Perceived Crisis-Management Training Adequacy and Wellbeing Indicators Among Seafarers During the Red Sea Crisis: An Exploratory Survey

Abstract

Global shipping carries over 80% of world trade, yet existing supply-chain-resilience theory models resilience at the level of the firm, node and network, without treating the workforce operating those nodes as a variable. This exploratory study addresses that gap using the 2023–2026 Red Sea/Houthi crisis as its empirical setting. A pilot-scale, cross-sectional online survey of 64 serving seafarers combined closed-ended items with qualitative content analysis of two open-text questions. Seafarers who had witnessed or been directly involved in a security incident rated STCW crisis-management training as less adequate for conflict-area risk than those without such exposure (Mann–Whitney U = 227.0, p = 0.007, r = 0.34), the only comparison that remained significant after Holm adjustment within its family of tests; Red Sea transit or Cape of Good Hope rerouting without an incident was not associated with lower ratings. Open-text responses identified situational information and realistic scenario-based drills, rather than generic compliance training, as the specific gap, and a minority questioned seafarers’ right to decline a high-risk voyage. Because fatigue and psychological distress could be captured only through single, non-validated items, findings on these outcomes are indicative only. Building on these results, the study proposes, as exploratory propositions for future testing rather than established relationships, two human-factor pathways through which geopolitical risk may affect operational resilience: a training-competence pathway and a retention/psychological-contract pathway. The study contributes hypothesis-generating evidence, gathered directly from an affected workforce, and derives practical and policy recommendations for STCW training content and seafarer consent protections.

1. Introduction

Global shipping carries over 80% of world trade by volume, and its efficient operation has traditionally been analysed as a problem of routing, capacity and cost. Over the past three years, however, a series of geopolitical shocks has exposed the limits of that framing. The Houthi campaign against commercial vessels in the Red Sea, which began in November 2023 and has recurred intermittently through 2026, forced the majority of container and tanker traffic away from the Suez Canal and around the Cape of Good Hope, adding roughly 3000 nautical miles and ten to fourteen days to a typical Asia–Europe voyage [1]. Industry and policy commentary on the crisis has focused overwhelmingly on its economic and logistical consequences: freight-rate volatility, insurance premiums, port congestion and carbon emissions. What this literature has been slower to register is that these disruptions are not experienced only by cargo, schedules and balance sheets, but by the seafarers who continue to crew the vessels that sail through, or are rerouted because of, an active conflict zone.

The Red Sea crisis is only the most recent and most acute in a sequence of exogenous shocks that have tested the shipping industry’s capacity to keep vessels crewed and operating. The COVID-19 pandemic offered an earlier, differently structured illustration of the same underlying vulnerability: an event-study analysis of shipping-company stock returns shows that markets priced the pandemic as a material shock to the sector, driven in large part by crew-change restrictions, port closures and the practical difficulty of rotating seafarers on and off vessels under conditions of extreme uncertainty [2]. That episode, like the Red Sea crisis, was ultimately experienced at the level of individual crew members stranded at sea for months beyond their contracted terms, even though it entered the academic and financial literature primarily as a shock to freight rates and equity valuations. This recurring pattern, an exogenous disruption that is modelled financially and operationally and only belatedly connected to seafarer welfare, motivates this study’s focus on the workforce dimension of the current crisis from the outset.

This is not a peripheral omission. Reporting from seafarers’ unions and maritime charities during the Red Sea crisis has documented crew members refusing to sail through the region, citing fear of missile and drone attack, the psychological burden of extended and rerouted voyages, and accumulated fatigue [3]. At least eight seafarers are known to have been killed in attacks on commercial shipping since the crisis began, and more than one hundred vessels have been targeted [4]. The human cost of geopolitical disruption is therefore not hypothetical; it is already visible in the willingness of experienced crew to leave the industry, in reported trauma symptoms among survivors of attacks, and in the operational difficulty shipping companies now face in crewing vessels for high-risk transits [5].

The long-term decline in the attractiveness of seafaring as a career has been a persistent concern for maritime nations. A 45% reduction in the number of Greek seafarers employed on Greek-owned vessels between 2000 and 2020 has been documented, highlighting a systemic challenge in recruiting and retaining qualified personnel [6]. Geopolitical threats appear to be accelerating this trend. Occupational-stress research in adjacent, shore-based sectors has repeatedly found that unmanaged exposure to chronic work-related stressors erodes both wellbeing and retention [7]; the mechanism such studies describe, sustained stress translating into attrition of exactly the experienced personnel an organization can least afford to lose, is the mechanism this study investigates in a maritime and geopolitical setting, where the stressor is acute and life-threatening rather than chronic and organizational.

A separate and long-established body of occupational-health research has documented the ordinary, non-conflict-related stressors of seafaring: social isolation, long and irregular working hours, disrupted sleep, separation from family, and exposure to accidents, stowaways and piracy [8,9]. This literature has produced validated instruments for measuring seafarer fatigue and psychological distress and has established fairly consistent risk factors across studies and fleets, but it has not yet extended this evidence base to the specific and escalating case of seafarers operating in or around active geopolitical conflict zones, because the relevant events are so recent. The present study is positioned at the intersection of these two literatures: it takes the theoretical and methodological apparatus developed for seafarer-wellbeing research in general and applies it to the emerging and under-studied context of geopolitical risk.

1.1. Problem Statement and Theoretical Gap

Recent scholarship on geopolitical disruption to global supply chains has begun to develop dedicated theoretical frameworks for understanding how conflict, sanctions and strategic rivalry propagate through supply networks and reshape their design [1]. These frameworks are a significant advance on earlier, more generic disruption models, but they remain, by design, structural: they theorize supply chains as networks of nodes, chokepoints and resource dependencies, and they explain resilience largely in terms of diversification, regionalization and governance. Even foundational definitions of supply-chain resilience, framed since [10] as a system-level capacity to absorb disturbance and recover, have tended to locate that capacity in structural and managerial choices rather than in the state of the people who operate the system. Ponomarov and Holcomb [11] extended this conceptualization by treating resilience as a multidimensional, dynamic capability embedded in supply-chain processes and relationships, spanning readiness, response and recovery, which opens conceptual space for factors such as workforce readiness and psychological preparedness that are not reducible to inventory buffers or network redundancy.

A parallel and more technical literature has developed around the mechanics of disruption propagation. Dolgui et al. [12] describe the “ripple effect” by which a localized disruption cascades through a supply network, modelling supply chains as networks of nodes and edges through simulation and control-theoretic methods. This tradition is overwhelmingly structural and quantitative and has systematically neglected the role of the human factor in mediating these flows; the workforce is treated as an automated resource whose availability, psychological state, fatigue and training are assumed to be constant. This limitation persists even in recent applied research: an optimization model for global wheat supply chains addressing extreme-weather and export-ban risk treats supply diversification, risk-management plans and backup suppliers as its key decision variables, again omitting workforce state [13].

Building on Caldara and Iacoviello’s [14] widely used, newswire-based index of geopolitical risk, Bednarski et al. [1] proposed the most theoretically developed account to date of what they term Geopolitical Supply Chain Disruptions, integrating Realist, Liberal and Constructivist perspectives from international relations with transaction-cost economics and resource-dependence theory. It explains how supply chains become “securitized,” how disruption propagates through chokepoints controlled by state or para-state actors, and how firms respond through reshoring, nearshoring and friendshoring. Its unit of analysis, however, remains the firm, the node and the supply network, not the individual whose labour keeps the node operating. This study takes that framework as its principal theoretical anchor while arguing that it requires extension to accommodate the human factor as a mediating variable.

This is a conceptual gap with direct practical consequences. A shipping company can diversify its routes, insure its cargo and reconfigure its network, but if its seafarers are unwilling to sail high-risk transits, are too fatigued to respond safely to an emergency or are inadequately trained for the specific demands of operating under threat of attack, the resilience of the wider system is compromised regardless of how well the structural response has been designed. The problem this study addresses is therefore twofold: first, that the mental health, fatigue and training needs of seafarers exposed to geopolitical risk have not been systematically examined, even though the adjacent literatures on seafarer wellbeing and on piracy-related trauma provide a clear methodological template for doing so; and second, that the theoretical treatment of supply-chain resilience does not yet incorporate this human dimension operationally.

1.2. The Human Element, Seafarer Wellbeing and Training Literature

A literature older than either the supply-chain-resilience or the seafarer-wellbeing literature has established that human and organizational factors, rather than equipment failure or environmental conditions alone, are the dominant cause of maritime accidents. Reason’s [15] “Swiss cheese” model of organizational accident causation argues that accidents result from the alignment of latent systemic weaknesses (poor design, maintenance issues, management decisions) with active failures at the point of operation. Applied to shipping, seafarer fatigue, stress or inadequate training function as latent conditions that increase the probability of an active failure under pressure, rather than as isolated welfare concerns unconnected to operational safety. Hetherington et al. [16], synthesizing twenty studies across fatigue, stress, health, situation awareness, teamwork, decision-making, communication, automation and safety culture, concluded that human and organizational factors represent the primary lever available for improving maritime safety performance and that a ship’s safety record is less a function of hardware or regulatory compliance than of its safety culture. This perspective is reinforced by the high-reliability-organizations literature, which shows that safety-critical industries must actively manage human factors to prevent catastrophic failure [17].

The empirical literature on seafarer mental health has grown substantially, moving from anecdotal accounts to systematic, peer-reviewed research. Oldenburg et al. [8] characterized seafaring as an occupation carrying distinct physical and psychological risks arising from social isolation, confinement, noise and vibration exposure, and long, irregular working hours. Brooks and Greenberg [9], in a systematic review of sixty-three studies (2012–2021), found consistent risk factors for poor seafarer mental health, including younger age, being single, poor physical health, feeling unsafe, high job demands, long working hours, poor team cohesion, poor social support and scheduling uncertainty. Several of these factors, feeling unsafe and scheduling uncertainty in particular, map closely onto the conditions produced by operating in or being rerouted because of an active conflict zone. Pauksztat et al. [18], studying 622 seafarers during COVID-19, found that pandemic-related disruption increased fatigue and mental-health problems, but this impact was significantly mitigated by onboard peer support, external family support and reliable internet access, using structural equation modelling to demonstrate these protective effects statistically. Slišković [19], using qualitative interviews, similarly found Croatian seafarers describing a compounding of ordinary occupational stressors by pandemic-era uncertainty about repatriation and contract extension, with reports of feeling abandoned and powerless.

Fatigue is treated in the literature both as an outcome and as a mediating variable between stressful working conditions and unsafe decision-making. Hystad et al. [20] found that safety climate and psychosocial work environment predicted fatigue in offshore-resupply seafarers independently of shift arrangement alone, implying that fatigue is not purely a function of hours worked. Zhao et al. [21], comparing four shipping companies across China and Europe, found that perceived fatigue and fatigue-mitigation approaches varied considerably even under comparable regulatory regimes, implying that company-level policy and culture, not regulation alone, substantially determine fatigue outcomes. Chung et al. [22] found that burnout among seafarers had measurable antecedents in organizational and job-related stressors and was associated with a higher incidence of safety-relevant incidents at sea, an empirical link that motivates this study’s interest in whether geopolitical-risk exposure carries analogous consequences.

The literature most directly analogous to the Red Sea context, in terms of the nature of the external threat, is the body of research on maritime piracy. Seyle et al. [23], surveying 101 former hostages and 363 seafarers without known piracy exposure across India, the Philippines and Ukraine, found using clinically validated scales that 25.8% of former hostages showed symptoms consistent with post-traumatic stress disorder, and that traumatic exposure measurably affected subsequent career decisions, including willingness to accept further high-risk contracts. Aleksandrov et al. [24], examining Bulgarian seafarers following release from pirate captivity, reported comparable psychiatric sequelae using structured clinical assessment. Doyle et al. [25], studying crew across 51 ships, found measurable variation in psychological resilience associated with identifiable organizational and individual factors rather than being randomly distributed, supporting the assumption that resilience, like fatigue and distress, is a variable that can be meaningfully measured. None of this literature was collected in the context of state-linked, sustained maritime conflict of the kind the Red Sea crisis represents, as distinct from episodic criminal piracy; whether a comparable order of psychological effect applies to seafarers exposed to missile and drone attack from an organized armed group is an open empirical question that this study’s primary data collection was designed to address.

Formal training in crisis response is mandated internationally through the STCW Convention’s Crisis Management and Human Behaviour provisions, which require designated crew to be trained in organizing shipboard emergency procedures and controlling the response to emergencies, historically centred on fire, man-overboard and abandon-ship scenarios [26]. Together with the welfare and repatriation protections in the Maritime Labour Convention, 2006 [27], this constitutes the regulatory baseline against which this study’s findings are assessed. Hetherington et al. [16] noted that training effectiveness in the sector has historically been evaluated more often through completion and certification rates than through measured psychological or behavioural outcomes, and the content and delivery of STCW crisis-management training were designed principally around passenger-ship emergencies rather than sustained exposure to external armed threat. An assessment of training adequacy from the perspective of those who have experienced such threat appears not yet to exist in the peer-reviewed literature; this study’s third research question was designed specifically to address that gap.

1.3. Conceptual Framework

This study connects the geopolitical-risk-and-resilience literature to the seafarer-wellbeing literature using the logic of the dynamic capabilities view [28], which holds that an organization’s capacity to sense threats, seize opportunities and reconfigure its resources under changing conditions determines its ability to sustain operational continuity through environmental turbulence. Geopolitical risk is treated as a form of environmental turbulence, and the firm’s response to it is a test of its dynamic capabilities. Geopolitical-risk exposure (operationalized through voyage routing, threat exposure and rerouting history) is proposed to affect seafarer fatigue and psychological distress (the first mediating variable). This effect is proposed to be moderated by organizational and peer support and by the adequacy of crisis-management training (the second mediating variable). These, in turn, are proposed to affect seafarers’ willingness and capacity to operate safely and to remain in high-risk service, the outcome variable most directly relevant to operational resilience, consistent with the burnout–incidents link reported by Chung et al. [22].

This framework does not claim to test or revise Bednarski et al.’s [1] structural model of geopolitical supply-chain disruption directly; rather, it proposes that the model’s account of firm-level resilience is incomplete without an explicit human-factor pathway, and it uses primary seafarer data to provide exploratory, indicative evidence on the early segments of that pathway (exposure, perceived training adequacy and perceived support). The downstream links to operational resilience, actual retention and safety performance are not measured in this study and are treated throughout as theoretical propositions. The pathway from external shock to seafarer wellbeing is itself likely moderated by voyage-specific operational and commercial context: Gavalas et al. [2] demonstrated that dry-bulk and shipping stock markets reacted negatively to the COVID-19 pandemic’s declaration while the tanker market exhibited a “pandemic-proof” reaction driven by a surge in floating storage, suggesting that vessel type, cargo and trading route condition how an external shock is experienced. Vessel type is reported in the sample profile (Section 3.1), although subgroup sizes did not permit vessel-type-specific comparisons.

1.4. Aim, Objectives and Research Questions

The aim of this study is to examine how geopolitical risk in the maritime supply chain, with particular reference to the Red Sea crisis, relates to seafarers’ perceived crisis-management training adequacy, perceived support and self-reported wellbeing indicators, as well as to explore how these human-factor effects may in turn condition the operational resilience of shipping companies. This aim is pursued through four objectives: (1) to synthesize the existing evidence on seafarer mental health, fatigue and trauma, and on geopolitical risk in global supply chains, into a single conceptual framework; (2) to explore seafarers’ self-reported security-related preoccupation, sleep or concentration difficulty and perceived safety when operating in, or being rerouted around, high-risk geopolitical zones, using primary data from serving crew; (3) to evaluate the adequacy of current crisis-management and psychological-resilience training, both STCW-mandated and company-specific, from the perspective of seafarers with direct experience of high-risk transits; and (4) to translate the findings into practical recommendations for shipping companies’ human-resource, training and welfare policy and into conclusions relevant to the wider academic debate on supply-chain resilience. Because validated fatigue and distress scales could not be administered (Section 2.3), mental health and fatigue are addressed in this study only through single-item indicators, and the empirical focus falls on perceived training adequacy and support.

These objectives are addressed through four research questions. RQ1: How do seafarers operating in, or rerouted because of, high-risk geopolitical areas describe their psychological stress and fatigue, and how does this compare with the established occupational-health literature on seafarer wellbeing? RQ2: Which factors (rank, nationality, voyage length, family contact, peer and company support, prior exposure to piracy or conflict) predict or mitigate this stress and fatigue? RQ3: How do seafarers assess the adequacy of current crisis-management and resilience training for geopolitically volatile conditions, and what gaps do they identify? RQ4: What relationship, as perceived by seafarers, exists between crew wellbeing and training on the one hand and the operational resilience of the wider supply chain on the other?

The academic contribution of this study is to extend an established seafarer-wellbeing research tradition into a context, an organized, state-linked geopolitical threat to commercial shipping, that has not yet been systematically studied, and to use the resulting evidence to argue for a more explicit incorporation of the human factor into supply-chain-resilience theory. Access to serving seafarers through a researcher embedded within a shipping company allows this study to gather primary data of a kind that most academic studies of seafarer wellbeing, reliant on convenience samples recruited through unions, ports or online panels, are unable to obtain. At a policy level, the findings may inform ongoing discussion at the International Maritime Organization and International Labour Organization on how the STCW Convention and the Maritime Labour Convention, 2006, should adapt to a period in which armed threats to commercial shipping appear to be recurring rather than exceptional.

2. Materials and Methods

2.1. Research Philosophy and Design

This study adopts a pragmatist research philosophy, which provides a flexible foundation for the mixed-methods design most appropriate to the research questions [29]. Pragmatism treats knowledge as generated through an ongoing process of inquiry, with the value of a method judged by its fit to the question at hand rather than by allegiance to a single ontological commitment; this is well suited to research questions that require both measurement (to establish the scale and distribution of a phenomenon) and interpretation (to understand its meaning to those who experience it), a rationale also employed by Ponomarov and Holcomb [11] in prior supply-chain-resilience research.

The study was originally designed as an embedded, qualitative-priority mixed-methods project, in which a pilot-scale questionnaire would be followed by semi-structured interviews with a subset of surveyed seafarers and with shore-based crewing and HSSE (Health, Safety, Security and Environment) managers. The interview phase was not carried out: no interviews were conducted with seafarers or with shore-based managers, and no interview data are reported or analysed in this article. The design as actually implemented is therefore a single-phase, cross-sectional online survey that combined closed-ended five-point items with two open-ended questions. Quantitative and qualitative evidence are both drawn from this one instrument, and the open-text responses are used to contextualize and help interpret the closed-item results rather than as a separate qualitative strand. The questionnaire data are treated as descriptive and exploratory; the sample is not powered for inferential generalization to the wider seafaring population, and no such claims are made.

2.2. Population and Sampling

The target population was seafarers currently or recently serving on vessels operated by multiple shipping companies; the companies are not named for confidentiality. A purposive sampling strategy was used, with a planned target of up to fifty questionnaire respondents, drawn to include variation in rank (senior officers, junior officers, ratings), nationality, and recent voyage routing (Red Sea transit, Cape of Good Hope diversion, or neither). In the event, 64 usable responses were received (Section 3.1). This is explicitly a pilot-scale, multi-company sample, treated as generating indicative, hypothesis-generating findings and a field-tested data-collection instrument suitable for a larger, multi-company follow-up study, rather than population-level estimates.

2.3. Instruments

The questionnaire was designed to combine the fifteen-item Occupational Fatigue Exhaustion Recovery scale (OFER-15) [30] and the twelve-item General Health Questionnaire (GHQ-12) [31] with a bespoke module on geopolitical-risk exposure (recent voyage routing, direct or witnessed exposure to a security incident, pre-departure risk information and understanding of security-incident procedures) and perceived training adequacy, including self-rated confidence in crisis-management training for an armed-threat scenario as distinct from the fire/evacuation scenarios on which STCW training traditionally focuses. Due to a licensing issue that emerged after the design stage and was not anticipated at planning, the questionnaire as actually administered replaced the OFER-15 and GHQ-12 with single bespoke items covering, respectively, frequency of thinking about incident risk and sleep/concentration difficulty. This substitution, and its consequences for the comparability of the present findings with the published literature, is addressed directly in Section 4.6 (Limitations) and is disclosed here rather than left implicit, consistent with the transparency expected of empirical reporting. All attitudinal items used a five-point response scale; background and exposure items used categorical response options. The full wording of every item as administered is reproduced in Appendix A.

Geopolitical-risk exposure was measured with three items (Appendix A, items 7–9). Item 7 asked: “In the past 12 months, have you served on a vessel that transited, or was scheduled to transit, the Red Sea/Gulf of Aden corridor?” (Yes/No/Not sure). Item 8 asked: “In the past 12 months, have you served on a vessel that was rerouted via the Cape of Good Hope specifically because of security risk?” (Yes/No/Not sure). Item 9 asked: “Have you personally witnessed, or been on a vessel that experienced, a security incident (e.g., attack, near-miss, distress call from another vessel, armed escort) in the past 12 months?” (Yes, directly involved/Yes, witnessed or nearby/No/Prefer not to say). Each item was coded 1 for any “Yes” response and 0 otherwise; “Not sure” (one response, to item 8) and “Prefer not to say” (one response, to item 9) were coded 0, because neither affirms exposure. Two exposure variables were then derived. Combined geopolitical-risk exposure was coded 1 if a respondent answered “Yes” to at least one of the three items, so that respondents reporting more than one form of exposure were counted once. Incident exposure was coded 1 if a respondent gave either “Yes” response to item 9. The two variables are therefore nested rather than mutually exclusive: every incident-exposed respondent is also classified as exposed under the combined definition, and the two sets of comparisons in Section 3.4 are not statistically independent. The overlap between exposure types is reported in Section 3.2.

The two single items that replaced the validated scales read: “How often do you think about the possibility of a security incident while on a voyage through a high-risk area?” (1 = Never to 5 = Always) and “I found it difficult to sleep or to concentrate on my work due to security concerns” (1 = Strongly disagree to 5 = Strongly agree). Neither item has established psychometric properties. Single-item measures can be acceptable for narrowly defined, concrete constructs [32,33], but they cannot capture the multidimensional constructs of occupational fatigue and recovery measured by the OFER-15 or of psychological distress measured by the GHQ-12; their reliability cannot be estimated from a single administration; and they provide no basis for clinical caseness or for comparison with published norms. In this article they are therefore interpreted only as indicators of security-related preoccupation and of self-reported sleep or concentration difficulty, not as measures of fatigue or mental health, and all results based on them are treated as indicative.

A semi-structured interview guide for seafarers and shore-based managers was also prepared at the design stage, and the questionnaire invited respondents to volunteer for a follow-up interview. As explained in Section 2.1, the interview phase was not carried out; the guide is therefore not described further. Optional contact details provided by volunteers were separated from the response data before analysis and were not used.

2.4. Data Collection and Analysis

The questionnaire was administered online through an anonymous, generic link (Google Forms; Google LLC, Mountain View, CA, USA; https://forms.google.com) between 2 and 8 August 2026, distributed through crewing channels so that non-participation would not be visible to the researcher. The first page set out the purpose of the study, the voluntary and anonymous nature of participation, and the right to skip questions or withdraw, and continuing to the questionnaire constituted consent. All 64 submissions received were retained for analysis. There were no missing values on the closed attitudinal items; one respondent did not report vessel type, and some open-text questions were left blank.

Given the pilot scale of the sample, closed-ended items are analysed primarily descriptively, reporting means, standard deviations and the proportions of low (1–2) and high (4–5) responses, and comparing them, where meaningful, with published seafarer samples [18,20]. Exposed and non-exposed respondents are compared with two-sided Mann–Whitney U tests, in preference to regression or structural-equation modelling, which would require a substantially larger sample to produce stable estimates. p-values are based on the normal approximation with correction for ties and continuity. For each comparison we report the U statistic, the standardized test statistic Z and the effect size r = |Z|/√N, interpreted against the conventional benchmarks of approximately 0.1 (small), 0.3 (medium) and 0.5 (large) [34]. Because ten comparisons are reported (five for each of the two exposure definitions), the Holm sequential procedure [35] was applied within each family of five tests to control the family-wise error rate, and the effect of applying it across all ten tests is also reported. Consistent with the exploratory, hypothesis-generating purpose of the study [36], unadjusted p-values are shown for transparency, but only results that remain significant after adjustment are described as statistically significant, and all inferential results are interpreted as indicative. Calculations were performed in Microsoft Excel (Microsoft 365; Microsoft Corporation, Redmond, WA, USA) and independently re-computed in Python (Python Software Foundation, Wilmington, DE, USA; https://www.python.org) using SciPy version 1.18.1.

The two open-ended questions were analysed using inductive qualitative content analysis [37], following the initial phases of Braun and Clarke’s [38] approach (familiarization, initial coding, and grouping codes into themes). Because most responses were brief (typically a phrase or a single sentence), the analysis is reported as category frequencies rather than as a fully reflexive thematic narrative. The first author (C.K.) read all responses to both questions, generated initial codes inductively from the response text, and grouped them into seven descriptive categories (see Section 3.5). Each response was assigned to every category it expressed, so a single response could contribute to more than one category; responses with no substantive content (blank, “N/a”, a single punctuation mark or unintelligible text) were left uncoded but retained in the denominator of 64. The second author (D.G.) reviewed the category definitions and every coding assignment against the raw response text; disagreements were resolved by discussion, and the agreed coding was recorded in the analysis workbook. Formal inter-coder reliability statistics were not calculated, which is noted as a limitation (Section 4.6).

2.5. Validity, Reliability and Trustworthiness

Because the fatigue and distress scales could not be administered, the questionnaire contains no multi-item scales; internal-consistency reliability (Cronbach’s alpha) is therefore not applicable, and each closed item is analysed separately. Content validity of the bespoke items is supported by their grounding in the literature reviewed in Section 1, but their construct validity has not been established (Section 2.3). For the open-text analysis, dependability and confirmability [39] were supported by second-author review of all coding decisions and by retaining the coding record alongside the raw responses. Because interviews were not conducted, the member-checking and interview-based triangulation envisaged at the design stage were not possible. Integration of the closed-item and open-text evidence is achieved by reading the comparisons in Section 3.3 and Section 3.4 alongside the categories in Section 3.5, and it is summarized against each research question in Section 3.6.

3. Results

3.1. Sample Characteristics and Response Profile

The questionnaire yielded 64 usable responses, exceeding the planned pilot-scale target of up to fifty. This higher-than-expected response may reflect the high level of interest and concern among seafarers regarding the Red Sea crisis. Table 1 summarizes the sample’s rank, sea-experience and nationality profile. The sample is weighted towards senior and junior officers (36 and 20 respondents, respectively), with a smaller group of 8 ratings and other roles, a distribution typical of seafarer samples. Sea experience is concentrated in the 11–20-year band (29 respondents, 45%). Nationality is dominated by Ukrainian (33, 52%) and Greek (19, 30%) respondents, with smaller Filipino (9, 14%), Russian and Chinese (3, 5%) groups. Vessel type is dominated by container (37, 58%) and tanker (22, 34%) vessels; one respondent did not report vessel type.

Table 1. Sample profile (n = 64). Percentages are rounded and may not sum to 100.

Characteristic	Category	n	%
Rank	Senior Officer (Master/CO/CE/2E)	36	56%
	Junior Officer (Deck/Engine)	20	31%
	Rating/Cadet/Other	8	13%
Sea experience	Under 2 years	6	9%
	2–10 years	21	33%
	11–20 years	29	45%
	Over 20 years	8	13%
Nationality	Ukrainian	33	52%
	Greek	19	30%
	Filipino	9	14%
	Russian/Chinese	3	5%
Vessel type	Container	37	58%
	Tanker	22	34%
	Other (bulk, LNG, offshore, cruise)	4	6%
	Not reported	1	2%

CO, Chief Officer; CE, Chief Engineer; 2E, Second Engineer; LNG, liquefied natural gas.

Over half the sample is Ukrainian, a nationality for whom “geopolitical risk exposure” plausibly includes proximity to the war in Ukraine and Black Sea shipping risk, which the questionnaire did not ask about directly, in addition to or instead of the Red Sea/Gulf of Aden risk the instrument was designed around. This is noted as a data-interpretation caveat carried forward into Section 4, since the questionnaire cannot distinguish the two sources of risk exposure retrospectively.

3.2. Geopolitical Risk Exposure in the Sample

To operationalize the key independent variable, respondents were asked about recent voyages and security incidents. Fourteen respondents (22%) reported serving on a vessel that transited or was scheduled to transit the Red Sea/Gulf of Aden corridor in the past 12 months. A larger group, 31 respondents (48%), reported a Cape of Good Hope diversion specifically attributed to security risk. Seventeen respondents (27%) reported personally witnessing or being directly involved in a security incident, of whom 5 reported direct involvement. Combining these three indicators, 42 respondents (66%) met the study’s working definition of geopolitical-risk exposure, while 22 (34%) did not. The three forms of exposure overlapped: 8 respondents reported both a Red Sea transit and a Cape diversion, and 12 of the 17 incident-exposed respondents also reported route or rerouting exposure. Under the combined definition (Section 2.3), the 42 exposed respondents therefore comprise 25 with route or rerouting exposure only, 12 with both route or rerouting and incident exposure, and 5 with incident exposure only.

3.3. Wellbeing, Support and Training-Confidence Descriptions

Table 2 reports the mean, standard deviation and the proportions of low (1–2) and high (4–5) responses for each five-point item. Confidence in conventional fire/evacuation training is high and narrowly distributed (mean 4.38, SD 0.81); confidence in responding to an armed threat is markedly lower and more dispersed (mean 3.67, SD 1.21), with 19% of respondents rating this confidence low (1–2) against only 3% for fire/evacuation. Perceived adequacy of war-risk/hardship compensation is the lowest-rated item (mean 3.39, SD 1.36), with 23% rating it low against 44% rating it high, indicating a genuinely divisive issue rather than a broadly shared complaint. Mean scores for having someone to talk to (3.94), the captain taking concerns seriously (4.39) and comfort in raising an inability to continue (3.95) suggest that the company’s welfare and support infrastructure is perceived as fairly effective by the average respondent; the high score for family contact helping coping (4.39) underscores the importance of communication with home, consistent with the COVID-19 pandemic literature [18].

Table 2. Descriptive statistics, all respondents (n = 64), 5-point scale. * For the sleep/concentration item, a low score (1–2) indicates disagreement (no reported difficulty) and a high score (4–5) indicates agreement (reported difficulty).

Item	Mean	SD	% Low (1–2)	% High (4–5)
Given clear pre-departure risk information	4.28	1.13	9%	84%
Understood procedure for security incident	4.52	0.87	5%	89%
War-risk/hardship compensation adequate	3.39	1.36	23%	44%
Frequency of thinking about incident risk	3.50	1.11	17%	45%
Someone on board to talk to if stressed	3.94	1.07	9%	69%
Captain/senior officers take concerns seriously	4.39	0.88	5%	88%
Reliable internet/communication access	3.83	1.20	17%	67%
Family contact helps coping	4.39	0.85	5%	86%
Comfortable raising inability to continue	3.95	1.15	12%	72%
Difficulty sleeping/concentrating (security-related) *	2.70	1.18	41%	25%
Confidence: fire/evacuation training	4.38	0.81	3%	86%
Confidence: armed-threat response training	3.67	1.21	19%	62%
STCW training adequate for conflict-area risk	3.67	1.27	19%	64%

3.4. Differences Between Exposed and Non-Exposed Seafarers

Two-sided Mann–Whitney U tests were used to compare respondents meeting the combined geopolitical-risk-exposure definition (n = 42) against those who did not (n = 22), as well as separately to compare incident-exposed respondents (n = 17) against those reporting no incident (n = 47). Results are summarized in Table 3 and Table 4, which report group means and medians, U, Z, unadjusted p-values, effect sizes (r) and Holm-adjusted p-values within each table.

Table 3. Combined geopolitical-risk exposure (Red Sea transit, Cape of Good Hope diversion or security incident) vs. non-exposed respondents (n = 42 vs. 22). Values are means with medians in parentheses. Z: normal approximation with tie and continuity correction; r = |Z|/√N (N = 64); p Holm: Holm-adjusted within the five tests of this table.

Outcome	Exposed Mean (Median)	Non-Exposed Mean (Median)	U	Z	p	r	p Holm
STCW adequacy for conflict-area risk	3.45 (4)	4.09 (4)	319.5	−2.08	0.037	0.26	0.186
Confidence: armed-threat response	3.52 (4)	3.95 (4)	369.5	−1.35	0.177	0.17	0.710
Frequency of thinking about incident risk	3.62 (4)	3.27 (3)	378.0	1.23	0.219	0.15	0.710
Sleep/concentration difficulty	2.62 (3)	2.86 (3)	418.5	−0.63	0.530	0.08	1.000
Compensation adequacy	3.33 (3)	3.50 (3.5)	432.5	−0.43	0.671	0.05	1.000

STCW, International Convention on Standards of Training, Certification and Watchkeeping for Seafarers.

Table 4. Incident-exposed (witnessed or directly involved) vs. no-incident respondents (n = 17 vs. 47). Values are means with medians in parentheses. Z: normal approximation with tie and continuity correction; r = |Z|/√N (N = 64); p Holm: Holm-adjusted within the five tests of this table.

Outcome	Incident-Exposed Mean (Median)	No-Incident Mean (Median)	U	Z	p	r	p Holm
STCW adequacy for conflict-area risk	3.00 (3)	3.91 (4)	227.0	−2.71	0.007	0.34	0.033
Confidence: fire/evacuation training	4.00 (4)	4.51 (5)	280.5	−2.01	0.045	0.25	0.179
Comfortable raising inability to continue	3.35 (3)	4.17 (4)	281.5	−1.88	0.059	0.24	0.179
Confidence: armed-threat response	3.24 (3)	3.83 (4)	305.5	−1.47	0.141	0.18	0.281
Sleep/concentration difficulty	2.53 (3)	2.77 (3)	356.0	−0.68	0.499	0.08	0.499

STCW, International Convention on Standards of Training, Certification and Watchkeeping for Seafarers.

After adjustment for multiple testing, one result remains statistically significant. Incident-exposed respondents rated STCW crisis-management training as less adequate for conflict-area risk than respondents reporting no incident (mean 3.00 vs. 3.91; U = 227.0, Z = −2.71, p = 0.007, r = 0.34, a medium effect), and this difference remains significant after Holm adjustment within Table 4 (p Holm = 0.033). It does not, however, remain significant when the Holm procedure is applied across all ten comparisons (adjusted p = 0.066), and it should therefore be regarded as a robust indication within this sample rather than a confirmed finding. The corresponding difference under the combined exposure definition (3.45 vs. 4.09; U = 319.5, p = 0.037, r = 0.26) does not remain significant after adjustment (p Holm = 0.186). A supplementary check shows that this combined-definition difference is carried by the incident-exposed respondents: when exposure is defined by Red Sea transit or Cape diversion alone (n = 37 vs. 27), mean STCW-adequacy ratings barely differ (3.59 vs. 3.78, p = 0.467). In these data, therefore, route or rerouting exposure in itself is not associated with lower perceived training adequacy, whereas direct or witnessed experience of a security incident is. Two further differences among incident-exposed respondents, lower confidence in conventional fire/evacuation training (4.00 vs. 4.51, p = 0.045, r = 0.25) and lower comfort in raising an inability to continue a voyage (3.35 vs. 4.17, p = 0.059, r = 0.24), are of small-to-medium magnitude but are not significant after adjustment (p Holm = 0.179 for both); they are reported as patterns that merit testing in a larger sample, not as findings.

A further descriptive pattern, not formally tested given small subgroup sizes but directly relevant to RQ3, is that mean confidence in armed-threat response rises steadily with reported receipt of company-specific (beyond-STCW) training: 2.80 among respondents who had not received such training, 3.20 among those unsure whether they had, and 4.02 among those who had (n = 15, 5 and 44, respectively).

3.5. Thematic Findings from Open-Text Responses

Two open-text items asked what respondents felt was missing from current training, and what would make them feel better prepared for a high-risk voyage. Table 5 reports theme frequencies; themes are not mutually exclusive. The largest theme in response to “what is missing” was that nothing is missing and current training is sufficient (36% of responses), which sits in tension with the statistically significant training-adequacy gap reported in Section 3.4 for exposed and incident-affected respondents specifically; this tension is interpreted in Section 4.1. Where gaps were identified, they clustered around information and situational intelligence rather than procedural or drill-based content. The “what would help you feel prepared” responses were more varied and concrete, splitting across realistic scenario-based training and drills (22%), armed protection such as naval escort (13%), increased or fairer compensation (14%), and avoiding high-risk routing altogether (11%).

Table 5. Thematic coding of open-text responses (themes not mutually exclusive).

Theme	‘What Is Missing’ (n = 64)	‘What Would Help’ (n = 64)
Nothing missing/current training sufficient	23 (36%)	5 (8%)
More practical, scenario-based training/drills	6 (9%)	14 (22%)
Clearer information, briefing or intelligence	11 (17%)	4 (6%)
Armed guard, naval escort or military protection	0 (0%)	8 (13%)
Higher or fairer compensation/hardship pay	1 (2%)	9 (14%)
Right to refuse a high-risk voyage without penalty	2 (3%)	2 (3%)
Avoiding or rerouting away from high-risk areas entirely	0 (0%)	7 (11%)

A small number of responses departed from this practical register and raised the issue in more fundamental terms. One respondent wrote that the central issue was not training at all but “whether a seafarer has a genuine right to refuse a high-risk voyage without pressure or career consequences.” Another, asked what would help, wrote that nothing would: “there is nothing to make you feel better… you are feeling the fear of being in a war that is not your own.” A third respondent, whose ratings across the questionnaire were the most negative in the sample, wrote that seafarers are “not army candidates” and asked that the company “stop treating us as paid missionaries.” These responses are a small minority of the sample but are analytically important because they resist the practical, trainable framing that both the questionnaire and most other responses adopt; they point to a deeper issue of consent and the psychological contract, discussed in Section 4.2.

3.6. Summary of Findings Against the Research Questions

Findings relevant to RQ1 are limited by the instrument deviation noted in Section 2.3: the single-item indicator shows variation in self-reported, security-related sleep or concentration difficulty (25% agreement, 41% disagreement) that was not associated with route or incident exposure in this sample, but it cannot be read as a measure of fatigue or psychological distress. Findings relevant to RQ2 are correspondingly limited; the clearest correlate identified is not of fatigue but of perceived training adequacy (RQ3). Findings relevant to RQ3 are the strongest in the data set: direct or witnessed incident exposure is associated with lower perceived adequacy of STCW crisis-management training for conflict-area risk (the only difference that remains significant after within-family adjustment), whereas route or rerouting exposure alone is not, and open-text responses independently identify situational information and realistic drills as the specific gap. RQ4 concerns perceived relationships that the study does not measure directly: the minority of responses questioning seafarers’ right to decline high-risk voyages, and the unadjusted pattern of lower comfort in raising an inability to continue among incident-exposed respondents, are consistent with the retention and willingness-to-serve pathway proposed in Section 1.3 but do not test it.

4. Discussion

4.1. Interpretation in Light of the Seafarer Mental-Health and Fatigue Literature

The clearest finding, that seafarers who have witnessed or been directly involved in a security incident rate their crisis-management training as less adequate for conflict-area risk than other seafarers, is consistent with, and extends, the risk-factor evidence reviewed in Section 1.2. Brooks and Greenberg’s [9] systematic review identified “feeling unsafe” and “scheduling uncertainty” as established predictors of poor seafarer mental health; the present findings suggest that direct or witnessed experience of a specific geopolitical threat, rather than exposure to the rerouting it causes, sharpens seafarers’ own assessment of whether their training has kept pace with that threat, even where confidence in more conventional emergency training (fire, evacuation) remains largely intact. This is a more targeted finding than the general wellbeing literature offers: the training-adequacy gap identified here is threat-specific rather than a general erosion of confidence in shipboard safety systems.

The apparent tension noted in Section 3.5, that 36% of open-text respondents said nothing was missing from their training while incident-exposed respondents rated STCW adequacy lower than other respondents, is not a contradiction but a reflection of different framings of the question. Many seafarers, especially in the non-exposed group, may feel that their general training is sound; when the question is made specific, “for the specific risks of transiting a conflict-affected area,” those who have actually faced that risk apply a more demanding benchmark. Individuals often rate their own competence and training quality as high in the abstract but become more critical when asked about a specific, high-risk scenario [15]; the occupational culture of seafaring, which emphasizes competence, self-reliance and a “can-do” attitude, likely inhibits the open expression of vulnerability in a general survey item [16], and the more specific, anonymous quantitative item may better penetrate this cultural barrier.

On trauma and threat exposure specifically, the piracy literature [23,24] reported clinically assessed post-traumatic symptom rates of around one in four among former hostages, using validated clinical instruments. The present data cannot speak to clinical caseness given the instrument limitation discussed in Section 4.6, but the pattern of lower comfort in raising an inability to continue a voyage among incident-exposed respondents, although not significant after adjustment for multiple testing, is directionally consistent with the piracy literature’s finding that traumatic exposure affects subsequent willingness to accept risk, suggesting the psychological effects observed in the piracy context may also be present, in some form, in a state-linked-conflict context.

4.2. Interpretation in Light of Geopolitical-Risk and Supply-Chain-Resilience Theory

Section 1.1 argued that the current geopolitical-supply-chain-disruption literature, including its most developed recent statement [1], theorizes resilience at the level of the firm, node and network without an explicit account of the workforce operating those nodes. The present findings provide exploratory, indicative support for that argument, although the study does not measure operational resilience, retention or safety performance directly. The training-adequacy gap reported in Section 3.4 can be understood, in the vocabulary of Reason’s [15] human-factors framework, as a latent condition: it would not by itself cause an incident, but would be expected to increase the probability that an active failure occurs when a geopolitical threat materializes. A resilience framework that accounts for chokepoints, rerouting costs and firm-level strategic response [1] but not for this kind of latent condition is, we propose, missing a mechanism through which geopolitical risk could degrade operational performance beyond route disruption alone; this is a theoretical proposition consistent with, but not demonstrated by, the present data.

The minority of qualitative responses questioning seafarers’ right to decline a high-risk voyage, and describing a sense of being treated as “paid missionaries” rather than employees whose consent is genuinely sought, point to a second and distinct mechanism that the structural resilience literature does not currently theorize: workforce retention and willingness-to-serve as a direct input into operational resilience, mediated not by training adequacy but by the perceived fairness of the employment relationship itself. This is well captured by Rousseau’s [40] concept of the psychological contract, the set of unwritten, reciprocal obligations an employee believes exist between themselves and their employer. Where a seafarer perceives that hazard pay, consultation and genuine choice have not kept pace with a unilaterally increased job risk, the psychological contract is breached, with plausible consequences including reduced retention, increased reluctance to accept further high-risk voyages, and erosion of the safety-critical vigilance that human-factors theory says organizational latent conditions depend on. We therefore propose that the human-factor pathway missing from current geopolitical-supply-chain-resilience theory may comprise not one pathway but at least two: a training/competence pathway and a retention/psychological-contract pathway, requiring different organizational responses (Section 4.3).

4.3. Theoretical Contribution: Extending the Conceptual Framework

The conceptual framework proposed in Section 1.3 linked geopolitical-risk exposure to seafarer fatigue and distress, moderated by organizational support and training adequacy, as the mechanism through which geopolitical risk affects operational resilience via willingness and capacity to operate safely. The present findings offer exploratory support for the training-adequacy segment of this framework, principally through the incident-exposure result, but suggest that the framework, as originally specified, under-differentiates the outcome side. Rather than a single outcome variable, the data point to two analytically distinct outcomes: a safety/competence outcome, most directly indicated by the STCW-adequacy result and, more tentatively, by the fire/evacuation-confidence pattern, and a retention/consent outcome, suggested by the (unadjusted) comfort-in-raising-concerns pattern and the qualitative minority discussed in Section 4.2, operating through the psychological-contract mechanism rather than through training. The revised framework therefore adds a second mediating pathway, running from geopolitical-risk exposure through perceived organizational fairness and consultation to retention and willingness-to-serve, in parallel to, rather than replacing, the training-adequacy pathway. This is offered as a refinement to be tested with a larger, multi-company follow-up sample, not as a finding this pilot-scale study can establish definitively. Neither outcome (safe operation or retention) was itself measured: both pathways are propositions derived from seafarers’ perceptions of training and support, not observed relationships with operational resilience.

4.4. Practical Implications for Shipping Companies

Three implications are suggested by the findings, subject to the exploratory status of the evidence. First, the training-adequacy gap and the specific content gap identified in the open-text themes suggest that STCW-mandated crisis-management training should be supplemented for seafarers rostered to high-risk-area transits, targeted at exposed crew rather than delivered uniformly, with content focused on situational information and realistic, scenario-based drills rather than generic compliance training. Second, the descriptive association between receipt of company-specific (beyond-STCW) training and confidence in armed-threat response is a directly testable proposition: a company is well-positioned to pilot such training and re-survey its crew to assess impact. Third, the psychological-contract mechanism suggests that consultation and perceived choice, not only compensation levels, matter for the minority of seafarers whose concerns go beyond training adequacy; a formal, visible, non-punitive mechanism for a seafarer to decline a specific high-risk transit without career consequence is suggested by the data as a response worth piloting and evaluating.

4.5. Policy Implications

At a sectoral and regulatory level, the findings speak most directly to the STCW Convention’s Crisis Management and Human Behaviour provisions [26], designed principally around passenger-ship and fire/evacuation emergencies rather than sustained exposure to external armed threat. The gap in perceived adequacy for conflict-area risk among incident-exposed seafarers, gathered directly from serving crew, is consistent with the argument that current STCW model-course content is due for review, including realistic, scenario-based modules on responding to armed threats and managing the associated psychological stress. On the retention and consent dimension, the findings are also relevant to the welfare and repatriation protections in the Maritime Labour Convention, 2006 [27], which do not currently address a seafarer’s right to decline a specific high-risk voyage; this issue merits attention in future revision cycles of the Convention, or in interim flag-state and industry-body guidance, to provide a consistent floor of protection rather than leaving the matter to individual company discretion.

4.6. Limitations

First, and most significantly, the questionnaire as actually administered replaced the validated OFER-15 fatigue scale and GHQ-12 psychological-distress scale specified in Section 2.3 with single bespoke items, owing to a licensing issue not anticipated at the planning stage. This means the findings on fatigue and psychological distress reported in Section 3 are indicative at best and not comparable with the published seafarer-wellbeing literature in the way the original research design intended; this is the most important limitation of the study and the first priority for follow-up work. Second, the sample’s nationality composition, with over half Ukrainian, means that “geopolitical risk exposure” as measured by this questionnaire may be confounded with proximity to the war in Ukraine and Black Sea shipping risk, which the instrument did not ask about separately from Red Sea/Gulf of Aden exposure; the exposure comparisons in Section 3 should therefore be read as indicative of geopolitical risk broadly rather than as isolating Red Sea risk specifically.

Third, the inferential comparisons are exploratory: ten tests were performed on a pilot-scale sample, only one remains significant after Holm adjustment within its family and none after adjustment across all ten, and the two exposure groupings are nested rather than independent. Fourth, the open-text responses were coded by one author and reviewed by the second without formal inter-coder reliability statistics, and the brevity of the responses limited the depth of thematic interpretation. Fifth, the semi-structured interviews with seafarers and shore-based managers envisaged at the design stage were not conducted, so the study lacks the in-depth and organizational perspective that would have allowed seafarers’ accounts to be triangulated. Sixth, the study measures perceptions of training, support and willingness to raise concerns; it does not measure operational resilience, actual retention or safety performance, and the links to these outcomes discussed in Section 4.2 and Section 4.3 are theoretical propositions. Finally, the cross-sectional, self-report design and purposive recruitment mean that the results may be affected by self-selection and common-method bias and cannot establish causal direction.

5. Conclusions

This exploratory study used survey data, including open-text responses, from 64 serving seafarers during the 2023–2026 Red Sea crisis to examine how geopolitical-risk exposure relates to perceived crisis-management training adequacy, perceived support and self-reported wellbeing indicators. Its central finding is that seafarers who had witnessed or been directly involved in a security incident rated STCW crisis-management training as less adequate for conflict-area risk than other seafarers (r = 0.34; Holm-adjusted p = 0.033), whereas rerouting without an incident was not associated with lower ratings; open-text responses located the gap in situational information and realistic, scenario-based drills. Because fatigue and psychological distress were captured only through single items, the study draws no conclusions about seafarers’ mental health or fatigue.

The study makes three contributions, each offered as hypothesis-generating rather than established. First, it extends seafarer-wellbeing research to state-linked armed threats to commercial shipping, using primary data from an affected workforce. Second, it identifies direct or witnessed incident exposure, rather than rerouting, as the factor associated with a perceived training gap. Third, it proposes that geopolitical risk may affect operational resilience through two distinct human-factor pathways (training/competence and retention/psychological contract), to be tested with validated measures and observed operational outcomes in future work.

For shipping companies, the findings recommend supplementing STCW-mandated training with targeted, scenario-specific briefing content for seafarers rostered to high-risk-area transits; piloting and evaluating a structured, non-punitive mechanism for a seafarer to decline a specific high-risk transit; and conducting a before-and-after evaluation of company-specific training to test its association with armed-threat confidence. For policymakers and industry bodies, the findings recommend treating the STCW Crisis Management and Human Behaviour model course content as due for review specifically for sustained, state-linked armed-threat scenarios, and considering a seafarer’s right to decline a specific high-risk voyage in future revision cycles of the Maritime Labour Convention, 2006, or in interim flag-state and industry-body guidance.

The most immediate priority for future research is a follow-up study using the validated OFER-15 and GHQ-12 scales as originally specified, once the licensing issue identified in Section 4.6 is resolved, to establish whether the training-adequacy findings reported here extend to measurable differences in fatigue and psychological distress. Further priorities include a multi-company, larger-sample study to test these pilot-scale findings for generalizability and to disentangle Red Sea-specific exposure from other sources of geopolitical risk, including the war in Ukraine; a programme of semi-structured qualitative follow-up interviews with seafarers and shore-based managers (as envisaged but not conducted in this study) to develop the psychological-contract and right-to-refuse themes in greater depth; a pre-and-post evaluation of a company-specific, scenario-based training intervention; further investigation of the protective role of reliable internet and communication access; studies linking seafarer-level measures to observed operational outcomes, such as retention, incident and near-miss records; and a longitudinal design tracking the same cohort of seafarers across multiple voyages and phases of geopolitical tension and de-escalation.