Study design
This study was conducted in Stockholm, Sweden, using a cross-sectional design as part of an overarching project about inclusion health for women experiencing homelessness. The project involved the development of an advisory board of women with lived experience of homelessness in the Spring of 2020 and the board contributed to this study by consulting with the researchers through a workshop focused on the interpretation of the research results. This study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist.
Participants
The researchers recruited a convenience sample comprised of the following three groups: (i) women experiencing homelessness (n = 37), (ii) RNs (n = 90), and (iii) nursing students (n = 138). The choice of participants was motivated by the study’s aim and enabled a comparison between women experiencing homelessness, as users of healthcare services, as well as RNs and nursing students involved in the provision of healthcare. The sample size was determined based on the aim of being able to detect a 10-point difference, assuming a standard deviation (SD) of 15 points, on the Caring Behaviours Inventory-24 (CBI-24) score between the women experiencing homelessness and the RNs and student groups using the Mann-Whitney U-test with a two-sided null hypothesis having α = 0.05 and power = 0.80). Utilising an allocation ratio of 2:1 for nursing students to RNs to allow for the larger pool of available students and attrition rates at 15–20%, the aim was to recruit at least 35 women experiencing homelessness, 35 RNs, and 70 nursing students. The sample size calculation was performed using G*Power 3.1 [30].
Women experiencing homelessness
Women experiencing homelessness were recruited from a primary healthcare centre in Stockholm, with support from a research assistant who had experience working with this population. The centre is privately-run with state funding and visits are free of charge. A variety of healthcare services are offered for persons experiencing homelessness and it has close collaborations with social services, primary and psychiatric care, and services for treatment of substance use disorder. Both referrals and walk-in appointments are accepted. The centre receives approximately 14,000 annual visits and provides care for 1300 persons, of which 40% are women. The research assistant approached women in the waiting room of the centre providing services for persons experiencing homelessness. The inclusion criteria were Swedish speaking women with experience of homelessness. Homelessness was defined according to the four categories of the European Typology of Homelessness and Housing Exclusion (ETHOS): (1) rooflessness; (2) houselessness; (3) living in insecure accommodation; and (4) living in inadequate accommodation [31]. The exclusion criteria were women who demonstrated violent and/or abusive behaviours, or who expressed severe anxiety or distress.
Registered nurses (RNs)
RNs were recruited at a clinical conference at a private university college with governmental funding in Stockholm. All preceptors (i.e., practice placement supervisors) in Stockholm who receive nursing students from the university college were invited to attend the conference, free of charge. Approximately 110 preceptors attended the conference. Potential participants were approached by researchers who shared information about the study, via a poster board and flyers located near the refreshments area. The information was also emailed to conference attendees who expressed interest in participating but did not have time to sign-up to participate. The inclusion criteria were Swedish speaking and licensed RNs at the time of data collection. The exclusion criterion was RNs who were not clinically active.
Nursing students
Nursing students were recruited through a group message on an online learning platform at a university college. The university college is the same location as where the conference was held to recruit the RNs. At the time of recruitment, it had approximately 600 enrolled bachelor nursing students, of which 400 of these would be eligible to participate due to them having completed a clinical placement. The inclusion criteria were Swedish speaking nursing students, who were enrolled in undergraduate nursing education. The exclusion criterion was if the nursing student had not yet experienced a clinical placement.
Data collection
Women experiencing homelessness
Data collection with the women experiencing homelessness began in October and November 2019. After a pause due to the Coronavirus (COVID-19) pandemic, it continued between September and December 2020. The women were provided with verbal and written information and given the opportunity to discuss and ask questions about the study.
With support from the research assistant, the face-to-face data collection occurred at the healthcare centre. Data were collected using pen-and-paper, Swedish versions of the Attitudes Toward Homelessness Inventory (ATHI) [32] and the Caring Behaviours Inventory-24 (CBI-24) instruments [33], in addition to general questions about background characteristics. The women were asked to answer the CBI-24 questions according to how they perceived nurses’ caring behaviours at their latest healthcare visit. As part of a larger research project, the women also answered questionnaires regarding general health, existential health, health literacy, and exposure to violence. The women received a grocery store voucher valued at approximately €10 at the end of the data collection.
Registered nurses (RNs)
During October 2019, data were collected with the RNs. The RNs were invited to participate, provided with verbal and written information about the study, and had the opportunity to discuss and ask questions.
Using SurveyMonkey [34], the RNs responded to the ATHI and the CBI-24 instruments, as well as general questions about background characteristics. The RNs were asked to answer the CBI-24 based on how they perceived the caring behaviours of RNs in their clinic. RNs received a lottery scratch card at the end of the data collection.
Nursing students
Between August and November 2019, data were collected with the nursing students. The students were provided with information about the study through a message on their learning platform. The message included contact information for the research group and a clickable link to access the questionnaires.
Using SurveyMonkey [34], the students responded to the ATHI and the CBI-24, in addition to general questions about background characteristics. The students were asked to answer the CBI-24 according to how they perceived nurses’ caring behaviours during their latest clinical placement. The students received a lottery scratch card at the end of the data collection.
Ethical considerations
This study was approved by the Regional Ethical Board in Stockholm, Sweden (Number 2019-021130). Participants were informed that their participation was voluntary and that they may withdraw at any time without any explanation required. Data were collected anonymously and written informed consent was obtained digitally in SurveyMonkey from all participants prior to data collection.
Instruments
Data were collected using the Swedish versions of the CBI-24 and ATHI instruments. Previous studies have indicated that these instruments are psychometrically acceptable for use in Swedish healthcare research [32, 33]. Psychometric testing of the reliability and validity of the Swedish version of the ATHI indicates acceptable item reliability [32]. The Swedish version of the CBI-24 has good face and content validity, good test–retest reliability, and high internal consistency when administered by RNs to patients in a hospital setting, or when used by nursing students [33].
The attitudes toward homelessness inventory (ATHI)
The ATHI is an 11-item instrument measuring multiple dimensions of attitudes towards people in homelessness [35]. The controversy on whether homelessness is caused by personal or societal factors stands at the core of ATHI. Participants answer (score) each item using a 6-point Likert scale (1 = strongly agree, 6 = strongly disagree). Higher total scores and domain scores indicate more favourable attitudes towards persons experiencing homelessness. The items are categorised into four domains, as follows:
-
1.
Personal Causation (PC, 3 items) measures the belief that homelessness is caused by personal deficiencies;
-
2.
Societal Causation (SC, 3 items) measures the belief that homelessness is caused by societal factors;
-
3.
Affiliation (AFFIL, 2 items) measures attitudes about willingness to associate with persons experiencing homelessness; and
-
4.
Solutions (SOLNS, 3 items) measures attitudes regarding viable solutions to homelessness.
The caring behaviours inventory-24 (CBI-24)
The CBI-24 is a 24-item instrument which assesses perceptions of caring behaviours among patients and nurses in diverse settings [33, 36]. Participants answer (score) each item according to a 6-point Likert scale (1 = never, 6 = always). Higher total scores and domain scores correspond to higher perceptions of care provision. The 24 items are categorised into the following four domains:
-
1.
Assurance (8 items) assesses how readily available nurses are to patients’ needs and security;
-
2.
Knowledge and Skill (5 items) assesses the nurses’ abilities to demonstrate their skills and competence;
-
3.
Respectful (6 items) assesses the degree to which nurses show interest and attend to patients; and
-
4.
Connectedness (5 items) assesses the nurses’ readiness and willingness to support patients.
Statistical analyses
Categorical data are presented as frequencies and percentages, n (%), while continuous and ordinal data are given as means with accompanying SDs. Correlations between ordinal variables were calculated using Spearman’s rank correlation ρ, with the strength of correlations classified according to the following for absolute values of the correlations: ≤ 0.20, very weak; > 0.20 to ≤ 0.40, weak; > 0.40 to ≤ 0.60, moderate; > 0.60 to ≤ 0.80, strong; > 0.80 very strong. Tests of equality between two independent correlations were performed using a Z-test applied to Fisher’s z-transformed correlations. All statistical analyses were performed in R ≥ 4.0.0 (R Foundation for Statistical Computing, Vienna, Austria), with P-values < 0.05 considered statistically significant.