*2.5. Analysis*

For the alternate forms reliability between the HESPER scale and HESPER Web, intraclass correlation coefficients (ICCs), two-way mixed and absolute agreement [16], of the total number of reported serious needs was calculated. To assess agreement on an item level and the percentage match between first priority needs in the HESPER scale and HESPER Web, Cohen's κ was used. Additionally, descriptive statistics for analyzing the feasibility questions and the reported needs were used. SPSS software (IBM Corp. Released 2016. IBM SPSS Statistics for Windows, Version 24.0. Armonk, NY: IBM Corp) was used to conduct the statistical analysis. A significance level of *p* ≤ 0.05 was used.

#### *2.6. Ethical Considerations*

Informed consent was obtained by each study's participants before participating in the interview and/or web survey. The study was approved by the Regional Ethical Committee in Sweden (ID 2017/481) and the National Commission for Science, Technology and Innovation (NACOST) in Kenya. Permission to develop and evaluate the HESPER Web was obtained from the WHO.

#### **3. Results**

In total, 308 individuals participated in the study: 50 in the alternate forms evaluation and 289 in the feasibility evaluation. Table 1 shows the demographics. There was no significant difference between the HESPER interview sample (*n* = 50) and the HESPER Web feasibility evaluation sample (*n* = 289) regarding gender (Chi 2 test, *p* = 0.33), age (Chi 2 test, *p* = 0.78) or present location (Chi 2 test, *p* = 0.99) but for country of origin (Chi 2 test, *p* < 0.001). There was no significant difference between the participants who participated in both the HESPER interview and the web survey, and those who dropped out from the survey (gender (Chi 2 test, *p* = 0.61), age (Chi 2 test, *p* = 0.50) or present location (Chi 2 test, *p* = 0.99) and country of origin (Chi 2 test, *p* = 0.97).


**Table 1.** Demographics of the study participants.

SD = Standard Deviation.

#### *3.1. Alternate Forms Evaluation*

Of the 50 participants recruited for the alternate forms reliability evaluation and who participated in the HESPER interview, there were 19 dropouts who did not answer the HESPER Web. The alternate form results were therefore based on 31 participants.

The ICC was 0.88 (CI 0.60–0.91) between the HESPER scale and HESPER Web. For the item-by-item evaluation between the HESPER scale and HESPER Web, Cohen's κ was calculated, and it varied between 0.43 (for the item concerning safety) and 1.0 (for the item relating to law and justice in the community and other serious problems), see Table 2. Regarding the first priority need rating, an overall match of 81% was found between the HESPER scale and HESPER Web.

**Table 2.** Persons reporting specific needs and Cohen's κ between the HESPER and HESPER Web, per item.



**Table 2.** *Cont.*

1 Items presented in the HESPER Web order; <sup>2</sup> Kappa value could not be calculated due to zero answers in one or more samples.

#### *3.2. Feasibility Evaluation*

Answering the HESPER Web survey was quicker than being interviewed for many of the study participants (*p* < 0.001, see Table 3). The questions asked in HESPER Web were considered to be easy to understand, and no participant reported experiencing harm caused by the survey. About 86% of all study participants could answer HESPER Web in privacy (see Table 4). An absolute majority of the participants used their own mobile phones to answer the survey (60%), followed by a significant number who used someone else's computer or tablet, including the school's or ICT center's (19%). About 13% used someone else's mobile phone or their own computers or tablets (4%).

**Table 3.** Time to answer the survey.


Chi 2 test, *p* = 0.00, Cramer's V = 0.503.

#### *3.3. Differences in Demographics between the Randomized Study Sample and Self-Selected Sample*

No significant difference in the total reported numbers of needs could be observed between the HESPER scale and HESPER Web study samples (*p* = 0.067, paired *t*-test) or when comparing HESPER Web (alternate forms) and HESPER Web (feasibility evaluation; two-sample *t*-test, *p* = 0.132). No significant difference in gender (Chi 2 test, *p* = 0.670) or age (two-sample *t*-test, *p* = 0.810) between the HESPER interview sample and the HESPER Web self-selected sample was observed.


**Table 4.** Feasibility evaluation questions for HESPER Web.

#### *3.4. Perceived Needs*

When reporting results on their perceived needs, a total sample of 320 people was used, including all study participants who answered HESPER Web (as part of the feasibility evaluation (*n* = 289) or the alternate forms evaluation (*n* = 31)), and not the ones who only participated in the HESPER interviews. When reporting results on their perceived needs, a total sample of 320 people was used, including all study participants who answered HESPER Web (as part of the feasibility evaluation (*n* = 289) or the alternate forms evaluation (*n* = 31)), and not the ones who only participated in the HESPER interviews.

The mean number of reported needs among the study participants in the HESPER scale was 4.52 (SD 3.2, range 1–15). The frequency of reported needs in total and sorted on gender is shown in Table 5.


**Table 5.** Reported serious needs, item by item.


**Table 5.** *Cont.*

Bolded number indicate a significant difference (*<sup>p</sup>* <sup>≤</sup> 0.05). a = calculated with the Chi<sup>2</sup> test or if indicated with <sup>b</sup> where the Fischer's exact test was used, or <sup>c</sup> where the Student's *t*-test was used.

> There was no significant difference between males and females regarding the mean of the total number of reported needs (Student's *t*-test: male mean 5.88, SD 3.9, (95% CI: 5.27; 6.48), range 0–21), female mean 6.43, SD 3.9, (95% CI: 5.50;7.0), range 1–19, *p* = 0.765), but there were some differences in what kind of needs were reported (see Table 5)

#### **4. Discussion**

HESPER Web was found to be reliable and usable for assessing perceived needs among refugees living in a large-scale humanitarian context such as the Dadaab refugee camp. The use of a web-based survey was positively experienced by the study participants, and the voluntary self-recruited study sample reported similar levels of needs and similar demographics regarding gender and age to the walking methods randomized study sample. The participants reported several unmet needs, and there were some differences in the kinds of needs identified depending on gender.

The alternate forms evaluation showed overall good correspondence between the HESPER scale and HESPER Web in general (ICC 0.88) and on an item by item level (Cohen's κ from 0.43 to 1.0) [17]. The item with the lowest consistency was the question on perceived problems caused by security issues. The reason for this might be that the current level of security varied a lot from day to day and from location to location within the Dadaab camp. Additionally, there was an observed difference when reporting on educational needs for children. However, this item was frequently reported in the larger sample (Table 5) and therefore, we could not explain the difference noted in the comparison between the HESPER and the HESPER Web. The association for the first priority rating was very good (81%) [17], showing that HESPER Web reliably can be used to assess the most serious perceived needs instead of or as a complementing data collection method to the HESPER interview. However, it should be noted that the timeframe between first and second data collection was short (from a few hours up to 3 days), and that might have influenced the results. It would have been preferred with a longer timeframe between the two measurements, but due to security regulations, repeated visits could not be conducted. The short timeframe may have resulted in that participants remembered their answers from the first data collection, which may have contributed to a slightly overestimated alternate forms reliability coefficient. Even when taking this into account the alternate forms reliability between the two forms of administration of the HESPER is good.

In the HESPER manual, strategies to perform data collection in order to ensure a proper study sample are described. When using web-based methods, the same procedures may be used with the difference that the study participant answers the web-based survey instead of taking part in an interview. If advertising the survey on social media or physical locations, the study sample will be a convenience sample. This study suggests that the study samples from the walk-around sampling method and the self-selected sample were similar, regarding both their demographics and the mean number of reported needs. However, it should be noted that the number of study participants differed between the samples, and the exact number of study participants needed for generalization of a web based, not randomized data collection cannot be concluded from this study.

When conducting HESPER interviews face to face, the interviewer could interact with the person and, if needed, provide specific advice or refer to, for example, psychosocial support. When using a self-administrated web-based survey, this is no longer an option. Therefore, it is of extra importance for a survey provider to state the limitations and to provide practical support and to state where the study participants should turn for help in case of an immediate need for such support. In addition, a web-based survey may offer new possibilities to direct people who report need of support, and guide them on where to turn for available support.

HESPER Web has shown potential in reducing several challenges that are common in disaster or humanitarian emergency health research related to the practical possibilities of physically reaching or visiting an area, security concerns and ethical considerations, such as the possibility of being anonymous [18–20]. HESPER Web can offer possibilities for conducting assessments and research studies that include populations that are rarely included in such evaluations, such as people who constantly move around, people evacuated from the study area or those who do not have access to a fixed address [20]. In addition, the tool may be used for longitudinal studies on perceived needs [11]. However, not all study populations or contexts are suitable for web-based needs assessment or research. The reasons may be several, including limited access to the Internet or a means for answering the survey, limited privacy when answering the survey or illiterate or severely traumatized populations where personal contact may be necessary to assess mental health or provide support. The responsibility of using a valid and proper instrument and data collection procedure and considering the context and study population is always the researcher's or the head of the organization's responsibility, and not that of the affected population.

In this survey, the study participants reported several needs, although, they were settled in a long-lasting state of displaceability. Web-based methods for assessing mental health have been suggested to provide a better picture of the actual situation while offering anonymity and reducing stigma in the interview situation [19]. Higher levels of perceived needs can significantly predict psychological distress and lower levels of functioning [6]. It has been suggested that further emphasis should be put on developing tools for community mental health providers to enhance reach and effects from mental health interventions in low- and middle-income populations [21]. To assess perceived needs and plan for mental health interventions also in populations with long lasting displaceability seems therefore reasonable. Additionally, it has been suggested to further explore the use of self-help digital mobile applications used in community based mental health interventions in for example refugee camps [21]. For such purposes, the HESPER Web could be a feasible tool, but need to be further evaluated.

This study had several limitations. It would have been preferable to let half of the study participants in the alternate form evaluation answer the HESPER scale first, and then HESPER Web, and the other half in the opposite order. Due to security reasons, that could not be done. Additionally, such a strategy was however considered to increase the risk of dropouts between the two data collections and therefore dismissed. The use of "random walk sampling" is usually not the preferably choice of the sampling method for research studies. However, it was considered as the best possible option, given the security environment and practical possibilities. The way the "random walk sampling" was used in this study can be described as a combination of a "spin the pen" sampling and a clustered sampling method and is recommended for research in humanitarian emergencies when other, traditional methods are not possible or suitable [22].

When conducting research in humanitarian emergency settings, the research needs to be done with, and for populations affected in order to determine interventions that are feasible and appropriate for the context [1]. In this study, several actors with extensive knowledge and involvement in local processes were involved in planning, practical data

collection and the analysis of this study, including local UN agencies, NGOs and academic partners. Partnerships with local individuals ensure a local perspective and add value to the interpretations of the results [23,24]. However, the study participants themselves were not actively engaged in parts other than the data collection. The active engagement of the people affected is essential to ensure that the response is based on their actual needs and supports their recovery [23]. Asking the refugees themselves for their perceived needs may, therefore, contribute to both community engagement and individual recovery [25]. However, little is known about refugee participation in the development of policies and programs that matter to their health and well-being. Such participation is fundamental for more sustainable and responsive projects [4], and a plan for the dissemination of the results should, therefore, be considered in future projects.

#### **5. Conclusions**

HESPER Web was found to be reliable and usable for assessing perceived needs among a population affected by a humanitarian emergency. The use of a web-based survey was positively experienced by the study participants, and the voluntary, self-recruited study sample reported similar levels of needs and similar demographics regarding gender and age to the randomized study sample. HESPER Web offers a reliable and feasible tool for assessment of needs in situations where web-based surveys are considered as practical and suitable. It offers new possibilities for conducting remote assessments and research studies that include humanitarian populations that are rarely included in such evaluations.

**Author Contributions:** Conceptualization, K.H., M.S., and M.H.; funding acquisition; K.H.; data collection, K.H. and C.N.; analysis, K.H., and M.H.; interpretation of results; K.H., C.N., M.S., M.H.; writing of original draft: K.H., M.S., M.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Elrha's Research for Health in Humanitarian Crises (R2HC) Programme, which aims to improve health outcomes by strengthening the evidence base for public health interventions in humanitarian crises. The R2HC programme is funded by the UK Government (DFID), the Wellcome Trust, and the UK National Institute for Health Research (NIHR). The funder did not participate in the design of the study, data collection, analysis, or in writing the manuscript.

**Institutional Review Board Statement:** The study was approved by the Regional Ethical Committee in Sweden (ID 2017/481) and the National Commission for Science, Technology and Innovation (NACOST) in Kenya. Permission to develop and evaluate the HESPER Web was obtained from the WHO.

**Informed Consent Statement:** Informed consent was obtained by each study participants before participating in the interview and/or web survey.

**Data Availability Statement:** The datasets analyzed during the current study are not publicly available due to the Swedish law on ethical approval for research but are available from the corresponding author on reasonable request.

**Acknowledgments:** The authors extend their gratitude to the UNHCR representation in Kenya, Nairobi, Kenya, for its support in facilitating the practicalities of the data collection.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

