Effectiveness of the Basic Infection Control Skills License (BICSL) Training Program on Nurses' Knowledge, Attitudes, and Practices in Makkah, Saudi Arabia

1.1. RESEARCH QUESTIONS

The following research questions guided this study; no formal directional hypotheses were prespecified:

1. To what extent do nurses' knowledge, attitudes, and self-reported practices regarding standard precautions change after completing the BICSL training program?
2. Are the observed pre-post changes in knowledge, attitude, and practice scores statistically significant when assessed using both parametric and non-parametric methods?
3. Which demographic, professional, and institutional factors are associated with post-training practice scores and improvement in practice after adjusting for baseline practice?
4. How should the observed improvements be interpreted considering the quasi-experimental one-group pre-post design and the absence of a control group?

2.1. Research Design and Study Setting

This multicenter, quasi-experimental, one-group pre-post study was conducted in 5 Ministry of Health public hospitals in Makkah, Saudi Arabia: Al Noor Specialist Hospital, Hira General Hospital, King Abdulaziz Hospital, King Faisal Hospital, and Maternity and Children Hospital. The study was implemented during the 2024 Hajj-related BICSL training period, when hospitals in Makkah experienced increased patient flow and heightened IPC preparedness requirements. All participants received the same intervention; there was no random allocation or concurrent comparison group. The study is reported in accordance with the TREND guideline for nonrandomized evaluations.

Because the study did not include randomization or a concurrent control group, the design assessed within-participant changes after training rather than establishing definitive causal effects. Potential threats to causal inference, including testing effects, maturation, secular trends, and concurrent IPC initiatives, were considered when interpreting the findings.

2.2. Target Population, Sampling, and Participants

The target population comprised registered nurses employed in the 5 participating hospitals who had not completed BICSL training during the current implementation cycle. Healthcare workers other than nurses, nurses who did not consent, and nurses without paired pre- and post-training data were ineligible for the paired analysis. The sampling frame included approximately 3,500 nurses in Makkah, and systematic random sampling was applied to the nursing staff lists, selecting approximately every 10th nurse. The minimum required sample size was 347; the calculation parameters were not available in the study documentation.

The study records documented 360 nurses who completed the pre-training questionnaire, the three-day BICSL training, and the immediate post-training questionnaire and who were included in the final paired analysis. The numbers initially approached or assessed for eligibility, excluded or declining before enrollment, and lost during follow-up were not separately documented. Protocol deviations were also not separately documented (Fig. 1).

2.3. BICSL Training Intervention

The BICSL intervention was delivered in person over 3 consecutive days by certified BICSL trainers from the participating hospitals' infection control departments. The standardized curriculum covered 7 core components: hand hygiene, personal protective equipment, biological spill management, needle-stick injury management, transmission-based precautions, respirator fit testing, and powered air-purifying respirators. Each module combined theoretical instruction, trainer demonstration, and participant return demonstration. Training was delivered in Arabic and English as needed. Because all participants received the educational intervention, neither participants nor trainers were blinded. Although the curriculum and procedures were standardized across hospitals, formal independent fidelity auditing of individual sessions was not conducted. Session group size was not recorded, and no incentives or other adherence-promoting activities were documented.

2.4. Data Collection and Measurement Instrument

Data were collected using a self-administered structured questionnaire adapted from previously used KAP instruments and aligned with CDC and WHO IPC guidance [6, 20, 21]. The questionnaire captured demographic and professional characteristics and assessed 3 co-primary outcome domains: knowledge, attitude, and practice. A pilot validity check was conducted with 30 nurses before the main study; these pilot observations were not included in the final analytic sample.

The knowledge scale included 13 dichotomous items scored 1 for a correct response and 0 for an incorrect or “do not know” response, giving a score range of 0-13. The attitude scale included three dichotomous items, scored 1 for a positive/correct response and 0 otherwise, giving a score range of 0-3. The practice scale included 9 self-reported items. For the primary analysis, “always” was coded as 1 to represent strict adherence, while “often,” “sometimes,” “seldom,” and “never” were coded as 0, yielding a score range of 0-9. This dichotomization was selected to identify consistent IPC compliance, but it also reduces ordinal information, which is noted as a limitation.

The pre-test was administered before BICSL training. The post-test was administered immediately after completion of the 3-day training within the same implementation cycle.

2.5. Statistical Analysis

Descriptive statistics summarized participant characteristics and KAP scores. The individual nurse was the unit of analysis. Paired t-tests were used to compare pre- and post-training KAP scores. Because the domain scores were based on dichotomous items and might not be normally distributed, Wilcoxon signed-rank tests were also reported. Domain-level effect sizes were calculated using Cohen's dz, and ceiling effects were assessed by reporting the proportion of participants who achieved the maximum score. No missing values were imputed; analyses were restricted to complete paired records. Data were analyzed using IBM SPSS Statistics version 31.0.2.0.

Internal consistency was assessed separately for each KAP subscale at pre-test and post-test using KR-20/ Cronbach's alpha and average inter-item correlations. Because the attitude scale contained only 3 items and all domains used dichotomous scoring, reliability estimates were interpreted cautiously.

To address concerns about post-test-only regression, the revised analysis used baseline-adjusted and change-score models. The baseline-adjusted model estimated the post-training practice score, adjusting for baseline practice, changes in knowledge and attitude, gender, nationality, qualification, department group, and hospital fixed effects. The change-score model estimated change in practice score and included baseline practice, knowledge, and attitude, changes in knowledge and attitude, and the same covariates. HC3 robust standard errors were used. Continuous predictors were mean-centered before creating interaction terms. A binomial GLM sensitivity analysis modeled the number of compliant practice items out of 9. Variance inflation factors (VIFs) were examined; age was evaluated but not retained in the final hospital fixed-effect models because diagnostics indicated instability when age and hospital indicators were included together. Baseline practice was retained in all adjusted models. Holm-Bonferroni correction was applied to exploratory pairwise subgroup comparisons.

2.6. Ethics

The study was approved by the Jawatankuasa Etika Penyelidikan Manusia Universiti Sains Malaysia (JEPeM-USM; protocol code USM/JEPeM/PP/23020204) and by the General Directorate of Research & Studies, Ministry of Health, Kingdom of Saudi Arabia (decision approval number 1122-836 076-K-02-H). Data were anonymized before analysis.

3.1. Participant Characteristics

Participant flow is presented in Fig. (1). A total of 360 nurses with complete paired pre- and post-training data were included. The mean age was 39.93 ± 11.00 years (range 22-59 years). Most participants were male (55.00%), Saudi (53.06%), staff nurses (42.78%), and employed in ICU departments (74.17%). Participant characteristics are presented in Table 1.

3.2. Pre- and Post-Training Kap Changes

KAP scores increased significantly across all domains after BICSL training (Table 2). Knowledge increased from 8.54 ± 2.03 to 10.71 ± 1.79, attitude from 2.30 ± 0.77 to 2.70 ± 0.51, and practice from 4.48 ± 1.91 to 5.90 ± 1.85. Paired t-tests and Wilcoxon signed-rank tests were statistically significant for all domains (p < 0.001). Cohen's dz indicated a moderate-to-large effect for knowledge (0.72), a small-to-moderate effect for attitude (0.43), and a moderate effect for practice (0.55). Ceiling effects were most evident for attitude, for which 72.5% of participants achieved the maximum post-training score.

3.3. Reliability of KAP Subscales

Internal consistency was low, especially for the three-item attitude domain (Table 3). Accordingly, the KAP domain totals should be interpreted as pragmatic summary scores of heterogeneous IPC competencies rather than as highly internally consistent psychometric scales.

3.4. Baseline-adjusted and Change-score Models

In the baseline-adjusted model (Table 4), change in knowledge was significantly associated with a higher post-training practice score (beta = 0.105, p = 0.001). In contrast, change in attitude was positive but not statistically significant (beta = 0.150, p = 0.154). Staff nurses had lower adjusted post-training practice scores than senior nurses (beta = -1.552, p < 0.001). ICU nurses had higher adjusted practice scores than nurses in other departments (beta = 0.583, p = 0.007). The model R2 was 0.251, and the adjusted R2 was 0.225.

The model used OLS with HC3 robust standard errors and hospital fixed effects. Reference groups were female, non-Saudi, senior nurse, non-ICU/other departments, and Hira General Hospital.

In the change-score model (Table 5), improvement in knowledge was significantly associated with improvement in practice (beta = 0.181, p < 0.001). Improvement in attitude was positive but not statistically significant (beta = 0.280, p = 0.114). Lower baseline practice was strongly associated with greater practice improvement (beta = -1.018, p < 0.001), consistent with greater room for improvement among participants with lower initial scores. The model R2 was 0.615 and adjusted R2 was 0.599.

3.5. Sensitivity and Multiple-comparison Analyses

The binomial GLM sensitivity analysis treated practice as the number of compliant “always” responses out of 9 (Table 6). Results were consistent with the OLS models: improvement in knowledge remained significantly associated with compliant practice responses (OR = 1.055, p < 0.001), whereas change in attitude was not statistically significant after adjustment (OR = 1.081, p = 0.139).

After Holm-Bonferroni correction, most exploratory subgroup differences were no longer statistically significant. For change-score subgroup comparisons, only the difference in attitude change between senior nurses and staff nurses remained statistically significant after correction (mean difference = -0.353, Holm-adjusted p = 0.041). Therefore, all subgroup findings are interpreted as exploratory (Table 7).