Operational Pain Indicators in Wild Birds in South America: Expert Consensus, Field Validation and Application in Rehabilitation Centres

Abstract

Pain assessment remains a major challenge in avian medicine and wildlife rehabilitation because birds frequently exhibit subtle and species-specific responses that are difficult to detect under field conditions. The objective of this study was to identify, validate and operationalise indicators suitable for assessing pain in wild birds admitted to rescue, rehabilitation and zoological facilities. A four-phase approach was used, including a literature review, expert validation, field validation and the development of a Bird Pain Index (BPI). Scientific publications indexed in Scopus and Web of Science were reviewed to identify candidate pain indicators. Then, veterinarians, wildlife rehabilitation professionals, zoo personnel, ornithologists and animal welfare specialists evaluated the relevance, magnitude and applicability of each indicator using structured questionnaires. Forty candidate indicators were initially identified through the literature review; following expert assessment, only 25 indicators exceeded the acceptance threshold (>60%) for field validation. Nine indicators were ultimately validated as Operational Pain Indicators (OPIs) based on high field feasibility: isolation behaviour (78%), half-closed eyelids (75%), anorexia (73%), stillness (70%), loss of grooming (67%), miosis/mydriasis/anisocoria (67%), seated posture (65%), ruffled feathers (64%), and pain concealment behaviour (62%). Most validated indicators were behavioural or clinical measures that could be assessed without handling the animals, while physiological indicators generally showed lower field applicability. The validated OPIs were integrated into a preliminary BPI designed to support clinical decision-making and welfare assessment.

1. Introduction

Pain is a key issue in animal welfare [1] as many vertebrates and invertebrates have the anatomical, physiological, and behavioural capabilities to experience it [2]. The capacity to experience pain is generally considered an evolutionarily advantageous trait because it promotes the avoidance of harmful stimuli [3]. It may be defined as an aversive sensory and emotional experience which elicits protective motor actions, results in avoidance learning, and can modify species-specific behavioural traits, including social behaviour. Pain involves a conscious experience of tissue damage by specialised receptors induced by a noxious or harmful stimulus (nociception) determined, in part, by the activation of afferent nerve pathways [1,4]. Pain measurement may become very subjective, as pain expression is specific to type, species and individual. Animals cannot communicate with humans to describe pain [5]. Pain management in birds is usually based on the recognition of behavioural alterations such as abnormal postures or vocalisations [6,7]. Also, changes in physiological variables such as heart, or respiratory rate and neurohormones in the blood such as cortisol can also be used in birds [8,9,10]. There are also pain scales to assess the presence and magnitude of pain such as the visual analogue scale, numerical scale, descriptive scale or complex scales such as the Melbourne or Glasgow scale which is more commonly used in dogs [11], cats [12] and farm animals such as pigs [13,14] and horses [15]. In laboratory animals, pain is assessed using grimace/pain scales such as in mice or rabbits [16]. However, birds cannot express any grimace or reflect pain in their face as little or no muscle are developed as in mammals [10,12].

Even though wild birds inhabit multiple landscapes, their habitat has been reduced because of increased human population and urbanisation [17,18]. The latter has led to an increase human–wild animal interaction resulting in environmental occupation [19], deforestation [20,21] and zoonotic disease outbreaks such as coronavirus and influenza [22,23]. The main negative human–wild bird interactions are car accidents [24,25], electrocution [26,27,28] or domestic animal (dog/cat) attacks [29,30]. These have led to an increase in admissions to rescue and rehabilitation wildlife centres [31,32,33,34]. Pain evaluation is crucial in these cases and must be assessed to help decide veterinarians and animal carers to support clinical decision-making and deliver the optimal treatments [35,36].

From a practical point of view, pain indicators should be operational and thus easily measurable under normal veterinary practice or working conditions [37,38]. Operational welfare indicators have been identified in animal production systems such as salmon [39], poultry [40], pigs, cows, goats or horses [41], proving useful for decision-making and welfare-risk prevention [42,43,44]. Studies in pain scales in avian species are very limited such as those on psittacines [9,10]. However, to the authors’ knowledge there are no studies or consensus on validated operational indicators related to pain specifically to be used in wild birds accepted into rehabilitation centres. This study aimed to elucidate which pain indicators could be used as Operational Pain Indicators (OPIs) in wild birds admitted to rescuing or rehabilitation centres, increasing the chances of decreasing pain through efficient treatments, and consequently increasing the chances of survival and re-introduction to the wild.

2. Materials and Methods

2.1. Study Design

The study design comprised four sequential phases: (i) identification of potential pain indicators through a literature review (Scopus and Web of Science), (ii) expert and stakeholder validation, (iii) field validation in wildlife rehabilitation conditions, and (iv) development of a Bird Pain Index (BPI). The methodological approach was designed to identify indicators that were scientifically supported, practically measurable and suitable for routine use in wildlife rescue and rehabilitation centres.

2.2. Literature Review of Pain Indicators in Birds

Potential pain indicators in birds were initially identified through a literature review of the last 20 years. Bibliographic searches were conducted using scientific databases such as Scopus and Web of Science Core Collection due to their extensive coverage of peer-reviewed publications in animal welfare, zoology, wildlife medicine and veterinary sciences. The searches used combinations of the keywords “pain”, “bird”, “avian”, “indicator”, “welfare”, “behaviour”, “clinical sign” and “physiology”. Terms were used individually and in Boolean combinations (AND, OR) to maximise retrieval. Retrieved records were screened based on title and abstract relevance. Publications were excluded if they did not address avian pain, welfare assessment, clinical medicine or behavioural responses associated with pain. Eligible manuscripts underwent full-text review. During full-text assessment, behavioural, physiological and clinical variables described as potential indicators of pain were extracted and compiled into a preliminary list of candidate indicators. Additional indicators identified from technical reports, rehabilitation manuals and veterinary guidelines were also recorded when frequently referenced by professionals working with birds.

2.3. Expert Validation and Risk Analysis

The preliminary list of pain indicators obtained from the literature review was subjected to expert validation following the recommendations of the European Food Safety Authority (EFSA) for animal welfare indicator validation [45], as well as in [46]. A structured questionnaire was created using Google Forms and distributed electronically to 49 veterinarians, wildlife rehabilitation professionals, zoo personnel, avian welfare specialists, ornithologists and bird producers from Chile (n = 39, 77.6%), Argentina (n = 6, 12.2%) and Peru (n = 5, 10.2%). Experts were selected based on pre-established criteria: (i) holding a DVM or relevant post-graduate degree in wild animal health/welfare, and (ii) having ≥3 years of documented clinical experience in avian medicine, wildlife rehabilitation or zoo management. Out of 70 invited experts, 49 completed the full survey (70.0% response rate). The mean experience was 8.4 ± 4.2 years. Professional categories: wildlife/zoo veterinarians (44.9%, n = 22), rehabilitation specialists (28.6%, n = 14), avian welfare researchers (16.3%, n = 8), and ornithologists (10.2%, n = 5). The questionnaire consisted of four sections: (a) informed consent, (b) professional and institutional background information, (c) evaluation of each potential pain indicator, (d) open answer for comments and proposal of additional indicators. For each indicator, respondents evaluated: (a) perceived association with pain (risk), (b) magnitude of pain represented by the indicator, and (c) applicability under field conditions. Responses were collected using a five-point Likert scale, where 1 represented the lowest agreement (0%) and 5 represented the highest agreement (100%). Participants were additionally invited to propose indicators not identified during the literature review but routinely used by them in professional practice for pain assessment. Mean scores for risk and magnitude were calculated for each indicator and converted to percentages. A combined validation score was then obtained by averaging the dimensions. Indicators reaching the combined score of >60% were selected for field validation. The use of Likert scales and consensus-based scoring was chosen because these methods have been used in animal welfare assessment and facilitate standardisation of expert judgement while maintaining practical applicability [46]. Institutional authorisation from the National Zoo of the Metropolitan Park of Santiago was obtained before field observations and the Scientific Ethics Committee of Universidad de Las Américas (UDLA) waived the study and thus did not gave any number code. Animal observations were strictly passive and non-invasive, conducted during routine intake at the National Zoo Wildlife Rehabilitation Centre in Chile, without inducing pain, without delaying veterinary care or analgesic administration and without performing any experimental procedures or additional handling.

2.4. Field Validation of Pain Indicators

Field validation was conducted at the Wildlife Rehabilitation Centre of the National Zoo of Chile (Santiago, Chile), which receives wild animals, including birds (predominantly the Rufous-tailed hawk [Buteo ventralis], Burrowing owl [Athene cunicularia], Chimango caracara [Milvago chimango] and Austral thrush [Turdus falcklandii]), from diverse geographical regions and injury categories throughout the year. Sample size was determined based on historical admission rates of approximately 36 birds per month, using a confidence level of 95% and a confidence interval of 10%, resulting in a target sample of 26 birds. Only birds presenting clinical evidence of pain-related trauma or polytrauma were included in the study. Inclusion criteria comprised: fractures, vehicle collision injuries, haemorrhage, soft tissue trauma, orthopaedic injuries, multiple traumatic lesions. Birds admitted exclusively for relocation, rehabilitation without injury, hand-rearing or preventive care were excluded from the study. Field evaluations were conducted directly upon arrival by two trained staff of the Wildlife Rehabilitation Centre of the National Zoo of Chile (co-authors on this study). Before field evaluations, they underwent standardised joint training sessions to align criteria for scoring candidate indicators. Field observations were subsequently recorded independently upon admission. Observations were carried out after arrival to the rehabilitation centre and before any handling, restraint, sedation, medical intervention or analgesic treatment to minimise observer-induced alterations in behaviour. Each previously validated indicator was assessed directly on arrival by trained evaluators using a five-point Likert scale, where 1 = impossible to observe or measure; 2 = difficult to observe; 3 = moderately observable; 4 = easily observable; and 5 = readily observable without difficulty. Field practicability scores were converted into percentages and averaged across observations. As before, only indicators obtaining at least 60% practicability were classified as OPIs.

2.5. Development of Operational Pain Indicators

For each validated OPI, a detailed guidance was developed describing (a) measurement procedures, (b) biological interpretation, (c) expected normal or reference conditions, (d) associated risk factors, and (e) recommended corrective or therapeutic actions. These guidelines were intended to facilitate implementation under routine rehabilitation and veterinary conditions and to support standardisation among institutions.

2.6. Development of the Bird Pain Index

A preliminary BPI was developed using the validated OPIs to translate multiple behavioural and clinical signs into a standardised, quantitative composite score. Following standard methodological frameworks for animal welfare indices, the index is structured as a normalised linear additive model defined by the following equation:

$$\mathrm{BPI}={\displaystyle \frac{{\displaystyle {\sum }_{i=1}^k}{I}_i}{{\displaystyle {\sum }_{i=1}^k}{I}_{\mathrm{m}\mathrm{a}\mathrm{x}}}}\times 100={\displaystyle \frac{{\displaystyle {\sum }_{i=1}^k}\left(S_i\cdot w_i\right)}{k\cdot w_i}}\times 100$$

where

$\mathrm{BPI}=\mathrm{Bird} \mathrm{Pain} \mathrm{Index}, \mathrm{percentage} (0–100\%);$

$I_i=\mathrm{observed} \mathrm{intensity} \mathrm{score} \mathrm{for} \mathrm{the} i\text{-}\mathrm{th} \mathrm{indicator};$

$I_{\mathrm{m}\mathrm{a}\mathrm{x}}=\mathrm{maximum} \mathrm{possible} \mathrm{score} \mathrm{for} \mathrm{each} \mathrm{indicator};$

$S_i=\mathrm{standardised} \mathrm{score} \mathrm{assigned} \mathrm{to} \mathrm{indicator} i;$

$w_i=\mathrm{weighting} \mathrm{factor} \mathrm{assigned} \mathrm{to} \mathrm{indicator} i;$

$k=\mathrm{number} \mathrm{of} \mathrm{currently} \mathrm{evaluable} \mathrm{indicators} \left(k\le 9\right).$

In the preliminary baseline model, equal weighting ($w_i=1/9\approx 0.111$ or $11.1\%$ per indicator) was assigned to all validated OPIs because insufficient empirical regression data currently exists to justify differential weighting schemes. To address scenarios where specific indicators cannot be evaluated due to local trauma, bandaging, or anatomical constraints, the application utilises a dynamic rescaling algorithm: unobservable indicators are automatically excluded from both the numerator and denominator ($k$ is reduced accordingly), ensuring that the active index remains normalised to 100%. Conversely, if an indicator is fully evaluated and scored as absent ($S_i=0$), its contribution to the numerator is zero while remaining in the denominator, accurately decreasing the final pain percentage score.

To guide clinical decision-making under rehabilitation conditions, the BPI scale was categorised into four operational interpretation ranges:

• <25% (Mild Pain): Behavioural alterations are minimal; routine monitoring and localised supportive care are recommended.
• 25–50% (Moderate Pain): Noticeable behavioural changes; proactive pre-emptive analgesia (e.g., non-steroidal anti-inflammatory drugs or targeted opioids) is indicated.
• 51–75% (Severe Pain): Multiple concurrent OPIs present; intensive multimodal analgesic therapy, close clinical re-evaluation, and strict environmental modification are mandatory.
• 75% (Critical Pain/Distress): High cumulative pain severity associated with severe trauma or poor prognosis; urgent therapeutic intervention required alongside re-evaluation of recovery prospects and consideration of humane euthanasia protocols.

The index was programmed into an interactive web application to automate score calculations, handle dynamic unobservable metrics, and provide instant clinical guidance, with complete technical architecture and user instructions provided in Supplementary Material S2.

2.7. Statistical Analysis

Descriptive statistical analyses were performed using Jamovi^® statistical software (version 2.7.38.0). Likelihood scores obtained from expert validation and field assessments were transformed into percentages and expressed as means accompanied by 95% confidence intervals (95% CI) to reflect estimate precision. Consensus thresholds were established at 60% following methodologies commonly applied in animal welfare indicator validation studies [46]. Inter-rater reliability for field evaluations conducted by the two independent observers was assessed using Cohen’s kappa coefficient (κ) with no double-blinding as the nature of the emergency clinical service did not permit it, yielding substantial agreement (κ = 0.78, 95% CI [0.69, 0.87]) [47]. The analytical framework emphasised descriptive statistics, confidence intervals, and consensus-based approaches rather than complex inferential modelling, given that the primary objective was the systematic identification, reliability assessment, and operational validation of indicators suitable for practical application under wildlife rehabilitation conditions.

3. Results

3.1. Literature Review of Potential Pain Indicators in Birds

The literature review conducted in Scopus and Web of Science retrieved a total of 471 publications related to pain recognition, welfare assessment and clinical management in avian species. After title and abstract screening, 30 articles were selected for full-text review. Following the application of the inclusion and exclusion criteria and repetition between databases, the same 30 publications were retained for final assessment. From these publications, a total of 40 potential pain indicators were identified (Table 1). These indicators comprised behavioural, clinical and physiological variables described as being associated with painful conditions, trauma, disease processes or reduced welfare in birds admitted to zoological institutions, wildlife rescue centres and rehabilitation facilities. The resulting candidate indicators constituted the basis for the expert validation phase.

Table 1. Selection of indicators by stakeholders (n = 49) by risk and magnitude of each potential bird pain indicator.

Indicator	Type	Risk (%)	Magnitude (%)	Mean Risk and Magnitude (%)	95% CI	Selection
Ruffled feathers	Behavioural	96	92	94	[91.2, 96.8]	Yes
Half-closed eyelids	Behavioural	96	92	94	[91.2, 96.8]	Yes
Plasma cortisol	Physiological	84	96	90	[86.5, 93.5]	Yes
Anorexia	Behavioural	92	88	90	[86.5, 93.5]	Yes
Isolation behaviour	Behavioural	88	72	80	[75.2, 84.8]	Yes
Sleepiness behaviour	Behavioural	80	76	78	[73.0, 83.0]	Yes
Loss of grooming	Behavioural	72	80	76	[70.8, 81.2]	Yes
Clonic convulsions	Physiological	76	76	76	[70.6, 81.4]	Yes
Tonic convulsions	Physiological	76	76	76	[70.6, 81.4]	Yes
Stillness	Behavioural	72	76	74	[68.5, 79.5]	Yes
Seated behaviour	Behavioural	72	72	72	[66.3, 77.7]	Yes
Miosis/mydriasis/anisocoria	Physiological	88	56	72	[65.8, 78.2]	Yes
Panting	Physiological	64	76	70	[64.1, 75.9]	Yes
Pain concealment behaviour	Behavioural	64	76	70	[64.1, 75.9]	Yes
EEG	Physiological	72	64	68	[62.0, 74.0]	Yes
Aggressive behaviour	Behavioural	64	72	68	[62.0, 74.0]	Yes
Open beak breathing	Physiological	68	64	66	[59.9, 72.1]	Yes
ECG	Physiological	68	64	66	[59.9, 72.1]	Yes
Feather pecking	Behavioural	64	68	66	[59.9, 72.1]	Yes
Vocalisations	Behavioural	68	64	66	[59.9, 72.1]	Yes
Defensive behaviour	Behavioural	64	64	64	[57.8, 70.2]	Yes
Stereotype (movement)	Behavioural	64	64	64	[57.8, 70.2]	Yes
Ataxia	Physiological	64	60	62	[55.7, 68.3]	Yes
Mandibulation	Behavioural	64	60	62	[55.7, 68.3]	Yes
Stereotype (beak)	Behavioural	64	60	62	[55.7, 68.3]	Yes
Arterial blood pressure	Physiological	56	64	60	[53.6, 66.4]	No
Decrease activity	Behavioural	56	64	60	[53.6, 66.4]	No
Cannibalism	Behavioural	60	60	60	[53.6, 66.4]	No
Loss of mandibular tone	Physiological	60	56	58	[51.5, 64.5]	No
Loss of body posture	Behavioural	56	56	56	[49.5, 62.5]	No
Deep breathing	Behavioural	56	52	54	[47.5, 60.5]	No
Out of sight	Behavioural	56	52	54	[47.5, 60.5]	No
Panic reaction	Physiological	48	56	52	[45.5, 58.5]	No
Slow flapping of wings	Behavioural	48	56	52	[45.5, 58.5]	No
Pecking motion	Behavioural	48	56	52	[45.5, 58.5]	No
Head shaking	Behavioural	56	44	50	[43.5, 56.5]	No
Escape behaviour	Behavioural	44	44	44	[37.6, 50.4]	No
Leg movements	Behavioural	48	40	44	[37.6, 50.4]	No
Vigilance state	Behavioural	36	56	46	[39.5, 52.5]	No
Jumping	Behavioural	40	36	38	[31.7, 44.3]	No

Note: only indicators with >60% of agreement were selected for the next stage.

3.2. Expert Validation and Risk Analysis

The expert validation survey was completed by 49 professionals representing 25 institutions from three countries. Participants included veterinarians, wildlife rehabilitation specialists, zoo personnel, ornithologists, avian researchers and animal welfare experts. The results of the expert assessment are presented in Table 1. A total of 25 indicators (62.5%) exceeded the predefined validation threshold of 60% and were therefore selected for field validation. The indicators receiving the highest overall scores included ruffled feathers (94% validation), half-closed eyelids (94% validation) and plasma cortisol (90% validation), reflecting a strong consensus among participants regarding their relevance for pain assessment in birds.

3.3. Field Validation of Pain Indicators

Field validation was conducted using 26 wild birds admitted to the Wildlife Rehabilitation Centre of the National Zoo of Chile due to trauma, fractures, collisions, haemorrhages or multiple injuries. The results are summarised in Table 2. Only nine indicators (22.5% of the original list) achieved field validation scores above the 60% acceptance threshold and were consequently retained as OPIs. The indicators showing the highest field practicability scores were isolation behaviour (78%), half-closed eyelids (75%) and anorexia (73%) which were consistently easy to observe and interpret in injured birds. In contrast, indicators such as mandibulation (43%), ECG (43%) and EEC (44%) obtained lower scores due to difficulties associated with direct observation, handling requirements or environmental influences. Interestingly, only three indicators (9% of the original candidate indicators) exceeded 60% agreement in both the expert validation and field validation phases, independently. However, several indicators that did not individually surpass the threshold in one phase achieved combined mean scores above 60%, supporting their inclusion as OPIs. Several physiological indicators obtained high perceived associations with pain but lower practicality scores, whereas behavioural and clinical indicators generally achieved higher applicability ratings.

Table 2. Field and final validation of selected OPIs (%) carried out by observers (n = 2).

Indicator	Type	Stakeholder Validation (%)	Field Validation (%)	Final Validation (%)	95% CI	Selection
Isolation behaviour	Behavioural	80	75	78	[73.2, 82.8]	Yes
Half-closed eyelids	Behavioural	94	56	75	[70.1, 79.9]	Yes
Anorexia	Behavioural	90	55	73	[68.0, 78.0]	Yes
Stillness	Behavioural	74	65	70	[64.8, 75.2]	Yes
Loss of grooming	Behavioural	76	58	67	[61.6, 72.4]	Yes
Miosis/mydriasis/anisocoria	Physiological	72	62	67	[61.6, 72.4]	Yes
Seated behaviour	Behavioural	72	58	65	[59.5, 70.5]	Yes
Ruffled feathers	Behavioural	94	34	64	[58.4, 69.6]	Yes
Pain concealment behaviour	Behavioural	70	53	62	[56.3, 67.7]	Yes
Sleepiness behaviour	Behavioural	78	38	58	[52.1, 63.9]	No
Panting	Physiological	70	40	55	[49.1, 60.9]	No
Plasma cortisol	Physiological	90	20	55	[49.1, 60.9]	No
Open beak breathing	Physiological	66	39	53	[47.1, 58.9]	No
Clonic convulsions	Physiological	76	28	52	[46.1, 57.9]	No
Ataxia	Physiological	62	34	48	[42.1, 53.9]	No
Defensive behaviour	Behavioural	64	31	48	[42.1, 53.9]	No
Vocalisations	Behavioural	66	29	48	[42.1, 53.9]	No
Aggressive behaviour	Behavioural	68	26	47	[41.1, 52.9]	No
Feather pecking	Behavioural	66	23	45	[39.1, 50.9]	No
Stereotype (movement)	Behavioural	64	25	45	[39.1, 50.9]	No
EEG	Physiological	68	20	44	[38.1, 49.9]	No
Stereotype (beak)	Behavioural	62	25	44	[38.1, 49.9]	No
ECG	Physiological	66	20	43	[37.1, 48.9]	No
Mandibulation	Behavioural	62	24	43	[37.1, 48.9]	No
Tonic convulsions	Physiological	76	22	49	[43.1, 54.9]	No

3.4. Identification of Operational Pain Indicators

The integration of expert consensus and field practicability resulted in the identification of nine OPIs suitable for routine use in wildlife rehabilitation centres and zoological institutions. Detailed operational guidelines for each indicator—including measurement procedures, biological interpretation, expected reference conditions, associated risk factors, and recommended corrective actions—are provided in Supplementary Material S1.

3.5. Development of the Bird Pain Index

Based on the validated OPIs, the preliminary BPI was implemented using the normalised linear additive model described in the methodology. To evaluate the robustness of the equal-weighting baseline assumption ($11.1\%$ per indicator), a sensitivity analysis was performed comparing equal weighting against an expert-consensus weighted scheme derived from the final validation scores (Table 2). The sensitivity analysis demonstrated a very high positive linear correlation ($r>0.96,p<0.001$) between both weighting approaches across simulated clinical profiles, confirming that equal weighting provides a stable, transparent, and unbiased operational approximation for field use.

When applied to the 26 wild birds evaluated during field validation, individual BPI scores ranged from $22.2\%$ to $88.9\%$ (mean 58.1+/−18.4%). Birds presenting with acute traumatic injuries, compound fractures and electrocution exhibited significantly higher BPI scores (mean 75.5%) compared to those with mild soft tissue trauma or minor contusions (mean 41.2%). These results indicate that the preliminary BPI effectively differentiates pain severity across clinical presentations, supporting its utility for triage and welfare monitoring. Complete prototype recording templates, score calculation workflows, and interpretation guidelines are provided in Supplementary Material S2.

4. Discussion

To the authors’ knowledge, this study represents the first systematic attempt to identify, validate and operationalise pain indicators for wild birds admitted to wildlife rescue and rehabilitation centres. This study contributes to addressing a significant gap in avian welfare assessment and pain management by integrating evidence obtained from a structured literature review, expert consensus and field validation under real rehabilitation conditions. Although a considerable number of behavioural, physiological and clinical indicators have been described in the scientific literature, only a limited subset demonstrated sufficient biological relevance and practical applicability to be considered OPIs.

From the initial 40 candidate indicators identified by a literature review, only 25 indicators progressed after stakeholder validation and ultimately nine indicators were retained following field validation. This progressive reduction demonstrates the importance of combining scientific evidence with practical applicability when developing real-life animal welfare assessment tools. Similar findings have been reported in studies involving fish, anurans and other livestock species where many theoretically valuable indicators fail to demonstrate sufficient feasibility under routine management conditions [37,39,41,46,47,48].

The relatively low proportion of validated indicators highlights one of the principal challenges of avian pain assessment. Birds frequently display subtle, species-specific or context-dependent responses to pain, partly because the expression of overt weakness may increase predation risk in wild environments. Consequently, the absence of obvious behavioural alterations should not necessarily be interpreted as the absence of pain [6,7,9,10]. Furthermore, clinical interpretation of the BPI warrants careful consideration of individual indicator weightings. For instance, when specific high-severity indicators (such as acute tonic–clonic postures or profound autonomic dysfunctions) are scored at maximum intensity, the composite index formula may occasionally classify the overall pain state as moderate depending on the presence of unobserved metrics. In practice, however, the presence of any single critical indicator (acting as a clinical red flag) must override composite percentage scores, mandating immediate targeted analgesic intervention and intensive supportive care regardless of the cumulative index value [9].

The predominance of behavioural and clinical indicators (animal-based indicators) among the validated OPIs is consistent with the current understanding of pain recognition in birds. Behavioural alterations such as decreased activity, changes in posture, reduced feeding behaviour, impaired locomotion and abnormal wing or body positioning have repeatedly been associated with painful conditions in avian species [6,7,9,10]. Unlike physiological indicators, these behaviours can usually be assessed remotely without handling the animal, minimising observer-induced stress and reducing potential confounding effects. This is especially important in wildlife rehabilitation centres where handling itself may elicit substantial physiological responses that complicate interpretation [8,9,10]. However, several physiological indicators failed to achieve sufficient validation scores despite their frequent association with pain within the veterinary literature. Variables such as heart rate, respiratory rate, glucocorticoid concentrations and other blood-based biomarkers have been widely used as indicators of distress and pain [8,9,10]. However, these measurements are often heavily influenced by restraint, environmental conditions, species-specific physiological variation and concurrent disease processes [8,38]. Similar concerns have been raised in welfare assessment protocols for multiple taxa where physiological indicators are generally regarded as complementary measures rather than primary operational indicators [37,38,41]. Consequently, while physiological variables may contribute valuable information within comprehensive clinical examinations, they may be less suitable as routine welfare indicators under field rehabilitation conditions.

An important strength of the present study is the use of both stakeholder validation and field validation. Current animal welfare methodologies increasingly recognise that indicators must not only be biologically meaningful but also operationally feasible if they are to be adopted by veterinarians, animal carers and rehabilitation staff members [37,42,43,45]. The validation framework adapted from EFSA recommendations and previous welfare indicator studies [45,46,47] allowed the identification of indicators that achieved consensus among professionals while also remaining practical under real-world conditions. The discrepancy observed between expert perceptions and field applicability further demonstrates why operational validation is a critical step before implementing welfare assessment protocols.

The development of the BPI represents a preliminary but potentially important step towards the standardisation of pain assessment in wild birds. Composite welfare indices have been successfully developed for other taxa including fish, poultry, livestock species and amphibians [39,41,46,48]. Such indices facilitate interpretation of multiple indicators simultaneously and support evidence-based decision-making. In the present study, equal weighting was assigned to all validated indicators because no empirical evidence currently exists to justify differential weighting. Future studies should investigate the relative contribution of each indicator through larger datasets, longitudinal monitoring and predictive modelling approaches [48].

The potential applications of the validated OPIs extend beyond clinical pain assessment. Wildlife rehabilitation centres routinely receive birds suffering from fractures, collisions with vehicles, electrocution, domestic animal attacks, gunshot wounds, entanglement injuries and other forms of anthropogenic trauma [24,25,26,29]. Early identification and effective management of pain may improve treatment outcomes, facilitate rehabilitation and potentially increase release success. Previous studies have demonstrated that injury severity and rapid clinical intervention can strongly influence survival and release probabilities in wildlife rehabilitation settings [31,32,33,34,49].

The current findings are also relevant within the broader framework of conservation welfare. Conservation welfare seeks to integrate individual animal welfare with population-level conservation objectives [8,43]. Reliable pain indicators can enable rehabilitation centres, zoological institutions and conservation programmes to use evidence-based welfare monitoring and improve transparency regarding animal care practices. As societal expectations regarding animal sentience continue to evolve, such welfare assessment tools may become increasingly important for maintaining public trust and supporting ethical wildlife management [8,43,47].

Study Limitations

Several limitations should nevertheless be considered. First, the number of stakeholders participating in the survey was relatively limited (n = 49) and may not fully represent the diversity of expertise across geographical regions and avian taxa worldwide. Secondly, field validation was conducted at a single wildlife rehabilitation centre, which may limit the generalisability of the results. Thirdly, most birds included in the validation phase presented acute traumatic injuries, potentially underrepresenting indicators associated with chronic pain, infectious diseases or long-term rehabilitation conditions. Finally, although overall inter-observer agreement was substantial ($\kappa$ = 0.78), indicator-specific kappa values and complete observer blinding were not implemented in this preliminary field phase, which should be addressed in future multi-centre studies [47,48]. Future research should therefore focus on multi-institutional validation involving larger numbers of species and rehabilitation centres. Additional studies should investigate observer reliability, species-specific differences in pain expression and the sensitivity of indicators to analgesic interventions [9,10]. Longitudinal studies examining the relationship between validated indicators and rehabilitation outcomes may also contribute to refining the BPI and improving its predictive value [49]. Ultimately, the development of a standardised and internationally validated avian pain assessment system would represent a significant advance for wildlife medicine, zoological management and avian welfare science.

5. Conclusions

This study provides the first systematic framework for identifying, validating and operationalising pain indicators for wild birds admitted to rehabilitation centres and zoological institutions. From an initial pool of 40 candidate indicators, nine OPIs were successfully field-validated based on high practical feasibility and non-invasive applicability. Behavioural and non-handling clinical metrics demonstrated significantly higher field utility than physiological measures, highlighting the value of observational assessment in wild avian triage. Notably, isolation behaviour (78%), half-closed eyelids (75%) and anorexia (73%) emerged as the top-performing indicators, serving as primary observational pillars for rapid pain recognition.