The revised HBDS-ON shows evidence of internal consistency reliability (Cronbach’s alpha > 0.7).¹²

The revised HBDS-ON has evidence of construct validity through exploratory factor analysis (based on factors’ eigenvalues) by showing a structure similar to the conceptual structure of the Health Belief Model.^8,9,12

The revised HBDS-ON has evidence of construct validity through convergence validity by demonstrating correlations between the revised HBDS-ON scores and the adherence to chemotherapy handling guidelines scores.¹²

The revised HBDS-ON shows evidence of construct validity through regression model testing of a model built based on the relationships in the Health Belief Model.¹²

A quantitative cross-sectional design was employed in this study. Online surveys were distributed to participants through REDCap (Research Electronic Data Capture), an app used for research data collection and management.¹³ The reporting guidelines, Strengthening the Reporting of Observational Studies in Epidemiology (STROBE), tailored for quantitative observational studies was used in reporting this study.¹⁴

Participants were recruited through convenience sampling from two medical centers in the Midwest United States. Inclusion criteria for participating were 1) being 18 years or older, 2) working in oncology settings as a nurse for at least 3 months, and 3) being responsible for handling chemotherapy drugs at work. Nurses in inpatient and outpatient infusion settings were eligible to participate. An email invitation that contained information about the study and a link to the consent form and the surveys in REDCap was sent to potential participants in these two medical centers. Additionally, participants were recruited through social media (i.e., Facebook) as described in Oudat and Bakas.¹⁵ A Facebook page was created for this study. The study flyer with the link to the study in REDCap was posted on the page and advertised to target oncology nurse population in the United States. Participants could view the study information and complete the surveys by accessing the link in the flyer. Data were collected between March 2023 and June 2023. Each participant was paid a $15 e-gift card as compensation for their time.

There is a variety of recommendations on the sample size for factor analysis. In general, larger samples yield more generalizable factors particularly when the number of factors cannot be estimated theoretically prior to analysis.^12,16 A simple rule is to use the number of respondents per item ratio. Tinsley and Tinsley ¹⁶ suggested a ratio of 5–10 subjects per item up to 300 subjects. To assess for sample adequacy, the Kaiser-Meyer-Olkin (KMO) test was used. A KMO value > 0.5 is recommended. Also, Bartlett’s Test of Sphericity was computed to evaluate if the data is suitable for factor analysis based on the relationships between the variables.¹⁷

Participants completed three surveys. First was the Revised Hazardous Drugs Handling Questionnaire. This questionnaire was used to measure the level of adherence to chemotherapy handling guidelines in terms of percent of time using a scale from 0–5 (5=Always, 4=76–99%, 3=51–75%, 2=26–50%, 1=1–25%, 0= Never). This questionnaire’s Cronbach’s alpha is 0.83.¹⁸ Second was the Revised Oncology Nurses Health Behaviors Determinants Scale. This is the scale that was psychometrically tested in this study. It contained 44 items and was used to measure the determinants affecting nurses’ adherence to chemotherapy handling guidelines. The response scale was a Likert type response scale from 1–5 (strongly disagree to strongly agree) with a “not applicable” option. Third was a demographic questionnaire, which collected data on participants’ age, gender, race, level of education, and years of experience.

The Statistical Package for Social Sciences (SPSS) version 29 was used to run the analysis.¹⁹ Descriptive statistics were computed to analyze the demographic characteristics of the participants as well as the other variables measured by the study surveys. The reliability of the revised HBDS-ON was tested through internal consistency reliability (Cronbach’s alpha).¹⁷ To evaluate meeting Cronbach’s alpha assumption of tau-equivalence (equal contribution of each item in explaining the latent variable), the dimensionality of the scale and subscales was assessed through Exploratory Factor Analysis, and the item-total correlations for each subscale were assessed through item analysis.^12,17 The construct validity of the scale was measured through: 1) exploratory factor analysis using principal axis factoring and varimax rotation, with a factor loading more than 0.4 being considered significant, 2) convergence validity testing of Pearson’s correlation between the revised HBDS-ON subscales’ scores and the Revised Hazardous Drugs Handling Questionnaire scores, and 3) regression model testing based on a model that was built from the relationships between constructs in the Health Belief Model (figure 1).^8,9,17

Based on the findings from factor analysis and reliability testing, items that did not psychometrically perform well were deleted. The items were deleted based on the criteria: 1) increased Cronbach’s alpha of the subscale if the item is deleted, 2) secondary or no significant item loadings on the extracted factors, and/or 3) high ceiling effects.

One hundred and twenty-three oncology nurses participated in the study. The mean age for participants was 36.9 years (SD=11.2), the mean of years of nursing experience was 12.2 years (SD=11.3), and the mean of years of oncology experience was 8.5 years (SD=8.8). Most participants were female (n=118, 95.9%), White (n=113, 91.9%), and holding Bachelor’s degrees (n=79, 65.3%). Table 1 provides further details on the sample characteristics. Based on the Revised Hazardous Drugs Handling questionnaire scores, the nurse participants reported adherence to chemotherapy handling guidelines at 54.5% of the time (SD=21.4; range: 0%−100%).

The results of KMO test indicated a suitable sample adequacy (KMO value= 0.74), and Bartlett’s Test of Sphericity showed that the variables in the dataset are correlated (χ² (630)=1995.43, p<.001). These results indicated that the data were suited for factor analysis.

The exploratory factor analysis results indicated that 9 factors have eigenvalues greater than 1. The scree plot method (visualizing the eigenvalues of the extracted factors through a line graph) suggested four- or seven-factor structure where the line started to flatten. After testing several factor structure solutions, the seven-factor structure was found as the solution that resonates conceptually. This seven-factor structure explained 59.9% of the variance related to factors affecting adherence to chemotherapy handling guidelines among oncology nurses. Based on table 2, the first factor includes 11 items and represents the variable, cues to adhere to chemotherapy handling guidelines, (eigenvalue=8.2, variance explained=22.9). Seven items related to perceived barriers to adhere to chemotherapy handling guidelines loaded in the second factor (eigenvalue=3.6, variance explained=10.1). The third factor includes 7 items that represent the perceived efficacy to adhere to chemotherapy handling guidelines (eigenvalue=2.9, variance explained=8.0). The fourth factor contains 6 items measuring the perceived threat of chemotherapy exposure complications (eigenvalue=2.0, variance explained=5.6). Two items representing institutional response to chemotherapy exposure incidents loaded in the fifth factor (eigenvalue=1.8, variance explained=4.9). The sixth factor includes two items that assess the variable PPE availability and accessibility (eigenvalue=1.6, variance explained=4.5). Only one item loaded in the seventh factor and represents equipment-related problems (eigenvalue=1.4, variance explained=3.9). This factor structure informed the structure of the subscales in the revised HBDS-ON. Table 3 presents the descriptive statistics for the revised HBDS-ON subscales.

Cronbach’s alpha was not calculated for the whole scale because the exploratory factor analysis results revealed a multidimensional structure. Thus, it was calculated only for the subscales. Item-total correlations had minimal variation for most of the items in each subscale, which supports the tau-equivalence assumption of Cronbach’s alpha. Cronbach’s alpha values from the first to the sixth subscales were 0.88, 0.81, 0.80, 0.82, 0.70, 0.59 respectively. The internal consistency reliability could not be assessed for the seventh factor because it contains only one item. Based on the results from factor analysis and reliability testing, eight items that did not psychometrically perform well were deleted from the revised HBDS-ON, two of these items were from the original HBDS-ON.

The correlations between the subscales’ scores and the adherence to chemotherapy handling guidelines show that five of the variables represented by the subscales have significant relationships with the adherence score. These variables are the cues to adhere to chemotherapy handling guidelines (r=0.52, p <0.001), perceived barriers to adhere to chemotherapy handling guidelines (r=0.27, p =0.004), perceived efficacy to adhere to chemotherapy handling guidelines (r=0.31, p= 0.001), institutional response to chemotherapy exposure incidents (r=0.20, p =0.042), and PPE Availability and Accessibility (r=0.19, p =0.044). Table 4 presents the correlations between each of the subscales and the handling score (Bivariate r).

The results of ANOVA test indicate that the regression model is a good fit for the data (F(7,97)=5.70, p<.001). The result of regression analysis indicates that the only variable that significantly predicts the adherence score is the cues to adhere to chemotherapy handling guidelines (B=1.12, p<.001). Other variables that represent the remaining subscales were not significant predictors of nurses’ adherence to chemotherapy handling guidelines. The model explains 29.1% of the variance in the adherence to chemotherapy handling score. The results of the regression model are presented in Table 4.

One of this study’s limitations is the social desirability bias that might influence oncology nurse participants to provide responses that are socially desirable. Additionally, selection bias might influence the study results; oncology nurses who were interested in providing information about their handling practices might not be representative of all oncology nurses. Other limitations are the small sample size, homogeneity of participants’ characteristics, recruiting participants through convenience sampling with the majority being from large medical centers in urban areas. This limits the generalizability of the study results to other populations or healthcare settings. Quantitative cross-sectional study design is another limitation because this study design limits the ability to draw conclusions on the cause-and-effect relationships between the variables.

Oncology nurses’ adherence to chemotherapy handling guidelines is critical to protect themselves from the hazards of chemotherapy exposure at workplace. Several factors have been found affecting nurses’ adherence to these guidelines. Healthcare institutions should consider these factors and relevance of the Health Belief Model when developing and implementing policies to promote oncology nurses’ occupational safety. Additionally, healthcare institutions should offer resources in the workplace to enhance oncology nurses’ readiness to adhere to the guidelines when handling chemotherapy.

The findings of this study highlight the importance of the cues to adhere to chemotherapy handling guidelines in the workplace in promoting nurses’ adherence to these guidelines. Even nurses possessing favorable health beliefs toward the adherence to safe chemotherapy handling practices would not be capable of following the handling guidelines when necessary equipment and workplace resources are unavailable. Within the workplace, a wide range of prompts can enhance oncology nurses’ readiness to adhere to safe practices when handling chemotherapy such as educational resources, hazard labels, monitoring nurses’ adherence to the handling guidelines, and the influence of coworkers.

Implications for future research include further testing of the revised HBDS-ON using confirmatory factor analysis to verify the structure of the instrument with a larger and more heterogenous sample of oncology nurses. Also, developing and testing interventions to increase adherence to safe practices when handling chemotherapy is warranted.