Adaptation and Validation of the Harvard Trauma Questionnaire (HTQ) and PTSD Checklist-5 (PCL-5) with Indian Women from Slums Reporting Gender-Based Violence

Background: Despite high rates of gender-based violence (GBV) in India, culturally sensitive measures that examine universal and culturally relevant trauma reactions are lacking. Although the Harvard Trauma Questionnaire (HTQ) has been used in India, no study has adapted the measure in full for use with this population. This study describes the adaptation, validation, and results from the adapted HTQ, containing the PTSD Checklist-5 (PCL-5), for Indian women from slums reporting GBV. Method: This study used the adaptation framework proposed by the HTQ measure developers. The adapted HTQ contained a (1) trauma screen relevant for stressors faced by Indian women from slums, (2) description of the index trauma, (3) description of any ongoing stressors, (4) universal trauma reactions (i.e., PTSD measured by the PCL-5), and culturally relevant trauma reactions (i.e., idioms of distress measured by a scale developed for the study). This measure was piloted on 111 women from Indian slums in face-to-face interviews. Trauma characteristics, types of ongoing stressors, and psychometric properties of the PCL-5 and idioms of distress scale were explored. These scales were validated against measures of depression (PHQ-9), anxiety (GAD-7), and somatic complaints (PHQ-9). Results: The majority of participants (77%) reported physical beatings, 18% reported unwanted sexual touch, and 28.8% reported indelity as the primary emotional abuse. Further, 96.7% of GBV was perpetrated by partner or family member and over half reported ongoing stressors (e.g., poverty-related strain). The PCL-5 embedded in the HTQ yielded good internal consistency (Cronbach’s alpha=.88) as did the idioms of distress scale with deletion of one item (Cronbach’s alpha=.80). Both scales were externally valid, yielding large correlations with depression, anxiety, and somatic complaints (rs between .54 to .80, ps < .05). Discussion: This is the rst study to develop a comprehensive measure of trauma exposure with universal and culturally relevant reactions in India. This study also enhances HTQ usage in India by delineating all the steps in the adaptation process. Results can inform the development of trauma-focused interventions for Indian women from slums. verbatim and later coded by both authors against the prompts on the CAPS-5, a gold-standard clinician-administered PTSD assessment. Results of the coding process and disagreements were resolved through discussion and consensus. the Harvard Trauma Questionnaire for Indian women from slums exposed to GBV. The trauma screen measured a wide range of GBV and non-GBV events, indicating high levels of trauma exposure across traumatic event categories in this sample. The Hindi PCL-5, embedded to measure universal trauma reactions of PTSD, appeared to be psychometrically useful and valid, although the hypervigilance and trauma-related amnesia items did not appear to measure PTSD. The Hindi PCL-5 demonstrated high convergence with measures of depression, anxiety, and somatic complaints. The idioms of distress scale, developed to assess culturally relevant reactions, also appeared to have promising psychometric properties. This scale yielded good reliability with the deletion of one item (i.e., believing that you deserved what happened because you are a woman) and also demonstrated convergent validity with known correlates of PTSD.


Introduction
The gang rape of a New Delhi medical student in 2012 garnered global attention and catapulted India into national protests and discourse about genderbased violence (GBV) (1). India has historically high rates of GBV (2)(3)(4)(5) that disproportionately affect socioeconomically marginalized women with less education (6-8). The risk of developing posttraumatic stress disorder (PTSD) following GBV is well established across cultures (9)(10)(11). However, culturally relevant trauma reactions, such as idioms of distress, are comparatively under-studied following trauma exposure. PTSD and idioms of distress are especially relevant to study in low-and-middle-income countries (LMICs) as most psychological research originates from high-income countries (HICs) (12,13). Research from HICs may not generalize to LMICs as HICs comprise westernized, educated, industrialized, rich, and democratic peoples (12). The current study thus investigates universal and culturally relevant trauma reactions (i.e., PTSD and idioms of distress) following GBV among Indian women from slums.

Background
Gender-based violence in India: Conceptualization, prevalence, and gaps in scholarship Gender-based violence in India is de ned by physical, sexual, emotional, and economic abuses that typically occur within the family setting. National lifetime rates of GBV range from 8% for sexual violence (2) to 41% for multiple types of violence (5). Signi cant regional variation exists, with rates up to 69% for lifetime GBV in certain Indian states (15). Although GBV prevails across all demographic groups, women who are less educated and socioeconomically marginalized are at signi cantly greater risk of GBV compared to their educated and wealthier peers (5,(16)(17)(18).
Variation in the prevalence of GBV is partly attributable to how GBV is de ned and assessed. Domestic violence (DV) is the most common type of GBV studied in India (19). The WHO de nes DV as "acts of physical, sexual, psychological abuse, and control by an intimate partner [or other member of the household] Study procedure and study sample Data were collected through verbal administration due to low literacy rates in Indian slums (65). Participants were administered a brief demographic form, followed by a PTEs and stressors (HTQ-I) screen. In total, 111 participants completed the HTQ-I and were included for the rst aim; of the 111 participants, the 99 who were able to link their PTSD symptoms and idioms of distress to an index trauma were included for analysis of the second and third aims.
Administration of HTQ-I was followed by a brief description of their index traumatic event (HTQ-II), any ongoing stressors (HTQ-III), and symptom reports on the PCL-5 (HTQ-IVa) and idioms of distress scale (HTQ-IVb). Lastly, participants were administered validated screeners for depression, anxiety, and somatic problems using the Patient Health Questionnaire (PHQ) series.
Participants were 34.13 (SD = 9.64) years old on average. They completed an average of 6.97 (SD = 3.76) years of education and subsisted on an average monthly per capita income of 57.99 USD (SD = 61.06 USD). The majority of participants were married (78.57%), of Dalit ethnicity (42.86%), and originated from a city (69.39%).
Adapted Harvard Trauma Questionnaire (HTQ) The HTQ was adapted following the established procedures of the measure developers in a recent study of Iraqi refugees (66). These methods were consistent with recommendations in cultural epidemiology for assessment tool adaptation (66,67). The content adaptation of the HTQ is described below by section, followed by adaptation procedures for translation. For a global view of all adaptations by section, see Figure 1 and Supplementary File 1.

Adaptation of HTQ-I: PTEs, stressors, and perpetrators
The HTQ was adapted to assess PTEs and stressors relevant in India by explicitly querying for PTEs that included GBV and non-GBV stressors. The PTEs and stressors were created from an extensive literature review about Indian women (20). The PTEs and stressors consistent with physical and sexual GBV include domestic violence perpetrated by partners and family members (20), sex tra cking (68), and dowry-related violence (69). Information on the perpetrators of GBV events was also collected by asking participants to select one or more perpetrators for each event from partner, family/in-laws, friend/acquaintance, and/or other. Similarly, as Indian women experience violence during pregnancy (70), pregnancy status was obtained for GBV events.

Adaptation of HTQ-II and HTQ-III: Index trauma and ongoing stressors
Participants were asked to brie y describe their index trauma. All symptom reports in HTQ-IV were linked to this index trauma to keep consistent with the intended administration of the PCL-5. Participants were also asked: "Under your current living situation, what is the worst event that has happened to you, if different from above. Please specify where and when these events occurred." Responses to this question were recorded as descriptive information and also coded dichotomously as '0' (when participants endorsed no ongoing stress) and '1' (when participants endorsed ongoing stress).

Adaptation of HTQ-IV: Symptom domains of HTQ
The HTQ was adapted to assess three types of trauma-related symptoms after participants endorsed a PTE or stressor. These included 20 PTSD symptoms in the DSM-5 derived from the PCL-5 (14) and 12 idioms of distress speci c to India and created for this study.
HTQ-IVa: Hindi PCL-5 within the HTQ A 20-item checklist, the PCL-5 assesses PTSD symptoms from the DSM-5 (14). Each symptom is rated for severity on a 5-point Likert scale ranging from 0 to 4, so total scores could range from 0 to 80. All participants were asked to anchor their symptoms to their most stressful event in the past 30 days. The Hindi PCL-5 was scored continuously. Any PCL-5 symptoms endorsed at or above moderate ('2') levels (i.e., rating of '2', '3', or '4') were considered clinically elevated symptoms (14).
HTQ-IVb: Idioms of distress within the HTQ A 12-item checklist, the idioms of distress scale contains idioms of distress gleaned from literature review and during discussions with two Indian key informants who were bilingual and familiar with trauma-related reactions. Examples of idioms included somatic complaints, such as orthostatic dizziness and pain in the nerves; culturally-laden cognitions, such as believing that the trauma was written in one's destiny; and affective reactions, such as tension. Each symptom is rated for severity on a 4-point Likert scale ranging from 1 to 4, so total scores could range from 12 to 48. As this study did not validate the idioms of distress scale against a clinician-administered measure, the level at which symptoms were clinically elevated could not be determined.
Patient Health Questionnaire-9 (PHQ-9) A 9-item screening instrument, the PHQ-9 (71) assesses the presence and severity of current depression and has been validated for use in India (72). This questionnaire queries all depression symptoms with response options ranging from 0 ('not at all') to 3 ('nearly every day') in the past two weeks, with total scores ranging from 0 to 27. The PHQ-9 demonstrated adequate internal consistency reliability in this study (Cronbach's α = .78).

Generalized Anxiety Disorder Screen (GAD-7)
A 7-item screening instrument, the GAD-7 (73), assesses the presence and severity of anxiety (including generalized anxiety, panic, and social anxiety) and has been validated in India (72). The GAD-7 has response options ranging from 0 ('not at all') to 3 ('nearly every day') in the past two weeks and total scores range from 0 to 21 (Spitzer et al., 2006). The GAD-7 demonstrated good internal consistency reliability in this study (Cronbach's α =.84).
Patient Health Questionnaire -15 (PHQ-15) A 15-item screening instrument, the PHQ-15, assesses the presence and severity of somatic complaints such as headaches, menstrual cramps, indigestion, and pain during sex. It has been validated in India. The GAD-7 has response options ranging from 0 ('not at all') to 3 ('nearly every day') in the past two weeks and total scores range from 0 to 21 (Spitzer et al., 2006). The GAD-7 demonstrated good internal consistency reliability in this study (Cronbach's α =.84).
Adaptation procedure: Translation process and comprehension check Consistent with other cross-cultural assessment tool adaptation, steps to translate and back-translate the instrument to ensure technical, semantic, and conceptual equivalence were undertaken (67). The assessment tool was rst translated into Hindi, then back-translated into English by two independent bilingual individuals. For the PCL-5 measure embedded into HTQ-IV, two English-speaking PTSD experts rated the original and back-translated versions for conceptual equivalence on a 5-point Likert scale (ranging from '0' denoting it was not at all the same meaning to '5' denoting it was exactly the same meaning). Items rated at or below 3 were then semantically adjusted iteratively so the nal back-translated version re ected the original meaning before being re-instated into the nal Hindi HTQ.
As the wording of PTSD symptoms may not have been comprehensible to participants who may not be familiar with PTSD in this cultural context, the Hindi PCL-5 was also subjected to a comprehension check. A subsample (12%) were administered the comprehension check. For items the subsample endorsed (rated '1', '2', '3', '4'), they were asked to provide examples of the symptoms. For the items the subsample did not endorse (rated '0'), participants were asked to explain the meaning of the prompt to ensure conceptual clarity. Answers were recoded verbatim and later coded by both authors against the prompts on the CAPS-5, a gold-standard clinician-administered PTSD assessment. Results of the coding process and disagreements were resolved through discussion and consensus. Note. Event endorsement was not mutually exclusive.

HTQ-I and II: Potentially traumatic events (PTEs) and ongoing stressors
On average, participants endorsed 3.97 (SD = 2.36) GBV events, 1.12 (SD =.94) non-GBV events, and 5.09 (SD = 2.80) total traumatic events. Nearly all (96.91%) reported that GBV events were perpetrated by their partners, family members, or in-laws and 62.92% of participants selected a GBV event as their index trauma. The most common GBV events were being kicked or beaten, directly witnessing others being kicked or beaten, and directly witnessing others attacked with a weapon. Participants also endorsed culturally relevant GBV, such as dowry-related violence and sex tra cking, respectively. The most common non-GBV events were the sudden loss of a loved one, followed by a major accident, and surviving a natural or manmade disaster.

HTQ-III: Ongoing stressors
More than half the sample (54.64%) reported ongoing stressors. Types of ongoing stressors included poverty-related stress (e.g., di culty meeting household, childcare, education expenses), caregiving stress (e.g., getting children married, caring for sick family members), and emotional abuse (e.g., humiliation by inlaws, marital in delity, desertion). Notably, 13 participants (i.e., 13.40%) reported that their ongoing stressor was a traumatic event that met the de nition of Criterion A (e.g., domestic violence, marital rape).
HTQ-Iva: Psychometric properties of the Hindi PCL-5 The full-scale reliability of the total PCL-5 scale was good (Cronbach's α =.88). The mean inter-item correlation for the full-scale PCL-5 was.27, indicating a satisfactory item agreement. The mean full-scale Hindi PCL-5 score was 29.88 (SD = 17.06). Table 2 delineates mean scores per item. Note. M = mean, SD = standard deviation. p < .05 *Indicates item is signi cantly positively skewed Intrusive memories, psychological reactivity, and persistent negative emotions were most frequently endorsed at clinically elevated levels and yielded the highest mean item scores. Conversely, trauma-related amnesia, self-destructive behaviors, and ashbacks were the least frequently endorsed and yielded the lowest mean item scores. Notably, the items for diminished interest, trauma-related nightmares, ashbacks, self-destructive behaviors, and trauma-related amnesia were signi cantly positive skewed. The internal consistency of the Hindi PCL-5 could be improved by deleting trauma-related amnesia and hypervigilance (see Table 2).
Twelve participants' comprehension of individual PCL-5 items was assessed. The subsample, who were interviewed about each PCL-5 item, appeared to understand the meaning of most items barring two items (see Table 3). Speci cally, ashbacks were di cult to understand for those participants who had never experienced them. This item resulted in the lowest comprehension (58.33% comprehension rate). Participants who misunderstood ashbacks typically reported general di culty with understanding it or described ruminative symptoms instead. Similarly, globalized negative beliefs about oneself, the world, and others also proved problematic to comprehend (75% comprehension rate). Participants typically misunderstood this item to mean wishing ill of others and were quick to deny this reaction.
HTQ-IVb: Psychometric properties of the Idioms of Distress Scale Table 4 Descriptive statistics, psychometric properties, and symptom frequency of the idioms of distress scale (N = 99)  Table 4). Notably, six items on the idioms of distress scale were signi cantly positively skewed: Fear of others learning about this hurtful experience, feeling that you are a burden on your family, feeling that you brought shame on your family, safed pani (vaginal discharge), nas mein dard (pain in the nerves), believing that you deserved what happened because you are a woman. These items were endorsed less frequently and with less intensity in this sample. Barring one item -"believing that you deserved what happened because you are a woman," -the idioms of distress scale retained good reliability by including the remaining skewed items.

Convergent validity of HTQ-IV: Hindi PCL-5 and idioms of distress scale
Posttraumatic stress disorder, idioms of distress, depression, somatic complaints, and anxiety yielded large positive correlations ranging from .54 (PTSD and somatic complaints, p <.001) to .80 (PTSD and idioms of distress, p <.001) (see Table 5).

Discussion
This study sought to adapt and validate the Harvard Trauma Questionnaire for Indian women from slums exposed to GBV. Reliability, validity, and comprehension of the Hindi PCL-5 Results indicated that the Hindi PCL-5 demonstrated good reliability estimates, adding to a body of literature from HICs (74)(75)(76) and LMICs (45,(77)(78)(79)(80)(81)(82). The Hindi PCL-5 yielded a very similar full-scale reliability to the 17-item PCL adapted for India in an older study (i.e., Cronbach's α = .89) (27). Overall, the full-scale reliability of the Hindi PCL-5 indicates that it assessed multi-faceted aspects of PTSD while also reliably measuring overall PTSD symptoms.
Similarly, results support the validity of the Hindi PCL-5. Medium-to-large positive correlations between full-scale PTSD, depression, and anxiety indicate that the Hindi PCL-5 assessed a construct that was related to depression and anxiety in the expected direction and magnitude. These ndings align with the validity literature from HICs (57,75) and LMICs (45,82,83) that demonstrate similar convergent validity estimates with depression and anxiety. Previous studies have found large positive correlations between PTSD, depression, and anxiety (81,83), such as a correlation of .78 between PTSD and depression (83). This study also found strong relationships between PTSD and somatic symptoms, adding to a growing body of cross-cultural work suggesting the relevance of assessing somatic targets for traumatized populations (84,85).
Certain items on the Hindi PCL-5 did not appear to measure PTSD in this cultural context. The hypervigilance item evinced nearly zero correlation with the remaining items on its arousal subscale, suggesting that it was not measuring PTSD-related arousal in this sample. Perhaps hypervigilance in the Indian slum context, where participants had to live more cautiously overall, did not function as an arousal symptom in this setting. Slum communities in Bombay are characterized by frequent adversities, including home destruction by the government, lack of sanitation, lack of privacy for bathing and defecating, water and electricity shortages, and rat infestations (86). It is possible that adaptive vigilance in a currently unsafe environment was confounded with hypervigilance symptoms linked to a prior traumatic event. Just over half the participants also reported ongoing stressors, including daily stressors, poverty-related strain, and current traumatic events. Experiencing ongoing stressors could elevate general stress and hypervigilance may not be linked with PTSD in this context.
Alternatively, participants could have interpreted the item as appropriate -rather than maladaptive -vigilance, given previous study ndings. The hypervigilance item, when translated literally into Portuguese for a study in Brazil (61), was also misunderstood by participants as a positive quality of appropriate alertness rather than the intended meaning of maladaptive hypervigilance. The hypervigilance item was also positively associated with alertness in another study in India (Charak et al., 2014b). These ndings suggest that the hypervigilance item could have been interpreted by participants facing ongoing stressors as adaptive and therefore unrelated to other PTSD symptoms.
Trauma-related amnesia also evinced nearly zero correlation with the remaining items on the NACM subscale. Participants responded to this item by emphasizing that they could not stop thinking of their stressful experiences and ruminated on the details. This item also received the lowest endorsement, with 14% of participants rating it at or above '2' on a 0 to 4 scale. Notably, trauma-related amnesia is endorsed less frequently in non-HIC cultural groups and it is considered an inconsistent item across cultures (28,87,88) and in India (26). In contrast, "thinking a lot" or "thinking too much" has emerged as a prominent complaint in myriad cultural settings following trauma and adversity, supporting the notion that rumination is more cross-culturally common compared to trauma-related amnesia (89,90). The current study's results extend the literature by con rming that trauma-related amnesia following trauma exposure are neither common nor indicative of the NACM symptom cluster of PTSD among Indian women from slums. Altogether, two items of the Hindi PCL-5 appear to operate distinctly in India and deleting both items increased the Hindi PCL-5's reliability to .89. This nding suggests that the hypervigilance and traumarelated amnesia items do not measure PTSD in this cultural context.

Reliability and validity of the Idioms Of Distress Scale
Idioms of distress are particularly relevant to assess given the problematic nature of two PTSD symptoms (i.e., trauma-related amnesia and hypervigilance) measured by the Hindi PCL-5. The idioms of distress scale yielded good reliability following deletion of one item about believing the traumatic event was deserved for being a woman. The highest mean rating assigned to idioms was for tension, believing the traumatic event was predestined, sar mein garmi (heat in the head), dil mein udasi (sadness in the heart), and dizziness. Tension emerged as the most strongly endorsed idiom, which is consistent with the existing literature that tension is a common idiom across linguistic groups indicating depression and anxiety (22,38,91,92). Tension has been documented as a primary complaint by women reporting marital and interpersonal problems (37,38,92), which corresponds with the nding that participants endorsed interpersonal trauma largely by husbands and in-laws. The idiom of believing the traumatic event was predestined is consistent with ndings from domestic violence survivors in South India (8). Tichy et al. (2011) proposed that predestiny may be culturally ingrained as a function of India's caste system legacy.
Although legally abolished, the caste system still shapes social and occupational life for Indians, who tend to view their position in the socioeconomic strata as relatively stationary and unchangeable (93). The intersectionality between one's caste and gender can layer social disadvantages, such that women from ethnic and religious minority backgrounds (e.g., Dalits and Muslims in this study) may be likelier to endorse a traumatic event as predestined. This hypothesis is speculative and needs veri cation in future studies.
The somatic idioms, such as sar mein garmi (heat in the head), dil mein udasi (sadness in the heart), and dizziness, are also consistent with the literature on somatization in India (39)(40)(41)94). These idioms' popularity strengthens the notion that distress is experienced and expressed physically for this population and somatization of distress is commonly reported across many cultures following trauma exposure (28,85,95). Altogether, the three most strongly endorsed idioms are distinct from the PTSD symptoms assessed using DSM-5 criteria. The fact that these three idioms diverged from PTSD symptoms suggests their potential clinical utility as trauma-related idioms to assess, and target for treatment, for Indian women from slums. This nding, however, is tempered by the nding that endorsement patterns were at low levels of severity. The level of severity may be partly explained by sample characteristics, as this study recruited a community sample.
The idioms of distress scale also showed large positive correlations with depression, anxiety, and somatic complaints. Considering that these idioms indicate meaningful reactions related to depression, anxiety, and somatic complaints, it is possible that the idioms might also indicate clinically elevated distress.
However, as the idioms of distress scale is a novel scale created for this study, future studies should validate it against clinician-administered interviews and measures of functional impairment to verify if the scale measures clinically elevated distress.

Limitations and strengths
This study is limited by a small sample size and by use of referral sampling rather than random sampling. Women who were referred by others or self-referred to the study may systematically differ from other women exposed to GBV. Although a fair representation of women living in urban poverty in India (96), the study sample is not generalizable to Indians from other socioeconomic groups or geographical regions. Another limitation relates to translation. Efforts were made to counterbalance lexical translation with conceptual equivalence that retained the spirit of each PTSD item. Despite such efforts, using an expert committee consensus method in which multiple experts translate the same assessment and compare their translations (97) could enhance future validation procedures. Similarly, using more idiomatic translation or translation into the local vernacular language could improve local comprehension. Finally, this study is limited by a lack of sensitivity and speci city analyses because clinician-administered measures were not used in this study. Sensitivity and speci city analyses can improve diagnostic accuracy, so future studies should expand upon this work by validating the subscales of the adapted HTQ against clinicianadministered interviews, including the idioms of distress scale.
Notwithstanding these limitations, this study also has strengths. This study is the rst to investigate PTSD in an under studied population of Indian women from slums exposed to GBV. Published studies from Indian slums mirror the demographic characteristics of the current study sample (86), indicating our ndings are applicable to women from urban Indian slums. This study also developed and tested a culturally sensitive assessment tool to screen broadly for all types of GBV in India. Correspondingly, this study is the rst to validate a commonly used assessment tool for DSM-5 PTSD symptoms in association with GBV for this population. This study is also one of the few to systematically record and report ongoing stressors in trauma-exposed populations from LMICs.
Although more nuanced measures of ongoing stress are needed for future studies in India, the descriptive information about ongoing stressors better contextualized study results, such as why the hypervigilance item may have proved problematic. Finally, this study also improves upon psychometric studies set in LMICs by reporting multiple reliability and validity indicators for universal (i.e., PTSD) and culturally relevant reactions (i.e., idioms of distress).

Conclusions
The adapted HTQ is a reliable and culturally valid assessment that includes a comprehensive trauma screen for India, an assessment of ongoing stressors, universal trauma reactions (i.e., PTSD measured by the Hindi PCL-5), and culturally relevant trauma reactions (i.e., idioms of distress measured by a new scale). As certain PTSD items may not be relevant for Indian women from slums, assessing idioms of distress is doubly relevant. Findings from this assessment can inform which treatment targets to focus on and sections of this measure may be abbreviated for use depending on the assessment needs in question. Declarations Ethics approval and consent to participate: All participants consented to study procedures through oral and written informed consent. Study procedures, including format of obtaining verbal consent, were approved for ethical research practice by the institutional review board of The University of Tulsa (IRB protocol  and the employees at the partnering non-governmental agency (CORO for Literacy).

List Of Abbreviations
Consent for publication: We consent to publishing this manuscript in BMC Women's Health.
Availability of data and material: all data, including the measure developed in this study, are available upon request.
Competing interests: We declare no competing interests.
Funding: The cost associated with this research was partially supported by The University of Tulsa's O ce of Research and Sponsored Programs and by the Center for Global Education.
Authors' contributions: All authors have contributed substantially to this manuscript. AP adapted the study measure, collected data, conducted statistical analyses, and wrote the manuscript. JR conducted a literature review and edited the manuscript. EN reviewed and edited the manuscript. All authors reviewed the nal manuscript.