Somatic dominance in clinical presentation

One striking and consistent finding was the near-complete absence of canonical psychological symptoms of depression in patients’ presenting concerns – such as low mood, anhedonia or cognitive complaints. Rather, depressive severity was overwhelmingly associated with physical symptoms. This reinforces the notion that, in this population, depression is primarily expressed and communicated through somatic idioms rather than via psychological language.

Among all symptoms, epigastric pain emerged as the most consistent and prominent correlate of depression across severity levels. Prior studies have demonstrated an association between depression and upper gastrointestinal symptoms, including epigastric discomfort, ^{Reference Jamil, Sarwar and Hussain24} supporting the interpretation that epigastric pain may function as a culturally salient marker of psychological distress. In this context, its recurrence across moderate-to-severe depression suggests that it may represent a dominant idiom of distress rather than coincidental comorbidity.

Constipation was the second most notable gastrointestinal symptom. Emerging literature suggests that constipation may act not only as a comorbid condition but also as an independent risk factor or prodromal feature of depression, possibly mediated through gut–brain axis dysregulation. ^{Reference Wang, Shen, Wang and Jia25,Reference Yun, Wang and Chen26} These findings underscore the importance of bidirectional screening: clinicians should assess mood symptoms in patients presenting with constipation, while also proactively evaluating bowel habits and medication use in patients diagnosed with depression.

Pain, fever and myalgias as markers of severity

Although bodily aches and myalgias were common across the cohort, their clinical significance varied by severity. In severe depression, bodily aches frequently occurred in the absence of fever, suggesting a direct association with depressive pathology rather than intercurrent infection. By contrast, patients with moderate and moderately severe depression often presented with fever alongside myalgia, probably prompting medical evaluation during which depression was subsequently identified through screening.

Exploratory overlap analyses indicated that bodily aches and myalgias maintained an independent association with depression, rather than representing secondary manifestations of febrile illness. Figure 8 reflects this distinction, which is clinically important because pain syndromes without objective inflammatory findings may serve as indicators of more advanced depressive illness in medical settings. ^{Reference Garcia-Cebrian, Gandhi, Demyttenaere and Peveler27}

Fig. 8

Overlap analyses of bodily aches and fever.

Progression of somatic burden with increasing severity

Symptom patterns showed a graded progression with increasing depression severity. Moderate depression was characterised by a limited cluster of somatic symptoms – epigastric pain, myalgia, fever, shortness of breath and weakness – suggesting that early depressive states present primarily as diffuse bodily and autonomic discomfort. With moderately severe depression, the somatic profile expanded to include headaches, abdominal pain and constipation, indicating increasing physiological burden. Severe depression was associated with a further accumulation of symptoms, most notably epigastric pain, weakness, headache, constipation and back pain, consistent with cumulative somatic load and neurovegetative dysregulation.

Although affective symptoms such as anhedonia, crying spells and suicidal ideation became more evident in severe depression, physical complaints remained predominant across all severity levels, underscoring the centrality of somatisation even in advanced illness.

Cultural context and somatisation

This pattern is consistent with prior studies from Pakistan and neighbouring regions, where psychological distress is often expressed through somatic symptoms. Cultural norms, stigma and limited mental health literacy probably contribute to this presentation, and the persistent prominence of epigastric pain may reflect a culturally mediated idiom of distress rather than a disorder-specific symptom.

Methodological contribution

Methodologically, this study employed a mixed-methods approach integrating NLP-based feature extraction with qualitative visualisation (word clouds and heatmaps) and quantitative modelling. Similar triangulated approaches have been used in prior work to contextualise symptom expression and semantic patterns in mental health research. ^{Reference Chen, Ngai, Cheng and Hu28} This design allowed us to quantify symptom patterns while preserving interpretive insight into how depression is expressed and documented in routine clinical encounters.

Limitations

Our study has several limitations. Although data were drawn from a large tertiary care hospital, the quality of clinical documentation was variable. PHQ-9 scores were missing for a subset of patient encounters and were thus predicted using predefined symptom mappings, which may have introduced measurement bias. Incomplete demographic or prescription data led to exclusion of some patient visits. The sample was gender-imbalanced, with a higher proportion of women, potentially limiting generalisability to male patients. Our study was conducted in a busy public hospital medical clinic in Pakistan, where attendance at healthcare services can itself be a source of distress. In this context, mild depressive symptoms (PHQ-9 scores of 5–9), which are both highly prevalent and more difficult to distinguish from background distress (i.e. a lower signal-to-noise ratio), were excluded. Additionally, the relatively small number of patients with severe depression reduced statistical power for comparisons across severity strata. Finally, our study focused on anxiety as a driver for depression and did not explore its clinical presentation independently. Among our study patients, 49% had moderate or severe anxiety (28 and 24%, respectively) and 17% did not have a GAD-7 score. We suggest that the wide confidence interval associated with anxiety warrants further focused study of anxiety in our population.

Implications and future directions

Future research should aim to incorporate the full spectrum of depressive symptom severity, with more balanced samples of the extremes of the spectrum (i.e. mild and severe PHQ-9 scores, respectively). The inclusion of these cases in larger data-sets would allow for more nuanced stratification. Additionally, although moderate depression was not excluded in this study, grouping of moderate and severe categories in certain analyses due to the relatively small number of severe cases may have obscured important clinical distinctions; future studies with larger sample sizes should examine these categories separately. Another important methodological consideration is the handling of repeated measures. Due to limitations in patient identification, repeated visits by the same individuals may have been treated as independent observations. Future work should adopt longitudinal or clustered analytical approaches to account for within-patient correlation. Encouragingly, as a part of our work, there has been successful implementation of unified hospital management systems across tertiary care hospitals in Punjab, which now enables reliable patient-tracking across visits, making such analyses feasible. Finally, larger, multicentre studies conducted in both tertiary and primary care general medical settings would provide more representative data and allow for a more granular understanding of how depression manifests in the Pakistani population.

Pakistan has approximately 0.19 psychiatrists per 100 000 population, far below World Health Organization recommendations. ²⁹ Given this constraint, most uncomplicated depression must be managed in primary care settings. Our findings suggest that routine depression screening should be implemented for patients presenting with unexplained somatic complaints – particularly gastrointestinal symptoms and pain syndromes – in general medical clinics. Validated, culturally adapted Urdu versions of PHQ-9 are available and should be incorporated into routine care. ^{Reference Ahmad, Hussain, Akhtar and Shah30}

With smartphone penetration exceeding 50–60% of the population, digital screening represents a feasible and scalable strategy. ³¹ Integration of PHQ-9 delivery at clinic check-in – potentially linked to national identity registration – could facilitate early detection. Strengthening such systems will require training primary-care clinicians in recognising and managing depression, an approach previously shown to improve outcomes in low-resource settings. Ultimately, culturally informed, mind–body-oriented models of care are essential to address the substantial burden of depression in Pakistan.