Distinguishing core identities

Khawaja Sira: This is the local, traditional, and culturally relevant umbrella term used in Pakistan to describe a recognised third-gender identity. The Khawaja Sira community generally includes individuals who identify as third gender, those who are transgender, and sometimes those with intersex variations, functioning within a specific kinship system that has deep historical roots in the subcontinent [8]. Although the Pakistani government, through the 2018 Transgender Persons (Protection of Rights) Act, has granted legal recognition to transgender individuals, the use of broad Western terminology often neglects the nuanced, deeply rooted social roles and historical marginalisation linked to the Khawaja Sira identity.

Intersex refers to individuals born with variations in sex characteristics, including chromosomal patterns, gonads, endocrine systems, or genitals, that do not fit the typical binary definitions of male or female. These variations, or Differences in Sex Development (DSD), are fundamentally biological [9].

Transgender describes individuals whose gender identity differs from the sex they were assigned at birth. Transgender identity is centred on psychological identity and personal experience, a distinction that must be maintained from intersex variations [10].

Ethics statement

The Institutional Review Board, Health Services Academy, approved the research No. 7–82/IERC-HSA/2022-16 The study follows the principles outlined in the Declaration of Helsinki. All participants were informed of the research aims and objectives. Informed consent was sought from all the participants. This was done with verbal consent, and where possible, written consent was sought to ensure that all participants fully understood that the data collection was only for research purposes.

The security measures for the collected information and protocols to preserve anonymity and confidentiality were observed. No names or identifiable details were included, and the participants were deidentified in the final dataset to protect their anonymity. All participants were informed that they could refuse or withdraw from further participation at any time, without the need to provide a reason.

Sample size and population estimates

The sample size calculation was based on the 2017 Pakistan Population Census figure of 133 transgender individuals in Islamabad [11]. However, we acknowledge this represents a significant underestimation. Community-based organizations (CBOs) in Islamabad and Rawalpindi report registration numbers exceeding 13,000 transgender women. Using the census figure to calculate our target sample (n = 99, based on 5% margin of error and 95% confidence level using RaoSoft calculator) compromises representativeness and limits generalizability.

We recognize that consultation with local CBOs and NGOs would have provided more realistic population estimates and enabled more representative sampling. This limitation affects the external validity of our findings.

Recruitment strategy

We recruited 91 participants through snowball sampling initiated via community leaders (“Gurus”). While this approach facilitated trust and access to a hidden population (response rate: 91.9%), it introduces substantial selection bias. Participants recruited through Gurus likely belong to more connected networks, potentially excluding isolated individuals or those unaffiliated with any Guru structure.

Methodological transparency: Snowball sampling does not minimize selection bias but rather manages access barriers to a stigmatized population. To reduce clustering effects, we:

• Limited referrals to maximum 3 per participant
• Recruited from 5 different Guru networks across Islamabad
• Conducted interviews in private settings in Urdu to reduce social desirability bias
• Trained interviewers in culturally sensitive communication

Alternative approaches considered: While targeted venue-based or time-location sampling could have diversified recruitment, the absence of consistent gathering venues and security concerns limited feasibility.

Survey instrument

We employed two validated instruments:

1. Hopkins Symptom Checklist (HSCL-25): 25-item scale measuring anxiety (10 items) and depression (15 items) with 4-point response scale (1 = “not at all” to 4=”extremely”). Internal consistency in our sample: overall α = 0.92; anxiety α = 0.83; depression α = 0.88. [12,13]
2. Health Protective Behaviour Scale (HPBS): 32-item scale across five dimensions (interpersonal support, general behaviour, self-knowledge, nutrition, healthcare) with 5-point response scale. Internal consistency: α = 0.89. [14]

Translation process: Instruments underwent forward translation (English to Urdu) by an independent bilingual translator, followed by back-translation by a second translator. Discrepancies were resolved through consensus.

Pilot testing: Questionnaires were pilot-tested with 10 participants (excluded from final sample) from similar community backgrounds. Based on feedback:

• Simplified 3 items with complex medical terminology
• Added visual aids for health awareness questions
• Confirmed comprehension for participants with low literacy

Cultural adaptation limitations: While translation ensured linguistic accuracy, we acknowledge limited cultural validation. Constructs like “interpersonal support” may not fully capture the lived realities of Khawaja Sira communities. Future research should employ cognitive interviewing and community validation.

Administration: Trained interviewers administered questionnaires verbally in Urdu to accommodate varying literacy levels. Interviews lasted 45–60 minutes and were conducted in private community spaces identified by participants.

Inclusion and exclusion criteria

• All those who identified themselves as Khawaja Sira, transgender, aged 18 years and above, and consented to participate in the study were included in this research.

Data management, collection and analysis

Five trained field researchers with prior experience in community-based research conducted interviews between January-March 2024. A quality assurance officer reviewed 10% of completed questionnaires daily for consistency and completeness.

We performed the following quality checks:

• Range checks for all numeric variables
• Logical consistency checks (e.g., employment status vs. income)
• Duplicate entry identification
• Missing data assessment (all < 5% per variable)

Descriptive statistics, including means, frequencies, and percentages, were computed using SPSS version 19. Basic bivariate analysis was conducted to explore associations between key variables; Chi-square tests (or Fisher’s Exact tests where cell counts were <5 were used for categorical variables (e.g., income adequacy vs regular health check-ups, employment status vs. HSCL-25 distress threshold); and Mann-Whitney U tests were used for continuous outcomes across two group comparisons. Results of these bivariate analyses are reported alongside descriptive statistics. Given the small, non-probabilistic sample, these analyses are intended to be hypothesis-generating rather than confirmatory and should be interpreted with caution. Missing data were minimal and were handled through case-wise deletion. No imputation or sensitivity analyses were performed. No inferential statistics were performed due to the exploratory nature of the study. Subgroup comparisons (e.g., by income, education, and gender identity) were examined descriptively but not tested for statistical significance. Missing data were minimal and were handled through case-wise deletion. No imputation or sensitivity analyses were performed. As this was a descriptive study with a small non-probabilistic sample, confidence intervals and multivariate adjustments were not calculated. However, the association between socio-economic factors (such as income and education) and health status was described only descriptively, suggesting potential confounding.

Out of 99 approached individuals, 91 agreed to participate (response rate: 91.9%). Eight declined, citing discomfort with sharing personal information.

All items in the survey were mandatory, resulting in little missing data. For any item left blank, the corresponding case was excluded from analysis for that specific variable only. The highest missing response rate was less than 5%.

Sociodemographic characteristics

The study involved 91 self-identified Khawaja Sira individuals recruited in Islamabad. Of 91 participants, 72 (79%) self-identified as intersex, 10 (11%) as female, 6 (7%) as male, and 3 (3%) as transgender (Table 1)

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Socio-demographic charcteristics.

https://doi.org/10.1371/journal.pgph.0005798.t001

Our study focused on individuals within the Khawaja Sira community network, which encompasses diverse gender identities beyond intersex. The term “Khawaja Sira” is a culturally specific umbrella category in Pakistan that includes third-gender individuals, transgender persons, and some with intersex variations. It is important to note that the high proportion of participants (79.1%) selecting the response option translated as “intersex” most likely reflects cultural self-identification with a socially recognised third-gender status—and possibly the Urdu equivalent “Khusra” which carries specific religious and legal legitimacy in Islamic jurisprudence—rather than indicating biological Differences in Sex Development (DSD), which occur in approximately 1.7% of the global population. The survey response option was derived from the 2018 Transgender Persons (Protection of Rights) Act’s recognized categories and was culturally validated with community representatives during piloting. Authors acknowledge this as a key limitation: the findings pertain to gender-diverse individuals within the Khawaja Sira social framework and should not be interpreted as data on biological intersex conditions. Future research should use clearer Urdu-language distinctions between cultural gender identities and biological sex variations.

Age distribution: Mean age was 30 years (SD = 5.2, range: 18–42). The narrow range (20–35 years for 80% of sample) suggests potential underrepresentation of older community members, possibly due to survival bias, social exclusion, or reluctance to participate.

Educational attainment: Half of the participants (51%, n = 46) never attended school. Educational levels varied widely: primary education (19%, n = 17), bachelor’s degrees (16%, n = 15), college attendance (7%, n = 6), and master’s degrees (4%, n = 4).

Employment and income: Unemployment affected 34% (n = 31), with 33% (n = 30) in part-time work, 16% (n = 15) students, 12% (n = 11) employed full-time, and 4% (n = 4) both students and employed. Income distribution showed 28% (n = 25) earning <25,000 PKR monthly, 19% (n = 18) earning 30,000–50,000 PKR, and 26% (n = 24) earning >75,000 PKR.

Ethnic composition: The sample comprised predominantly Punjabi participants (74%, n = 67), with representation from Pashtun (11%, n = 10), Sindhi (8%, n = 7), Balochi (4%, n = 4), and Afghan (3%, n = 3) communities. This distribution likely reflects network-based recruitment bias rather than true population proportions.

Health status and healthcare access

Physical health conditions: Thirty-five percent (n = 32) reported no diagnosed health conditions. Among those with conditions, hypertension was most common (20%, n = 18), followed by diabetes (9%, n = 8) and asthma (7%, n = 6).

Healthcare utilization: Hospital visit frequency varied: annually (43%, n = 39), every six months (21%, n = 19), every three months (19%, n = 17), and monthly (18%, n = 16).

Healthcare comfort levels: Comfort visiting hospitals varied considerably: 44% (n = 40) felt comfortable only sometimes, 29% (n = 26) never felt comfortable, 18% (n = 16) felt comfortable, and 10% (n = 9) never visited due to discomfort. Mean comfort score was 2.3 on a 4-point scale (SD = 0.9), indicating generally low comfort levels possibly reflecting experiences of discrimination, stigma, or inadequate provider sensitivity.

Substance use

Participants reported use of: tobacco (57%, n = 52), marijuana (7%, n = 6), alcohol (3%, n = 3), and heroin (2%, n = 2). Additional substances were reported by 21% (n = 19). The high tobacco prevalence may reflect coping mechanisms for social exclusion, economic stress, or limited access to mental health support.

Mental Health Assessment (HSCL-25)

Mean HSCL-25 score was 2.1 (SD = 0.6, range: 1.2-3.8). Using the standard clinical cut-off of 1.75, 68% (n = 62) of participants exceeded the threshold for significant psychological distress, indicating probable anxiety and/or depressive disorder.

Anxiety symptoms among the participants

The assessment of anxiety symptoms shows different levels of distress among participants (Table 2). For example, 28.6% did not experience feeling scared for no reason at all, while others felt it a little (45.1%), quite a bit (18.7%), or extremely (7.7%), with an average score of 2.05. Similar patterns were seen with fear, dizziness, nervousness, heart racing, trembling, tension, headaches, panic, and restlessness. Headaches were found to be most prevalent (mean = 2.57, SD = 0.89); 83% experienced at least “a little”. Restlessness reported with a mean = 2.20 (SD = 0.93). Similarly Heart racing was reported mean = 2.11 (SD = 1.01), followed by Faintness/dizziness with a mean = 2.10 (SD = 0.95) and Tension mean = 2.08 (SD = 0.96)

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Mental Health Assessment of Khawaja Sira Participants via Hopkin Symptom Checklist (HSCL-25).

https://doi.org/10.1371/journal.pgph.0005798.t002

Depression symptoms among the participants

Based on the assessment, participants experienced various symptoms of depression. Many reported feeling low energy, with some experiencing extreme tiredness, while others felt only slightly tired or not at all. The data indicates that worrying excessively was the most common concern, with a mean score of 2.45 and a standard deviation of 0.91, and 88% reported experiencing at least some level of worry. The average score for suicidal ideation was 2.01 (SD = 1.0), and feelings of hopelessness had a mean of 2.42 (SD = 1.04). Sleep difficulties and loneliness were also prominent, with mean scores of 2.36 and standard deviations of 1.01 and 1.02, respectively. Additionally, feelings of worthlessness had a mean score of 0.9, and 27% of participants reported experiencing at least some thoughts of worthlessness. (Table 2)

Gender differences: Female-identified participants (n = 10) reported higher mean HSCL-25 scores (2.6, SD = 0.5) compared to intersex-identified participants (2.0, SD = 0.6), though small subgroup sizes limit interpretation.

Health awareness

Table 3 presents data on health awareness among intersex individuals, highlighting their awareness and practices related to health monitoring and financial stability. Low health awareness was evident across multiple indicators. 41% (n = 37) of participants were aware of blood pressure, while 59% (n = 54) were unaware. Regarding blood sugar, 28% (n = 25) had knowledge, whereas 73% (n = 66) lacked awareness. Only 25% (n = 23) reported regular health check-ups, compared to 75% (n = 68) who did not. When it came to income sufficiency for basic needs, 33% (n = 30) said yes, while 67% (n = 60) said no. For disaster and emergency preparedness knowledge, 26% (n = 24) were informed, whereas 74% (n = 67) were not.These findings indicate substantial gaps in health literacy and preventive care engagement, compounded by financial constraints limiting access to routine health monitoring.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Health awareness among Khawaja Sira individuals.

https://doi.org/10.1371/journal.pgph.0005798.t003

Health protective behaviours

Table 4 shows that many people do not follow healthy habits. Almost half rarely use vegetable oils instead of animal fats. The HPBS uses a 5-point response scale (1 = Never to 5 = Always), with higher scores indicating greater engagement in health-protective behaviours; a score of 3.0 represents the midpoint (“half a time”). Mean HPBS score was 2.5 (SD = 0.4, range: 1.6-3.8), indicating low to moderate engagement in health-protective practices, falling below the scale midpoint. Specific behaviours showed concerning patterns:. Specific behaviours showed concerning patterns:

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Health protective behavior among Khawaja Sira individuals.

https://doi.org/10.1371/journal.pgph.0005798.t004

Nutrition and dietary practices

Approximately 30% of individuals report frequently or always consuming 0.5 to 1 pound of vegetables daily, with a mean score of 2.15 and a standard deviation of 1.0. Similarly, only 14% regularly or always consume fruits daily, reflected by a mean of 2.40 and a SD of 1.0. Control over salt intake is maintained by 25% of the population, with a mean score of 2.56 and SD of 1.1, while 22% frequently or always control their sugar intake, indicated by a mean of 2.43 and SD of 1.0.

Physical activity and sleep

Moderate exercise for more than 30 minutes daily was reported as occurring “frequently” or “always” by 15% of respondents, with a mean score of 2.29 and a standard deviation of 1.1. Adequate sleep was similarly reported by 29% of participants as “frequently” or “always,” with a mean score of 2.96 and a standard deviation of 1.1.

Tobacco exposure

Approximately 15% of individuals frequently or always persuade others to quit smoking, with a mean score of 2.08 and a standard deviation of 1.2, while 41% never attempt to do so. Similarly, regarding avoidance of secondhand smoke, 15% do so frequently or always, with a mean of 1.95 and a standard deviation of 1.3, whereas half of the respondents never avoid it.

Healthcare adherence

Following medication instructions, 37% of respondents report doing so “frequently” or ‘always,’ with a mean score of 3.02 and a standard deviation of 1.2. Regarding the disposal of expired medicines, 40% indicate they do this “frequently” or ‘always,’ with a mean of 3.04 and a standard deviation of 1.4.

Low engagement in health-protective behaviours may reflect multiple barriers including financial constraints (66% insufficient income), limited health literacy, and restricted access to resources rather than knowledge deficits alone.

Bivariate analyses

Basic bivariate analyses were conducted to explore associations between key socioeconomic and health variables. Given the small, non-probabilistic sample, these analyses are intended to be hypothesis-generating rather than confirmatory, and all findings should be interpreted with caution.

Income adequacy and regular health check-ups. Among the 91 participants with complete data on both variables, participants who reported sufficient income for basic living expenses were significantly more likely to undergo regular health check-ups than those with insufficient income (40.0% vs. 18.3%). A Chi-square test with Yates’ continuity correction confirmed a statistically significant association (χ²(1) = 3.86, p = 0.049; OR = 2.97, 95% CI: 1.11–7.92).

Employment status and psychological distress. Participants without any paid employment (unemployed or student only; n = 46) were more likely to exceed the HSCL-25 clinical threshold for significant psychological distress (≥1.75) compared to those in any form of paid work (n = 45): 80.4% vs. 55.6%. This association was statistically significant (χ²(1) = 5.39, p = 0.020; OR = 0.30, 95% CI: 0.12–0.78), indicating that not being employed was associated with nearly 3.4 times higher odds of exceeding the distress threshold.

HSCL-25 total scores by employment status. HSCL-25 scores were significantly higher among participants without paid employment (Median = 2.21, IQR: 1.90–2.53) compared to those in paid work (Median = 1.97, IQR: 1.55–2.34), with a Mann-Whitney U test confirming the difference (U = 770.5, p = 0.036, r = 0.22, indicating a small-to-moderate effect size).

HSCL-25 total scores by educational attainment. Participants with no formal education (n = 46) reported significantly higher HSCL-25 scores (Median = 2.38, IQR: 1.90–2.75) than those with any formal schooling (n = 45; Median = 1.95, IQR: 1.56–2.30). This difference was statistically significant (U = 685.5, p = 0.006, r = 0.29, a moderate effect size). These findings are consistent with existing literature linking socioeconomic marginalisation, specifically unemployment and limited educational attainment—to poorer mental health outcomes in gender-diverse populations (Table 5).

Download:

• PNG
  larger image
• TIFF
  original image

Table 5. Summary of bivariate analyses (n = 91).

https://doi.org/10.1371/journal.pgph.0005798.t005

Study limitations

The study has a number of limitations. Firstly, sampling bias arises from snowball recruitment through Gurus, likely overrepresenting connected network members and excluding isolated or marginalized community segments. Our ethnic composition, with 74% Punjabi, probably reflects network clustering rather than the actual population distribution. Additionally, using census figures (n = 133) instead of CBO estimates (>13,000) for sample calculation likely compromised representativeness. The cross-sectional design prevents establishing causal relationships between socioeconomic factors and health outcomes. The small sample size limits subgroup analyses and statistical testing of associations. Sensitive topics such as substance use and suicidal ideation may be underreported due to social desirability bias, despite interviewer training and private settings. While instruments were translated, limited cultural adaptation may impact construct validity for this population. Lastly, the findings are specific to Islamabad’s sociocultural context and may not be generalizable to other regions of Pakistan or beyond.

Implications for policy and practice

Healthcare system reforms include implementing mandatory cultural competency training for healthcare providers on gender-diverse care. Additionally, establishing dedicated gender-affirming health services with trained staff can improve patient outcomes. Integrating routine mental health screening alongside low-cost counseling services further supports community well-being. Developing community health worker programs that embed trusted Khawaja Sira community members ensures culturally sensitive care.

In terms of social protection, expanding income support programs targeting unemployed or underemployed Khawaja Sira individuals is crucial. Creating employment quota systems with effective enforcement mechanisms can promote inclusivity. Furthermore, establishing anti-discrimination complaint mechanisms supported by legal aid helps address grievances and protect rights.

Education and empowerment efforts should focus on implementing inclusive policies that guarantee access for gender-diverse youth. Developing adult literacy programs within Khawaja Sira community spaces can facilitate greater educational opportunities. Additionally, creating health literacy initiatives that emphasize preventive care and chronic disease management empowers individuals to take control of their health.

Research priorities involve conducting larger, probability-based studies with improved population estimates. Employing longitudinal designs enables the examination of health trajectories and the impact of interventions over time. Incorporating participatory research approaches that center community voices ensures relevance and cultural sensitivity. Lastly, investigating the intersectionality of gender identity with ethnicity, religion, and regional variation can provide a more comprehensive understanding of diverse experiences.