatment NPS and REEDA scores between the first, second and third  day and also between subgroups (laser or no laser) by analyzing daily measurements described. A comparison between before and after PBM moments for NPS was also done within the subgroup of participants who accepted PBM. For comparisons about measurements taken across the three days of hospitalization, we also chose to compare REEDA and NPS scores among three other subgroups (that emerged from participants’ spontaneous choice): participants who chose not to undergo the laser on any day; participants who underwent only one laser session throughout their three days of hospitalization; and women who underwent two laser sessions throughout three days of hospitalization (Lines 182-198).

 

Despite this challenge, the migration provided a rare opportunity in observational studies to compare outcomes between users and non-users of the laser therapy. This migration highlights inherent methodological challenges in pragmatic clinical studies within real-world settings, where treatment choices are guided by clinical need rather than research protocols. Although such studies require complex analysis, they generate highly applicable results by evaluating interventions under actual clinical conditions. This approach assesses not just effectiveness, but also feasibility, acceptability, and impact on healthcare workflows (Lines 543-550).

While randomized controlled trials (RCTs) operating under strictly controlled ideal conditions remain the gold standard for efficacy analysis, this 'real-world' observational design offers a good external validity by demonstrating the effectiveness and feasibility of PBM within a routine hospital workflow. This approach respects patient autonomy and pragmatic clinical practice, thereby aligning the results with clinical effectiveness. Although RCTs excel at isolating the specific effects of a therapy, our observational study contributes insights often overlooked by controlled trials, as it evaluates the intervention precisely as it occurs in a university hospital setting, facilitating a direct and rapid translation of findings into clinical practice. By allowing patients to transition between groups (migration between laser and conventional treatment), the study captures the acceptability and perceived need for the therapy among women—nuances that the forced randomization of an RCT might mask. Furthermore, whereas RCTs often employ highly restrictive inclusion criteria, this study demonstrates that PBM is effective in a diverse population with varying degrees of laceration extent and baseline pain levels. It also confirms that PBM implementation caused no adverse events and was successfully integrated with conventional therapies (analgesics and anti-inflammatories) without requiring supplemental analgesia. Moreover, to reinforce the robustness of these conclusions within a real-world setting, Propensity Score Matching (PSM) was utilized to balance the main baseline covariates between groups, thereby minimizing the selection bias inherent in observational designs. By demonstrating that pain reduction was statistically significant even after propensity score matching, our results suggest a robust association that approaches the internal validity of an RCT while maintaining the external validity of an already implemented clinical protocol (lines 566-587).

The PSM is a very robust, rigorous, and more reliable test in situations where the sample size is small or the groups being compared are unbalanced. PSM offers advantages over traditional ANOVA, particularly in small samples where ANOVA assumptions are rarely met, including the flexibility to use different probability distributions and covariates analysis. After matching, treatment effects were estimated using Welch's t-test and ordinary least squares regression, validated with Monte Carlo simulations. The PSM was conducted in R (version 4.4.3) using R Studio, with code written by CSSF author. All propensity scores matching comparisons, along with comprehensive documentation of the PSM analytical process (including all calculations, tables, and graphs), are available in Supplementary File 1 (Lines 225-234).

 

 

Following your suggestion we discussed these aspects in the manuscript (lines K)

 

 

 

1. Short-term follow-up: The study is limited to the hospital stay (max 3 days). How do the authors justify the assessment of "tissue healing" in such a short window, considering that remodelling and complete healing of Grade 2 and 3 lacerations take significantly longer

 

Dear reviewer, thank you for this valuable comment.

 

Naturally we acknowledge that p-values above the conventional threshold do not constitute conclusive evidence and should be interpreted with caution as suggestive findings, possibly limited by statistical power, heterogeneity of exposure, and the short observation period and sample. Moreover, the follow-up period limited to hospitalization, up to 72 hours, does not allow evaluation of tissue remodeling or complete healing of second- and third-degree perineal lacerations, which in fact occur over several weeks. This assessment was primarily to analyze the early aspects of the healing process, such as early inflammatory signs, edema, ecchymosis, hyperemia, wound edge approximation, and pain evolution, which are clinically relevant parameters in the immediate postpartum period and can be captured by instruments such as the REEDA scale and the NPS. In addition, the early treatment assessment was precisely to demonstrate that there was no worsening of inflammatory signs in patients treated with PBM, as these signs may worsen in the first days in the absence of specific wound care. We acknowledge that this short-term assessment of tissue healing should be interpreted only as an initial signal of the inflammatory phase and the early repair process rather than as an endpoint of definitive healing. A clinically meaningful time horizon for evaluating perineal healing would be a long-term follow-up until complete wound healing, since the first 24–72 hours, to approximately 6 weeks postpartum, aligning with routine postpartum review and the expected timeline of tissue repair. Clinically relevant complications also include wound infection, suture dehiscence, persistent pain with functional impact, dyspareunia, urinary symptoms, and anorectal symptoms. Therefore, in addition to clinical instruments such as the REEDA scale, future studies should include patient-reported outcomes, quality-of-life and treatment satisfaction questionnaires, and evaluation of return to sexual activity and urinary/anorectal symptoms Despite these limitations, these preliminary results show a trend toward lower REEDA scores (indicating better outcomes) in the laser group, as suggested by the negative coefficient—an effect that could be confirmed in larger studies.

 

Following your suggestion we discussed these aspects in the manuscript (Lines 514-540)

 

1. Clarification of significance: The results for the REEDA scale showed a p-value of 0.076. By standard scientific convention (p < 0.05), this is not statistically significant. How do you justify the statement in the Abstract and Conclusion that PBM was effective in improving healing? Please revise the text to describe this as a trend toward improvement or a non-significant clinical observation.

 

Dear reviewer, thank you for this valuable comment. Following your suggestion, we modify this statement  throughout the manuscript (Lines 43-46, 527-540)

 

1. Session variability: There was a range of 1 to 3 sessions due to early hospital discharges. Did the authors perform a sub-analysis to see if there was a dose-response relationship (e.g., did 3 sessions show significantly better results than 1)?

 

Dear reviewer, thank you for your comment.

 

During the short period of hospitalization in our hospital, the laser is offered daily to patients. Most parturient women are discharged 48 to 72 hours after giving birth.  We evaluated the outcomes before and after each of these daily applications (as shown in Figure 1) and evaluated these results. The primary objective of the study was to evaluate PBM effectiveness under real-world conditions rather than efficacy under controlled exposure; therefore, classification based on the initial treatment decision was considered the most appropriate approach to reflect pragmatic clinical use. Exploratory analyses considered the cumulative number of PBM sessions received to assess potential dose–response effects, acknowledging variability in exposure due to hospital discharge timing and clinical evolution.

 

In the intra-subgroups comparison of NPS values according to the number of laser sessions received over the three days, a repeated-measures ANOVA found significant differences for women who underwent one session (n=17, *p*=0.016) and for those who underwent two sessions (n=73, *p*<0.0001) across the three days assessment period. No significant difference was found for the subgroup receiving zero sessions (n=12). These results are detailed in Table 4 and Supplementary material Figure S1. (Lines 299-304).

The PSM comparison between subgroups receiving two laser sessions and one session revealed no statistically significant difference in pain reduction (*p* = 0.306) over the three-day follow-up period (Supplementary Table S2).  (Lines 333-336).

 

Those who received only one session did not differ significantly from controls, suggesting either a cumulative therapeutic effect associated with two PBM sessions or, as evidenced in the systematic review by Kurnaz et al [18], an influence of the timing of laser introduction on the clinical response to pain. Regarding the results of our study, unfortunately, due to the variable sample sizes between groups and across assessment days, it was not possible to determine whether there was a difference in the evolution of participants who received laser treatment only on the first day compared to those treated only on the second day (Lines 457-464).

 

 

1. Analgesic interference: Since both groups received standard pharmacological care, was a sensitivity analysis conducted to ensure that the use of conventional analgesics did not mask or confound the isolated effect of the PBM?

 

Dear reviewer, thank you for your comment and the opportunity to discuss this topic.

 

We agree that the use of standardized analgesia in the hospital may have influenced the evolution of pain, however, all patients, regardless of the use of Laser, received the same analgesic protocol, standardized in the hospital (Lines 122-125). Both groups received the standard institutional pharmacological analgesic regimen, administered at scheduled times and at the same doses according to the hospital care protocol, ensuring uniform pain management across groups. Therefore, conventional analgesic use was not considered a relevant confounding factor, since exposure to medications was equivalent, and pain scale analyses were performed using both intergroup and intragroup comparisons over time.

 

This aspect is also described as a limitation of our analysis (Lines 601-607).

 

Unfortunately, we not to carry out an analysis about other covariates such as age, type and number of previous labor, delivery type, parity, analgesic medications used, or laser use on the first or second day. Future post-hoc analyses of these factors may clarify their influence on treatment response and provide new directions for research. We chose not to include all these adjustments in the models presented in this manuscript in order to parsimoniously preserve caution in the presentation of results. First, increasing the number of covariates in the statistical model elevates the risk of Type I errors (false positives) due to multiple testing, particularly when these analyses were not pre-specified. Second, given the observational nature of our study and the fact that our sample size was not predetermined but rather reflected the real-world characteristics of the service and patient flow during the study period, we must exercise caution when expanding the number of covariates in our models. With a relatively small sample size, the inclusion of numerous covariates would lead to model overfitting, resulting in unstable estimates and inflated standard errors. This could produce spurious associations that are not reproducible. In observational studies with limited samples, each additional covariate reduces the degrees of freedom and increases the risk of Type I errors, particularly when these analyses were not pre-specified in a protocol. Our sample size was determined by the natural patient flow (convenience sample), meaning it provides adequate power only for the primary analyses and the limited set of covariates originally planned. Expanding the model post-hoc to include multiple new covariates would violate key statistical assumptions and could generate misleading conclusions. Therefore, to maintain the statistical rigor and interpretability of our study, we have opted to retain the focus on the covariates defined in our original protocol. (Lines 488-510).

 

1. Internal consistency: There is a discrepancy between the Abstract (which claims effectiveness) and the Results section (which shows non-significant p-values for healing). Please align the entire manuscript to reflect the statistical reality of the data.

 

Dear reviewer, thank you for this valuable comment. Following your suggestion, we modify this statement  throughout the manuscript (main text and abstract).

 

1. Originality vs clinical context: While PBM is documented for pain, its application in Grade 3 lacerations in a real-world setting is valuable. Please expand the Discussion to compare your findings with existing literature, specifically for high-grade tears

 

The management of perineal pain and the promotion of wound healing following episiotomies and lacerations are critical components of postpartum care, with evidence supporting a combination of pharmacological and non-pharmacological strategies. Non-pharmacological approaches such as cryotherapy and transcutaneous electrical nerve stimulation (TENS) have been identified as promising conservative therapies for mana-ging early postpartum pain, offering alternatives with limited adverse effects [11]. Emerging evidence also supports the use of specific technologies. Ultimately, optimal outcomes depend on accurate diagnosis, appropriate surgical repair techniques, and the integration of these evidence-based interventions into a comprehensive, individualized care plan. A recent meta-analysis by Kurnaz et al. [18] highlights that interventions performed within the first 24 h after episiotomy did not reduce pain. However, the effects of the interventions were observed on the second day, with cold application identified as the most effective method. Additionally, interventions did not affect healing during the first three days, but a more pronounced improvement was noted in the intervention group by the fifth day. Healing began around the 7th-10th days even without inter-vention.  The REEDA (redness, edema, ecchymosis, discharge, and approximation) score decreased most significantly in the patients that received perineal education (diet, Kegel exercise, infection symptoms, and perineal hygiene).  Infrared therapy has been shown in phase II studies to significantly improve episiotomy wound healing and reduce pain compared to standard care alone. Constant et al. [19] compared photobiomodulation and cryotherapy in the im-mediate postpartum period among women with grade I and II lacerations and/or episiotomy, observing superiority of PBM in pain reduction and improved healing after 24 hours. These results suggest that laser therapy can have similar or even superior results to other non-pharmacological therapies, and that dosimetry and timing of application are key determining factors in obtaining the best results. Nonetheless, data in obstetrics remain limited. The heterogeneity of protocols (including dose parameters, timing, and treatment frequency) and the lack of controlled studies comparing laser therapy with other conventional techniques have been identified as possible explanations for these inconclusive results, highlighting the need for standardization in future studies to ge-nerate higher-quality evidence and support the development of informed clinical gui-delines. This study is among the first to evaluate PBM in a real-world clinical setting based on an already implemented care protocol, providing statistical evidence of its therapeutic benefit in an obstetric population with a substantial sample size of 183 women. Furthermore, statistical analysis revealed positive effects on pain and healing over three days, for group of PBM treated patients (Lines 367-401).

 

The literature has well-documented the effects of PBM in various other clinical contexts. The therapeutic effects of 808 nm PBM on perineal pain and tissue repair are mediated by complex biomodulatory pathways. At this infra-red wavelength, photons penetrate deeply into the tissue and are primarily absorbed by cytochrome c oxidase in the mitochondrial respiratory chain. This interaction enhances ATP synthesis and leads to the photodissociation of nitric oxide (NO), which promotes local vasodilation and improves oxygen delivery to the damaged tissue. For analgesia, 808 nm PBM modulates the inflammatory response by downregulating pro-inflammatory cytokines (e.g., TNF-α and IL-6) and increasing anti-inflammatory mediators like IL-10, effectively reducing perineal edema and pressure-induced nociception. Regarding tissue regeneration, the therapy stimulates fibroblast proliferation and activates the TGF-β signaling pathway, which is essential for collagen deposition and extracellular matrix remodeling. These combined mechanisms explain the significant reduction in NPS scores and the improved healing rates observed, particularly in patients receiving multiple sessions. Previous studies such as those by Chougala & Mahishale [23] and Constant et al. [19] corroborate the findings of this study, reinforcing the efficacy of PBM for pain and perineal healing in the obstetric context  (Lines 431-447).

 

Author Response File: Author Response.pdf

Round 2

Reviewer 4 Report

Comments and Suggestions for Authors

Thank you for your revision

Reviewer 5 Report

Comments and Suggestions for Authors

Dear Editor,
I have reviewed the authors' responses and the revised manuscript. The authors have meticulously addressed the concerns raised during the review process, particularly regarding the methodological challenges of their observational design and the statistical interpretation of their findings.

Key improvements that satisfy the revision requirements. In my view, the revisions have significantly strengthened the manuscript’s internal consistency and clinical relevance. I recommend the manuscript for publication.

Best regards,