Sănătatea mintală a personalului aeronautic

Introduction

Aviation is a safety-critical sociotechnical system in which human performance remains consequential even in highly automated operations. Pilots must integrate technical information, changing environmental conditions, crew communication, automation status and operational constraints, often under time pressure and uncertainty. Mental health is therefore relevant not only as a clinical issue but also as one component of a broader system of human performance and risk management.

For psychiatrists who do not routinely work with aviation personnel, aviation psychology may initially appear to be mainly concerned with pilot selection. In practice, the field is broader. Aviation psychology addresses cognitive performance, personality, decision-making, stress and coping, interpersonal behavior, crew functioning, human factors, psychological selection, training and mental-health support. Crew Resource Management (CRM), for example, evolved from a narrow focus on cockpit interpersonal problems toward a broader model of team and system performance^(1,2).

The central conceptual distinction for clinical practice is that psychiatric diagnosis, psychological functioning and aeromedical fitness overlap, but are not interchangeable. A psychiatrist primarily asks whether a mental disorder is present, how severe it is, how it should be treated, and what the prognosis is. An aviation psychologist asks how the individual functions psychologically in relation to aviation demands. The aeromedical examiner or the aeromedical center integrates relevant clinical and functional evidence within the applicable certification framework.

European regulation makes this distinction explicit. For commercial air transport flight crew, operator psychological assessment before commencing line flying is intended to identify psychological attributes and suitability in relation to the work environment and to reduce the likelihood of negative interference with safe aircraft operation. The associated guidance states that this operator assessment should not be considered or conducted as a clinical psychological evaluation⁽³⁾. Current European aeromedical rules, by contrast, require a comprehensive mental-health assessment in the initial Class 1 examination and specify circumstances requiring psychiatric evaluation⁽⁴⁾.

The present paper has two linked objectives. Firstly, it provides psychiatrists with a clinically oriented overview of aviation psychology and the psychiatrist-aviation psychologist-aeromedical interface. Secondly, it reports an empirical analysis of decision-making in a real-world aviation applicant sample to demonstrate why functional psychological performance should not be inferred from a diagnosis, a personality score, or a single cognitive indicator.

Aviation psychology and psychiatry: different questions, shared safety goals

Aviation psychology is primarily a functional discipline. Its practical concern is not whether a score is unusual in isolation, but how a pattern of characteristics interacts with the demands of a particular role. This perspective is consistent with the regulatory expectation that psychological assessment be linked to the particularity, complexity and challenges of the operating environment, and includes cognitive abilities, personality traits, operational/professional competencies and social competencies⁽³⁾.

This functional orientation is important, because the same observable behavior can have different mechanisms. Slow responding may reflect careful checking, but it may also reflect indecision, cognitive inefficiency, anxiety-related checking, perfectionistic overcontrol or fatigue. Rapid responding may reflect impulsivity, but it may also reflect efficient rule identification and well-developed expertise. Similarly, a high personality trait score may represent a resource in one context and a vulnerability in another.

The idea that fitness for aviation involves more than the absence of psychiatric disease has a long history in aerospace psychiatry. Jones and Marsh emphasized that aerospace psychiatric assessment may require consideration of positive functional qualities such as motivation, ability, and stability, as well as the absence of significant mental disorder⁽⁵⁾. The contemporary European framework similarly combines clinical mental health requirements with operational psychological assessment and proactive support systems^(3,4).

AME – aeromedical examiner; AeMC – aeromedical center. The roles are complementary and may overlap when a professional holds more than one relevant qualification.

Figure 1. Proposed functional model for psychiatric and psychological assessment in aviation
Note: The clinical condition or vulnerability influences symptoms and treatment effects, which shape cognitive, emotional and behavioral functioning. These functional effects interact with aviation task demands and may acquire safety relevance, leading to management through treatment, monitoring, support, aeromedical follow-up and return-to-work decisions.

From psychiatric diagnosis to functional aviation assessment

A diagnosis is clinically meaningful, but it does not by itself describe the functional risk. Two individuals with the same diagnosis may differ substantially in symptom severity, insight, recurrence history, treatment response, sleep, cognitive interference, adherence and willingness to seek help. Conversely, a person may not meet the criteria for a psychiatric disorder but may show functionally relevant limitations in sustained attention, decision-making, stress regulation, impulse control, or interpersonal behavior.

For aviation purposes, a useful conceptual sequence is: clinical condition or vulnerability -> symptoms and treatment effects -> cognitive, emotional and behavioral functioning -> interaction with aviation task demands -> potential safety relevance -> treatment, monitoring, support and aeromedical management (Figure 1). The framework does not transform every symptom into a safety event. Rather, it requires evidence about functional consequence, context and probability of recurrence.

This approach also helps avoid two opposite errors. Diagnostic determinism assumes that a diagnosis automatically implies unsafe functioning. Diagnostic reassurance assumes that absence of a diagnosis establishes absence of relevant vulnerability. Neither inference is sufficiently precise for a safety-critical environment.

Mental health in aviation personnel

Research on pilot mental health has expanded substantially, particularly since 2015. A systematic review by Ackland and colleagues identified considerable methodological heterogeneity across studies, including variation in sampling, instruments, case definitions and reporting conditions⁽⁶⁾. This limits the use of any single prevalence estimate as a description of the pilot population as a whole.

An anonymous web-based survey by Wu and colleagues found that 12.6% of responding airline pilots met the study’s PHQ-9 threshold for probable depression and 4.1% reported suicidal thoughts during the preceding two weeks⁽⁷⁾. The convenience-sampling design limits generalization, but the study demonstrated that clinically relevant symptoms and suicidality can occur in active aviation populations.

An equally important issue is healthcare avoidance. In a survey of 3765 pilots, 56.1% reported some history of avoiding healthcare because of concern about aeromedical certificate loss; substantial proportions also reported informal care-seeking or withholding information⁽⁸⁾. These findings should not be interpreted as universal behavior across jurisdictions, but they illustrate a structural problem: if seeking care is perceived as automatically threatening professional identity and income, early disclosure may be discouraged.

European pilot peer-support research reinforces the role of organizational context. In a 2024 study of 4494 pilots, attitudes toward approaching peer support were associated with perceived just culture and employment/support conditions; pilots with symptoms consistent with depression or anxiety were less willing to approach such programs⁽⁹⁾. Recent regulatory commentary has similarly argued that mental health systems should reduce unintended healthcare avoidance while preserving legitimate safety protections⁽¹⁰⁾.

Fatigue and sleep are additional areas of overlap between psychiatric and operational functioning. Short- and long-haul pilots may experience sleep restriction, circadian disruption, fatigue, stress and reduced well-being; these factors can also amplify or mimic psychiatric symptoms⁽¹¹⁾. Clinical assessment should therefore distinguish primary psychiatric disorder, primary sleep disorder, occupational circadian disruption, insufficient sleep, substance or medication effects and combinations of these mechanisms.

Empirical study: decision-making under uncertainty in an aviation applicant sample

Study design and data source

The empirical component was designed as a retrospective secondary analysis of routinely collected occupational psychological assessment data. The source database was accumulated over approximately two years during psychological assessments of aviation personnel applying for employment with several commercial air operators. The assessments were conducted as routine professional services rather than as a dedicated research protocol; however, at data collection, the participants were asked for and provided consent for secondary use of their assessment data for research purposes. No additional assessment, intervention, or participant contact was undertaken for the present study.

The occupational selection context increases the ecological relevance of the data because the assessments reflect real high-stakes aviation recruitment rather than laboratory participation. At the same time, the context introduces potential selection and self-presentation effects. Applicants may be more motivated than general-population participants, and personality or symptom self-report may be influenced by perceived employment consequences.

Participants

The analytical sample comprised 256 aviation personnel aged 21 to 64 years old (M=43.08; SD=11.29); 240 were men, and 16 were women. Flight-hour data were available for 253 participants (M=6217.30; SD=6434.99; median=4500; range=20-37; 764 hours). Participants were undergoing psychological assessment in relation to recruitment or employment in commercial aviation. The sample was not constructed to estimate psychiatric prevalence and should not be interpreted as representative of all pilots or aviation personnel.

Measures

Decision-making ability test

Decision-making was assessed using a standardized 14-item Decision-Making Ability Test included in a Romanian battery of cognitive-aptitude tests⁽¹²⁾. The manual describes the test as an adaptation and standardization of experimental decision tasks and conceptualizes performance in terms of rational decision-making and reduced sensitivity to common heuristics and biases. The test is administered under a nominal 7-minute time limit. Normative responses receive 1 point; non-normative responses, indecision and unattempted items receive 0 in the conventional total score. The conceptual background includes representativeness and probability judgment, framing effects, ambiguity aversion and broader individual differences in decision competence^(13-16).

For the present analysis, the 14 items were grouped according to the decision mechanisms represented in the item set: nine representativeness/probability-judgment items, two framing items, and three Ellsberg-type ambiguity items. These component scores were treated as bias-specific performance indicators rather than equivalent psychometric subscales. This distinction is important, because the item families differ in number, position, content and internal consistency.

Four complementary performance indicators were calculated: total number of items attempted (task coverage), proportion of normative answers among attempted items (accuracy), total number of normative answers and actual item-solving time. LimeSurvey also recorded total decision-section time, time spent completing demographic information and time spent on the instruction page. For bias-specific analyses, accuracy was examined only among participants who reached the last item of the corresponding component, reducing confounding between an incorrect choice and lack of opportunity to reach a late item.

Mathematical Calculation Test

The Mathematical Calculation Test assesses numerical ability through the speed and correctness of simple arithmetic operations and access to acquired mathematical knowledge⁽¹²⁾. It contains 15 addition, subtraction, multiplication and division exercises with four response alternatives and a 5-minute time limit. The validated total score used in the analysis ranged from 0 to 15. The instrument manual explicitly defines the construct in terms of speed and correctness of simple numerical processing.

Mathematical Reasoning Test

The Mathematical Reasoning Test contains 20 ordered number series. Participants identify one or more mathematical rules and select the pair of numbers that correctly completes each sequence from four alternatives; the time limit is 10 minutes⁽¹²⁾. In the present sample, one item showed no variance, and it was excluded automatically from the reliability estimate; internal consistency for the remaining 19 items was alpha=0.795.

Personality

Personality was assessed with the 240-item NEO PI-3, yielding the five Five-Factor Model domains: Neuroticism, Extraversion, Openness, Agreeableness and Conscientiousness⁽¹⁷⁾. Internal consistency was high in the present sample: alpha=0.916, 0.824, 0.862, 0.844 and 0.922, respectively. Previous European civil-pilot research has supported the psychometric utility of Five-Factor personality assessment in aviation while also cautioning against treating a group-level pilot profile as a deterministic individual criterion⁽¹⁸⁾. Updated meta-analytic evidence suggests that personality measures have statistically detectable but generally modest relationships with pilot-training success, with effects depending on criterion and measurement context⁽¹⁹⁾.

Procedure and ethics

Measures were completed individually in electronic format during occupational psychological assessments. For the research analysis, the working dataset was separated from direct personal identifiers, and the results are reported at group level. No participant was contacted for the secondary analysis, and the analyses reported here do not identify individual applicants or air operators.

At the time of data collection, the participants were asked for and provided consent for the use of their assessment data for research purposes. The present study is a retrospective secondary analysis of those routinely collected occupational assessment data. The analytical dataset was de-identified before research analysis and results are reported at group level. Formal ethical review of the secondary-analysis protocol is pending. The final submission should report the reviewing body and reference/approval number exactly as issued by that body.

Statistical analysis

Analyses were redesigned after the additional LimeSurvey timing variables became available. Total decision-section time was decomposed into demographic-entry time, instruction-viewing time and actual time spent on Items 1-14. The decomposition was internally consistent to within one second in 252 of 256 records; four discordant timing records were flagged as probable logging or session anomalies. Core time-performance correlations were therefore examined in the full sample and repeated as sensitivity analyses in the 252 internally consistent records.

Because actual item-solving time showed pronounced heaping around the nominal seven-minute limit among participants who did not finish the task, it was not treated as an unconstrained continuous processing-speed variable. Task completion (14/14 items versus fewer than 14) was analyzed as a separate binary outcome. A multivariable logistic regression predicted full completion from age, log flight experience, mathematical calculation, mathematical reasoning and the five NEO PI-3 domains; continuous predictors were standardized, and nested likelihood-ratio tests evaluated the cognitive and Big Five blocks. Among full completers, the association between actual item-solving time and accuracy was examined separately, including a sensitivity analysis restricted to completion times of 425 seconds or less.

Bias-specific accuracy was analyzed in full-exposure subgroups: representativeness (n=124), framing (n=175) and ambiguity (n=73). Spearman correlations examined actual item-solving time, mathematical abilities and Big Five domains; Benjamini-Hochberg false discovery rate correction was applied within each bias family. Item-level generalized estimating equations (GEE) used a binomial logit link, robust covariance estimates, participant clustering and item fixed effects; age and actual item-solving time were included as contextual covariates. Relationships among the three bias-specific indicators were evaluated among the 73 participants who reached all 14 items. Partial Spearman correlations controlled age, actual item-solving time, mathematical calculation, mathematical reasoning and all five NEO PI-3 domains.

Results

Overall decision performance

Participants attempted a mean of 11.15 of 14 decision items (SD=2.52) and produced a mean of 6.66 normative answers (SD=2.17). Mean accuracy among attempted items was 0.599 (SD=0.145). The original total decision-section time averaged 608.79 seconds (SD=171.53), but the additional LimeSurvey metadata showed that this variable included activities occurring before the decision items themselves.

Using actual item-solving time, the relationship with task coverage became substantially stronger than in the original total-time analysis (rho = -0.511; p<0.001). Actual item-solving time was also negatively associated with the total number of normative answers (rho = -0.363; p<0.001), but remained unrelated to accuracy among attempted items (rho = -0.027, p=0.669). In the 252 records with internally consistent timing decomposition, the corresponding correlations were rho = -0.508, -0.364 and -0.035, respectively, showing that the findings were not driven by the four discordant timing records.

LimeSurvey timing decomposition and task completion

In the 252 internally consistent records, mean total decision-section time was 610.39 seconds (SD=171.78). This comprised a mean of 143.86 seconds for demographic completion, 55.20 seconds for the instruction page and 411.33 seconds for the 14 decision items. Thus, based on component means, actual item solving accounted for approximately two-thirds of total section time. Total section time correlated more strongly with demographic-entry time (rho=0.845) and instruction-viewing time (rho=0.694) than with actual item-solving time (rho=0.370), demonstrating that the original total-time variable was not a pure measure of decision speed.

Response-volume and accuracy profiles

The sample medians were 11 attempted items and with accuracy of 0.583. Crossing these dimensions generated four exploratory profiles. The high-volume/high-accuracy group had the strongest overall performance, whereas the low-volume/low-accuracy group had the weakest. The high-volume/low-accuracy and low-volume/high-accuracy groups produced similar total numbers of normative answers (6.63 versus 6.44; p=0.403) despite opposite performance architectures. Their actual item-solving times differed; however (397.24 versus 423.68 seconds; Welch p=0.009; d = -0.48), consistent with the low-volume/high-accuracy group frequently reaching the time ceiling rather than simply choosing to stop early.

Bias-specific accuracy after equalizing exposure

Because unattempted items receive zero in the conventional raw score and ambiguity items occur late in the fixed task order, raw component scores combine choice quality with opportunity to respond. After equalizing exposure, actual item-solving time showed a modest negative bivariate association with representativeness accuracy (rho = -0.217; p=0.015), but not with framing accuracy (rho=0.092; p=0.229) or ambiguity accuracy (rho = -0.112, p=0.345). The representativeness time association did not survive false discovery rate correction and was not significant in the multivariable GEE model, so it should not be interpreted as a robust independent effect.

High volume – more than 11 attempted items; high accuracy – accuracy > 0.583. Actual item-solving time is subject to the nominal 7-minute ceiling;
low-volume groups therefore contain substantial right-censoring. The profile classification is exploratory, and it is not a normative diagnostic system.

rho – Spearman correlation; HV/LA – high volume/low accuracy; LV/HA – low volume/high accuracy. GEE – generalized estimating equations; LR – likelihood-ratio test. Actual item-solving time, rather than total decision-section time, is used in the updated timing analyses. The results are not certification thresholds and should not be used diagnostically.

Ambiguity accuracy was positively associated bivariately with mathematical reasoning (rho=0.279; p=0.017) and mathematical calculation (rho=0.239; p=0.042). Lower Neuroticism (rho = -0.246; p=0.036) and higher Conscientiousness (rho=0.251; p=0.032) also showed nominal bivariate associations with ambiguity performance. However, after false discovery rate correction across the eight prespecified associations within the ambiguity family, these effects were attenuated (q approximately 0.083 for reasoning, calculation, Neuroticism and Conscientiousness), being therefore best regarded as convergent but exploratory signals.

Representativeness showed no consistent association with the five NEO domains, mathematical calculation, or mathematical reasoning. Framing correlations were also small and nonsignificant in the bivariate analysis. At item level, calculation retained a positive coefficient for framing (OR=1.25 per SD; 95% CI; 1.02-1.53; p=0.033), but the cognitive block as a whole remained marginal; Wald chi-square(2)=5.64, p=0.059. This result is therefore interpreted cautiously.

For ambiguity, the updated GEE model included actual item-solving time, which is interpretable in this subgroup, because all 73 participants reached the final item. The joint cognitive block was significant, Wald chi-square(2)=6.69, p=0.035, although the individual mathematical-reasoning coefficient was only trend-level after simultaneous adjustment (OR=1.49 per SD; 95% CI; 0.98-2.26; p=0.062). The Big Five block was not significant, Wald chi-square(5)=8.05, p=0.154. Taken together, the updated analysis supports a more cautious conclusion: cognitive resources show a clearer signal than broad personality traits in ambiguity performance, but individual effects require replication.

Relationships among the three bias-specific indicators

Among the 73 participants who reached all 14 items, representativeness was unrelated to framing (rho = -0.061; p=0.608) and ambiguity (rho = -0.050, p=0.675). Framing and ambiguity were moderately associated (rho=0.325, p=0.005; bootstrap 95% CI approximately 0.125-0.505). The association remained after controlling age, actual item-solving time, mathematical calculation, mathematical reasoning, and all five Big Five domains (partial rho=0.326, p=0.005; bootstrap 95% CI approximately 0.091-0.534).

The pattern does not support a simple unitary “bias-proneness” construct. It suggests partial commonality between framing and ambiguity, while representativeness/probability-judgment performance is more heterogeneous. Internal consistency in the full-exposure subgroups was alpha=0.064 for representativeness, -0.265 for framing and 0.636 for ambiguity. Accordingly, these indicators are interpreted as performance on bias-specific problem families rather than equivalent latent scales.

What the empirical findings mean for psychiatry

The updated empirical results are relevant to psychiatry, because they show how easily an observed behavioral measure can be misidentified as a psychological mechanism. The original “decision time” variable was not a pure decision-speed measure: it included demographic completion and instruction viewing, and total section time correlated more strongly with these nondecision components than with actual item-solving time. Once actual item-solving time was isolated, greater time remained unrelated to accuracy. A clinician should therefore be cautious about interpreting a long test duration as evidence of carefulness, indecision, psychomotor slowing, anxiety, or cognitive impairment without knowing which process consumed the time.

The time limit adds a second layer of interpretation. Out of 256 participants, 183 did not complete all 14 items, and 158 of these 183 had actual item times between 419 and 425 seconds. Overall, 175 participants clustered in this narrow interval around the nominal 7-minute limit. For these participants, the observed item time is largely a ceiling value rather than an unconstrained measure of processing speed. The clinically analogous lesson is that behavioral data must be interpreted in relation to task constraints before being attributed to an internal psychological state.

The four response profiles provide a third example. High-volume/low-accuracy and low-volume/high-accuracy participants achieved nearly identical total correct scores, yet the former processed many more items with lower precision, and the latter achieved higher precision while frequently reaching the time ceiling. A total score therefore concealed distinct performance architectures. Clinical assessment often faces an analogous problem: similar symptom scores can arise from different diagnoses, compensatory strategies, levels of insight, or functional consequences.

The updated models also refine the role of cognition and personality. Mathematical reasoning strongly predicted the probability of completing all 14 items within the available interval (OR=1.77 per SD; 95% CI; 1.28-2.45; p=0.001). Adding calculation and reasoning significantly improved the completion model, LR chi-square(2)=11.66, p=0.003, whereas the five Big Five domains did not add a significant block contribution, LR chi-square(5)=8.80, p=0.117. A negative individual coefficient for Conscientiousness in the full model is not interpreted substantively, because the personality block was nonsignificant and the domains are intercorrelated. This pattern argues against deterministic personality interpretations and supports a stronger role for specific cognitive resources in efficient task progression.

Finally, naturally occurring interindividual differences should not be equated with experimentally induced speed-accuracy trade-offs^(20-22). In the present task, many slower observations were censored by the time ceiling. Among the 73 participants who completed all 14 items, actual item-solving time was still not significantly related to accuracy (rho = -0.162; p=0.172); restricting the analysis to completion times <425 seconds produced the same conclusion (rho = -0.158; p=0.195). Faster performance may therefore reflect efficient rule extraction, whereas slower performance may reflect conflict, checking, or difficulty without an accuracy gain. Professional intuition itself is reliable only under conditions that permit valid learning from a sufficiently regular environment⁽²³⁾.

Psychiatric conditions in the aviation context

Psychiatric expertise becomes central when variation in mood, anxiety, behavior, cognition, or personality reaches clinically significant levels. Current European Class 1 aeromedical rules require comprehensive mental-health assessment, drug and alcohol screening, and specialist psychiatric evaluation for specified mental and behavioral conditions, including relevant mood, neurotic, personality, substance-related, self-harm and other psychiatric histories⁽⁴⁾.

Anxiety and panic

For anxiety disorders, the diagnosis alone does not establish the degree of operational impairment. Relevant functional questions include whether symptoms have occurred during flight, whether there is anticipatory avoidance, whether internal sensations capture attention, whether panic is predictable, whether sleep is impaired, whether the person can recognize and manage symptoms and whether the treatment is stable. A pilot with successfully treated anxiety, good insight, and no residual cognitive interference represents a different clinical situation from a pilot with unpredictable panic attacks and avoidance.

Depression and suicidality

In depression, functional assessment should address concentration, psychomotor change, fatigue, sleep, motivation, hopelessness, decision-making, occupational functioning, recurrence and suicidality. The aeromedical significance of depression is partly determined by severity and longitudinal stability rather than by the diagnostic label alone. Mental health working-group recommendations in aerospace medicine emphasize trust, specialist input and a balance between confidentiality and public-safety concerns⁽²⁴⁾.

Bipolar-spectrum and psychotic disorders

Bipolar-spectrum disorders require careful longitudinal assessment of mood episodes, sleep reduction, risk-taking, impulsivity, insight, treatment adherence and recurrence. Psychotic symptoms require specialist assessment because of their potential relevance to perception, reality testing, judgment and behavior. Regulatory decisions in these conditions are governed by the applicable aeromedical framework and should not be inferred from psychological testing alone⁽⁴⁾.

ADHD and executive functioning

Adult ADHD illustrates the distinction between diagnosis and functional performance. A diagnostic formulation should be supplemented, where clinically and aeromedically appropriate, by evidence concerning sustained attention, executive control, organization, impulsivity, medication, compensatory strategies and occupational history. A symptom checklist cannot by itself quantify performance in a complex flight environment.

Personality: trait, pathology and function

Normal-range personality assessment is common in aviation selection, but trait scores should not be confused with personality disorder. High Neuroticism does not establish an anxiety or depressive disorder. Low Agreeableness does not establish personality pathology. High Conscientiousness does not guarantee operational reliability. A personality inventory describes tendencies that must be integrated with history, behavior, cognition, occupational performance and clinical context.

There is also no single “ideal pilot personality”. Aviation tasks may require apparently opposing tendencies: procedural discipline and adaptability; confidence and self-correction; autonomy and cooperation; assertiveness and receptivity to challenge; persistence and willingness to abandon a deteriorating plan. CRM research reinforces the importance of team behavior, communication, coordination, and adaptation rather than a simple trait maximum^(1,2).

When personality pathology is suspected, psychiatric or clinical psychological assessment addresses a different question from occupational trait measurement: pervasiveness, rigidity, distress, impairment, interpersonal consequences, impulse control, reality testing and longitudinal stability. The two forms of assessment can inform one another but should not be collapsed into a single score.

Psychotropic medication, sleep, fatigue and substance use

Psychotropic medication

Clinical efficacy and aeromedical compatibility are separate questions. A medication may effectively treat a disorder while producing sedation, activation, cognitive effects, sleep disturbance, orthostatic symptoms, discontinuation effects, or interactions with fatigue. Conversely, untreated psychiatric illness may itself present greater risk than appropriately managed treatment. Historical aeromedical literature on selective serotonin reuptake inhibitors illustrates how regulatory approaches have evolved as these competing risks have been reconsidered⁽²⁵⁾.

The psychiatrist’s most useful contribution is therefore a detailed description of medication, dose, duration, stability, response, adherence, adverse effects, and recurrence risk. The final certification decision belongs within the competent aeromedical system unless the psychiatrist is specifically acting within that system.

Sleep and fatigue

Sleep disturbance may be a primary sleep disorder, a symptom of depression or anxiety, a medication or substance effect, an occupational circadian consequence, or a combination. Aviation schedules can include early starts, night duty, time-zone transitions, and irregular recovery. The resulting fatigue may affect vigilance, processing speed, working memory, emotional regulation and decision-making⁽¹¹⁾. Clinical assessment should therefore identify mechanism rather than treating “poor sleep” as a unitary symptom.

Alcohol and psychoactive substances

Substance-related assessment should distinguish isolated exposure, occasional use, hazardous use, harmful use, substance-use disorder, dependence and sustained remission. European Class 1 examination includes drug and alcohol screening and substance-related mental or behavioral disorders require satisfactory psychiatric evaluation after treatment before a fit assessment can be considered⁽⁴⁾.

A useful psychiatric formulation includes the substance, quantity, frequency, context, loss of control, tolerance or withdrawal, consequences, periods of abstinence, prior treatment, comorbidity, coping function, insight, relapse risk and willingness to engage in monitoring. A laboratory result alone does not provide the full longitudinal risk formulation.

Confidentiality, disclosure and peer support

Aviation mental health systems face a disclosure paradox. Safety requires identification and management of conditions that may impair performance. Yet if disclosure is perceived as automatically leading to loss of license, income or professional identity, the same system may discourage early treatment. Healthcare-avoidance data suggest that this concern is clinically relevant⁽⁸⁾.

European operator support-program requirements explicitly attempt to address this problem. They emphasize prevention, early support, professional mental health involvement, trained peers, return-to-work support, management of fear of license loss and confidential handling of personal data. Guidance states that disclosure to the operator should ordinarily be anonymized and aggregated, while procedures must exist for serious safety concerns⁽²⁶⁾.

Peer support should not be mistaken for informal psychiatry. Peers do not diagnose, prescribe, assess suicidality independently or determine fitness. Their value lies in accessibility, shared occupational language, early contact, normalization of help-seeking and facilitated referral. Reviews and recent empirical work suggest that peer support is most credible when embedded in a just culture and linked to professional care^(9,27).

Note. The report should remain within the psychiatrist’s competence and applicable confidentiality/legal duties. Aeromedical fitness is determined within the relevant regulatory framework.

For psychiatrists, the practical implication is that under-reporting may reflect fear of professional consequences as well as poor insight or deception. Clear explanation of confidentiality and its limits, careful documentation and proportionate communication with aeromedical professionals can support both treatment and safety.

What should a psychiatric report for aviation purposes contain?

A useful psychiatric report for an aviation referral should do more than state a diagnosis. Unless the psychiatrist is specifically authorized to make the aeromedical certification decision, a detailed clinical-functional formulation is generally more useful than an unsupported statement such as “fit to fly.” The report should provide evidence that the AME, AeMC, licensing authority and aviation psychologist need to integrate the case within their respective competences.

Return to flying as a process

Return to flying after a psychiatric condition should not automatically be framed as a permanent binary outcome. Depending on diagnosis, treatment, jurisdiction and aeromedical requirements, the process may include clinical recovery, a period of stability, psychiatric reassessment, psychological reassessment, medication review, aeromedical evaluation, monitoring and staged return. Historical military aviation studies demonstrate that psychiatric treatment or even hospitalization did not inevitably preclude later return to flying duties^(28,29). These historical findings cannot be transferred directly to contemporary civil certification, but they counter the assumption that help-seeking necessarily ends an aviation career.

A proportionate return-to-work model can also improve disclosure. If pilots understand that clinically appropriate treatment and documented recovery may lead to a structured reassessment pathway rather than automatic permanent exclusion, the perceived cost of seeking help may decrease.

Discussion

The present paper integrates a psychiatric perspective on aviation mental health with original functional data from an aviation applicant sample. The central argument is that aviation safety decisions should not be reduced to diagnostic categories or isolated psychometric scores. Diagnosis, symptoms, cognitive functioning, personality, treatment effects, occupational demands and longitudinal stability represent different layers of information.

The empirical results provide a concrete demonstration. Response time was strongly related to task coverage but not to accuracy. Two groups with opposite response strategies achieved similar total scores. Representativeness, framing and ambiguity did not behave as interchangeable manifestations of one bias trait. Mathematical reasoning showed its clearest relationship with ambiguity, while general personality domains provided a more limited independent contribution. A single “decision score” would conceal much of this structure.

For psychiatry, the analogy is direct. A single depression score does not describe suicidality, cognitive slowing, sleep, treatment adherence, recurrence risk or occupational functioning. A personality profile does not establish personality disorder. A psychiatric diagnosis does not establish the aeromedical outcome. Conversely, the absence of psychiatric diagnosis does not guarantee optimal functioning in every safety-critical cognitive domain.

The results also argue against common stereotypes about pilot personality and decision speed. Faster responding was not associated with lower accuracy at the between-person level. Personality effects were not sufficiently strong or uniform to justify deterministic interpretations. Updated personality meta-analysis in aviation similarly indicates small relationships rather than a single predictive profile⁽¹⁹⁾. This supports multimethod assessment, particularly when decisions carry substantial consequences for both safety and career.

The occupational context of the dataset is both a strength and a limitation. Its strength is ecological relevance: participants were genuine aviation applicants under real selection conditions. Its limitation is that high stakes may influence effort, self-presentation and sample composition. The sample is likely healthier and more occupationally selected than a psychiatric population. Therefore, the study does not estimate prevalence of mental disorder and should not be used to define clinical or aeromedical cut-offs.

Aviation mental health policy must also manage the tension between detection and disclosure. Regulations and support programs appropriately protect public safety, but systems that are experienced as indiscriminately punitive may promote concealment. European support program guidance explicitly recognizes the need to manage fear of license loss and maintain confidentiality while retaining procedures for serious safety concerns⁽²⁶⁾. This is compatible with a clinical model based on early intervention, professional referral, and proportionate risk management.

Limitations

Several limitations require explicit consideration. Firstly, the sample was an occupational applicant sample rather than a random or epidemiological sample. Selection effects limit generalization to the entire aviation workforce and especially to clinical populations.

Secondly, the assessments were conducted in a high-stakes employment context. Although this improves ecological validity for selection practice, self-presentation and motivation may differ from research settings without occupational consequences.

Thirdly, the empirical study is retrospective and cross-sectional. Associations do not establish causal direction, and no inference can be made that a measured cognitive variable predicts accidents, incidents, psychiatric illness, or future aeromedical outcomes.

Fourthly, the standardized decision task is not a simulator and should not be equated directly with cockpit decision-making. It isolates selected aspects of judgment under uncertainty rather than reproducing the full operational environment, including crew interaction, workload, automation, time-critical consequences and domain expertise.

Fifthly, LimeSurvey timing data revealed both a strong ceiling effect around the nominal 7-minute decision limit and a small number of discordant timing records. For participants who did not complete the task, actual item-solving time is therefore right-censored and cannot be interpreted as an unconstrained continuous measure of processing speed. The main timing associations were stable after excluding the four internally inconsistent timing records, but future studies should record item-level latency or experimentally manipulate time limits when process-level temporal inference is required.

Sixthly, the three bias-specific item families are psychometrically unequal. The framing component contains only two items, internal consistency was very low for representativeness and framing, and the ambiguity full-exposure analysis included only 73 participants. Bias-specific and personality findings should therefore be regarded as exploratory and require replication.

Seventhly, although the participants provided consent at data collection for the use of their assessment data for research purposes, the present study remains a retrospective secondary analysis of data collected in an occupational selection context.

Conclusions

Aviation psychology and psychiatry address different but complementary dimensions of human functioning. Psychiatry provides expertise in psychopathology, differential diagnosis, treatment, suicidality, substance-related disorders, psychopharmacology and prognosis. Aviation psychology contributes functional assessment of cognition, personality, decision-making, coping, interpersonal behavior and adaptation to operational demands. Aeromedical professionals integrate these findings within certification requirements.

The empirical data show why such integration is necessary. Decomposed LimeSurvey timings demonstrated that total section time was not equivalent to decision-processing time. Actual item-solving time was strongly related to whether participants progressed through the task but not to the accuracy of attempted decisions. Mathematical reasoning predicted full task completion, while broad Big Five personality domains added limited independent information. Total scores also concealed different response strategies, and bias-specific problem families were not interchangeable. Psychological performance is therefore multidimensional even within a selected aviation applicant sample.

For psychiatrists, the practical message is that diagnosis and aviation fitness should not be treated as synonyms. The most useful psychiatric contribution is a careful description of diagnosis, severity, longitudinal course, suicidality, treatment, medication effects, insight, functional impact and prognosis. For aviation psychology, psychiatric expertise becomes indispensable when normal variation crosses into clinically significant symptoms or psychopathology.

A system that combines early help-seeking, effective treatment, functional assessment, confidentiality with clearly defined safety exceptions, and proportionate aeromedical management is more defensible than either diagnostic exclusion or uncritical reassurance. The shared objective is safe functioning while preserving appropriate access to mental healthcare.

Statements and declarations

Conflict of interest: The first author provides professional psychological assessment services to commercial aviation operators. The occupational assessments from which the de-identified analytical dataset was derived were conducted under professional service agreements. The second and third authors practice as aeromedical examiners at Clinica LAMED. No air operator or aeromedical organization sponsored the preparation of this manuscript, had a role in defining the research questions or statistical analyses, interpreted the results, drafted the manuscript, or influenced the decision to submit it.

Funding: No external research funding was received for the preparation of this manuscript.

Ethics approval and informed consent: The empirical component is a retrospective secondary analysis of data originally collected during routine occupational psychological assessments. At the time of data collection, participants were asked for and provided informed consent for the use of their data for research purposes.

Author contributions (credit): Constantin Roangheși – Conceptualization; Methodology; Investigation; Data curation; Formal analysis; Visualization; Writing – original draft; Project administration. Leonard-Marin Lupu – Conceptualization; Validation; Writing – original draft. Camelia-Lucia Bakri – Conceptualization; Validation.

Conflict of interests: none declared

Financial support: none declared

This work is permanently accessible online free of charge and published under the CC-BY.