Enhancing Laboratory Animal Welfare through Motivational Training of Facility Personnel: A One Welfare Approach

Abstract

Measuring the effect of laboratory animal facilities employees’ motivational training on laboratory animal welfare is essential to study the impact of the human factor on experimental results. To test this, 15 volunteers were initially recruited, with 14 completing the motivational training, and its impact on animal welfare was evaluated in 174 Swiss mice, specifically primiparous pregnant females. The mice were grouped into a control group (n = 90), observed before the motivational training, and a treated group (n = 84), observed immediately after training. Animal welfare was assessed through various measures, including nesting ability, scored from 0 to 5, time to integrate into nest test, and maternal behavior (considering rates of maternal care such as licking/grooming and staying with the pups). Each volunteer’s motivational level was evaluated using the Utrecht Work Engagement Scale, PERMA (positive emotion, engagement, relationships, meaning, and accomplishment) Profiler, and The Authentic Happiness Inventory. The motivational training was based on Yale University’s course “The Science of Well-Being,” the book The How of Happiness, and “Motivation and Action.” Results indicated that the methods used to assess mouse welfare might have been influenced by environmental differences at each research center, resulting in a lack of reproducibility. The motivational training effect varied across research centers, suggesting context-dependent outcomes potentially influenced by environmental or institutional factors for nesting scores and maternal behavior tests. However, no effect of motivational training was observed on the time to integrate into nest test, although facility-related differences were significant. The questionnaires revealed improvements in personnel well-being, although the effect on happiness was attenuated in one facility, indicating possible institutional barriers to intervention efficacy.

Introduction

The use of animals in education and research is essential for biomedical progress. According to the annual statistical report on the number of animals used in scientific, medical, and veterinary research in Great Britain, 2.68 million procedures involving animals were conducted in 2023. Most of these procedures (95%) were performed on mice, fish, birds, or rats.¹ In Brazil, such data are not yet available.

Given these numbers, animal models clearly remain an important component of research. However, science currently faces a reproducibility crisis, with >60% of medical researchers and 70% of biologic researchers reporting failures in reproducing the results of studies conducted by others. Many believe this lack of reproducibility to be related to the selective presentation of information.² In biomedical research, these issues may be influenced by incomplete data reporting, particularly regarding environmental, management, and behavioral factors affecting the animals.

These deficiencies in the disclosure of animal-related information significantly impact experimental results with respect to research reproducibility. Numerous studies demonstrate that factors such as prenatal stress and maternal behavior influence offspring behavior.^3–6 For experiments replicating techniques that assess behavior, outcomes can vary depending on the stress experienced by animals and their parents in different animal facilities.⁷

The human factor can also influence results. Professionals working in laboratory animal science are affected by the protocols used in experiments and may develop compassion fatigue, as well as stress from performing euthanasia.^8,9 Those involved in euthanasia may develop psychological defense mechanisms, either by reducing empathy and respect for animals or by experiencing grief and sadness over their deaths.¹⁰ Consequently, there is increasing recognition that laboratory animal personnel may face substantial psychological burdens, such as euthanasia-related stress, compassion fatigue, and burnout, and that indicators of animal well-being are closely linked to the emotional state and overall well-being of the caretakers.^11–13 Despite the global relevance of this issue, there are currently no studies directly assessing whether professional motivation among laboratory animal personnel affects the welfare of research animals or evaluating motivational training as a targeted intervention. Improving staff motivation may enhance human–animal interactions during routine husbandry, thereby reducing stress in the animals, a factor known to affect experimental reliability and reproducibility.^2,14,15

To address this knowledge gap, we developed a study aimed at evaluating whether motivational training for laboratory animal personnel could influence behavioral indicators of animal welfare. Pregnant mice were selected as the experimental model because maternal behavior, including nesting quality and pup retrieval, is a well-established, sensitive indicator of stress, health, and welfare in laboratory rodents.^15–17 Maternal behavior is influenced by both prenatal stress and environmental factors, and subtle variations in care or management can alter nest-building performance and maternal responses, with consequences for offspring development and behavioral test outcomes.^4,6,15–17

Based on this rationale, our study aimed to (1) evaluate whether a structured motivational training program improves the motivation and psychological well-being of laboratory animal personal; (2) assess whether such training influences behavioral indicators of mouse welfare, particularly nesting performance and maternal care; and (3) determine the applicability of a motivational questionnaire as a tool to monitor staff engagement. We hypothesized that motivational training would enhance staff engagement and that improved caregiver motivation would translate into more stable and positive welfare outcomes in mice.

Materials and Methods

Ethical review.

This study was submitted to the IACUC (nos. 11280521-0, 013.21, 065/2022) of each animal facility at each of the 3 institutions participating in the project. The study began only after the project was approved by these committees. In accordance with the provisions of the Brazilian Guide for the Production, Maintenance, or Use of Animals in Teaching or Scientific Research Activities, this study protocol included the use of humane withdrawal criteria described in CONCEA (National Council for the Control of Animal Experimentation) normative resolution no. 37: provision of treatment to relieve pain, discomfort, or distress; interruption of a painful procedure; removal of the animal from the study; and humane euthanasia of the animal.¹⁰

This study was also submitted for evaluation by the Human Research Ethics Committee (no. 4.823.430) of the Faculdade de Zootecnia e Engenharia de Alimentos of Universidade de São Paulo (School of Animal Science and Food Engineering at the University of São Paulo) through the Plataforma Brasil. The certificate of approval is part of the project documentation, and the study only began after this approval. All project participants signed an informed consent form before the study began.

Experimental human.

Participants.

A total of 15 laboratory animal facility (LAF) employees were enrolled in the study, comprising 4 veterinarians, 2 managers, and 9 animal technicians. Participants were employees from 3 different research institutions, with 4 volunteers from institution A (2 women and 2 men), 7 from institution B (3 women and 4 men), and 4 from institution C (1 woman and 3 men). One participant formally withdrew prior to the initiation of the treated phase, resulting in a final sample of 14 individuals for that stage.

Eligibility criteria for participation included active and direct involvement in animal handling within the facility, and voluntary informed consent. Sex and age were not used as inclusion or exclusion criteria. Participants were excluded if they had recently experienced psychologic trauma unrelated to the workplace or expressed discomfort with study participation. Withdrawal criteria encompassed voluntary resignation from the workplace, exposure to traumatic events that could adversely affect motivation, or a personal decision to discontinue involvement at any stage of the study.

Regarding years of experience, participants from institution A had between 2 and 5 years of experience in laboratory animal care (3 individuals with 2 years and 1 with 5 years). In institution B, experience ranged from 2 to 12 years, with 6 employees having ∼2 years of experience and 1 with 12 years. Participants from institution C reported between 2 and 10 years of experience.

Volunteers were not shared across institutions; each participant belonged exclusively to a single LAF. Variations in sex distribution, age range, and years of professional experience may influence individual motivational scores and were therefore recorded for analysis.

Study phases.

The experimental design consisted of 2 sequential phases involving the same cohort of volunteers (Table 1). The control phase represented the preintervention period, during which participants performed their routine animal-handling activities before receiving any motivational training. This was followed by a 30-day intervention period, during which the motivational training program was implemented.

Table 1.

Study Timeline

Groups	Moment	Action
Control	T0	Motivational assessment of the LAF staff
		Assessment of the welfare of pregnant females (nest score)
	T1	Assessment of the welfare of pregnant females (TINT and nest score)
	T29	Assessment of the welfare of lactating females (maternal behavior, pup recovery test, and postnatal performance)
		End of control group assessments
Treatment	T30	Start of motivational training of the LAF staff
	T60	End of motivational training of the LAF staff
Treated	T61	Motivational assessment of the LAF staff
		Assessment of the welfare of pregnant females (nest score)
	T62	Assessment of the welfare of pregnant females (TINT and nest score)
	T90	Assessment of the welfare of lactating females (maternal behavior, pup recovery test, and postnatal performance)
		End of evaluations

During this 30-day period, volunteers continued performing their regular duties at their respective LAFs; however, the animals included in the study were not handled or manipulated by those volunteers during the intervention. Routine husbandry and handling activities involving other animals (those not participating in the experimental assessments) were maintained. This approach ensured that the welfare and behavioral measurements collected in the subsequent phase were not influenced by interim handling of the study animals, while avoiding interruption of normal staff workflows.

The treatment phase began immediately after completion of the training program, with the same volunteers reevaluated while performing their usual handling tasks, now potentially influenced by the motivational strategies acquired during the intervention.

Table 1 presents a timeline of the study phases, where “moment” indicates each specific assessment point. In this context, T0 corresponds to day 0, marking the beginning of the study and the baseline (preintervention) evaluation, whereas T90 corresponds to day 90, marking the end of the study and the postintervention evaluation. Intermediate labels (eg, T1) represent additional assessment points within this timeline.

Motivational training intervention.

The motivational training program was adapted from “The Science of Well-Being” and The How of Happiness, incorporating evidence-based practices aimed at enhancing well-being and workplace engagement.^11,14 The intervention comprised a series of daily and weekly activities designed to foster positive affect, strengthen interpersonal connections, and promote healthy lifestyle habits. The proposed activities included: practicing gratitude; performing acts of kindness; engaging in social interactions; performing physical exercise at least 3 times per week for a minimum of 30 minutes; practicing meditation at least 5 times per week for 5-10 minutes; optimizing sleep by achieving at least 8 hours of rest twice per week; setting personal and professional goals; cultivating positive thinking; interacting with domestic animals, such as the volunteers’ own dogs or cats or those belonging to family members; reducing time spent on social media; and promoting workplace recognition initiatives. Participants documented the completion of each activity daily, recording its duration, date, and their mood immediately afterward, rated on a scale from “bad” to “unbelievably excellent.”

Volunteers received in-person training on how to perform the proposed activities, except for the personnel from LAF C, who were trained online due to the COVID-19 pandemic. During the in-person training sessions, volunteers from the same LAF interacted with one another while learning the procedures and expectations of the program. Each participant received an activity log sheet to record which activities were performed, their frequency and the duration, as well as their mood immediately afterward. Although all activities were standardized and a minimum recommended frequency was provided, adherence varied among volunteers according to individual engagement. No participants abstained from the intervention, and all volunteers performed the activities to some degree.

Motivational assessment.

Motivational levels were assessed at 2 time points: prior to the intervention (pretraining) and after the intervention (posttraining). Three validated instruments were applied: (1) the Utrecht Work Engagement Scale (UWES), which evaluates work engagement through 3 dimensions, vigor, dedication, and absorption, using a scale from 0 (“never”) to 6 (“always”)¹⁸; (2) the positive emotion, engagement, relationships, meaning, and accomplishment (PERMA)-Profiler, which measures well-being across 5 domains, positive emotions, engagement, relationships, meaning, and accomplishment, on a scale from 0 (“never” or “not at all”) to 10 (“always” or “completely”). Each question presented in the PERMA-Profiler is equivalent to a parameter, indicated by a letter. Questions related to positive emotions are represented by the letter P. Those relating to engagement correspond to the letter E. Questions that represent the relationship are described as the letter R. Those pertinent to the meaning represent the letter M. The last letter, the letter A, concerns issues related to achievement. Some questions are related to health and are therefore given the letter H. Questions that cover negative emotions are represented by the letter N. A global assessment of the results is also carried out, which is the global PERMA¹⁹; and (3) the Authentic Happiness Inventory, which provides a global measure of subjective happiness, scored from 0 to 5.²⁰

Experimental animals.

Animals and group allocation.

A total of 174 primiparous pregnant Swiss female mice were used, with 90 in the control phase and 84 in the treated phase. Each volunteer managed 6 cages per phase, with each cage containing one breeding pair (male and female) and their litter; only females were evaluated, as the focus was maternal behavior.

Inclusion criteria were healthy, primiparous, pregnant females aged 50-60 days, in good nutritional condition, and with SPF status in LAFs A and B, or conventional status in LAF C. The SPF colonies (LAFs A and B) were determined to be free of key pathogens, including Mycoplasma pulmonis, Rodentibacter pneumotropicus, Citrobacter rodentium, Salmonella spp., mouse hepatitis virus, Sendai virus, pneumonia virus of mice, murine norovirus, mouse parvovirus, ectromelia virus, and other agents assessed. Animals from LAF C were maintained under conventional health status, without exclusion of commensal or opportunistic microorganisms.

No animals met the exclusion criteria (preexisting clinical conditions) or removal criteria (illness unrelated to welfare, interference with assessments, or humane endpoint). Randomization was stratified by age, followed by random allocation to volunteers.

The study was conducted in 3 LAFs in different Brazilian states: A and B maintained SPF colonies from the Universidade Estadual da Paraíba and Universidade Federal do Ceará, and C housed conventional animals from Anilab. A total of 48 animals were housed in LAF A (24 control, 24 treated), 78 in LAF B (42 control, 36 treated), and 48 in LAF C (24 control, 24 treated). Each site followed the protocol for 90 days.

Animals were kept in individually ventilated microisolation cages, in pairs, under controlled temperature (20-26 °C), humidity (40%-60%), and a 12-hour light/12-hour dark cycle. Bedding was changed every 10 days, and ventilation was provided by a heating, ventilation, and air conditioning system. Animals received sterilized commercial diet (Nuvilab; Quimtia, Columbo, Brazil) and filtered water ad libitum. Upon arrival in the experimental area, the animals underwent an 11-day acclimatization period to the rooms and to routine handling procedures (including gentle handling during cage change and brief daily exposure to the researcher), aiming to reduce stress before behavioral evaluations began. After pairing, animals were allowed a 24-hour adaptation period before the first assessment. No additional environmental enrichment was provided besides nesting material. Additional enrichment was intentionally excluded to avoid introducing behavioral confounders in nest and maternal assessments.

Animal handling.

Animal handling procedures were standardized across all facilities, using cupped-hand restraint to minimize stress and prevent potential injuries associated with tail lifting. Daily health monitoring assessed general condition, including physical appearance, locomotor activity, respiratory pattern, and skin and ocular health. Body weight was recorded with a calibrated scale to monitor expected weight gain according to species, age, strain, and reproductive status.

General health monitoring.

Animals underwent at least one daily welfare inspection throughout the study period to identify clinical or behavioral alterations. Observations included locomotor abnormalities, atypical vocalizations, social withdrawal, tremors, signs of respiratory distress (eg, dyspnea), and other indicators of compromised health or well-being.

Nest-building behavior.

Nest-building behavior was assessed by providing each cage with 10 g of nesting material (weighed using a calibrated scale) placed in a standardized location.²¹ The following day, nests were scored per cage according to the following scale: score 0, material remained untouched; score 1, no defined central region, with material scattered across the cage; score 2: defined central region but flat, without walls; score 3: slightly concave nest, with wall height less than half of the dome that would cover the mouse; score 4, walls reaching half the dome height; score 5, walls exceeding half the dome height.

All nest assessments were performed by the same trained evaluator, who completed standardized training before the beginning of the study. The evaluator used photographs taken from each cage to score nests, following the methodology described by Hess et al.²¹ Nests were always scored from images, never by cage-side inspection, ensuring consistent scoring conditions. Although the evaluator was blinded to which volunteer handled each cage, full blinding to treatment group was not possible due to the sequential nature of the study; however, the evaluator was unaware of the identity of the staff responsible for each cage.

Time to integrate to nest test (TINT).

The TINT was used to evaluate the incorporation of new material into the nest within a 10-minute observation period. A square piece of cotton was placed on the side of the cage opposite the existing nest, and the animals were recorded via video in the absence of the evaluator¹⁷: positive TINT, material incorporated into the nest within 10 minutes; negative TINT, material remained in its original position after 10 minutes. In addition, the latency to the first interaction with the material was recorded for each cage.

All TINT evaluations were performed by the same trained evaluator, using video recordings only. The evaluator underwent formal training for TINT scoring before study initiation and was blinded regarding which volunteer handled each cage, although not fully blinded to study phase for logistical reasons. The binary scoring criteria minimized subjectivity, and all analyses were performed retrospectively from the videos.

Maternal skills.

Females were evaluated for 20 minutes at 3 different times of day (morning, afternoon, and evening) on the following items: nursing, the female adopted the nursing position, or at least one pup was visibly nursing/with milk in its stomach, with the duration and frequency (how often this position was adopted) recorded; grooming/cleaning the pups, the female licked one or more pups, with the duration and frequency of this activity recorded; staying with the pups, the female remained in the nest in contact with the pups, with the duration and frequency of this activity recorded. For this test, animals were filmed to avoid evaluator interference with their behavior.

All maternal behavior scoring was performed retrospectively from video recordings by the same evaluator who conducted the nest and TINT analyses. The evaluator completed a standardized training session prior to the experiment, following predefined ethograms and scoring criteria. As with other behavioral tests, the evaluator was blinded to which volunteer was responsible for each cage, although complete blinding to intervention phase was not possible. The use of a single trained evaluator ensured consistency and eliminated interobserver variability; therefore, interobserver reliability analysis was not applicable, and this is acknowledged as a methodological consideration.

Statistical methods.

The effects of motivational training on employees and its potential impact on maternal behavior in mice were analyzed separately. For motivational outcomes, a longitudinal design was used, with pre- and posttraining measures analyzed by linear mixed model (LMM) or cumulative link mixed models, depending on whether the response was treated as continuous or ordinal. Training and animal facilities (LAFs) were included as fixed effects, with participants as a random effect. For maternal behavior, repeated measures from the same females were analyzed using mixed-effects models, and when assumptions were not met, nonparametric tests for repeated-measures factorial designs were applied. All analyses adopted a 5% significance level and were conducted in R.

Results

Motivational assessments of laboratory animal technicians.

The motivational measures assessed across different questionnaires exhibited strong intercorrelations, indicating redundancy of information. This pattern was particularly evident among the PERMA domains, which displayed very high correlations with the global score (0.80 ≤ r ≤ 0.97), except for the negative emotions’ domain (PERMA N), which showed a moderate inverse correlation (r = −0.61).

Therefore, the PERMA global score condenses most of the variability contained in the individual domains, meaning that analyzing highly correlated variables separately would only reproduce the inferences obtained from the global index. Moreover, relevant correlations were observed between different instruments, such as PERMA health, Authentic Happiness, and the UWES subscales, further reinforcing the overlap among different motivational assessment approaches.

PERMA profile.

Global PERMA.

Distributions remained symmetrical but shifted consistently toward higher values in the trained group (Figure 1A). Among the 16 participants, only 1 exhibited a reduction in global PERMA after training. The dispersion of scores remained largely unchanged, suggesting that training mainly influenced central tendency rather than variability.

Figure 1.

Distribution and Inferential Analysis of Motivational and Well-Being Assessments Among Laboratory Animal Facility Staff. Global PERMA scores: (A) density plots and conditional boxplots before (control) and after (trained) motivational training, (B) linear mixed model (LMM) results for training and facility effects, (C) interaction testing between training and facility, (D) estimated marginal effects across facilities, and (E) model diagnostics. PERMA health: (F) distribution and conditional boxplots by training and facility, (G) LMM results for training and facility effects, (H) interaction testing, (I) estimated marginal effects, and (J) model diagnostics. PERMA N (negative emotions): (K) score distributions by training and facility, (L) LMM results indicating no significant training or facility effects, (M) interaction testing, and (N) model diagnostics. Authentic Happiness Inventory: (O) score distributions by training and facility, (P) LMM results showing a significant training effect and a training-by-facility interaction involving facility C, (Q) estimated marginal effects across facilities, and (R) model diagnostics. Utrecht Work Engagement Scale (UWES): (S) vigor scores with distribution and conditional boxplots and (T) corresponding LMM results showing no significant effects; (U) dedication scores with distribution by training and facility and (V) LMM results indicating a significant training effect.

This pattern is further illustrated in the boxplots (Figure 1A). Both median and interquartile range increased after training, and most individual trajectories showed an upward shift. When analyzed by facility, participants from LAF A generally presented higher global PERMA scores compared with those from facilities B and C, which showed intermediate distributions (Figure 1A). However, these differences did not substantially affect the overall shape or symmetry of the distributions.

Inferential analysis using an LMM confirmed the effect of motivational training. The estimated increase in global PERMA was 0.9181 (Figure 1B; Wald test: estimate = 0.9181, SE = 0.2222, df = 12.28, t = 4.13, P = 0.0013). In contrast, baseline differences among facilities did not reach statistical significance, although the contrast between LAF A and LAF B showed a statistical trend (P = 0.0769). Importantly, these facility effects were properly controlled within the model and did not confound the training effect.

The possible interaction between training and facility was tested through a likelihood ratio comparison of models with and without the interaction term. No evidence of interaction was found (χ² = 0.044, df = 2, P = 0.978), confirming that the positive effect of training on global PERMA scores was consistent across all facilities (Figure 1C). The inferences described are illustrated and represent the estimated biologic effect of training on the global average PERMA (Figure 1D).

Model validation further supported the reliability of these inferences. Residuals showed no violations of model assumptions, confirming good fit and supporting the robustness of the conclusions (Figure 1E).

PERMA health.

The distribution of PERMA health scores appeared approximately symmetrical in both control and trained groups, and this characteristic was preserved across facilities (Figure 1F). The dispersion of the distribution also remained stable, indicating that motivational training did not substantially alter variability. Nevertheless, the central tendency of the data suggests an increase in PERMA health scores after training.

This tendency is further illustrated in the boxplots (Figure 1F), with higher medians observed in the trained group. Individual trajectories (dashed lines) demonstrate that most participants showed increased scores following training. Across facilities, values tended to be lower in LAFs B and C compared with LAF A, although considerable overlap was present in the score distributions (Figure 1F).

Inferential analysis with an LMM confirmed the positive effect of training, estimating an average increase of 0.8476 in PERMA health scores after training (Figure 1G; Wald test: estimate = 0.8476, SE = 0.2263, df = 11.7631, t = 3.7446, P = 0.0029). While comparisons among facilities did not reach statistical significance (P > 0.20), the model adequately accounted for baseline differences. Importantly, no interaction between training and facility was detected (Figure 1H; likelihood ratio test: χ² = 2.628, df = 2, P = 0.26), indicating that the training effect was consistent across all facilities (Figure 1I).

Model validation (Figure 1J) confirmed that residuals followed expected theoretical distributions, with no evidence of nonnormality, dispersion bias, or outlier influence. Residuals were evenly spread against model predictions, supporting the adequacy of model fit.

PERMA N.

The distribution of PERMA N scores showed no clear trend associated with motivational training (Figure 1K). The density plots indicated similar concentration ranges between control and trained groups. Boxplots further confirmed the lack of consistent effect: the number of participants with increased scores after training was balanced by those with decreased scores. When analyzed across facilities, distributions remained largely comparable, with overlapping ranges of variation (Figure 1K).

Inferential analysis using an LMM confirmed these observations (Figure 1L). The estimated effect of training on PERMA N was not statistically significant (estimate = −0.7595, SE = 0.5451, df = 13.1507, t = −1.3932, P = 0.1867). Likewise, no significant differences were detected between facilities (P > 0.32).

The possibility of an interaction between training and facility was also rejected (Figure 1M). The likelihood ratio test indicated no significant interaction (χ² = 0.811, df = 2, P = 0.667), demonstrating that variation in PERMA N scores could not be explained by the combined effects of these factors.

Finally, model validation confirmed the reliability of these null findings (Figure 1N). Residual diagnostics showed compatibility with theoretical expectations, no evidence of bias in variance or dispersion, and uniform distribution of residuals across predicted values. Taken together, these results reinforce that variation in PERMA N scores was largely random and insensitive to motivational training or facility differences.

The Authentic Happiness Inventory.

The distribution of Authentic Happiness scores shifted markedly after motivational training (Figure 1O). Compared with the control condition, the trained group showed a higher concentration of responses at the upper end of the scale, with reduced dispersion. Boxplots (Figure 1O) illustrate this upward shift, with medians increased in the trained group and most individual trajectories (dashed lines) pointing to improvement. Across facilities, distributions showed considerable overlap (Figure 1O).

Inferential analysis confirmed the significant impact of training (Figure 1P). The estimated effect corresponded to an average increase of 1.0313 points on the Authentic Happiness scale (SE = 0.2560, df = 9.99754, t = 4.0279, P = 0.0024). Facility differences independent of training were not significant (P > 0.54). However, an interaction between training and facility was detected: in LAF B, the training effect was not significantly different from that in LAF A, while in LAF C, the effect was attenuated and reached statistical significance in the opposite direction (estimate = −0.9271, SE = 0.3621, df = 9.9754, t = −2.5605, P = 0.0284). These findings indicate that the positive effect of training on happiness was consistent in facilities A and B but diminished in facility C (Figure 1Q).

Model validation supported the reliability of these inferences (Figure 1R). Residual diagnostics showed compatibility with theoretical expectations, no evidence of bias, and uniform distribution across predicted values, confirming the adequacy of the model fit.

UWES.

The UWES was applied to assess vigor, dedication, and absorption. The analyses focused on vigor and dedication (Figure 1S-V).

For vigor, the distribution of scores was mostly symmetrical and only marginally affected by motivational training (Figure 1S and U). Boxplots indicate a slight upward shift in scores after training, although dispersion remained similar. Across facilities, considerable overlap was observed, suggesting that facility effects were not systematic.

The LMM did not confirm a statistically significant effect of training on vigor (estimate = 1.1524, SE = 0.7229, Z = 1.5942, P = 0.1109), and no facility-related differences reached significance (Figure 1T).

For dedication, the application of motivational training was associated with an increased concentration of scores in the higher range of the scale. Boxplots illustrate this tendency, with most individual trajectories indicating higher posttraining values. Facility differences were less evident, with distributions showing partial overlap. Inferential analysis confirmed a significant training effect on dedication (Figure 1V; estimate = 2.2184, SE = 0.9719, Z = 2.2825, P = 0.0225). This result indicates that motivational training increased the likelihood of higher dedication scores across employees. No significant facility effect was detected (P > 0.23).

Finally, predictive probabilities illustrate the consistency of higher scores after training, regardless of facility, and highlight the predominance of improved dedication across the sample.

Mouse welfare assessments.

In the assessment of mouse welfare, overall patterns indicated limited effects of motivational training on the measured indicators of mouse welfare, with results being primarily influenced by facility-level differences rather than the intervention itself.

Nest score.

Nests were evaluated 24 hours after the provision of nesting material, receiving scores from 0 (lowest quality) to 5 (highest quality). After motivational training, there was a reduction in the proportion of nests with intermediate scores (3-4) and an increase in those classified with the maximum score (5) (Figure 2A). Marked variation was also observed between facilities (Figure 2A): in facility A, more than half of the nests reached score 5, whereas in facilities B and C, nests were predominantly classified with lower values, especially score 2.

Figure 2.

Distribution and Inferential Analysis of Laboratory Mouse Welfare Assessments. Nest score: (A) distribution of nest quality scores (0-5) before (control) and after (trained) motivational training, stratified by facility; (B) cumulative link mixed model (CLMM) results showing a positive training effect in facility A and negative effects in facilities B and C, in addition to baseline facility differences; (C) predicted probabilities of nest scores across facilities and training conditions, highlighting increased likelihood of maximum-quality nests (score 5) in facility A and predominance of lower scores in facilities B and C after training; and (D) model comparison for training and facility effects on nest score. Nest integration test (TINT): (E) distribution of positive (cotton moved) and negative (cotton not moved) responses across training conditions and facilities; (F) logistic regression (Wald test) indicating no significant training effect and higher odds of positive TINT outcomes in facility C compared with facility A; (G) predicted probabilities of positive TINT by facility and training condition; and (H) model diagnostics confirming adequacy of fit. Maternal behavior (nursing time): (I) distribution of nursing duration by training condition, facility, and time of day; (J) nonparametric repeated-measures ANOVA showing significant effects of training, facility, time of day, and their interactions; and (K) predicted marginal means of nursing time across facilities and training conditions, stratified by circadian period, illustrating longer nursing durations in trained animals, particularly during evening assessments.

Inferential analysis revealed that facility effects were already present at baseline, with lower odds of higher nest scores in facilities B (estimate = −1.67272, P < 0.001) and C (estimate = −1.68152, P = 0.00159) compared with facility A (Figure 2B). Training significantly increased nest quality in facility A (estimate = 1.29728, P = 0.02717), but this effect was reversed in facilities B and C, where the odds of achieving higher scores decreased (estimates = −2.25078 and −2.16243, P = 0.00230 and P = 0.00483, respectively).

The probability curves derived from the model (Figure 2C) confirmed these effects, showing increased probability of score 5 in facility A after training, while in facilities B and C the predicted distribution shifted toward lower scores, particularly score 2.

Finally, the inclusion of volunteer as a blocking factor did not improve the model fit (likelihood ratio test: likelihood ratio statistic = 6.354, df = 12, P = 0.897), indicating that variability among employees did not account for additional differences in nest quality (Figure 2D).

TINT.

The TINT was used as a binary measure of maternal behavior, considered positive when females moved the cotton and negative when it remained in place. Across facilities, the distribution of positive and negative outcomes varied markedly, whereas training did not visually alter these proportions (Figure 2E). Facility A showed the lowest proportion of positive TINT responses, facility B intermediate levels, and facility C the highest.

Statistical inference corroborated these observations. Training had no significant effect on TINT outcome (estimate = 0.18175, SE = 0.31075, Z = 0.58488, P = 0.55863; Figure 2F). In contrast, facility differences were relevant: the probability of a positive TINT was significantly higher in facility C compared with facility A (estimate = 0.95740, SE = 0.42576, Z = 2.24869, P = 0.02453), while the contrast between facilities A and B remained only a statistical trend (estimate = 0.64254, SE = 0.38613, Z = 1.66404, P = 0.09610).

The predicted probabilities derived from the model (Figure 2G) illustrate these effects, highlighting that in facility C the likelihood of a positive outcome surpassed 50%, while in facility A it remained closer to 30%. Within each facility, however, training did not change these probabilities substantially.

Model validation showed no deviations from model assumptions (Figure 2H), supporting the robustness of the null training effect and the facility differences.

Maternal behavior.

Nursing time was evaluated in 20-minute sessions, with each female monitored in the morning, afternoon, and evening to capture circadian variability. Most sessions displayed a binary pattern (no nursing compared with full-interval nursing). Motivational training was associated with fewer sessions without nursing and more sessions with full nursing (Figure 2I). When average nursing duration per female was considered, modest increases in central tendency and greater dispersion were observed in the trained group compared with baseline. Facility effects were also evident, with distinct distributions but high internal homogeneity (Figure 2I).

Regarding circadian variation, evening sessions showed consistently longer durations, while morning sessions displayed shorter nursing times across facilities (Figure 2I). A repeated-measures nonparametric ANOVA confirmed significant effects of training, facility, and time of day, as well as most interactions (Figure 2J). Predicted marginal means illustrated these trends (Figure 2K). In facility A, training effects overlapped with circadian variation, with shorter times in the morning. In facility B, training reversed the baseline decline across the day, producing progressively longer nursing durations. In facility C, however, nursing time remained shorter and largely unaffected by training or time of day.

Discussion

This study explored the influence of motivational training of LAF employees on both personnel well-being and the welfare of laboratory animals. The results demonstrated a clear improvement in several aspects of employee motivation and subjective well-being, particularly in the PERMA global score, PERMA health, and the Authentic Happiness Inventory, after a 30-day evidence-based intervention. However, the translation of these motivational changes to measurable improvements in animal welfare, as assessed through nesting, TINT, and maternal behavior, appeared to be context-dependent and strongly influenced by the specific characteristics of each research center. These facility-dependent effects suggest that structural, environmental, and organizational differences between LAFs played a substantial role in shaping the outcomes.

The consistent increase in PERMA global and health scores following the intervention supports the hypothesis that structured motivational programs can enhance employees’ psychologic well-being and engagement at work. These findings align with studies showing that practices such as gratitude, physical activity, meditation, and social connection, core components of our intervention, positively affect happiness, resilience, and job satisfaction.^18,22,23 The improvement in health-related well-being suggests a potential buffering effect against occupational stress, which is especially relevant in laboratory animal science, where professionals are routinely exposed to emotionally demanding tasks.^8,9

Interestingly, the PERMA N domain did not change significantly, suggesting that while positive aspects of well-being can be strengthened through targeted practices, reducing negative affect may require longer or more individualized interventions. This finding is consistent with prior research indicating that reductions in negative affect often lag behind increases in positive emotions during well-being interventions.^19,24 Furthermore, the heterogeneity observed across facilities may reflect differences in institutional culture, workload, management style, or team cohesion, factors known to modulate the effectiveness of motivational interventions.²⁵ Participants reported that interpersonal dynamics, leadership style, and daily operational pressure differed markedly between LAFs, which may explain why improvements in subjective well-being did not translate uniformly into animal-related outcomes.

The Authentic Happiness Inventory corroborated the PERMA results, with a significant increase in self-reported happiness posttraining. However, the attenuation of this effect in one of the facilities (LAF C) suggests that environmental or organizational conditions can limit the generalizability of motivational outcomes. This variation emphasizes that training alone cannot compensate for structural or cultural deficiencies within an institution. This is because high job demands and internal conflicts can negatively affect the employees’ well-being.²⁶

Although the motivational training improved personnel well-being, its measurable effect on animal welfare was modest and variable. The observed dependence of results on the research center indicates that environmental and management differences between facilities, such as cage type, noise, lighting, or routine procedures, may have outweighed subtle behavioral changes induced by improved human motivation. This aligns with prior evidence that reproducibility in behavioral studies is strongly affected by local environmental factors and human–animal interactions.^6,7,15,27

While nesting scores and maternal behavior appeared to differ by facility, no consistent direct effect of motivational training was detected across all sites. This suggests that although human emotional state and handling style can influence animal responses, these effects were likely subtle in this study and may have been overshadowed by substantial environmental variability between facilities.

In addition, the assessment of animal welfare immediately after the 30-day intervention may not have allowed sufficient time for the behavioral changes of personnel to consolidate and reflect in the animals’ daily management. Future studies might benefit from longer follow-up periods and from assessing physiologic markers of stress, such as cortisol levels or heart rate variability, which can capture subtler human–animal interaction effects.

Another possible explanation for the limited transfer of human motivation to animal outcomes is the ‘ceiling effect’ of well-managed facilities. Because all participating LAFs already followed high welfare standards, the margin for measurable improvement may have been narrow. In such settings, motivational enhancement might contribute more to sustaining consistent welfare practices and reducing burnout than to producing measurable gains in animal behavior per se.

The strong intercorrelations among the motivational instruments used (PERMA, UWES, and Authentic Happiness) indicate that they capture overlapping dimensions of well-being and engagement. While this redundancy confirms internal consistency, it also highlights the importance of selecting a concise set of complementary measures in future research. Moreover, because facility effects were often stronger than training effects in animal welfare assessments, multicenter studies should employ harmonized environmental monitoring and statistical approaches capable of partitioning variance attributable to site-level factors.

It is also relevant to note that the intervention was based on self-reported adherence to activities, which may have introduced variability in engagement levels among participants. Feedback collected informally from participants indicated that activities such as gratitude practice, meditation, and physical exercise were perceived as the most beneficial. Several participants reported improved mood regulation, better sleep, and enhanced interpersonal relationships at work, which may have contributed to the observed improvements in well-being. Future studies could benefit from objective measures of participation, periodic reinforcement sessions, and integration of behavioral feedback mechanisms to enhance compliance.

This study provides novel evidence that motivational training based on positive psychology can be feasibly implemented in laboratory animal facilities and can significantly enhance employee well-being. Although direct improvements in animal welfare were not consistently observed, promoting human well-being is intrinsically valuable and may contribute to a healthier, more empathetic, and ethically aware research environment. The findings support the growing recognition that the ‘human factor’ is a critical, yet often overlooked, component of reproducibility and animal welfare in biomedical research.^8,18

By investing in personnel motivation, institutions may indirectly strengthen animal welfare outcomes through improved consistency of care, reduced human-induced stress, and enhanced team communication. From a practical standpoint, facilities aiming to implement similar programs should (1) provide structured, weekly motivational activities; (2) offer short, guided sessions on gratitude, meditation, and physical activity; (3) integrate group discussions to reinforce team cohesion; and (4) designate a staff member or, possibly, psychologist to monitor adherence and provide support. Integrating motivational strategies into continuing education programs and occupational health policies represents an important step toward a more holistic approach to laboratory animal science, one that considers the welfare of both animals and humans as interdependent elements of research integrity.

Conclusions

Motivational training improved the well-being and engagement of LAF personnel, confirming the potential of structured positive psychology interventions in the research environment. However, translating these human-level improvements into measurable animal welfare outcomes requires longer-term studies with stricter environmental standardization and larger sample sizes. Importantly, these findings offer practical implications for laboratory settings. Facilities implementing motivational training should ensure management involvement, provide structured guidance for the activities, and adopt simple reinforcement strategies, such as periodic check-ins, peer-support practices, or integration into occupational health programs, to promote adherence and long-term impact. Even if immediate changes in animal behavior are not observed, enhanced staff well-being can improve consistency of care, reduce handling-related stress, and strengthen human–animal interactions. Future research should explore how sustained motivational interventions, combined with organizational support and management engagement, can promote a culture of well-being that benefits both humans and animals in laboratory settings.

Conflict of Interest

Some of the authors serve or have served as coordinators or collaborators of the animal facilities involved in this study. However, none received financial or personal benefits related to this research, and their institutional roles did not influence the study design, data collection, analysis, or interpretation. The authors have no conflicts of interest to declare.

Funding

This study was supported by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior–Brasil (CAPES), Finance Code 001. Additional support was provided by the Faculdade de Zootecnia e Engenharia de Alimentos–Universidade de São Paulo, Universidade Estadual da Paraíba, Universidade Federal do Ceará, Biotec Controle Ambiental, and Ourofino Saúde Animal.

Author Contributions

B.F.R., conceptualization, methodology, investigation, writing—original draft; L.M.G.d.M.B., investigation; T.Z.d.L., statistical analysis; L.G.A.S., investigation; M.L.S.d.S., investigation representing Biotec Controle Ambiental; V.B.F.M., investigation representing Ourofino Saúde Animal; L.L.d.S.M., data organization, writing—original draft; P.R.F., data organization, writing—original draft; D.d.S.M., conceptualization; V.L.d.A.R., supervision, conceptualization.

Protocol registration

A protocol including the research question, key design features, and analysis plan was prepared prior to the study and was publicly registered on Plataforma Brasil (Ethics Committee approval no. 4.823.430). In addition, the protocol was reviewed and approved by the IACUCs of the participating institutions: CEUA-NPDM (Universidade Federal do Ceará, protocol n 11280o.521-0), CEUA-UEPB (Universidade Estadual da Paraíba, protocol no. 013.21), and CEUA of Ourofino Saúde Animal (protocol no. 065/2022). The study was initiated only after all ethical approvals had been obtained. All procedures were conducted in accordance with international guidelines for the care and use of laboratory animals, as well as the ethical standards established by the Brazilian National Council for the Control of Animal Experimentation (CONCEA).