Knowledge, attitudes and behaviours toward evidence-based practice among healthcare professionals in Kuwait: a cross-sectional study

Abstract

Objectives

To assess and compare evidence-based practice (EBP) knowledge, attitudes and behaviours among physicians, occupational therapists, physical therapists and nurses in Kuwait and to identify modifiable correlates of EBP implementation.

Design

Cross-sectional survey.

Setting

Six governmental hospitals, one in each of Kuwait’s six governorates; data were collected from May to September 2025.

Participants

365 licensed clinicians with at least 1 year of clinical experience (occupational therapists n=81; physical therapists n=108; nurses n=94; physicians n=82), recruited through hospital department heads.

Primary and secondary outcome measures

The primary outcomes were the five domains of the Evidence-Based Practice Profile Questionnaire (EBP²) (relevance, sympathy, terminology, practice and confidence), standardised to 0–100. Secondary outcomes were differences by profession, English fluency, EBP training and gender (Welch’s tests) and independent predictors of domain scores (multivariable linear regression).

Results

Standardised mean scores were highest for relevance (93.9; 95% CI 93.2 to 94.6) and lowest for practice (46.4; 95% CI 43.9 to 48.6). Professions differed significantly on all five domains (p≤0.004), with the largest effect for practice (η² = 0.194). Physicians and occupational therapists outperformed physical therapists and nurses, particularly on terminology, confidence and practice. After adjustment, English fluency independently predicted higher terminology, confidence and practice scores; prior EBP training predicted higher sympathy but lower practice scores; and longer duration of practice predicted lower relevance scores. Sensitivity analyses supported these findings.

Conclusions

EBP utilisation in Kuwait was low to moderate, with the largest gaps in actual practice; even physicians did not reach the high-performance threshold on most domains. English fluency was the most consistent modifiable correlate of EBP knowledge, confidence and behaviour, whereas prior EBP training was linked to higher sympathy but not to more frequent EBP behaviours. Institution-wide implementation strategies, clinically integrated EBP training and multilingual evidence resources are needed.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• A multidisciplinary sample of 365 clinicians from six governmental hospitals, one in each of Kuwait’s governorates, allowed direct comparison of four professions within a single hospital-based sample.

• The Evidence-Based Practice Profile Questionnaire (EBP²) is a validated trans-professional instrument, and effect sizes were reported alongside p values for all reported comparisons.

• Multivariable regression adjusted for potential confounders, and non-parametric and bootstrap sensitivity analyses tested the robustness of the parametric findings.

• The cross-sectional design and convenience sampling limit causal inference and external generalisability beyond Kuwaiti governmental hospitals.

• All measures were self-reported, and the survey was administered in English only, raising the possibility of social desirability and response bias.

Introduction

Evidence-based practice (EBP) is the conscientious, explicit and judicious integration of the best available research evidence with clinical expertise and patient values to optimise care.¹ It is now widely regarded as a foundational competency across the health professions and a driver of safety, quality and efficiency in modern health systems.^{2 3} The translation of EBP from concept to routine clinical behaviour, however, remains uneven, and substantial gaps persist between what is recommended and what is delivered at the bedside.^{4 5}

Interprofessional differences in EBP implementation are particularly important because contemporary care is increasingly multidisciplinary. Physicians, occupational therapists, physical therapists and nurses each bring distinct educational pathways, scopes of practice and decision-making contexts to the same patient pathways. Differences in prelicensure curricula, exposure to research methods, time pressures and access to research resources have all been associated with disparities in EBP uptake.^{4 6} Earlier comparative work has suggested that physicians report more positive attitudes and more frequent EBP behaviours than nurses or allied health professionals,⁷ although findings vary by setting, instrument and the structural support available to clinicians.^{8 9}

In the Gulf region, EBP has gained policy and curricular traction, but barriers persist. Nurses’ use of EBP is associated with both individual and organisational factors,¹⁰ and nurses in Saudi primary care rated organisational barriers, such as inadequate facilities and lack of cooperation from physicians, as the most important¹¹; language may be an additional barrier in the region because most evidence sources are published in English. Surveys from neighbouring countries, mostly of single professions, paint a consistent picture of favourable attitudes coexisting with limited knowledge and implementation: among physical therapists in Saudi Arabia¹² and the United Arab Emirates,¹³ anaesthesiologists and anaesthesia technologists in Saudi Arabia¹⁴ and nurses in Oman.^{15 16} In Kuwait, profession-specific surveys have documented high awareness of EBP but limited operationalisation among physical therapists^{17 18} and primary-care physicians,¹⁹ while comparable national-level data for occupational therapists and nurses are scarce. Multiprofessional comparisons in the region are rare. A recent survey across the Eastern Mediterranean Region compared nurses with physicians, physical therapists, dentists and pharmacists,²⁰ and an Egyptian study compared physicians, pharmacists and nurses,⁶ but neither included occupational therapists, and both relied on online convenience or snowball recruitment across heterogeneous settings rather than a defined hospital workforce. To our knowledge, no study in Kuwait or the wider Gulf has examined EBP knowledge, attitudes and behaviours across physicians, occupational therapists, physical therapists and nurses within a single hospital-based sample using a validated, transprofessional instrument. This evidence gap limits the design of targeted, equitable EBP interventions and the benchmarking of profession-level performance.

The present study addressed this gap using the Evidence-Based Practice Profile Questionnaire (EBP²), a validated instrument that captures five complementary EBP dimensions—relevance, sympathy, terminology, practice and confidence—across the health professions.²¹ The objectives were: (1) to estimate EBP knowledge, attitudes and behaviours among physicians, occupational therapists, physical therapists and nurses in Kuwait; (2) to compare profession-level EBP profiles; and (3) to identify modifiable correlates of EBP implementation, including EBP training, English fluency, gender and years of practice. We hypothesised that EBP scores would be low to moderate overall, and that profession, EBP training, and English fluency would be associated with significant differences in EBP domain scores.

Materials and methods

Study design and reporting

A cross-sectional, descriptive-comparative survey design was used. Self-reported indicators of EBP knowledge, attitudes and behaviours were collected from a multidisciplinary sample of healthcare professionals working in governmental hospitals in Kuwait.

Setting and participants

The target population comprised licensed physicians, occupational therapists, physical therapists and nurses delivering inpatient or outpatient care in six Kuwaiti governmental hospitals. Kuwait has seven main governmental hospitals; we purposively selected one large hospital in each of the country’s six governorates so that all geographical regions were represented. Eligible participants were qualified clinicians in one of the four professions with a minimum of 1 year of clinical experience. Trainees, students and administrative staff without active clinical caseloads were excluded. Convenience sampling was used to recruit participants, with study invitations distributed through hospital department heads.

An a priori sample-size calculation was conducted using G*Power V.3.1.9.7.²² Assuming α=0.05, power=0.95 and a medium effect size according to Cohen’s conventions (d=0.50 for two-tailed independent-samples t-tests, f=0.25 for one-way analysis of variance (ANOVA) and w=0.30 for χ² tests),²³ the minimum required sample sizes were 210 for independent-samples t-tests, 280 for one-way ANOVA and 191 for χ² tests. The achieved sample (N=365) exceeded these requirements for all planned analyses.

Instrument

EBP knowledge, attitudes and behaviours were measured with the EBP²,²¹ a 58-item self-report instrument developed for use across the health professions. The superscript in the conventional abbreviation reflects the two P’s of ‘Practice Profile’; it does not indicate a second version of the questionnaire. The instrument comprises five domains: relevance (14 items, range 14–70), sympathy (7 items, range 7–35), terminology (17 items, range 17–85), practice (9 items, range 9–45) and confidence (11 items, range 11–55). Items are scored on a 5-point Likert scale. Conceptually, the relevance domain reflects the value attributed to EBP, the sympathy domain reflects the perceived compatibility of EBP with one’s professional role, the terminology domain reflects comprehension of common research terms, the practice domain reflects the application of EBP in clinical decision-making and the confidence domain reflects the perceived level of EBP skills. Higher scores indicate a more favourable EBP profile on every domain: greater perceived relevance of EBP, greater sympathy toward EBP (negatively worded sympathy items are reverse-scored in accordance with the developers’ scoring instructions), better understanding of research terminology, more frequent EBP behaviours and greater confidence in EBP skills. The EBP² has demonstrated strong internal consistency (Cronbach α=0.96), test–retest reliability (intraclass correlation coefficients 0.77–0.94), and convergent validity (domain-level Pearson r=0.54–0.80) in its original validation.²¹ In the present sample, Cronbach’s α was 0.92 for relevance, 0.71 for sympathy, 0.94 for terminology, 0.89 for practice and 0.95 for confidence. For descriptive interpretation, raw scores were converted to a standardised 0–100 scale using the formula (observed score − minimum) × 100/(maximum − minimum), and were broadly characterised as low (<50), moderate (50 to <70) or high (≥70). These cut-points are pragmatic thresholds adopted for descriptive purposes only; they are not derived from validation studies or criterion referencing, and no validated thresholds for the EBP² currently exist. All inferential analyses therefore used the continuous domain scores, which is appropriate for summed Likert-type scales.²⁴ Demographic items captured age, gender, profession, years of clinical practice, prior EBP training, and self-rated English fluency.

Procedure

The questionnaire was administered in English using Google Forms. The landing page presented the study aims, eligibility criteria, voluntary nature of participation, anonymity safeguards and contact details for the study team; clicking ‘Start’ constituted informed consent. The survey link was sent to the heads of the medicine, occupational therapy, physical therapy and nursing departments at participating governmental hospitals in Kuwait, who distributed it to eligible staff. Data were collected between May and September 2025, and weekly reminders were sent to department heads while the survey was open.

Statistical analysis

Analyses were conducted in IBM SPSS Statistics V.29 (IBM Corp., Armonk, New York, USA). There were no missing data because the online survey design required participants to complete all questions before submission. Descriptive statistics summarised demographic characteristics and EBP² domain scores using means, SD, frequencies and percentages.

The assumptions underlying the parametric tests were examined before analysis. Normality of domain scores and of regression residuals was assessed with the Shapiro-Wilk test, skewness and kurtosis statistics and visual inspection of histograms and normal Q–Q plots; homogeneity of variance across professions was assessed with Levene’s test; and independence of observations was supported by the design, as each participant responded once and anonymously, although participants were clustered within hospitals (see Limitations). In the full sample, Shapiro-Wilk tests were significant (p<0.001), as is expected at this sample size. For terminology, confidence, practice and sympathy, skewness (−0.25 to 0.11) and kurtosis (−0.74 to 0.02) were well within the thresholds commonly used to indicate substantial non-normality in samples larger than 300 (absolute skewness >2 or kurtosis >7),²⁵ whereas relevance scores were negatively skewed (skewness −2.30; kurtosis 6.88), reflecting a ceiling effect. Parametric tests were retained as the primary analyses because the conditions of the central limit theorem—independent observations from a distribution with finite variance, which is guaranteed for bounded scale scores—were met, so that the sampling distribution of the mean approaches normality at the present group sizes (smallest group n=81)^{26 27}; because the F-test is robust to non-normality when groups are large and of similar size²⁸; and because violations of residual normality have little effect on regression estimates when the number of observations per predictor is large.²⁹ Levene’s test indicated homogeneity of variance for practice (p>0.05) but not for relevance, terminology, confidence or sympathy (p<0.05); for consistency, Welch’s ANOVA and Welch’s t-tests, which do not assume equal variances, were used for all domains. As sensitivity analyses, between-profession differences were re-examined with the Kruskal-Wallis test, differences by English fluency, EBP training and gender with the Mann-Whitney U test, and regression coefficients with bias-corrected and accelerated (BCa) bootstrap 95% CIs (2000 resamples) (online supplemental tables S2-S4).

Between-profession differences in domain scores were examined using Welch’s ANOVA with Games-Howell post hoc comparisons; effect sizes were reported as eta-squared (η²) and omega-squared (ω²), computed from the one-way ANOVA sums of squares, and interpreted using Cohen’s benchmarks of 0.01 (small), 0.06 (medium) and 0.14 (large),²³ and for pairwise comparisons, as Cohen’s d (difference in means divided by the pooled SD of the two groups). Welch’s t-tests compared EBP² scores by English fluency, prior EBP training and gender; effect sizes were reported as Cohen’s d, calculated as the mean of the group coded 1 (fluent, trained or female) minus the mean of the group coded 0, divided by the pooled SD and interpreted as 0.20 (small), 0.50 (medium) and 0.80 (large).²³ Associations between categorical variables (eg, profession and English fluency) were examined with the χ² test, with effect sizes reported as Cohen’s w = √(χ²/N) and interpreted as 0.10 (small), 0.30 (medium), and 0.50 (large).²³

Multivariable linear regression was used to examine whether profession independently predicted each EBP² domain after adjustment for gender, years of practice, EBP training and English fluency. Age was not entered because it was highly correlated with years of clinical practice; years of practice was retained as the more direct measure of professional experience. Profession was dummy-coded with physicians as the reference category, and binary covariates were coded 0=male, no prior training or not fluent and 1=female, prior training or fluent in English. All covariates were specified a priori as potential confounders and were retained in each model regardless of statistical significance, because data-driven removal of covariates can bias the remaining coefficients and invalidate their standard errors and CIs.³⁰ Unstandardised coefficients (B) with 95% CIs, standardised coefficients (β) and p values are reported for every predictor, and R² and adjusted R² for each model. Regression assumptions were assessed: the ratio of observations to predictors exceeded 50:1, and the Shapiro-Wilk test indicated non-normality of residuals (p<0.001), which was addressed with the bootstrap sensitivity analysis described above. Multicollinearity was acceptable (all bivariate r among the entered predictors <0.70; maximum variance inflation factor 2.29; minimum tolerance 0.44). Standardised residuals fell within ±3 except for limited points in the relevance, terminology and practice domains; Cook’s distances were <1 (maximum 0.08), indicating no influential outliers. The threshold for statistical significance was set a priori at p≤0.05 (two-sided).

Patient and public involvement

No patients or members of the public were involved in the design, conduct, reporting, or dissemination of this research.

Ethical considerations

The study was conducted in accordance with the Declaration of Helsinki. Participation was voluntary and anonymous; no personally identifying information was collected. Informed consent was obtained electronically before questionnaire access.

Results

Sample characteristics

A total of 365 healthcare professionals completed the online survey: occupational therapists (n=81; 22%), physical therapists (n=108; 30%), nurses (n=94; 26%) and physicians (n=82; 22%). Of the participants, 208 (57%) were women, and 157 (43%) were men. Age ranged from 23 to 58 years (mean±SD, 36.0±7.0 years), and years of practice ranged from 1 to 35 (mean±SD, 13.0±7.0 years). Sample characteristics by profession, including gender, prior EBP training and self-rated English fluency, are presented in table 1.

Table 1. Sample characteristics by profession (N=365).

Characteristic	Overall (N=365)	OT (n=81)	PT (n=108)	Nurse (n=94)	Physician (n=82)	P value	Effect size
Age (years) mean±SD	36.0±7.0	30.7±6.9	38.5±6.4	36.4±5.3	37.5±7.1	<0.001*	0.169
Years of practice mean±SD	13.0±7.0	7.8±6.8	15.6±6.4	13.5±5.2	14.2±7.1	<0.001*	0.165
Gender, n (%)						<0.001†	0.412
Male	157 (43)	28 (35)	42 (39)	22 (23)	65 (79)
Female	208 (57)	53 (65)	66 (61)	72 (77)	17 (21)
Prior EBP training, n (%)						<0.001†	0.478
Yes	132 (36)	42 (52)	26 (24)	8 (9)	56 (68)
No	233 (64)	39 (48)	82 (76)	86 (91)	26 (32)
Self-rated English fluency, n (%)						<0.001†	0.270
Fluent	198 (54)	60 (74)	51 (47)	36 (38)	51 (62)
Not fluent	167 (46)	21 (26)	57 (53)	58 (62)	31 (38)

Continuous variables are presented as mean±SD; categorical variables as n (%).

^*Welch’s ANOVA (effect size η²).

^†χ2 test (effect size Cramér’s V, equivalent to Cohen’s w for 2 × 4 tables).

ANOVA, analysis of variance; EBP, evidence-based practice; OT, occupational therapist; PT, physical therapist.

Overall and profession-level EBP domain scores

For the overall sample, raw mean scores were 66.6±4.0 for relevance, 54.5±14.6 for terminology, 36.9±9.1 for confidence, 25.7±8.3 for practice and 21.9±4.5 for sympathy (table 2; see online supplemental table S1 for additional results). Standardised on a 0–100 scale (table 3), participants’ scores were high for relevance (93.9; 95% CI 93.2 to 94.6), moderate for confidence (58.9; 95% CI 56.6 to 60.9), moderate for terminology (55.1; 95% CI 52.8 to 57.2), moderate for sympathy (53.2; 95% CI 51.4 to 54.6) and low for practice (46.4; 95% CI 43.9 to 48.6). When standardised scores were examined by profession (table 4), all four groups scored high on relevance; physicians and occupational therapists scored in the moderate range for terminology, confidence, practice and sympathy (52.5–69.4), while nurses and physical therapists scored low to moderate (33.6–54.4) on the same domains.

Table 2. Raw EBP² domain scores by profession and Welch’s ANOVA results (N=365).

Domain (range)	OT (n=81)	PT (n=108)	Nurse (n=94)	Physician (n=82)	Welch’s F (df)	P value	ω²
Relevance (14–70)	66.0±5.0	66.7±3.6	66.2±4.1	67.7±2.8	4.5 (3, 193)	0.004	0.020
Terminology (17–85)	61.1±11.4	50.4±16.0	48.3±12.7	60.4±12.6	24.2 (3, 199)	<0.001	0.145
Confidence (11–55)	41.6±7.3	34.1±10.9	34.9±8.5	38.2±6.3	14.9 (3, 200)	<0.001	0.096
Practice (9–45)	27.9±7.7	24.2±7.9	21.1±7.3	30.9±6.8	31.7 (3, 197)	<0.001	0.187
Sympathy (7–35)	21.9±5.1	21.6±5.0	20.8±3.7	23.7±3.4	9.9 (3, 196)	<0.001	0.044

Values are mean±SD raw domain scores.

Between-profession differences were tested by Welch’s ANOVA (df in parentheses).

ω² = omega-squared, computed from the Welch’s ANOVA sums of squares (0.01 small, 0.06 medium, 0.14 large).

ANOVA, analysis of variance; EBP2, Evidence-Based Practice Profile Questionnaire; OT, occupational therapist; PT, physical therapist.

Table 3. Raw mean scores, standardised mean scores (0–100) and descriptive levels for EBP² domains (N=365).

Domain	Raw mean (95% CI)	Standardised mean (95% CI)*	Descriptive level
Relevance	66.6 (66.2 to 67.0)	93.9 (93.2 to 94.6)	High
Terminology	54.5 (52.9 to 55.9)	55.1 (52.8 to 57.2)	Moderate
Confidence	36.9 (35.9 to 37.8)	58.9 (56.6 to 60.9)	Moderate
Practice	25.7 (24.8 to 26.5)	46.4 (43.9 to 48.6)	Low
Sympathy	21.9 (21.4 to 22.3)	53.2 (51.4 to 54.6)	Moderate

Descriptive levels: low <50; moderate 50 to <70; high ≥70; these are pragmatic thresholds used for description only. Welch’s ANOVA results are given in table 2.

^*Standardised on a 0–100 scale calculated as (observed score – minimum) × 100 / (maximum – minimum).

ANOVA, analysis of variance; EBP2, Evidence-Based Practice Profile Questionnaire.

Table 4. Standardised mean EBP² domain scores (95% CI) and descriptive levels by profession (N=365).

Domain	OT (n=81)	PT (n=108)	Nurse (n=94)	Physician (n=82)
Relevance	92.8 (90.8 to 94.8) High	94.0 (92.8 to 95.2) High	93.1 (91.6 to 94.6) High	95.9 (94.8 to 97.0) High
Terminology	64.8 (61.1 to 68.5) Mod.	49.1 (44.6 to 53.6) Low	46.0 (42.2 to 49.8) Low	63.8 (59.7 to 67.9) Mod.
Confidence	69.4 (65.7 to 73.1) Mod.	52.4 (47.7 to 57.1) Mod.	54.4 (50.4 to 58.4) Mod.	61.7 (58.6 to 64.8) Mod.
Practice	52.5 (47.8 to 57.2) Mod.	42.2 (38.0 to 46.4) Low	33.6 (29.4 to 37.8) Low	60.8 (56.6 to 65.0) Mod.
Sympathy	53.3 (49.3 to 57.3) Mod.	52.1 (48.7 to 55.5) Mod.	49.2 (46.5 to 51.9) Low	59.4 (56.7 to 62.1) Mod.

Scores standardised on a 0–100 scale.

Descriptive levels: low <50; moderate 50 to <70; high ≥70.

EBP2, Evidence-Based Practice Profile Questionnaire; Mod, moderate; OT, occupational therapist; PT, physical therapist.

Between-profession comparisons

Welch’s ANOVA showed significant between-profession differences for all five domains (table 2): relevance (F(3, 193) = 4.5, p=0.004, η² = 0.028), terminology (F(3, 199) = 24.2, p<0.001, η² = 0.152), confidence (F(3, 200) = 14.9, p<0.001, η² = 0.104), practice (F(3, 197) = 31.7, p<0.001, η² = 0.194) and sympathy (F(3, 196) = 9.9, p<0.001, η² = 0.052). Effect sizes were largest for the practice domain, indicating substantial profession-related variability in actual EBP behaviours. Games-Howell post hoc comparisons (table 5) showed that physicians outperformed nurses and physical therapists across most domains, with large effects for terminology versus nurses (Cohen’s d=0.96) and for practice versus nurses and physical therapists (d=1.39 and 0.90), and a medium effect for terminology versus physical therapists (d=0.68). Occupational therapists scored significantly higher than nurses and physical therapists on terminology (d=1.06 and 0.75, respectively), confidence (d=0.83 and 0.79) and practice (d=0.91 and 0.47), and higher than physicians on confidence (d=0.50), but lower than physicians on relevance (d=0.44) and practice (d=0.41). Physical therapists scored higher than nurses on practice (d=0.40). Kruskal-Wallis tests yielded the same conclusions for all five domains (online supplemental table S2).

Table 5. Games-Howell post hoc comparisons among healthcare professions across EBP² domains (significant pairs only).

Domain	Group A (mean)	Group B (mean)	Mean diff.	P value (Games-Howell)	Cohen’s d	95% CI (d)	Magnitude
Relevance	Physician (67.7)	Nurse (66.2)	1.59	0.015	0.45	0.15 to 0.75	Small
	Physician (67.7)	OT (66.0)	1.76	0.032	0.44	0.13 to 0.75	Small
Terminology	Physician (60.4)	Nurse (48.3)	12.11	<0.001	0.96	0.65 to 1.27	Large
	Physician (60.4)	PT (50.4)	9.98	<0.001	0.68	0.39 to 0.97	Medium
	OT (61.1)	Nurse (48.3)	12.79	<0.001	1.06	0.74 to 1.38	Large
	OT (61.1)	PT (50.4)	10.66	<0.001	0.75	0.45 to 1.05	Medium
Confidence	Physician (38.2)	Nurse (34.9)	3.22	0.023	0.43	0.13 to 0.73	Small
	Physician (38.2)	PT (34.1)	4.09	0.008	0.44	0.15 to 0.73	Small
	OT (41.6)	Physician (38.2)	3.39	0.010	0.50	0.19 to 0.81	Medium
	OT (41.6)	Nurse (34.9)	6.62	<0.001	0.83	0.52 to 1.14	Large
	OT (41.6)	PT (34.1)	7.49	<0.001	0.79	0.49 to 1.09	Medium
Practice	Physician (30.9)	Nurse (21.1)	9.80	<0.001	1.39	1.06 to 1.72	Large
	Physician (30.9)	PT (24.2)	6.71	<0.001	0.90	0.60 to 1.20	Large
	Physician (30.9)	OT (27.9)	3.00	0.046	0.41	0.10 to 0.72	Small
	OT (27.9)	Nurse (21.1)	6.80	<0.001	0.91	0.60 to 1.22	Large
	OT (27.9)	PT (24.2)	3.71	0.008	0.47	0.18 to 0.76	Small
	PT (24.2)	Nurse (21.1)	3.09	0.022	0.40	0.13 to 0.68	Small
Sympathy	Physician (23.7)	Nurse (20.8)	2.86	<0.001	0.80	0.49 to 1.11	Large
	Physician (23.7)	PT (21.6)	2.06	0.005	0.47	0.18 to 0.76	Small

Mean values are raw EBP2 subscale scores; Group A is the group with the higher mean. p values are adjusted for multiple comparisons by the Games-Howell procedure. Cohen’s d was calculated with the pooled SD of the two groups; magnitude: small, d=0.20–0.49; medium, 0.50–0.79; large, ≥0.80.

diff., difference; EBP2, Evidence-Based Practice Profile Questionnaire; OT, occupational therapist; PT, physical therapist.

Effects of English fluency, EBP training and gender

Welch’s t-tests (table 6) showed that participants who were fluent in English scored significantly higher than non-fluent participants on the terminology (p<0.001, d=1.76, large), confidence (p<0.001, d=1.34, large), practice (p<0.001, d=0.64, medium) and sympathy (p=0.035, d=0.22, small) domains. Participants who had previously received EBP training scored significantly higher than those without training on the terminology (p<0.001, d=0.56), confidence (p=0.004, d=0.29), practice (p=0.028, d=0.22) and sympathy (p<0.001, d=0.79) domains. Women scored significantly lower than men on terminology (p<0.001, d = −0.40), practice (p<0.001, d = −0.44) and sympathy (p=0.008, d = −0.28). A χ² test indicated a significant association between profession and English fluency, χ²(3) = 26.70, p<0.001, w=0.27, with occupational therapists being the most fluent (74%) and nurses the least (38%). Mann-Whitney U tests yielded the same conclusions for 14 of the 15 comparisons (online supplemental table S3); the exception was relevance by English fluency, which differed on the Mann-Whitney U test (p=0.004) but not on Welch’s t-test (p=0.272), most likely reflecting the ceiling effect and unequal variances in relevance scores. Given the negligible mean difference (0.5 points; d = −0.11), we do not consider this difference meaningful.

Table 6. Differences in EBP² domain scores by English fluency, EBP training and gender (N=365).

Variable	Relevance	Terminology	Confidence	Practice	Sympathy
English fluency
Not fluent, mean±SD	66.9±2.1	44.0±12.3	31.4±7.7	23.0±7.8	21.4±3.8
Fluent, mean±SD	66.4±5.0	63.3±9.8	41.5±7.4	28.1±8.0	22.4±4.9
P value	0.272	<0.001	<0.001	<0.001	0.035
Cohen’s d (95% CI)	−0.11 (−0.32 to 0.10)	1.76 (1.51 to 2.00)	1.34 (1.11 to 1.56)	0.64 (0.43 to 0.85)	0.22 (0.01 to 0.42)
Prior EBP training
No, mean±SD	66.4±4.3	51.6±14.7	35.9±9.7	25.1±9.0	20.7±4.3
Yes, mean±SD	67.1±3.4	59.5±12.9	38.6±7.5	26.9±6.7	24.1±3.9
P value	0.084	<0.001	0.004	0.028	<0.001
Cohen’s d (95% CI)	0.18 (−0.04 to 0.39)	0.56 (0.34 to 0.78)	0.29 (0.08 to 0.51)	0.22 (0.01 to 0.44)	0.79 (0.57 to 1.01)
Gender
Male, mean±SD	66.9±4.0	57.7±13.6	37.5±9.4	27.8±8.2	22.6±4.5
Female, mean±SD	66.4±3.9	52.0±14.9	36.4±8.8	24.2±8.0	21.4±4.4
P value	0.191	<0.001	0.236	<0.001	0.008
Cohen’s d (95% CI)	−0.14 (−0.35 to 0.07)	−0.40 (−0.61 to −0.19)	−0.13 (−0.33 to 0.08)	−0.44 (−0.65 to −0.23)	−0.28 (−0.49 to −0.08)

Comparisons by Welch’s t-test. Cohen’s d = (mean of fluent, trained or female participants − mean of non-fluent, untrained or male participants) / pooled SD; positive values indicate higher scores among fluent, trained or female participants. Magnitude: small, |d|=0.20–0.49; medium, 0.50–0.79; large, ≥0.80.

EBP, evidence-based practice; EBP2, Evidence-Based Practice Profile Questionnaire.

Multivariable predictors of EBP domains

Multivariable linear regression results are summarised in table 7. The models explained 4.6% (relevance) to 49.8% (terminology) of the variance in domain scores (adjusted R² 0.028–0.488). Years of practice independently predicted only relevance, with slightly lower scores among more experienced clinicians (B = −0.075 per year, 95% CI −0.143 to −0.007, p=0.030). English fluency independently predicted higher terminology (B=17.49, p<0.001), confidence (B=9.39, p<0.001) and practice (B=3.63, p<0.001) scores, but not relevance or sympathy. Prior EBP training independently predicted higher terminology (B=2.63, p=0.044) and sympathy (B=2.97, p<0.001) scores but lower practice scores (B = −2.54, p=0.005), although trained participants had scored higher on practice in the unadjusted comparison (table 6). Gender was not an independent predictor in any model.

Table 7. Multivariable linear regression models of EBP² domain scores (N=365): B with 95% CIs and β for all predictors.

Predictor	Relevance	Terminology	Confidence	Practice	Sympathy
	B (β) 95% CI P value	B (β) 95% CI P value	B (β) 95% CI P value	B (β) 95% CI P value	B (β) 95% CI P value
Years of practice	−0.075 (−0.133) −0.143 to −0.007 0.030*	0.079 (0.038) −0.102 to 0.261 0.390	0.078 (0.060) −0.051 to 0.206 0.237	−0.061 (−0.052) −0.187 to 0.064 0.336	−0.069 (−0.108) −0.142 to 0.003 0.061
Gender (female vs male)	−0.456 (−0.057) −1.413 to 0.500 0.349	−1.370 (−0.047) −3.925 to 1.185 0.292	0.560 (0.031) −1.250 to 2.370 0.543	−1.160 (−0.069) −2.924 to 0.605 0.197	−0.778 (−0.086) −1.797 to 0.241 0.134
Prior EBP training (yes vs no)	0.331 (0.040) −0.626 to 1.288 0.497	2.628 (0.087) 0.072 to 5.185 0.044*	0.324 (0.017) −1.487 to 2.136 0.725	−2.542 (−0.147) −4.308 to −0.777 0.005*	2.968 (0.319) 1.948 to 3.987 <0.001**
English fluency (fluent vs not fluent)	−0.610 (−0.077) −1.455 to 0.235 0.157	17.489 (0.598) 15.231 to 19.747 <0.001**	9.391 (0.516) 7.791 to 10.991 <0.001**	3.632 (0.218) 2.073 to 5.191 <0.001**	0.466 (0.052) −0.435 to 1.367 0.309
Profession (ref. physician)
Occupational therapist	−1.920 (−0.202) −3.236 to −0.603 0.004*	0.159 (0.005) −3.359 to 3.676 0.929	2.580 (0.118) 0.088 to 5.072 0.042*	−3.725 (−0.187) −6.154 to −1.296 0.003*	−1.359 (−0.126) −2.761 to 0.044 0.058
Physical therapist	−0.733 (−0.085) −2.000 to 0.534 0.256	−5.759 (−0.180) −9.149 to −2.368 < 0.001**	−2.869 (−0.144) −5.270 to −0.467 0.019*	−6.738 (−0.371) −9.079 to −4.397 < 0.001**	−0.261 (−0.027) −1.614 to 1.091 0.704
Nurse	−1.339 (−0.148) −2.731 to 0.053 0.059	−5.544 (−0.166) −9.263 to −1.824 0.004*	−1.045 (−0.050) −3.679 to 1.590 0.436	−9.854 (−0.520) −12.422 to −7.286 < 0.001**	−0.586 (−0.057) −2.070 to 0.898 0.438
R²; adjusted R²	0.046; 0.028	0.498; 0.488	0.349; 0.336	0.259; 0.244	0.150; 0.134

Each column is a separate model with the domain score as the dependent variable and all listed predictors entered simultaneously; coefficients are in raw domain-score units. Age was not entered because it was highly correlated with years of practice.

Binary predictors were coded 0 = male, 630 no prior training or not fluent and 1 = female, prior training or fluent; profession was dummy-coded with physicians as the reference category.

*p≤0.05; **p≤0.001.

B, unstandardised coefficient; EBP2, Evidence-Based Practice Profile Questionnaire; EBP, evidence-based practice; ref., reference; β, standardised coefficient.

Compared with physicians, and after adjustment for the other covariates, occupational therapists had significantly lower relevance (B = −1.92, p=0.004) and practice (B = −3.73, p=0.003) scores but significantly higher confidence scores (B=2.58, p=0.042); physical therapists had significantly lower terminology (B = −5.76, p<0.001), confidence (B = −2.87, p=0.019) and practice (B = −6.74, p<0.001) scores; and nurses had significantly lower terminology (B = −5.54, p=0.004) and practice (B = −9.85, p<0.001) scores. No profession differed significantly from physicians on sympathy.

BCa bootstrap CIs supported the primary regression findings (online supplemental table S4). All statistically significant associations remained robust except that between prior EBP training and terminology, whose BCa interval included zero (−0.006 to 5.446). Conversely, the BCa interval for occupational therapists versus physicians on sympathy (B = −1.36, p=0.058) marginally excluded zero (−2.657 to −0.054); this borderline association should be interpreted with caution.

Discussion

This is, to our knowledge, the first study in Kuwait or the wider Gulf region to examine EBP knowledge, attitudes and behaviours among physicians, occupational therapists, physical therapists and nurses within a single hospital-based sample covering all of Kuwait’s governorates, using a validated, transprofessional instrument. Three findings stand out. First, while all four professions strongly endorsed the relevance of EBP, actual EBP behaviours (the practice domain) were the lowest-scoring dimension overall, with the largest between-profession effect size (η² = 0.194). Second, even physicians, who scored highest on relevance, practice and sympathy, did not reach the high-performance threshold on most domains. Third, English fluency emerged as the most consistent modifiable correlate of EBP knowledge, confidence and behaviours, whereas, after adjustment, prior EBP training was associated with higher sympathy but lower practice scores, and longer practice was associated with slightly lower EBP relevance.

The pattern of high relevance and low practice converges with international evidence that endorsement of EBP precedes—and substantially outpaces—its operationalisation.^{4 7 8} The same asymmetry has been reported in other studies using the EBP²: final-year students across four Norwegian health disciplines had a standardised relevance mean of 81.2, with every other domain at or below 54,³¹ and among allied health participants from five disciplines in Australia, most of them students, domain scores varied with profession, EBP exposure, stage of training and age.³² Our participants’ relevance scores were higher than those of the Norwegian students, whereas their terminology, confidence and practice scores fell within a similar low-to-moderate band, suggesting that the attitudinal foundation for EBP in Kuwait is at least as strong as elsewhere and that the shortfall lies in knowledge and enactment. Regionally, our findings align closely with the Eastern Mediterranean multiprofession survey, in which nurses held favourable attitudes but lacked the knowledge and support needed to adopt EBP and faced more barriers than other professionals,²⁰ with the Egyptian survey, in which understanding of EBP terminology and use of EBP in decision-making were highest among physicians and lowest among nurses,⁶ and with the Saudi anaesthesia-team survey, in which anaesthesiologists outscored anaesthesia technologists on knowledge and use despite similar attitudes¹⁴; internationally, they mirror the Taiwanese nationwide survey, in which physicians reported greater knowledge, skills and implementation than nursing and allied health personnel.⁷ Single-profession surveys of Gulf nurses^{15 16} and physical therapists^{12 13 17 18} likewise report positive attitudes coexisting with limited knowledge and implementation, and nurses in Saudi primary care rated organisational barriers as the most important.¹¹ Our finding that physicians outperformed nurses and physical therapists on most domains is therefore consistent with the regional and international literature, and it is plausibly attributable to longer pre-licensure curricula, earlier and more sustained exposure to research methods, and clinical contexts that demand explicit guideline-based decision-making in time-critical situations such as diagnosis, prescribing and risk stratification. By contrast, the rehabilitation professions typically operate over longer treatment horizons and within models of care that emphasise functional restoration, where the immediate consequences of evidence non-use are less visible, even though the long-term clinical and economic stakes are substantial.

A counter-intuitive finding was that occupational therapists in our sample tracked physicians more closely than physical therapists did, particularly on terminology, confidence and practice. In a previous Australian study using the same instrument, in which most participants were students, physical therapy participants scored higher than occupational therapy participants on relevance and terminology, although occupational therapy participants scored higher on practice.³² Differences in experience are unlikely to explain this pattern: although occupational therapists had fewer years of practice than physical therapists (mean 7.8 vs 15.6 years), years of practice was independently associated only with slightly lower relevance scores, which may reflect greater reliance on experiential knowledge and consolidated routines among more senior clinicians.⁴ This association is not universal: among Omani nurses, more years of experience were associated with greater EBP use and more positive attitudes,¹⁵ suggesting that the relationship between seniority and EBP is profession-specific and context-specific. By contrast, prior EBP training and English fluency were unevenly distributed across professions: 52% of occupational therapists and 68% of physicians reported prior EBP training, compared with 24% of physical therapists and only 9% of nurses and English fluency followed a similar gradient. Kuwaiti physical therapists have themselves reported lack of time, research skills and resources as the main barriers to engaging in research.³³ Because most peer-reviewed evidence, databases and appraisal tools are predominantly in English, language access plausibly mediates EBP behaviour through its effects on understanding of terminology and confidence in evidence appraisal. However, the regression coefficients for occupational therapists remained more favourable than those for physical therapists on terminology, confidence and practice after adjustment for training, fluency and years of practice (table 7), suggesting that profession-specific factors, such as the emphasis given to EBP in pre-licensure curricula and professional culture, also contribute.

The association between prior EBP training and practice deserves comment. Trained participants reported more frequent EBP behaviours in unadjusted comparisons, but after adjustment for profession, English fluency and the other covariates, training was associated with higher sympathy scores and lower practice scores. Because training was concentrated among physicians and occupational therapists, much of the unadjusted advantage probably reflects profession rather than training itself; the negative adjusted association may also indicate that clinicians who have been taught EBP apply a stricter standard when rating their own practice. In either case, the finding is consistent with evidence that standalone EBP teaching improves knowledge but not behaviour, whereas teaching integrated into clinical practice improves knowledge, skills, attitudes and behaviour.³⁴

These findings have implications for both individual clinicians and health systems. At the individual level, English fluency and EBP training are modifiable, although previous evidence suggests that training must be integrated into clinical work if it is to change behaviour.³⁴ At the system level, organisational facilitators—leadership commitment, protected time for evidence appraisal, accessible point-of-care evidence summaries and EBP champions—are needed to translate positive attitudes into routine behaviours.^{35 36} The observation that physicians, despite scoring highest on most domains, remained in the moderate range for terminology, confidence, practice and sympathy is consequential: it indicates that EBP capacity is incomplete even in the highest-performing group and that institution-wide EBP implementation strategies should not bypass physicians on the assumption that physician EBP competence is already optimal.

Three priorities follow for occupational therapy, allied health and medical education in the Gulf. First, embedding longitudinal, clinically integrated EBP content—with formal instruction in critical appraisal, guideline interpretation and shared decision-making—across pre-licensure and continuing education curricula. Second, investing in language-accessible evidence resources, including bilingual guideline summaries, structured journal clubs and mentorship that builds research-terminology fluency. Third, designing targeted refreshers for more experienced clinicians, whose relevance scores were slightly lower, and for nurses and physical therapists, who showed the largest gaps in actual practice.

Limitations

Several limitations should be acknowledged. First, the cross-sectional design and convenience sampling limit causal inference and external generalisability beyond the six participating governmental hospitals, and because the survey link was distributed by department heads, the number of clinicians who received it is unknown and a response rate could not be calculated. Although these hospitals covered all of Kuwait’s governorates, they were purposively rather than randomly selected, so the findings may not represent specialised hospitals, primary care or the private sector, and generalisation to the wider Gulf should be cautious; governmental hospitals across the Gulf Cooperation Council countries may, however, share a similar workforce composition, training pathways and language environment. Second, all measures were self-reported, raising the possibility of social-desirability bias, particularly for practice items. Third, the survey was administered in English; although English is the working language of much clinical documentation in Kuwait, the absence of an Arabic version may have introduced response bias against participants with lower self-rated English fluency. Fourth, participants’ graduating institution and country of pre-licensure training were not collected, which may modulate EBP exposure, and prior EBP training was recorded as a single yes/no item that did not capture its type, duration or recency. Fifth, the cut-points used to describe standardised scores as low, moderate or high were pragmatic thresholds chosen for descriptive purposes; they are not derived from validation studies or criterion referencing, so readers should interpret the descriptive levels as approximate, although all inferential analyses used continuous scores. Sixth, relevance scores showed a ceiling effect, which limits discrimination between participants on this domain. Seventh, the five domains were tested without correction for multiple outcomes, so associations with p values close to 0.05, such as those between years of practice and relevance and between prior training and terminology, should be interpreted cautiously; in addition, participants were clustered within hospitals, which was not accounted for in the analysis and may have led to slightly underestimated SEs. Eighth, the study was confined to the four professions that constitute the core interprofessional team on the medical and rehabilitation wards from which participants were recruited; clinical pharmacists, dietitians, radiographers and other allied health professionals—who are integral to EBP in multidisciplinary care—were not included, and their EBP profiles in Kuwait remain unknown. Finally, the unequal distribution of EBP training and English fluency across professions is itself a structural feature of the Kuwaiti workforce and limits the ability to disentangle profession-specific from training-specific effects.

Future research

Future studies should (1) replicate these comparisons using bilingual (Arabic–English) instruments and stratified random sampling across Gulf Cooperation Council countries; (2) test the impact of structured, profession-specific, clinically integrated EBP training programmes using pragmatic randomised or stepped-wedge designs; (3) examine organisational determinants of EBP behaviour using multilevel modelling that nests clinicians within hospitals and units; (4) investigate whether digital EBP supports (clinical decision-support systems, point-of-care evidence summaries) attenuate the language and time barriers documented here; and (5) extend the comparison to clinical pharmacists, dietitians, radiographers and other allied health professions so that EBP capacity can be benchmarked across the whole multidisciplinary team.

Conclusions

EBP utilisation among physicians, occupational therapists, physical therapists and nurses in Kuwait was low to moderate, with the largest gaps in actual practice and the largest profession-level differences in the practice and terminology domains. Physicians and occupational therapists outperformed physical therapists and nurses in most domains, but no profession reached the high-performance threshold on any domain other than relevance. English fluency was the most consistent modifiable correlate of higher EBP performance, whereas, after adjustment, prior EBP training was associated with higher sympathy but lower practice scores. Closing the EBP knowledge-to-practice gap in Kuwait will require coordinated investment in interprofessional, clinically integrated EBP training, language-accessible evidence resources and institutional implementation supports that include, rather than exempt, physicians.

Data availability statement

Data are available upon reasonable request.