Effectiveness of ICT-based intimate partner violence interventions: a systematic review

In total, 259 articles were identified, among which 105 articles were duplicates. Out of the 154 unique articles, 125 were excluded based on the content of their abstracts. The inclusion criteria were then applied to the remaining 33 articles after reading their full text. Four articles were then excluded, and 25 articles were kept for analysis [9, 55‐78] (Fig. 1).

Table 4 (see Appendix) lays out the studies in terms of population, intervention, comparison groups, and outcomes (PICO). Table 1 presents the authors, publication year, study country, study type, recruitment space, theme, outcomes, sample size, sample size per arm, control group, and the type of ICT Used for the 25 studies. Out of the 25, 23 (92%) took place in North America (20 studies (80%) in the United States and 3 (12%) in Canada), 1 study (4%) took place in Australia, and 1 (4%) in New Zealand.

Most studies focused on women with potential vulnerability to, past experience of, and/or current experience of intimate partner violence, with the exception of one [74], which included both men and women as study participants. Four studies included women who were pregnant [56, 58, 69, 72]; two of these studies included women up to 3 months postpartum who had history of IPV [56, 72]. Two studies focused on women with a history of IPV and who were active substance(s) users [59, 65], and 1 study on women who were at risk of HIV through unprotected intercourse [65].

Out of the 25, 17 studies (68%) were solely desktop- or laptop-based [9, 55, 57, 58, 60, 61, 63, 65‐67, 69, 70, 72‐74, 76‐78], 2 studies (8%) were solely tablet-based [56, 62], 1 study (4%) used computer and telephone [77], 1 study (4%) used tablet and telephone [59], 1 (4%) implemented a kiosk system [75] and 3 (12%) were not reported and supposed any type of ICT [64, 68, 71].

Table 2 shows the characteristics of the included studies. The 25 studies included 16 randomized controlled trials (12 two-arm and four three-arm studies), four pre-post designs, two cross-sectional studies, two prospective studies, and one diagnostic case-control study (i.e. retrospective data with known disease-positive and disease-negative cases [79]).

Control groups varied widely, and wait-list controls were used in five RCT studies [55, 69, 73, 74, 78]. Four studies allowed control groups to access websites with static, or non-interactive, or non-tailored content [9, 57, 64, 68], while two studies used irrelevant information for control groups [59, 61], seven control groups used face-to-face (or paper-based self-reported) screening [56, 60, 62, 65, 70, 71, 75], four had the intervention group play the role of control (i.e. pre-post design) [58, 63, 66, 67], and three studies had no control groups [72, 76, 77]. The sample size in the RCT studies varied extensively from 32 participants to a high of 2416.

The 25 interventions implemented had various foci. ICT was used for screening and disclosure in 13 (52%) of the studies. Five studies (20%) aimed at IPV prevention, four (16%) studies used ICT to address the mental health of female victims of IPV, and two (8%) studies used ICT to provide support for decision aid. Only one (4%) study assessed mainly the suitability of ICT for use in an IPV context.

The 25 studies had five types of interventions and varied study settings. In terms of settings, 14 studies were conducted in medical services facilities [55, 58, 62, 63, 67, 69‐76, 78] such as emergency departments, clinics, community health centers, trauma treatment centers, and family practices. Six studies were conducted in the community [9, 56, 57, 61, 64, 68], four in social services facilities [59, 60, 66, 77], and one in legal services facilities [65].

The 25 studies represent a range of uses of ICT in the context of IPV, addressing screening and disclosure, IPV prevention, ICT suitability, empowerment and support, and women’s health.

Table 5 (Appendix) summarizes the outcomes measured by each study. Our review revealed a wide variation among studies in terms of outcomes measured for studies that address the same focus. In total, 27 measurement tools were used in the 25 studies (see Table 5 in Appendix).

Among the 12 studies that address screening and disclosure, five studies used a simple disclosure count [55, 56, 69, 75, 78]. Two studies used non-validated questionnaires [58, 72], and two studies used the Partner Violence Screen (PVS) and the Abuse Assessment Screen (AAS) [73, 74]. Three studies had no common outcome measurement tools.

The four studies [63, 64, 67, 77] that focus on mental health used eight different outcome measurement tools; only the PTSD Checklist (PCL) was common to two studies [63, 67].

In terms of suitability of ICT, the Systems Usability Scale (SUS) was used in one study only to assess software usability [59].

Out of the five studies [9, 57, 62, 65, 70] focusing on IPV prevention, three studies [57, 62, 65] used the Revised Conflict Tactics Scale (CTS2). The two studies that addressed support [66, 68] had no common measurement tools.

Our review revealed the emerging nature of ICT use in IPV research. While there is a growing interest in the use of ICT in IPV interventions, there are virtually no studies examining its challenges.

While most of the studies used ICT to enhance screening and increase the disclosure rate, few studies targeted IPV prevention and even fewer aimed at improving support. Suitability of ICT was seldom assessed in a formal way using a validated usability scale (e.g. Systems Usability Scale) or methodology [80].

In addition, while most of the studies used RCT design, the number of arms, the population, the control groups used, the sample sizes, and the outcome measures varied widely among the studies, which makes it hard to compare those results. With the exception of two large sample sizes that were used in two non-RCT studies (one that accessed electronic health records for an artificial intelligence application [76], and another that used the emergency department [75]), the sample sizes per arm were generally low. The sample size per arm was less than 30 in four studies. Only six studies had a sample size per arm between 100 and 300, and only four studies had a sample size per arm between 300 and 805. This suggests that the current ICT-based IPV interventions have limited generalizability and comparability—especially because only six studies were conducted in the community.

Twenty-three (92%) of the studies were conducted in North America, 20 (80%) of which were in the United States, which is an additional limitation to the generalizability of the findings since they lack diversity in terms of ethnicity, race, language, and cultural backgrounds. Diversity is crucial in IPV. Research shows that foreign-born immigrant as well as indigenous women are more likely to experience IPV [81, 82] and intimate partner homicide than other women [83, 84]; hence, addressing diversity in IPV is critical. It is encouraging that one recently published RCT protocol laid out a plan for culturally tailored intervention targeting immigrant, refugee, and indigenous survivors of IPV [85].

Technology is costly in terms of hardware, software and data plan costs. Consequently, while access to ICT by women experiencing IPV is a challenge in high income countries, including the United States [86, 87], it is even more difficult in low- and middle-income countries (LMICs). This creates inequity in access to technology, and a digital divide among women subject to IPV. This inequity challenge and its impact on outcomes has long been observed in electronic health (eHealth) [88, 89] and needs to be addressed in ICT-based IPV interventions; it was not addressed in the studies covered by our review. Also, involvement of users in software design is a well-known need that is effective in producing software that works for users and aligns with their priorities and is suitable for their environments [47, 90‐94]. Hence, involving women experiencing IPV in the research team and in the ICT software design process is paramount to ensure usability and accessibility of the software and as a matter of equity [95]. There is a lack of research in this area in the studies covered by our review.

A recent study protocol is promising that ICT will ensure lower access barriers [96], which is the traditional unchecked point of view; this is another demonstration of the need to shed a critical light on the use of ICT for women experiencing IPV, analyzing equity as well as the safety and ethical challenges involved.

Our review shows that 8 studies [55, 56, 59, 63, 64, 70, 72, 75] reported that women found ICT interventions suitable for IPV disclosure; three of those studies found it particularly suitable in terms of confidentiality, usefulness, and satisfaction [56, 63, 72]. Stigma is an important factor associated with intimate partner violence [97] limiting agency in help-seeking for IPV [98]; ICT seems to be a tool that provide an opportunity for women subject to IPV. With the exception of one in which participants preferred a face-to-face discussion [62], IPV disclosure through ICT was found to be most appropriate in most of the studies compared to face-to-face disclosure and was perceived as non-judgemental and more anonymous than face-to-face discussion, which facilitated more disclosure.

The increase in phone ownership and internet access in low- and middle-income countries [51], coupled with the ability to use ICT to target individuals through health informatics tools that targets individuals (i.e. consumer health informatics) [99] such as apps, makes ICT a flexible tool to address IPV in multiple languages, embedding different cultural cues, and overcoming the cultural stigma related to disclosing IPV from the convenience of a personal ICT device (e.g. cell phone, smart phone). Simple ICT tools such as cell phones are available in rural areas and proved to be successful tools in the health domain (e.g. chronic disease management) [100‐102]. However, it is important to note that one challenge of ICT-based interventions is that only women with basic literacy and IT knowledge can benefit; also, some victims may not have access to ICT, and some abusers may restrict their partners’ access to ICT. Therefore, in addition to the traditional security considerations related to the use of ICT, such as maintenance of privacy [103] and confidentiality [99], there are ethical issues related to the unintended consequences of ICT [104, 105], including safety risks.

In the IPV domain, sharing cell/smart phones at home or with neighbors is a common practice [106, 107], which might increase the risk of IPV if the perpetrators notice that women are using these devices to address IPV [108]. The studies covered by this review were located in high income countries; there is little to no examination of the problem of access to ICT (i.e. cost), nor of the risks inherent in the use of ICT (e.g. sharing devices, ability to access browsing history) in addressing IPV programming in a variety of contexts. Ethical challenges related to the safety of women increase when women are sharing cell/smart phones with perpetrators; in such contexts special considerations should be taken care of, including “safety by design” [109].

Safety challenges involved in the use of ICT in health have recently attracted much attention [105, 110]. Moreover, recent reflections related to the ethical challenges of using web-based RCT show the need to equip participants with information about Internet safety [111]; likewise, identifying and managing safety risks within ICT-based IPV research remains a perspective to be explored. This raises ethical questions related to the use of ICT, for example in the case of referral embedded in the IPV programming, as was the case in three studies included in this review [55, 73, 78]. Poor quality services are well documented in low-resource and rural areas [112], so referring women to such services might have negative consequences for them. While this is not an ICT issue, ICT facilitates communication of information and has the potential to exacerbate current challenges. This is part of the well-known unintended consequences of the use of information technology in health [113‐115].

Of the 19 studies that explicitly mentioned their settings, 14 were in urban settings, only three were in urban and suburban areas, and two were in suburban settings, suggesting a need to test ICT use for IPV in rural settings [67, 77] and uncover any particularities compared to the urban context.

It is also worth noting that our systematic review has not included search terms regarding the user of ICT tools to address IPV for women with disabilities. However, in a quick assessment, when we searched in PubMed for research that addresses the use of ICT to address violence in the context of women with disabilities, our search revealed only two papers [116, 117]. The use of ICT to address IPV for this particular group of women is important to address in a separate study, as ICT accessibility may be challenging for women with certain types of disabilities, especially since there is evidence that IPV occurs at higher rates in this population compared to the general population [118‐123], and that ICT can play a major role in empowering people with disabilities [124]. The use of ICT tools to address IPV for women with disabilities, and the accessibility of these tools, remains an important area for future studies.

Moreover, our review indicated that there is a paucity of research addressing ICT use for IPV prevention and IPV treatment. There is a clear need for more research on ICT-based interventions to prevent IPV and to address post-IPV challenges, such as mental illness and the integration and coordination of mental and social services (e.g., employment, housing), which has never been addressed in the reviewed literature. In this context, virtual communities may play an important role in integrating and coordinating mental health services and social services [125]. While the studies showed different aspects of ICT use for IPV, a more integrative approach can be taken if researchers approach IPV using a virtual community framework. A virtual community (VC) is defined as a community of individuals cooperating using online tools to attain an objective [126]. Health VCs have been used in healthcare to provide patients with education, health education, and remote support; that proved to be an enabling and empowering factor, which allowed patients to become active participants in managing their health conditions [127, 128]. Support was not provided solely by health professionals; instead, health VCs connected individuals with common experiences (e.g., similar health conditions), which enabled them to interact and mutually support each other [129]. Healthcare providers could provide validated evidence-based health information, coupled with strategies for effective chronic disease management [130‐132]. Ample evidence exists demonstrating that virtual tools are effective and efficient for addressing health issues experienced by patients with various health conditions (chronic kidney disease, pulmonary hypertension, cancer) [126, 132‐134]. There is also ample evidence that health VCs are effective in engaging individuals managing their own health condition [131, 132]. Moreover, VCs can be patient-centred, customizable to individual preferences, and responsive to individuals’ needs and values [135]. In terms of mental health, an important factor for women experiencing IPV, VCs provide a secure, private way for women to communicate privately and securely and to access information tailored to their situation in a personalized manner. This privacy facilitates access and assists in overcoming stigma, especially for women from visible minority groups [136]. VCs have a proven potential to engage participants [137]. There is ample evidence that health VCs are associated with positive mental and social benefits, such as reduced loneliness and increased emotional well-being, self-esteem, and self-empowerment [129, 138, 139]. It is important to explore an integrative approach to ICT-based intervention in IPV using VCs, especially since VCs enable a community dimension that facilitates mutual support and empowerment among its members (e.g., abused women).

In a study that screened for IPV, while women preferred computerized over face-to-face disclosure, computerized screening did not increase prevalence, so ICT did not lead to increase in disclosure. Also, when women disclosed by answering paper-based questionnaires, the self-completed paper-based questionnaires had less missing data collected than both computer-based and face-to-face interviews [70], which shows the advantage of having for paper-based screening (i.e. less missing data).

Likewise, while ICT allowed considerably higher IPV detection, this did not always lead to charting for IPV or to a follow-up by treating physicians [74]; more research is needed to understand the factors, such as continuing medical education [140], that increase the chances of charting and follow-up. Detection is not enough.

We have noted above the lack of research regarding equity, safety, and the ethical challenges involved in the use of ICT, as well as the lack of culturally, ethically, and racially sensitive ICT programs. ICT might be able to support and enhance more traditional on-the-ground program delivery; however, ensuring that effective ICT-based interventions reach the most vulnerable in equitable, ethical, and safe ways remains a research agenda to be undertaken.

Current results suggest that face-to-face and paper-based approaches should not be discarded, and that the computer-based software design must be user-centred and must follow usability principles [141, 142].

Limiting the search to English language is one of the limitations of this study. Another limitation was the difficulty to compare the results, since the tools used to measure the same outcome varied widely between the studies. Various questionnaires were used to detect IPV, assess decisional conflict, assess mental health challenges, assess treatment efficacy, and assess different primary and secondary outcomes. An illustrative example is the varied questionnaires that researchers used to measure IPV [55, 57, 60, 62, 65, 69, 73‐75], which included the use of an artificial neural network to identify IPV automatically via analysis of the notes stored in the electronic health records [76]. Our review shows that there are limits for comparing the effectiveness of the interventions in terms of mental health (e.g., reduction in stress, anxiety, or depression levels), given the great variety of mental-health-related measurement tools that have been used.