benchmarking

…The guidelines outlined herein serve as a practical reference for conducting systematic, evidence-based measurement of animal welfare impacts, facilitating decision-making and benchmarking across sectors. This core documentation should be cited…

ual error pattern (with the lowest AIC preferred among the possible covariance matrices). The denominator degrees of freedom in the covariance pattern repeated measures models were calculated using the Satterthwaite approximation. Model suitability was as- sessed by examination of residual plots, preliminary data exploration for normality or non- normality, tests for normality and results from homogeneity of variance tests. Data were not transformed with the exception of Creatine Kinase which was log-transformed. Categorical urinalysis data were compared (proportions of animals/group in each category) using Fisher’s Exact Test across sex. Urinalysis data reported on a numerical scale (pH and specific gravity) were compared by Kruskal–Wallis test with exact p-Values estimated using Monte Carlo estimation. Animals 2021, 11, 869 7 of 17 2.3. Study 2—Safety and Efficacy of Tri-Solfen® Administration to Calf Disbudding Wounds under Field Conditions Animal experiments for Study 2 were completed between November and December 2019 on commercial dairy farms in Gloucester, New South Wales, Australia. Seventy-four (74) young female Holstein-Friesian and Jersey calves 2–6 weeks of age were selected from two similar dairy herds located within 20 km of each other. Calves were identified as suit- able for selection based on confirmed overall good health and presence of normal, healthy horn buds. Animals were weighed, stratified by weight and herd of origin and randomly allocated into two treatment groups—placebo (n = 36) and Tri-Solfen® (n = 38)—using “draw from a hat” methodology. Prior to disbudding and treatment, all calves were exam- ined for safety assessments and sham-treated. Sham treatment involved restraining and handling as per disbudding but without

4.04.0 Build 735; R Core Team, 2024). The experimental unit for performance, physiological, and inflammatory outcomes was the individual piglet, while histopathological data were analyzed at the biopsy level (n = 5 piglets per treatment per timepoint). PGE2 and haptoglobin concentrations were analyzed using generalized linear mixed models with a Gamma distribution and log link. Fixed effects included treatment, timepoint (day), and their interaction. Random intercepts accounted for piglets nested within litters. Baseline concentrations of the biomarkers were included as covariates, along with piglet weight at enrollment, baseline infrared body surface temperature (IRTd0), sow parity, and litter characteristics (born alive, stillbirths, and mummified piglets). Baseline haptoglobin levels were a strong positive predictor of post-treatment concentrations (P < 0.001), supporting their inclusion in the model as a covariate. Infrared thermography and piglet body weight data were analyzed using linear mixed-effects models with the same fixed and random effect’s structure. Litter was modeled as a random intercept. Pre-weaning mortality was compared across treatments using Chi-square or Fisher’s exact tests, as appropriate. Histopathological scores were analyzed using non-parametric Kruskal–Wallis tests followed by Dunn’s post hoc comparisons. A linear regression model was also fitted to the total histological score to explore associations with treatment, day, treatment x day interaction, and covariates. Least-squares means Page 7/20 (LSMeans) were estimated, and pairwise comparisons were adjusted for multiple testing using Holm’s method. Statistical significance was set at α = 0.05. RESULTS Piglet Performance Sow reproductive parameters or piglet pre-weaning

4.04.0 Build 735; R Core Team, 2024). The experimental unit for performance, physiological, and inflammatory outcomes was the individual piglet, while histopathological data were analyzed at the biopsy level (n = 5 piglets per treatment per timepoint). PGE2 and haptoglobin concentrations were analyzed using generalized linear mixed models with a Gamma distribution and log link. Fixed effects included treatment, timepoint (day), and their interaction. Random intercepts accounted for piglets nested within litters. Baseline concentrations of the biomarkers were included as covariates, along with piglet weight at enrollment, baseline infrared body surface temperature (IRTd0), sow parity, and litter characteristics (born alive, stillbirths, and mummified piglets). Baseline haptoglobin levels were a strong positive predictor of post-treatment concentrations (P < 0.001), supporting their inclusion in the model as a covariate. Infrared thermography and piglet body weight data were analyzed using linear mixed-effects models with the same fixed and random effect’s structure. Litter was modeled as a random intercept. Pre-weaning mortality was compared across treatments using Chi-square or Fisher’s exact tests, as appropriate. Histopathological scores were analyzed using non-parametric Kruskal–Wallis tests followed by Dunn’s post hoc comparisons. A linear regression model was also fitted to the total histological score to explore associations with treatment, day, treatment x day interaction, and covariates. Least-squares means Page 7/20 (LSMeans) were estimated, and pairwise comparisons were adjusted for multiple testing using Holm’s method. Statistical significance was set at α = 0.05. RESULTS Piglet Performance Sow reproductive parameters or piglet pre-weaning

tor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

tor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

statistically equivalent to or greater than the algorithm with all pain-altered behaviors was understood as the best refinement. Scientific Reports | (2025) 15:7161 | https://doi.org/10.1038/s41598-025-91551-6 8 www.nature.com/scientificreports/ Short pain scale The pain-altered behaviors contained in the algorithm with the best refinement were added together to make the Short UPAPS total sum, with each pain-altered behavior computed value ‘1’ in the total sum regardless of its classification on the original scale. An optimal cut-off point was established for Short UPAPS based on the high 95% confidence interval of the YI, as described in the previous session. Finally, the AUC of the Short UPAPS was compared with the AUC of the UPAPS using the DeLong test. These steps were done with the testing base. Data availability The R script (Appendix file S1) and databases (Appendix file S2, S3, and S4) used for the current study are available in supplementary information files without restrictions. The R code was commented using # specifying each step following the heading sequence described at the statistical description section. In addition, the code is available in ​h​t​t​p​s:​ ​​/​/​g​i​t​h​​u​b.​ ​c​o​m​​/​P​e​t​e​r​​T​r​i​n​d​​a​d​e​/​P​r​​o​p​o​s​i​n​​g​-​a​-​s​h​​o​r​t​-v​ ​​e​r​s​i​o​n​​-​of​ ​-​t​h​​e​-​U​n​e​s​​p​-​B​o​t​​u​c​a​t​u​-​​P​i​g​-​A​c​​u​t​e​-​P​ a​​i​n​-​Sc​ ​a​l​e​.​g​i​t. Received: 15 February 2024; Accepted: 21 February 2025 References 1. Fao, F. Food and agriculture organization of the United Nations. Rome

statistically equivalent to or greater than the algorithm with all pain-altered behaviors was understood as the best refinement. Scientific Reports | (2025) 15:7161 | https://doi.org/10.1038/s41598-025-91551-6 8 www.nature.com/scientificreports/ Short pain scale The pain-altered behaviors contained in the algorithm with the best refinement were added together to make the Short UPAPS total sum, with each pain-altered behavior computed value ‘1’ in the total sum regardless of its classification on the original scale. An optimal cut-off point was established for Short UPAPS based on the high 95% confidence interval of the YI, as described in the previous session. Finally, the AUC of the Short UPAPS was compared with the AUC of the UPAPS using the DeLong test. These steps were done with the testing base. Data availability The R script (Appendix file S1) and databases (Appendix file S2, S3, and S4) used for the current study are available in supplementary information files without restrictions. The R code was commented using # specifying each step following the heading sequence described at the statistical description section. In addition, the code is available in ​h​t​t​p​s:​ ​​/​/​g​i​t​h​​u​b.​ ​c​o​m​​/​P​e​t​e​r​​T​r​i​n​d​​a​d​e​/​P​r​​o​p​o​s​i​n​​g​-​a​-​s​h​​o​r​t​-v​ ​​e​r​s​i​o​n​​-​of​ ​-​t​h​​e​-​U​n​e​s​​p​-​B​o​t​​u​c​a​t​u​-​​P​i​g​-​A​c​​u​t​e​-​P​ a​​i​n​-​Sc​ ​a​l​e​.​g​i​t. Received: 15 February 2024; Accepted: 21 February 2025 References 1. Fao, F. Food and agriculture organization of the United Nations. Rome

Foundational studies in niche fields may have low citations but be the only quantitative source for a specific parameter. - Unexpected or inconvenient results. Findings that contradict your priors should increase scrutiny, not trigger automatic exclusion. - Methodological weakness on one parameter. A paper that is weak on pain duration may still provide the best available data on cortisol response. Record what the paper **can** contribute. - Manufacturer sponsorship alone. These papers are capped at Evidence Class 3, not excluded. Tag INCLUDE with a note in the NOTES column. ## Output format For each paper, produce one row of a Markdown table with these columns: - **PAPER_ID** — leave blank, researcher assigns P-001, P-002, etc. - **TITLE** — full paper title - **FIRST_AUTHOR_YEAR** — e.g. "Ranheim 2005" - **JOURNAL** — journal name - **DOI** — if available in the abstract metadata - **AI_SCREEN** — INCLUDE / EXCLUDE / UNCERTAIN - **EXCLUSION_REASON** — if EXCLUDE, one of: wrong_species / no_quant_data / conf_abstract / not_peer_reviewed / retracted / out_of_scope. If UNCERTAIN, write the specific ambiguity. If INCLUDE, leave blank. - **EVIDENCE_TYPE** — RCT / observational / review / expert — your best guess from the abstract - **PARAMETER_TYPE** — duration / intensity / prevalence / efficacy (multiple allowed) - **SPECIES_AGE_GROUP** — e.g. "neonatal piglets", "dairy calves 4–6 weeks" - **NOTES** — cross-use flags, e.g. "Also contains tail docking data — cross-tag for future SHE_TDOCK module" ## After the screening table Produce a short summary: - **Total screened:** X - **INCLUDE:** Y - **EXCLUDE:** Z (broken down by exclusion reason) - **UNCERTAIN:** W ## Verification reminder End every output with this exact line: > ⚠ AI abstract

e pos- sible combinations between the three experience level groups. The heteroskedasticity of the linear model was tested with the Breusch Pagan test (olsrr::ols_test_breusch_pagan). Multilevel negative binomial modeling (lme4::glmer.nb) was employed to investigate the influence of timepoint and experience level (explanatory variables) on UPAPS total score PLOS ONE | https://doi.org/10.1371/journal.pone.0309684 September 4, 2024 4 / 15 PLOS ONE Less experienced observers assess piglet acute pain differently than experienced observers: A pilot study (response variable). The best combination of fixed and random effects were identified accord- ing to the lowest Bayesian information criterion (stats::BIC) using preliminary models. Best-fit preliminary model did not consider the interaction between predictive variables, in other words, the dynamic of the UPAPS total sum for each experience level group was same for all timepoints. Piglets nested within each litter were considered as random effects. Bonferroni correction was used for adjusting the multiple comparisons in the post-hoc test (lsmeans:: lsmeans and multcomp::cld). Multilevel binomial modeling (lme4::glmer) was employed to investigate the influence of experience level group (explanatory variable) on UPAPS behaviors as dummy variables (bino- mial response variable). Dummy variables were created using fastDummies::dummy_columns for each score level. For example: Posture item was transformed into four items: Posture 0, Posture 1, Posture 2 and Posture 3, and each one of them was a binary variable of ‘0’, for when the score is not given, and ‘1’, for when the score is given. Bonferroni correction was also used for adjusting the multiple comparisons in the post-hoc test (lsmeans::lsmeans and multcomp:: cld).